Proceedings of the 17th Offshore Mechanics and Arctic Engineering, July 1998 and Journal of Offshore Mechanics and Arctic Engineering, September 1998
EFFECT OF SLOW-DRIFT LOADS ON NONLINEAR
DYNAMICS OF SPREAD MOORING SYSTEMS
by
Michael M. Bernitsas, Ph.D.
Professor, ASME Member
and
Boo-Ki Kim
Research Assistant, Ph.D. Candidate
Submitted for presentation at the
17th International OMAE Conference
Lisbon, Portugal
July 5-9, 1998
and
for publication in
Journal of OMAE, ASME Transactions
Department of Naval Architecture and Marine Engineering
The University of Michigan
Ann Arbor, MI 48109-2145, USA
TECHNISCHE WW1ERSITEII Laboratorium 'icor Scheepshydromechanlca
Mekatweg 2,2628 CD Deift
i& 0t-Nd5?-Fs 015 181S33B
ABSTRACT
Spread Mooring Systems (SMS) may experience large am-plitude oscillations in the horizontal plane due to slow-drift loads. In the literature, this phenomenon is attributed to
resonance. In this paper, it is shown that this conclusion is only partially correct. This phenomenon is investigated using nonlinear stability and bifurcation analyses which re-vea.! an enhanced picture of the nonlinear dynamics of SMS. Catastrophe sets are developed in a parametric design space to define regions of qualitatively different system dynam-ics for autonomous SMS, including mean drift forces. Lim-ited time simulations are performed to verify the qualitative conclusions drawn on the nonlinear dynamics of SMS using catastrophe sets. Slowly varying drift forces are studied as an additional excitation on the autonomous SMS and simu-lations reveal that slow drift may cause resonance or bifur-cations with stabilizing or destabilizing morphogeneses. The mathematical model of SMS is based on the slow-motion maneuvering equations in the horizontal plane (surge, sway, yaw) including hydrodynaniic forces with terms up to third-order, nonlinear restoring forces from mooring lines, and
en-vironmental loads due to current, wind, and wave-drift.
i INTRODUCTION
During the past two decades, considerable research effort has been devoted to the development of ocean mooring sys-t.ems for station-keeping of tanker and ifoating production, storage and offloading (FPSO) facilities and for oil/gas off-shore operations. Several types of mooring configurations are available such as Single Point Mooring (SPM), Two Point Mooring (TPM), and Spread Mooring Systems (SMS) (API, 1987). SPM and TPM systems aie widely used for relatively short periods of operation during which there is little change in environmental conditions. For longer periods of
opera-EFFECT OF SLOW-DRIFT LOADS ON
NONLINEAR DYNAMICS OF SPREAD MOORING SYSTEMS
Michael M. Bernitsas and Boo-Ki Kim
Department of Naval Architecture and Marine Engineering
The University of Michigan
Ann Arbor, Michigan
tion, the direction of environmental excitation changes sig-nificantly, thus making a SMS more appropriate for
station-keeping.
Time-domain simulation has been used widely to deter-mine the nonlinear dynamical behavior of any specific moor-ing configuration. Results of this approach can be found, for example, in van Oortmerssen et al. (1986), Chakrabarti and Cotter (1989), Kat and Wichers (1991), and Brook (1992). Simulation alone, however, is not adequate for the analy-sis and design of mooring systems because it cannot reveal the complete picture of the nonlinear phenomena associated with system dynamics. A comprehensive design methodol-ogy for mooring systems based on catastrophe theory has been developed by Papoulias and Bernitsas (1988), Chung and Bernitsas (1992), and Garza-Bios and Bernitsas (1996). They developed nonlinear stability analysis and bifurcation theory for mooring systems to produce catastrophe sets in the design space defining regions of qualitatively different system dynamics. This design methodology eliminates both trial and error and extensive nonlinear simulations typically used in the design process. Catastrophe sets have been de-rived numerically by systematic search for bifurcations (Pa-poulias and Bernitsas, 1988; Chungand Bernitsas, 1992) or through analytical expressions of loo of static and dynamic loss of stability (Garza-Bios and Bernitsas, 1996).
Current, wind, and waves are the main sources of vessel excitation in ocean. Papoulias and Bernitsas (1988), Chung and Bernitsa.s (1992), Nishimotoet aI. (1995), Beraitsas and Garza-Rios (1996), and Fernades and Aratanha (1996) have studied extensively the nonlinear dynamics of SPM, TPM, or SMS, but did not include waves in their numerical appli-cations. In this paper, slow-drift loads are considered, along with current, in SMS dynamics. Slow-drift loads are of im-portance in the design of mooring systems because it has been observed that these second-order low-frequency excita-tions may result in large amplitude horizontal plane moexcita-tions
of the moored vessel.
The SMS model is introduced in Section 2. It consists of the maneuvering equations in the horizontal plane includ-ing hydrodynamic terms up to third-order, nonlinear restor-ing forces from moorrestor-ing lines, environmental loads due to current, wind, and wave-drift. In Section 3, solutions by simulation are used to reveal the richness of SMS dynam-ics. In Section 4, the SMS is modeled in a six-dimensional state space and nonlinear stability analysis is introduced. The SMS is autonomous and catastrophe sets are developed in a parametric design space to define regions of qualita-tively different dynamics. The effect of slow-drift loads on SMS dynamics is studied in Section 5. It is shown that even mean drift forces due to head seas may push SMS into al-ternate equilibria or limit cycles. On the other hand, mean drift forces may have stabilizing effect by reducing the un-stable manifold dimension. A slowly varying drift spectrum is imposed as an additional excitation on the corresponding SMS. Resonance - to which large amplitude SMS motions are attributed in the literature - is only one of the possible qualitatively different responses of SMS. Stabilization, ap-pearance of alternate limit cycles or alternate equilibria are possible as well.
2 PROBLEM FORMULATION
The SMS model consists of the mooring line system, the horizontal plane slow motions, and the mooring line model. Their models are presented in the three subsections below.
2.1
Spread Mooring System
Coordinate System
Consider a vessel moored to a spread mooring system,
as shown in Figure 1.
Let O-XYZ be the body-fixed coordinate system with its origin located at the center ofgravity (G) of the moored vessel. The Z-axis is positive upwards and the X-axis points toward the bow. The o-zyz system denotes the inertial coordinate system fixed to the
earth with the origin located at the mooring terminal 1.
The z-y plane coincides with calm water level and the z-axis is positive upwards.
Geometric Relations
The following geometric properties are defined to describe the arrangement of a spread mooring system, as shown in
Figure 1: (z, y,)) are the coordinates of the ith mooring
line fairlead in the body-fixed coordinate system; (zw, y)
and (x, y) are the positions of the ith mooring
termi-nal and attachment coordinates of the ith mooring line with respect to the earth-fixed coordinate system, respectively; 7(') is defined as the angle between the z-axis and the ith mooring line counter-clockwise; ¡(s) is the length of the ith mooring line; is the yaw angle.
The geometric relations of the mooring lines are (i) (i)
COS7' , (1)
DIRECTION OF WiNO ANO WAVES
Figure 1: Geometry of Spread Mooring System
(t) (s) (s) Ihn YT
smy =
=
i,,/(z2-
z)2 +
4)
= z +zcos'y»sin*,
()
()
.()
YT =Y+Z1,
sm&+y
cos1i.For convenience, is defined as
= 2r
+ '.
Kinematic relations
The slow motions of the vessel in the horizontal plane -surge, sway, yaw - are considered. The horizontal compo-nents of vessel velocity (u, y, r) relative to the surrounding water in surge, sway and yaw, respectively, are related to the inertial coordinates (z, y) by the kinematic relations
i = u cos ' - y sin - U cos o, (7)
y =usin,,b+vcos,,b+Usina,
(8)(9)
where U is the magnitude of the current velocity. The direc-tion of the current is measured with respect to the positive z-axis counter-clockwise, as shown in Figure 1.
