BEHAVIOUR

OF OFF-SHORE
STRUCTURES
Proceedings of the
Third International Conference
Volume 1
Edited by
Chryssostomos Chryssostomidis
and
Jerome J. Connor
both of
Massachusetts Institute of Technology
o
HEMISPHERE PUBLISHING CORPORATION
Washington New York London
DISTRIBUTION OUTSIDE THE UNITED STATES
McGRAW-HILL INTERNATIONAL BOOK COMPANY
Auckland Bogota Guatemala Hamburg Johannesburg Lisbon
London Madrid MellCico Montreal New Delhi Panama Paris
San Juan sao Paulo Singapore Sydney Tokyo Toronto
I
THE STATICS AND DYNAMICS OF THE
MOORING LINES OF A GUYED TOWER
FOR DESIGN APPLICATIONS
H. S. Triantafyllou G. Kardomateas A. Bliek
Massachusetts Institute of Technology
SUMMARY
The statics and dynamics of the supporting cables of compliant structures are of primary im
portance in studying the behavior of the overall system. In this study efficient analytic
solutions are presented for the guyed tower that reduce significantly the computational effort
required. and allow a wide parametric search, which is essential in the early design phases.
First, the cahle statics are presented and the importance of the concept of the effective
tension is discussed. Analytic solutions including the elasticity effects are derived.
Next, the cable dynamics are studied and analytic solutions are developed using perturbation
techniques. In particular, the effects of elasticity and catenary shape are presented since they
can alter the nature of the dynamic response. The form of the solutions allows a simple treatment
of the dynamics of multi-leg systems.
The dynamics of the guyed tower are studied and the effect of the clump weight is assessed
i' 'e form of a softening spring. The inertia and drag forces of the clump weight are identified
a, urces of dynamic amplification and phase lag. The use of perturbation techniques allows an
e t ! ~ c i e n t study of the dynamic behavior of the system.
The effect of nonlinearities is discussed and analytic techniques to account for such ef
fects are presented.
The theory presented is illustrated by studying a specific guyed tower in 1,500 ft. of water.
INTRODUCTION
The dynamics of structures supported by mooring lines are complicated because of the blend-
of the dynamics of the structure with those of the cables. The need for efficient tools is
pressing at the early design phases, when no detailed system configuration exists so
as to exercise it by using numerical techniques, such as finite elements or finite differences.
At the same time, the most important parameters are to be selected at this phase, with very sig
nificant consequences for the subsequent design phases. '
The of the dynamics of a single line is such that simplistic models lead
to unreliable nredictions. The fact that the governing are nonlinear partial differen
tial equations with variable coefficients forces many investigators to employ numerical techniques
at a very early stage, with significant consequences for the cost and flexibility of the design
procedure.
This study is presented so as to indicate some siMPlifying facts