2.2
Vessel Motions and Loads
Slow-Motion Maneuvering Equations
The mathematical model is based on the application of
Newton's laws of motion for linear and angular momentum in the earth-fixed coordinate system o-z yr. Forces and moment acting on the vessel are defined in the body-fixed coordinate system O-XYZ. Thus, transformation from the o-z yr sys-tem to the O-XYZ syssys-tem is required to write the equations
of motion in the body-fixed coordinate system. This trans-formation results in the Euler equations of motion for a rigid body
du
- rv) =
dvm(
+ ru) = F,
where m is the mass of the vessel; Izz is the moment of
in-ertia about the Z-axis; Fx, Fy and M are the total forces and moment acting on the vessel in surge, sway, yaw,
re-spectively. In the body-fixed coordinate system O-XYZ, equations (10)-(12) can be written as
(m - X4) - mrv = XH + + Xw + Xwv,
i=1 (13)
(m - Y - Y + mru = YH + + Yy + Ywv,
=1 (14)
n
-Noi' + (Izz - N
= NH + E(zT2ay -
3,;T2rge)+Nw+Nwv,
(15)where X, Y,., and N,. are added masses and moment of
inertia in surge, sway arid yaw directions, respectively; Y,.
is the added mass in sway direction due to yaw motion;
N,, is the added moment of inertia in yaw direction due to
sway motion; and 2)O denote the tension in the th
mooring line in surge and sway directions, respectively; n is the total number of mooring lines. Subscripts H, W and WV indicate hydrodynamic forces due to vessel motions, wind loads, and second-order slow-drift loads, respectively.
Hydrodynamic Forces and Moment
The hydrodynamic forces and moment due to vessel mo-tions can be expressed in terms of the slow motion deriva-tives. The following third-order Taylor expansion is used (Abkowitz, 1969):
XH = X0 + Xu +!Xuuu2+ + X,,,,v2
+ X,.rr2 + !X5852 + Xvuv2u +
+ X55,.62u X,,,.vr + X,.,6v5 + X,.6r6
+ Xvruv'ru + X,,óv5u + XrórSu,
= Y0 + Yo,.0
+You2 + ')
+!YvvV32
+ Y,,,.vr2 + Y,55v5
Y,.,vu + Y,,vu
+ Yrr + Yrrrf3 + Yrvvrv2 + Yr66'52 + Y,.0ru
+ Yruuru2 + 1'5 +!y5663+ Y6,,,5v2 + Yorr8r2 + Y5u + Y5óu2 + Y,,,.,çvrô, s?
2 1
Nif = N0 + N,,,u + N0u + N,,v + N,,,,,,v
+ lN,.,.tìf2 + N,,6vS2 + Nvuvu +
!N,,vu2
+ N,.r + !Nrrrr3+!N,.,,,,Tv2 + !Nr65r52+ N,.Tu+!Nruuru2 + N55 + N6555 + 1N6,.,,,,5v2
+ Nórr6r2 +No6u + No,.6u2 + N,,,.5vró, (18)
where subscripts o and 5 denote propeller and rudder angle effect, respectively. The resistance of the vessel R at velocity u is related to the first four terms in equation (16) by
-R
Xe + X,u + Xu2 +
(19)The hydrodynaniic forces and moment acting on the vessel
depend not only on the instantaneous vessel motions but
also the history of the surrounding fluid motion. The effect of the history of the surrounding fluid motion is referred to as the hydrodynamic memory effect. This will be studied in a separate paper.
Mooring Line Tension
The tension in the ith mooring line for the surge and sway directions can be represented as
ge =
+Tcosw = Rcos&+ Rsin,
= -T
sin w' = RS»cos- R»sin '.
The components of the mooring line tension R», R» in the r-y plane are given by
= cosy , (22)
) .
()
= smi'
Current Load
Current loads are implicitly included in the expressions
of equations (16),
(17), and (18) by using the relative
velocities of the vessel with respect to the surrounding
water. Therefore, current loads are accurate to third-order.
This approach is more accurate than the usnaI method based on projected area and drag coeflicients. rEWith no
loss of generality, the current is assumed to be directed to
the negative x-axis in the earth-fixed coordinae system
(a = 00).
Wind Load
1gSteady wind forces and moment exerted on the vessel are expressed as
(23)
Xw
=
paU,Cxw(ar)Ar,
Yw= -POU,CYW(OT)AL,where p,, is the density of air; U,,, is the mean wind velocity relative to the air at a standard height of ten meters above the calm water level; AT and AL are the maximum trans-verse and lateral projected areas of the above-water part of the vessel, respectively; L is the length of the vessel. The relative angle a,. of wind direction is defined as
0,. = Oo
-where Oo is the absolute angle of wind direction in the (z, y) coordinate system. The coefficients Cxw(a,.), Cyw(ar), Czw(ar) depend on the vessel type, superstructure loca-tion, loading condiloca-tion, and the relative angle a,. of the wind direction. These coefficients can be obtained from experimental and full-scale measurement data.
Slow-Drift Load
The common approach to formulation of non-linear wave-body interaction problems in ship and offshore hydrodynam-ics is based on potential flow theory using the wave ampli-tude to length-ratio perturbation parameter. The solution of the second-order problem results in three types of wave excitation forces: mean, difference-frequency (slowly vary-ing), and sum-frequency (springing) forces. Mean and slowly varying drift forces are of importance in mooring problems. Newman's approximation (1974) for slowly varying drift forces has been used extensively for many applications in ship hydrodynamics. The important practical consequence of this approach is that slowly varying drift forces depend merely on evaluation of second-order mean drift forces, and this can reduce the computational time significantly. The formula for slowly varying drift forces is
sv__ T
F1 = (Jtwv, Ywv, Nwv)
=
A3AkTkcos[(tk w,)t+( e,)]
j=1 k=1 (24)
where the coefficient 7k =
0.5(D + D)
and i = 1,2, 6indicate surge, sway and yaw directions, respectively. A,
w, and e are the wave amplitude, circular frequency, and random phase angle of component number j (or k) in the
total number of wave components N. The random phase
angle .e are uniformly distributed between O and 2v and
remaiçonstant in time. The wave amplitude A (or A)
can bomputed from the wave spectrum S(w) as
p,
A, = 2/S(t,)iw,
(25)where ¿w is the difference between adjacent frequencies. The iwo-parameter Bretschneider wave energy spectrum for-mulatz used to express S(w). The two parameters are the significant wave height H,,,i3 and the mean wave period T2.
The time mean value over the diagonal terms (when k = j) in equation (24) gives the mean drift forces and the mean of the off-diagonal terms (when k j) vanishes. Mean drift
forces can be written as
The transfer functions D' for mean wave drift force are ap-proximated as D'
=
= PWgLCXD cos3(Oo -Ywv = PWgLCYD sin'(Oo-
---D6 sin 2(90 -C 2where ,, is the wave amplitude, Pv is the density of water, o is the absolute angle of wave direction, and g is gravi-tational acceleration. The coefficients CXD,CYD,CZD can be obtained from the complete linear (first-order) solution of the seakeeping problem. Further details about the methods of predicting mean drift forces can be found, for example, in the study of Ogilvie (1983). In this paper, it is assumed that these can be given simply in terms of the ship particulars (Hirano et al., 1980).
2.3
Mooring Line Model
In our line of work, we use three types of quasi-static moor-ing line models: nylon or polyester rope, chain, or steel cable. The stability of mooring systems is affected significantly by the line model (Bernitsas and Papoulias, 1990). Chains are usually fully submerged in water and only slightly
exten-sible. They experience large drag force and undergo two-dimensional deformation in the vertical plane modeled by catenary equations. Steel cables are modeled by a three-dimensional, large-strain, finite element model which takes into account drag force, weight, and extensibility. They are lighter than chains and have less drag but are more
extensi-ble. Elastic ropes are light, considerably extensible, nearly buoyant in water. Nonlinear behavior is due to their extensi-bility, and they are modeled as follows (McKenna and Wong,
1979):
ii
q(30)
where T is the actual tension in the rope, Sb is the average breaking strength, p and q are empirically determined on-stants, and 1 is the working length of the unstrained rse. A wet nylon rope of 120 mm diameter (Sb = 680, 625 ¡bf, p = 9.78, q = 1.93) is selected for the numerical applications ipre-sented in this paper.
Initial pretension is applied to one or more mooring lines. The pretension T1 can be defined in terms of the initial sLiain Ei of the line as T1 = Sbp(EJ)", (31) il - ¡w (32) 'w D2 =
V (Xwv,Ywv,Nwv)T = (26) where ¡ denotes the initial length of the pretensioned moor-ing line.
Table 1: Principal Particulars of Vessels and Mooring Line Arrangements used in Simulations
3 RICHNESS OF SMS DYNAMICS
A barge with stabilizing skegs and a tanker without
propeller axe selected for numerical applications. Their principal particulars and mooring line arrangements used
in simulations are summarized in Table 1. Configuration FinAs denotes a (m+s)-Line SMS with m lines attached on the moored vessel forward of the center of gravity and with s lines aft of the center of gravity. Pretension is imposed on the forward mooring lines and is calculated as a function of initial strain E1. Slow motion derivatives of the two vessels
are provided in Bernitsas and Kekridis (1986).
Nonlinear dynamic behavior of mooring systems such as stable or unstable focus near an equilibrium, limit cycle,
and chaotic motion have been demonstrated by systematic and extensive simulations (Papoulias and Bernitsas, 1988;
Chung and Bernitsas, 1992; Bernitsas and Garza-Rios, 1996). In this section, the richness of SMS dynamics
and the effect of slow-drift loads are demonstrated by
simulation in Figures 2 to 6. The fourth-order Runge-Kutta time-stepping method is implemented because it is relatively simple and sufficiently accurate. The geometric points GI to G4 used to define SMS configuration in Figures 2 to 6 do not cover the entire range of design practice. They were selected to demonstrate SMS dynamics. In Section 5, these
simulations are studied further in relation to catastrophe
sets in Figures 7 to 11.
Barge with Skegs
Figure 2 shows a periodic oscillation (limit cycle) of a
2-Line barge with skegs SMS about equilibrium A which has zero yaw angle. The amplitude of the yaw angle increases when mean drift forces are applied in the current direction
(head seas 80 = 0°). The system still undergoes a limit
cycle and the increased amplitude of yaw angle due to
mean drift forces results in higher tension in the mooring
lines. A different angle of wave direction (8 = 30°) leads to a non-symmetric equilibrium position with non-zero yaw angle. Figure 3 shows that the dynamical behavior of mooring systems can be different qualitatively depending on
003 002
¡
0_01 t003 0.02 o 25 50 75 Im, Nm,,m,s4on.i Tm,. (Ut&) 125 150Figure 2: Effect of Mean Drift Forces on 2-Line SMS (FiAi, 103/L
0.3,x/L = 0.3,E1
0.013), Barge with Skegsthe mean drift force. As can be seen in Figure 3, the 2-Line barge with skegs SMS experiences stable focus behavior and converges to equilibrium A when only current load is applied. A limit cycle near equilibrium A appears when mean drift forces are applied. This simulation shows that an SMS may exhibit large amplitude slow periodic motions due to mean drift forces. As will be explained in Section
4, this is the result of dynamic loss of stability and not
resonance. In Figure 4, slowly varying drift forces have a stabilizing effect on the SMS; actually higher slowly varying drift forces result in a stable equilibrium A. This shows that there is no resonance in yaw. In surge, however, Figure 5 shows resonant motion on top of the periodic motion due to loss of stability.
Tanker without PropeUer
Figure 6 shows that a 3-Line tanker without propeller SMS converges to equilibrium B in head current without waves; equilibrium A becomes unstable and the system converges to equilibrium B which has a non-zero yaw angle. When mean drift forces are added to the excitation, equilibrium A becomes stable. In this case, mean drift forces have a
favorable effect on the system stability. This is the opposite effect to that observed in the barge with skegs SMS in Figure 3. The system is unstable and the yaw angle fluctuates about equilibrium B when slowly varying drift forces are applied in head seas.
4 NONLINEAR STABILITY ANALYSIS
4.1
State Space Representation
The SMS model presented in Section 2 consists of the equations of motion (13)-(15) and kinematic relations (7)-(9). It can be modeled by six first-order nonlinear coupled ordinary differential equations using the following six state
variables (z1 = u,52 = v,X3 = f,X4 = x,x = y,zs = i'). In
System 2-Line SMS 3-Line SMS
Configuration FIAI F1A2
Vessel Barge with Skegs Tanker without Propeller
L 191.56 ft 1066.3 ft B 35.00ft 173.gOft T 10.42 ft 71.30 ft CB 0.855 0.831 E1 0.013 0.04 U 2 knots 2 knots H113 4.5 ft 8.0 ft
0.03
0.02
10.06
r26
0260
t_/L.O4 x1L=O.3 (G2A E,. aol3
V
'II
160I
i
t
II
660 200 260 P1di1=0à0r Tm,. (Uk) t - ,,1 * 6th66 L 400Figure 3: Effect of Mean Drift Forces on 2-Line SMS
(FiAi, 1/L = 0.6,x/L = 0.3,Ej = 0.013),
Barge with SkegsFigure 4: Effect of Slow-Drift Loads on 2-Line SMS
(FiAi, 1/L = 0.4,z/L = 0.3,E1 = 0.013),
Yaw, Barge with Skegs
Figure 5: Effect of Slow-Drift Loads on 2-Line SMS
(FiAi, 100/L = 0.4,z/L = 0.3,E1 = 0.013),
Surge, Barge with Skegs
Figure 6: Effect of Slow-Drift Loads on 3-Line SMS
(F1A2, 100/L = 0.15, x/L = 0.52, Ej = 0.04), Tanker without Propeller
Canchy standard form, they become = i +on - X, {mx2r3 + Xw(x6) + Xwv(x6)}, +
(IZZ_S){
D mxlr3+Yw(r6)+Ywv(z6)} +yPR»)cosz6}
i=1 n(mYo)
+ D{N(zi,z,x3) +
(r,'14 - i4,'R)cosxs}
= i n (m - Y,,){>(4'R +
i4
D +(m Yb)
{N(x6) + Nwv(z6)}, 14 z1c06x6X2SiflX6Uco6a, 15 z1 sinz6+z2cosz6+Usina, SIflX6 } (33)where D = (IzzN)(mY,)NY,.. Evolution equations
(33}.(38) can be written in the form of one six-dimensional
n
_(tzzNr)
-D (34)-
+ +{Nw(x6) + N(z6)},
i=113 = {YH(x1,x2,x3)+ >2(Rf'coszs
R»sinx6)}
N,,
first-order vector differential equation as
x = f(x),
f E C',
f: R6 R6 (39)where * denotes the derivative of state vector x with respect to time t, lR is the six-dimensional Eucidean space, and C' is the class of continuously differentiable functions.
Since the vector field f in equation (39) does not depend explicitly on time t, the system is said to be autonomous.
The mean drift forces as expressed by equation (26) are au-tonomous. The slowly varying drift forces given by equation (24), however, depend explicitly on time and would render the SMS model non-autonomous. In general, stability anal-ysis of a non-autonomous nonlinear system is not possible. Time simulation of a non-autonomous system which includes slowly varying drift forces, however, can be performed with-out any difficulty and is used in Sections 3 and 5 to assess the effects of non-autonomous terms on the stability of the corresponding autonomous system.
4.2
Equilibria of SMS
Equilibria of the vector field (39) are the points x E IR6
such that
O=f(5c),
i=(î,,±2,
16). (40)These are stationary solutions. Equilibria can be interpreted
as intersections of the null clines. The equilibria of SMS, which are the solutions to a nonlinear equation (40), depend on the mooring line and system configuration, as well as on
the environmental conditions.
4.3
Stability in the sense of Lyapunov
The behavior of a nonlinear system near an equilibrium
point can be determined via linearization with respect to that point. Since linearization is an approximation in the
neighborhood of an equilibrium point, it can predict only
the local behavior of the nonlinear system in the vicinity of that point (Guckenheimer and Holmes, 1983; Seydel, 1988).
Let x be an equilibrium point of the nonlinear system (39). Expanding the right-hand side of equation (39) into
its Taylor series about point x, we obtain
df
x=f()+>6.__ x=t
+
Ix=a +H.O.T.i,., = I
where deviation ¿ from equilibrium is defined as
¿(t) = x(t) - x.
(42)Since X is an equilibrium point, we have f() = O. Then, the state equation can be written as
(43)
6
If attention is paid to a sufficiently small neighborhood of the equilibrium point so that the higher-order terms are negligi-ble, the nonlinear state equation (39) can be approximated
by the linear state equation
= [A),
¿ E R6, [A] E R6X6 (44)where [A] = Df(i) is called the Jacobian matrix evaluated
at the equilibrium point X. If the eigenvalues of [A], ,, have zero real parts (usually called hyperbolic or non-degenerate), then it can be expected that the trajectories of the nonlinear system in a small neighborhood of an
equilib-rium point are close to the trajectories of its linearization about that point.
4.4
Bifurcation and Catastrophe Sets of SMS
A bifurcation is a qualitative change in a limit set of the system as a parameter is infinitesimally perturbed. Bifurca-tions can be interpreted as the appearance or disappearance of a limit set and the change in the stability type of a limit
set. A bifurcation also occurs when a non-stable limit set
of the system remains non-stable but undergoes a change in the unstable manifold dimension. Saddle-node, pitch-fork, and Hopf bifurcation have been found in the nonlin-ear dynamics of mooring systems (Papoulias and Berthtsas, 1988; Chung and Bernitsas, 1992; Bernitsas and Garza-Rios, 1996). A catastrophe set is the set of all loci of bifurcations in a parametric design space. It defines the boundaries be-tween regions of qualitatively different system dynamics and the morphogeneses occurring as boundaries are crossed.
In the numerical applications below, catastrophe sets are constructed for several SMS configurations of a barge with
skegs and a tanker without propeller. The effect of mean
drift forces on bifurcation of SMS dynamics is demonstrated in the catastrophe sets of Figures 7 to 11. Five different SMS with symmetric mooring configuration are studied and their catastrophe sets are constructed for equilibrium A. The five SMS configurations and their geometries are summarized in Table 2. The pretension is set in the forward mooring lines in terms of the initial strain E1. The angle of wave direction
00 is set equal to O degree, i.e., head seas and wind loads exerted on the vessel are not considered.
Four regions of qualitatively different SMS dynamics are identified in Figures 7 to 11.
Region I: All six eigenvalues of the system have negative
real parts. Equilibrium A is stable, and trajectories
(41) converge asymptotically to equilibrium A.
Region II: There exists one eigenvalue with positive real part (one-dimensional unstable manifold). Equilibrium A is unstable, therefore, trajectories deviate from equi-librium A.
Region III: There exists a complex conjugate pair of eigenvalues with positive real parts (two-dimensional unstable manifold). Equilibrium A is unstable, and
trajectories asymptotically reach periodic oscillations (limit cycle).
Table 2: SMS Configurations and Geometries used in Catastrophe Sets
Region IV: There exists a real eigenvalue and a com-plex conjugate pair of eigenvalues with positive real parts. Therefore, equilibrium A is unstable with a
three-dimensional unstable manifold which is the mini-mum dimension for onset of chaos (Seydel, 1988).
Pitchfork bifurcations appear at the boundary between
Regions I and II. Static loss of stability occurs when
crossing from Region I to Region II. Periodic solutions
appear in Region III, and dynamic loss of stability occurs when crossing from Region I to Region III. Region IV is characterIzed by the merging of unstable Regions II and III.
Barge with Skegs
The effect of mean drift forces on the FiAi and F2AO configuration is shown in Figures 7 and 8, respectively.
Figures 7 and 8 show three regions of qualitatively different SMS dynamics about equilibrium A. A stationary pitchfork
bifurcation occurs between Regions I and II. A dynamic
(Hopf) bifurcation occurs between Regions I and III. This boundary has the form of a fold singularity which encloses Region III where periodic solutions exist. As can be seen in Figures 7 and 8, mean drift forces expand the unstable Region III as the significant wave height increases. Mean drift forces have a significant effect on increasing the Hopf
bifurcation domain for the barge SMS, while they have
little effect on the stationary bifurcation locus. Geometric
point G2 (see Figure 7) is in Region I in the presence of very small or no mean drift forces. For higher realistic
mean drift forces, G2 falls in the dynamically unstable Region III. Geometric point Gi remains in Region III.
These observations are verified by simulation in Figures 2
and 3. Figure 9 shows the effect of mean drift forces on
the dynamics of the F2A1 configuration Only Regions I
and II appear in the catastrophe set of Figure 9; therefore,
Figure 7: Catastrophe Set
of 2-Line SMS (FiAi,
(yr/B
= 0.0, E1=
0.0 13), Barge with SkegsFigure 8: Catastrophe Set
of 2-Line SMS (F2AO,
yr/B
=
0.0, =5°), Barge with Skegsthis system undergoes only a stationary bifurcation in
the design parameter range. This stationary bifurcation
depends weakly on the mean drift forces. This conclusion is similar to the one obtained from Figures 7 and 8.
Tanker without Propeller
Figures 10 and 11 show the catastrophe sets for a
tanker without propeller SMS in the F1A2 and the F2A2
configuration, respectively. As the significant wave height
increases, the set of bifurcation boundaries moves to the
left. As a result, stable Region I and chaotic Region IV move to the left, unstable Region II shrinks, and Hopf
bifurcation domains expands significantly. Geometric point
G4 (1/L
= 0.15, z,/L=
0.52) in Figure 10 is in unstable Region H for low mean drift forces and falls in the stable Region I as mean drift forces increase. This is verified by simulation in Figure 6. 0. 0.4-02: -II ... / +G2 /u'
u'TT.
-
2.01! . 05 SMS Corifig. Line 4Y2-Line FiAi i , >0
rr, 0
Vp1 =0y1O
2 Vp 0 (2)....,,.
F2AO i w>0
2p 0
Vp O-(')=2c
2 w = w Z, p Vp Yp
3-Line F1A2 i
,, >0
r, 0
y, =0
"y(')=O2
Vp 0
2)=ir_f 3 ..y(3)=,r+ 2 F2A1 i w >0p 0
Vp <O()=2rÇ2
2 .(2) 3y, =0
()=r
4-Line F2A2 i,. >0
xp' 0
Vp' 0 (12)
2 .(2). ç 3Vp 0
rf
4 w w Xp Vp =YpFigure 9: Catastrophe Set of
3-Line SMS (F2A1,
yr/B
= 0.5,
= 15°, Ej = 0.013), Barge
with Skegs
Figure 10: Catastrophe Set of 3-Line SMS (F1A2,
yr/B = 0.5,
= 5°, Ej = 0.04), Tanker
without Propeller
Figure 11: Catastrophe Set of 4-Line SMS (F2A2,
y/Bja = 0.0, y,/B0j
0.5, 2 5°,E1 = 0.03), Tanker without Propeller
5
EFFECT OF SLOW-DRIFT LOADS ON
SMS DYNAMICS
5.1
Effect of Mean Drift Forces
Nonlinear stability and bifurcation analyses are used to study the effect of mean drift forces on SMS dynamics. This is possible because the system is autonomous. Based on the previous observations, we can summarize the effects of mean drift forces on SMS dynamics as follows:
Mean drift forces have a small detrimental effect on static loss of stability for barge SMS's. Unstable Re-gion II expands at the expense of ReRe-gion I, as shown in Figures 7, 8, and 9.
Mean drift forces may have a detrimental effect on static loss of stability for tanker SMS's. Specifically, Figures
10 and 11 show that Region III may expand at the
expense of Region Il. This implies that a tanker SMS whose trajectories converge to equilibrium B may ex-hibit large amplitude oscillations in limit cycles around equilibrium A. Obviously, this is not resonance. On the other hand, mean drift forces may significantly stabilize tanker SMS's statically. As shown in Figures
10 and 11, Region I expands into Region II, and
Re-gion Ill expands into ReRe-gion IV. That is, the pitchfork bifurcation is deferred significantly.
Mean drift forces have dramatic detrimental effect on dynamic loss of stability for barge SMS's. As shown in Figures 7 and 8, Region III expands at the expense of Region I. Thus, trajectories that converge to equilib-rium A reach a limit cycle around equilibequilib-rium A. Again, these large-amplitude oscillatory slow motions are not instigated by resonance.
Similar is the effect of mean drift forces on dynamic loss of stability of tanker SMS's. Figures 10 and 11 show that Region III dramatically expands into Regions II and IV. Thus, trajectories converge to equilibrium B, and chaotic motion around equilibrium A may become periodic around equilibrium A.
5.2
Effect of Slowly Varying Forces
When the spectrum of slowly varying drift forces is added
to the external excitation of a SMS, the system becomes
non-autonomous. There is no general stability theory to analyze the non-autonomous system subject to spectral
excitation. A preliminary investigation of the effect of slowly
varying drift forces on SMS dynamics is presented here
using simulations of selected SMS systems. The following observations can be made:
-(a) Slowly varying drift forces may stabilize oscillatory mo-tions instigated by mean drift forces. Figure 4 shows significant reduction in yaw angle oscillations. Actu-ally, slowly varying drift forces of higher significant wave
height totally eliminate yaw angle oscillations.
50 4.0 II M--4_5" -z. /L 1.5 1.0 -0.5 . IV . i ::---' ni 444.Øft 5,.l.Oft -Il
-of Moored Vessels in Irregular Seas," Proceedings -of the 6th International Conference of Behavior of Offshore Structures, Vol. 1 , Imperial College of Science, Technology and Medicine, London, pp. 251-264.
Chakrabarti, S.K., and Cotter, D.C., 1989, "Motions of Articulated Towers and Moored Floating Structures,"
Journal of Offshore Mechanics and Arctic Engineering, Vol. 111, pp. 233-241.
Chung, J.S., and Bernitsas, M.M., 1992, "Dynamics
of Tow-Line Ship Towing/Mooring Systems: Bifurcation,
Singularities of Stability Boundaries, Chaos," Journal of
Ship Research, Vol. 36, No. 2, pp. 93-105.
Fernades, A.C., and Aratanha, M., 1996, "Classical As-sessment to the Single Point Mooring and Turret Dynamic Stability Problems," Proceedings of the 15th International Conference on Offshore Mechanics and Arctic Engineering, Vol. l.A., Florence, Italy, pp. 423-430.
Garza-Rios, L.O., and Bernitsas, M.M., 1996, "Analytical
Expressìons of the Stability and Bifurcation Boundaries for General Spread Mooring Systems," Journal of Ship
Research, Vol. 40, No. 4, pp. 337-350.
Guckenheimer, J., and Holmes, P., 1983, Nonlinear Os-cillations, Dynamical Systems, and Bifurcations of Vector Fields, Springer-Verlag, New York, Inc..
urano, M., Takashina, J., Takaishi, Y., and Saruta, T., 1980, "Ship Turning Trajectory in Regular Waves," Transactions of the West-Japan Society of Noval Architects, Vol. 60.
Kat, J.O., and Wichers, J.E.W., 1991, "Behavior of a Moored Ship in Unsteady Current, Wind, and Waves,"
Marine Technology, pp. 251-264.
McKenna, H.A., and Wong, R.K., 1979, "Synthetic Fiber
Rope, Properties and Calculations Relating to Mooring
Systems," Deep Water Mooring and Drilling, Ocean Engi-neering, Vol. 7, pp. 189-203.
Newman, J.N., 1974, "Second-Order Slowly Varying Forces on Vessels in Irregular Waves, Proceedings of Inter-national Symposium on Dynamics of Marine Vehicles and Structures in Waves, London, pp. 182-186.
Nishjmoto, K., Brinati, H.L., and Fucatu, C.H., 1995, "Analysis of Single Point Moored Tanker using
Maneu-vering Hydrodynamic Model," Proceedings of the 14th
International Conference on Offshore Mechanics and Arctic Engineering, Vol. I.B, Copenhagen, Denmark, pp. 253-261. Ogilvie, T.F., 1983, "Second Order Hydrodynamic Effects on Ocean Platforms," Proceedings of International Workshop on Ship and Platform Motions, pp. 205-265.
Papoulias, F.A, and Bernitsas, M.M., 1988, "Autonomous Oscillations, Bifurcations, and Chaotic Response of Moored Vessels," Journal of Ship Research, Vol. 32, No. 3, pp.
220-228.
Seydel, R., 1988, From Equilibrium to Chaos, Elsevier Science Publishing Co., Inc., New York.
van Oortmerssen G., Piukster, J.A., and van den Boom
H.J.J., 1986, "Computer Simulation of Moored Ship Be-havior," Journal of Waterway, Port, Coastal and Ocean
Engineering, Vol. 112, No. 2., pp. 296-308. On the other hand, slowly varying drift forces produce
the resonant beating behavior observed in Figure 5 for the surge motion. This behavior is indicative of narrow band excitation near a natural frequency.
Finally, Figure 6 shows that slowly varying drift forces
may render a stable equilibrium A unstable, insti-gate a pitchfork bifurcation, and take a tanker SMS
through transient oscillations to non-resonant oscilla-tions around a stable equilibrium B.
CLOSING REMARKS
A methodology based on stability and bifurcation analy-ses has been developed to perform a comprehensive analysis of the effect of mean drift forces on the nonlinear horizontal plane motions of SMS. Following that, slowly varying drift forces were considered as an additional excitation on the cor-responding autonomous system and their effect on SMS was studied. It was shown that the notion that large amplitude slow motions of SMS are due to resonance of a mooring sys-tem natural frequency with slowly varying drift represents only one of the mechanisms that can instigate such motion-s. Mean drift forces may also cause large amplitude oscillations. Further, slowly varying drift forces may dramatically reduce and even eliminate such motions. Actually, a wide variety of bifurcations and morphogeneses may be caused by mean or slowly varying drift forces.
ACKNOWLEDGMENTS
This work was sponsored by the University of Michi-gan/Industry Consortium in Offshore Engineering.
Indus-try participants include Amoco, Inc.; Conoco, Inc.; Exxon Production Research; Mobil Research and Development; Shell Companies Foundation; and Petrobras, Rio de Janeiro, Brazil.
REFEREN CES
Abkowitz, M.A., 1969, Stability and Motion Control of Ocean Vehicles, MIT Press, Cambridge, Massachusetts.
API, 1987, "Analysis of Spread Moorinç Systems for
Floating Drilling Units," API Recommended Practice 2P (RP2P), Second Edition.
Bernitsas M.M., and Garza-Rios. -L.O, 19.9, "Effect of
Mooring Line Arrangement on the Dynamics of Spread
Mooring Systems," Journal of Offshore Mechanics and
Arctic Engineering, Vol. 118, No. 1, pp. 7-20
Bernitsa.s M.M., and Kekridis, N.S., 1986, "Simulation and Stability of Ship Towing," International Shiçbuilding Progre.ss, Vol. 32, No. 369, pp. 112-123.
Bernitsas M.M., and Papoulias, F.A, 1990, "Nonlinear
Stability and Maneuvering Simulation of Single Point Mooring Systems," Proceedings of Offshore Station Keeping Symposium, SNAME, Houston, Texas,pp. 1-19.
TECHNISCHE UNVERSU hratOriUm voor
M2MSrnitsas
"'"' hProtessor,
Mem ASME B. -K. Kim Research Assistant, Graduate Student.Department of Naval Architecture and Marine Engirreerin. The University of Michigan, 2600 Draper Road, Ann Arbor. MI 48109-2145
I Introduction
During the past t o decades, considerable research effort has been devoted to the development of ocean mooring systems for
station-keeping of tanker and floating production. storage and offloading FPSO facilities and for oil/gas offshore operations. Several types of mooring configurations are available such as single-point mooring (SPM). two-point mooring (TPM). and spread mooring systems (SMS) (API, 1987). SPM and TPM
systems are widely used for relatively short periods of operation during which there is little change in environmental conditions.
For longer periods of operation. the direction of environmental excitation changes significantly. thus making an SMS more appropriate for station-keeping.
Time-domain simulation has been used widely to determine the nonlinear dynamical behavior of any specific mooring con-figuration. Results of this approach can be found, for example. in van Oortmerssen et al. (1986). Chakrabarti and Cotter (1989), Kat and Wichers (1991). and Brook (1992).
Simula-tion alone, however, is not adequate for the analysis and design
of mooring systems because it cannot reveal the complete
pic-ture of the nonlinear phenomena associated with system
dynam-ics. A comprehensive design me1hodolog for mooring systems based on catastrophe theory has been developed by Papoulias and Bernitsas (1988). Chung and Bernitsas (1992). and Garza-Rios and Bernitsas (1996). They developed nonlinear stability
analysis and bifurcation theory for mooring systems to produce catastrophe sets in the design space defining regions of
qualita-tively different system dynamics. This design methodology eliminates both trial and error and extensive nonlinear
simula-tions typically used in the design process. Catastrophe sets have
been derived numerically by systematic search for eigenvalue
bifurcations (Papoulias and Bernitsas. 1988: Chung and
Bernit-sas. 1992) or through analytical expressions of loci of static
and dynamic loss of stability (Garza-Rios and Bernitsas. 1996).
Current, wind, and waves are the main sources of vessel
excitation in ocean. Papoulias and Bernitsas (1988). Chung and Contributed b the OMAE Di'.ision for publication in the JOURNAL OF Orr:
SHOR.E MECHANICS AND ARCrIC ENGINEERING. Manuscnpt received by the OMAE
Disision, March 24. 1998: re'.ised manuscript receised May 8. 1998. Technical
Editor S. Liu
Effect of Slow-Drift Loads on
Nonlinear Dynamics of Spread
Mooring Systems
Spread mooring s-cslems (SMS) may experience large-amplitude oscillations in the horizontal plane due to slow-drift loads. In the literature, this phenomenon is
artrib-wed io resonance. In this paper, it is short n thai this conclusion is on/v partial/v correct. This phenomenon is investigated using nonlinear stability and bifurcation
analyses which reveal an enhanced pictu re of the nonlinear dvna,nics of SMS. Catas-trophe sets are developed in a parametric design space to define regions of
qualita-tively different system dynamics for autonomous SMS. including mean drift forces. Limited time simulations are performed to verift the qualitative conclusions drawn
on the nonlinear dynamics of SMS using catastrophe sets. Slowly varying drift forces are studied as an additional excitation on ¡he autonomous SMS and simulations re-teal
that slow drift may cause resonance or bifurcations trith stabilizing or destabilizing
morpho geneses. The mathematical model of SMS is based on the s/ott-motion man eu-vering equations in the horizontal plane (surge. s-tray. vaw,i, including hydrodnamic
forces with terms up to third-order, nonlinear restoring forces from mooring lines,
and environmental loads due ro current, wind, and wave-drift.
Bernitsas (1992). Nishimoto et al. (1995). Bernita and
Garza-Rios i 1996). and Femandes and Aratanha t 19% I have
studied extensively the nonlinear dynamics of SPM. TP\1. or
SMS. but did not include waves in their numerical applications. In this paper. slow-drift loads are considered, along v ith current.
in SMS dynamics. Slow-drift loads are of importance in the design of mooring systems because it has been observed that these second-order low-frequency excitations may result in
large amplitude horizontal plane motions of the moored vessel.
The SMS model is introduced in Section 2. It consists of the maneuvering equations in the horizontal plane including hydrodynarnic terms up to third-order, nonlinear restoring
forces from mooring lines, environmental loads due to current.
wind, and wave-drift. In Section 3. solutions by simulation are used to reveal the richness of SMS dynamics. In Section 4. the
SMS is modeled in a six-dimensional state space and its
nonlin-ear stability analysis is introduced. The SMS is autonomous
and catastrophe sets are developed in a parametric design space to define regions of qualitatively different d namics. The effect
of slo-drif1 loads on SNIS dynamics is studied in Section 5. It is shown that even mean drift forces due to head seas may
push SMS into alternate equilibria or limit cycles. On the other hand, mean drift forces may have stabilizing effect by reducing the unstable manifold dimension. A slowly varying drift spec-trum is imposed as an additional excitation on the corresponding
SMS. Resonanceto which large-amplitude SMS motions are attributed in the literatureis only one of the possible qualita-tively different responses of SMS. Stabilization, appearance of
alternate limit cycles or alternate equilibria are possible as well.
2 Problem Formulation
The SMS model consists of the mooring line system. the horizontal plane slow motions, and the mooring line model. Their models are presented in the following three subsections.
2.1 Spread Mooring System
Coordinate S stem. Consider a sessel moored to a spread mooring system. as shown in Fig. 1. Let O-XYZ be the body-fixed coordinate system with its origin located at the center of
gravity (G) of the moored vessel. The Z-axis is positive up ards
Journal of Offshore Mechanics and Arctic Engineering NOVEMBER 1998, Vol. 120 / 201
DIRECTiON OF WINO AND WAVES
(x,y)
Fig. i Geometry of spread mooring system
and the X-axis points toward the bow. The o-xv: system denotes
the inertial coordinate system fixed to the earth with the origin
located at the mooring terminal 1. The x- plane coincides with calm water level and the z-axis is positive upwards.
Geometric Relations. The following geometric properties are defined to describe the arrangement of a spread mooring
system. as shown in Fig. I: (x, v) are the coordinates of
the ith mooring line fairlead in the body-fixed coordinate
sys-tem; (x». y) and (x. v) are the positions of the ith
mooring terminal and attachment coordinates of the ith mooringline with respect to the earth-fixed coordinate system, respec-lively: 'y is defined as the angle between the x-axis and the ith mooring line counterclockwise; 1' is the length of the ith mooring line; i/i is the yaw angle.
The geometric relations of the mooring lines are
Kinematic Relations. The slow morions of the vessel in the horizontal planesurge. sway. yaware considered. The horizontal components of vessel velocity tu. y, r) relative to the surrounding water in sur2e. sway and yaw, respectively, are related to the inertial coordinates (x. 'i by the kinematic relations
r = u cos ¿b - L sin i' - U cos a (7)
= u sin ¿li + reos¿r + Usin a 8)
(9) where U is the magnitude of the current velocity. The direction of the current is measured with respect to the positive x-axis counterclockwise, as shovn in Fig. I.
2.2 esseI Motions and lAds
Slow-i1orion tfwi4i rj,l I,1nwít ':' The mathematical model is based on the applicarion of It)n' lass of motion
for linear and angular momentum in the earth-fixed coordinate
system o-xvz. Forces and morneni acting on the sessel are de-fined in the body-fixed coordinate stem ()-X}Z. Thus, trans-formation from the o-xv: system to the O-X7 system is
re-quired to write the equations of motion in the bodv-fned coordi-nate system. This transformation results in the Euler equations
of motion for a rigid body f du
m(
-\ dt(dt
-\ dt-
= F + ru) = dr - = Mwhere m is the mass of the vessel; I is the moment of inertia about the Z-axis; F, F, and M are the total forces and moment acting on the vessel in surge. sway, yaw, respectively. In the body-fixed coordinate system O-XYZ. Eqs. ( l0)( 12) can be
written as
(m - X ) i - mr = XH 1 + Xw + (13)
(mYjU Y,,'+mru=
YH+TV+YW+YWV
(14)N,r) + (J
- N,),'= NH +
(x'
- y
Tjgc) + N + Nwv (15)where X, Y. and N, are added masses and moment of inertia
in surge. sway. and yaw directions, respectively; Y, is the added
mass in sway direction due to yaw motion; N is the added moment of inertia in yaw direction due to sway motion; TJ5,
and denote the tension in the ith mooring line in surge
and sway directions, respectively: n is the total number of
moor-ing lines. Subscripts H. W, and W V indicate hydrodvnamic forces due to vessel motions, wind loads, and second-order
slow-drift loads, respectively.
Hvdrodvnamic Forces and Moment. The hydrodynamic forces and moment due to vessel motions can be expressed in terms of the slow motion derivaties. The following third-order Taylor expansion is used (Abkowitz, 1972):
XH = X, + Xu + Xu2 ± ±
+ X,52 + Xyu
Xru + X,,62u + X,vr
- X,r6 -
X,r6 -i- X,,uru ± X.,,uóu + X,r6u 16)= Y, + Y,u + Y,u
Yc + }Ç.u1+ Y,,vr2 +'4- Yvu + Y,tu2 + Y,r +
+ Y,rv2
+ Y,r62 + Y,,ru - Yru ± Y6 +
*
Yr,6r - Yòu +
Y,.,»5u + Yr,,vri5 (¡7)N= Pi, + N,,u + N,u2 + Nv +
V,u3 + N,,vr2-
-'- N LU -Vuu + N,r
-UI-
(1) = (2) VIi)-¡U) =
- x) + (y
-
yI)2
(3)x =x + x cos ¿Ii - y»sin ¿b (4)
= y + x sin ¿b + y,'' cos çl' (5)
For conenience. P is defined as
= 2r -
+ ii, (6)+ + + N,,ru + + Nb
+
+ N)òv +
+ Nóu+ N6u + N,6uró
(18)where subscripts o and b denote propeller and rudder angle effect, respectively. The resistance of the vessel R at velocity u is related to the first four terms in Eq. (16) by
R X + Xu +
+ (19)The hydrodynamic forces and moment acting on the vessel
depend not only on the instantaneous vessel motions, but also
the history of the surrounding fluid motion. The effect of the history of the surrounding fluid motion is referred to as the hvdrodvna,nic memory effect. This willbe studied in a separate paper.
Mooring Line Tension. The tension in the ith mooring line for the surge and sway directions can be represented as
Tg. = +T" cos »'
R' cos i/i + R'' sin ¿1' (20)= T' sin
.°' = RY' cos ii'R'
sin ib (21)The components of the mooring line tension Rk". RY' in the x-y plane are given by
R'» = T" cos y'
(22)R» = T' sin y' (23)
Current Load. Current loads are implicitly included in the expressions of Eqs. (16). (17), and (18) by using the relative velocities of the vessel with respect to the surrounding water. Therefore, current loads are accurate to third order. This ap-proach is more accurate than the usual method based on pro-jected area and drag coefficients. With no loss of generality. the current is assumed to be directed to the negative x-axis in the earth-fixed coordinate system (a = O deg).
Wind Load. Steady wind forces and moment exerted on the vessel are expressed as
Xw =
= PUC)W(a,)AL = PUUCZ(a,)L4L
where p is the density of air; U is the mean wind velocity
relative to the air at a standard height of 10 m above the calm
water level; Ar and AL are the maximum transverse and lateral
projected areas of the above-water part of the vessel, respec-tively: L is the length of the vessel. The relative angle a, of wind direction is defined as
a, = a, çt'
where a i th absolute angle of wind direction in the (x. y) coordinate system. The coefficients C(a,). C}W(a,). Cz4(a,)
depend on the vessel tpe. superstructure location, loading
con-dition. and the relative angle a, of the wind direction. These coefficients can be obtained from experimental and full-scale
measurement data.
Slow-Drift Load. The common approach to formulation of nonlinear wave-body interaction problems in ship and offshore hydrodynamics is based on potential flow theory using the wave
amplitude to length-ratio perturbation parameter. The solution of the second-order problem results in three types of wave
excitation forces: mean, difference-frequency (slowly varying).
and sum-frequenc ) springing) forces. Mean and sims h
vary-ing drift forces are of importance in moorvary-ing problems.
Nemans approximation (1974) for slowly varying drift
forces has been used extensively for many applications in ship
hydrodnamics. The important practical consequence of this approach is that slowly varying drift forces depend merely on
evaluation of second-order mean drift forces, and this can
re-duce the computational time significantly. The formula for slowly varying drift forces is
FSV
iv
y si rk v.
= AAT cas ((Wk w)t + (
-
)J (24)j-1 k=I
where the coefficient TL
0.5(D + D)
and i = 1. 2. 6indicate surge. sway, and yaw directions, respectively. A.
and are the wave amplitude, circular frequency, and random phase angle of component numberj(ork) in the total number of wave components N. The random phase angles are uniformly
distributed between O and 2 and remain constant in time. The
wave amplitude .4 (or A can be computed from the wave spectrum S(.') as
A,=2vS(w)&
(25)where is the difference between adjacent frequencies. The two-parameter Bretschneider wave energy spectrum formula is
used to express S( .). The two parameters are the significant wave height H, and the mean wave period T.
The time mean ' alue over the diagonal term' (whenk = j
in Eq. 24) gives the mean drift forces and the mean of the
off-diagonal terms (when k j) vanishes. Mean drift forces can be written as
F'
(Xwy. Ywv,Nwv)T= A'D (26)The transfer functions D' for mean wave drift force are approxi-mated as
D' = - = p,gLC,0 cos3 (9 - i/i) (27)
(28)
D6 N PgLCZDsin 2(8e ib) (29)
where is the s ave amplitude. p, is the density of water. 8
is the absolute angle of wave direction, and g is gravitational acceleration. The coefficients C0. Cr0. C can be obtained from the complete linear (first-order) solution of the
sea-keep-ing problem. Further details about the methods of predictsea-keep-ing mean drift forces can be found, for example. in the study of Ogilvie (1983). In this paper. it is assumed that these can be given simply in terms of the ship particulars (Hirano et al..
1980).
2.3 Mooring Line Model. In our line of work, we use
three t pes of quasi-static mooring line models: nylon or polyes-ter rope. chain, or steel cable. The stability of mooring systems is affected significantly by the line model (Berriitsas and
Papou-lias. 1990). Chains are usually fully submerged in water and
only slightly extensible. They experience large drag force and
undergo two-dimensional deformation in the vertical plane
modeled by catenarv equations. Steel cables are modeled by a
three-dimensional, large-strain, finite element model which takes into account drag force, weight. and extensibility. They
are lighter than chains and have less drag. but are more extensi-ble. Elastic ropes are light, considerably extensible, nearly
Table i Principal particulars of vessels and mooring line arrangements used in simulations
ant in water. Nonlinear behavior is due to their extensibility. and they are modeled as follows (McKenna and Wong. 1979):
(/ _1)
where T is the actual tension in the rope. Sb is the average
breaking strength. p and q are empirically determined constants. and i, is the working length of the unstrained rope. A wet nylon
rope of 120 mm diameter(Sb = 680. 625 lbf. p = 9.78. q = 1.93) is selected for the numerical applications presented in this paper.
Initial pretension ïs applied to one or more mooring lines. The pretension T, can be defined in terms of the initial strain E, of the line as
T, S5p (31)
E,
-
IS (32)where i, denotes the initial length of the pretensioned mooring
line.
3 Richness of SMS Dynamics
A barge with stabilizing skegs and a tanker without propeller are selected for numerical applications. Their principal
particu-lars and mooring line arrangements used in simulations are summarized in Table I. Configuration FmAs denotes a (m +
s)-line SMS with m lines attached on the moored vessel forward
of the center of gravity and with s lines aft of the center of gravity. Pretension is imposed on the forward mooring lines
and is calculated as a function of initial strain E,. Slow motion derivatives of the two vessels are provided in Bernitsas and
Kekridis (1986).
Nonlinear dynamic behavior of mooring systems such as
sta-ble or unstasta-ble focus near an equilibrium, limit cycle. and cha-otic motion hase been demonstrated by systematic and exten-sive simulations (Papoulias and Bernitsas. 1988: Chung and Bernitsas. 1992: Bernitsas and Garza-Rios. l996}. In this sec-tion. the richness of SMS dynamics and the effect of slow-drift
loads are demonstrated by simulation in Figs. 2 to 6. The
fourth-order Runge-Kutta time-stepping method is implemented be-cause it is relatively simple and sufficiently accurate. The geo-metric points GI to G4 used to define SMS configuration in Figs. 2 to 6 do not cover the entire range of design practice. They v.ere selected to demonstrate SMS d>namics. In Section
5. these simulations are studied further in relation to catastrophe
sets in Figs. 7 to Il.
3.1 Barge With Skegs. Figure 2 shows a periodic
oscilla-tion (limit cycle) of a two-line barge with skegs SMS about equilibrium A. which has zero 'aw angle. The amplitude of the aw angle increases shen mean drift forces are applied in the
T = Shp (30)
current direction (head seas 9,, = O deg). The system still
under-goes a limit cycle and the increased amplitude of yaw angle
due to mean drift forces results in higher tension in the mooring
lines. A different angle of wave direction (&,, = 30 deg) leads
to a nonsymmetric equilibrium position with nonzero yaw angle. Figure 3 shows that the dynamical behavior of mooring systems can be different qualitatively, depending on the mean drift force.
As can be seen in Fig. 3. the two-line barge with skegs SMS
experiences stable focus behavior and converges to equilibrium A when only current load is applied. A limit cycle near equilib-rium A appears when mean drift forces are applied. This simula-tion shows that an SMS may exhibit large-amplitude slow
peri-odic motions due to mean drift forces. As will be e'splained in Section 4. this is the result of dynamic loss of stability and not
resonance. In Fig. 4. slowly varying drift forces have a
stabiliz-ing effect on the SMS: actually, higher slowly varystabiliz-ing drift
forces result in a stable equilibrium A. This shows that there is no resonance in yaw. In surge. however. Fig. 5 shows resonant
motion on top of the periodic motion due to loss of stability.
3.2 Tanker Without Propeller. Figure 6 shows that a three-line tanker without propeller SMS converges to equilib-rium B in head current without waves: equilibequilib-rium A becomes unstable and the system conterges to equilibnum B. which has a nonzero yaw angle. When mean drift forces are added to the excitation. equilibrium A becomes stable. In this case, mean
drift forces have a favorable effect on the system stability. This
is the opposite effect to that obseed in the barge with skegs
SMS in Fig. 3. The system is unstable and the aw angle
fluctu-ates about equilibrium B when slowly varying drift forces are applied in head seas.
4 Nonlinear Stability Analysis
4.1 State Space Representation. The SMS model pre-sented in Section 2 consists of the equations of motion (13)-(15) and kinematic relations (7)(9). It can be modeled by
six first-order nonlinear coupled ordinary differential equations using the following six state variables: (X1 = u. .r i. x, = r.
r4 = t. .r5 = y. r6 = ib). In Cauchy standard form, they become
=
mX,
{XH(XI.X..X) tRY Cos.t - R sin.v5))m.x..r, -
+ X(x6)
J m - X,(1N)
= D f YH(xr r-. x,) +(Rcosx - R' sin.r5)J
+ (- V
Ott1.v I.r) D -f(x,R
+ v;'Ry') sin .r J +\'
(x6) + N141 (.)J{YH(.VI. X:..r) + (R' cos x - sin r5)
+ f
nrr ± },ix)
Y,( .v)fv, R, ) cos r5)
204 / Vol. 120, NOVEMBER 1998 Transactions of the ASME
System 2.Line SMS 3-Line SMS Configuration FIAi F1A2
\eseI Barge with Skegs Tanker without Propeller
L 191.56 ft 1066.3 ft B 35.00 ft 173.90 ft T 10.42 ft 71.30 ft CB 0.855 0.831 E1 0.013 0.04 U 2 knots 2 knots H413 4.5 ft 8.0 ft.
+ -
t:. .t) (.r DFig. 2 Effect of mean drift forces on two-line SMS (FIAI, I_IL = 0.3, x,IL = 0.3, E, 0.013), barge with skegs
0.03
0.02
.-0.01
1/L=0.6, x/L=0.3 (G2), E1 = 0.0)3
w/o mean drift
w mean drift
lt
I $I
I I t i , , I i , ,I Ihlilili, t
, , , n 150 200 250 300 350 400Nondimensional Time (Ut&)
Fig. 3 Effect of mean drift forces on two-line SMS (FIAi, I_IL = 0.6, x,,/L = 0.3, E, = 0.013), barge with skegs
X4 = X1 COS X,, - X2 Sifl X - U cos a 36
= X1 Sifl X X: COS X, + U sin a ( 37
= 3S)
where D = (J - N,)(m - Y) - NY,. Evolution Eqs.
(33-(38) can be written in the form of one six-dimensional first-order vector differential equation as
= f(x).
fC'. f:6-
(39)where x denotes the derivative of stale vector x with respect to
Journal of Offshore Mechanics and Arctic Engineering NOVEMBER 1998, Vol. 120 / 205
C D . 0.00 Q) C e-0.01 -0.02 -003 t I I I I i 0 50 100 D {,VLV. X', x) .v
R:' - v R) cos
J(m - Y)
(,rR
+ vRY) sin x J L) (m Y){N(.) + N5())
(35)0.03
Fig. 4 Effect of slow-drift loads on two-line SMS
(FIAi, I,/L = 0.4, x,/L = 0.3, E,= 0.013), yaw, barge with skegs
/L=0.4, x/L=0.3 (G3), E, = 0.013
Fig. 5
Effect of slow-drift loads on two-line SMS (F1A1,Ç/L = 0.4, x,,/L = 0.3, E, 0.013), surge, barge with skegs
time t. L6 is the six-dimensional Euclidean space. and C' is the class of continuousl differentiable functions. Since the vector field tin Eq. 139 does not dependexplicitly on time i. the system is said to be auro,i,mous. The mean drift forces as expressed by Eq. (26) are autonomous. The slov,lv varying drift forces given b Eq. i24). howe er. depend explicitly on time and would render the SMS model nonautonomous.In general. stability analysis of
a nonautonomous nonlinear system is not possible. Time
simula-fon of a nonautonomous s stem hich includes slowly varing
drift forces. hoeer. can be pertbrmed ithout any difficulty
and is used in Sections. and 5 to assess the effects of
nonautono-mous terms on the stability of the corresponding autonomous system.
4.2 Equilibria of SMS. Equilibria of the vector field (39) are the points X E R such that
O = f(X). X = (., .t.. ). (40)
These are stationary solutions. Equilibria can be interpretedas intersections of the null clines. The equilibria of SMS. which
0.02 0.01 (e V cc 0.00 (e >-0.01 -0.02 -0.03 /L=0.4, x/LO.3 t (G3), , E, = I 0.013
i)i
- I $ J:tJ $ w meandnft W slowty-vasying $i.
50 100 150 200 250Nondimensional Time (ULt)
206 / Vol. 120, NOVEMBER 1998
0.04 0.03 (e 0.02 0.01 0.00 -0.01 0
1,/L=0.i5, x,/L=O.S2 (G4), E, = 0.04 w/o mean drift
w mean drift w slowly-varying
50 100
Nondimensional Tui,e (Utt)
150 200
Fig. 6 Effect of slow-drift loads on three-line SMS (FIA2, I,.,/L = 0.15, x0/L = 0.52, E, = 0.04), tanker without propefler
+G2
wlo mean drift
H,0 = 2.0 ft H,0 = 4.5 ft
IL
Fig. 7 Catastrophe set of two-line SMS (FIAI, (y,/B = 0.0, E, = 0.013), barge with skegs
are the solutions to a nonlinear Eq. (40 i. depend on the mooring Let be an equilihnum point of the nonlinear system (39) line and system configuration. as well as on the environmental Expar.iing the nght-hand side of Eq. (39) into its Taylor series
conditions, about point . we obtain
4.3 Stability in the Sense of Lyapunov. The behaviorof
a nonlinear system near an equilibrium point can be determined via linearization with respect to that point. Since linearization is an approximation in the neighborhoodofan equilibnum point, it
can predict only the local behavior ofthe nonlinear system in
the vicinity of that point (Guckenheimer and Holmes. 1983: SedeI. 1988). df X
= f
) + -dx,'-:';
±
± HOT. -.where deviation from equilibrium is defined as
(41)
Journal of Offshore Mechanics and Arctic Engineering NOVEMBER 1998, VoI. 120 / 207
1.0 0.8 0.6
tì
0.4 0.2 0.000 I H I I :' 0.1 0.2 0.3 0.4 05Fig. 8 Catastrophe set of two-line SMS (F2AO, yIB = 0.0 fl = 5 deg), barge with skegs
Fig. 9 Catastrophe set of three-line SMS (F2AI, y,/B 0.5. 11 = 15 deg, E = 0.013), barge with skegs
the nonlinear state Eq.(39)can be approximated by the linear
state equation
=[AJ.
[A) (44)where [Al Df() ts called the Jacobian matrix evaluated at the equilibrium point . If the eigenvalues of [A]. X,. have nonzero real parts (usually called hperbolic or
nondegener-ate),then it can be expected that the trajectories of the nonlinear
system in a small neighborhood of an equilibrium point are close to the trajectories of its linearization about that point.
5.0 4.0 3.0 2.0 1.0 An '00 I
w/o mean drift
I'1,3=2.Oft - H,=4.5ft
-0.1 0.2 IL 0.3 0.4 05208 I Vol. 120, NOVEMBER 1998 Transactions of the ASME
E(I) = x(t) -
(42)Since is an equilibrium point, we have f() = O. Then. the state equation can be written as
+ HOT. (43)
If attention is paid to a sufficiently small neighborhood of the equilibrium point so that the higher-order terms are negligible.
1.5
1.0
0.5
w/o mean drift
H,,3 = 4.0 ft I-Ç3 = 8.0 ft II Iv -St'S--. I III
Fig. 10 Catastrophe set of three-line SMS (FIA2,y0/B = 0.5, lì = 5 deg, E, = 0.04), tanker without propeller
Fig. 11 Catastrophe set of four-line SMS (F2A2, y,,/B,,,,, = 0.0, y0/B.ft = 0.5, 1? = 5 deg, E, 0.03), tanker without propeller
4.4 Bifurcation and Catastrophe Sets of SMS. A bifur-cation is a qualitative change in a limit set of the s stem as a parameter is infinitesimally perturbed. Bifurcations can be interpreted as the appearance or disappearance of a limit set and the change in the stability type of a limit set. A bifurcation also occurs hen a nonstable limit set of the system remains nonstable. but undergoes a change in the unstable manifold dimension. Saddle-node, pitchfork. and Hopf bifurcation have been found in the nonlinear dynamics of mooring systems
(Pa-poulias and Bernitsas. 1988: Chung and Bernitsas. 1992:
Berntt-sas and Garza-Rios. 1996). A catastrophe set is the set of all loci of bifurcations in a parametric design space. It defines the boundaries beteen regions of qualitatiel dIfferent system
dynamics and the morphogeneses occurring as boundaries are
crossed.
In the numerical applications below, catastrophe sets are
con-structed for several SMS configurations of a barge with skegs and a tanker without propeller. The effect of mean drift forces
on bifurcation of SMS dynamics is demonstrated in the
catastro-phe sets of Figs. 7 to 11. Five different SMS with symmetric
Journal of Offshore Mechanics and Arctic Engineering NOVEMBER 1998. Vol. 120 / 209
0.35 0.40 0.45 0.50 IL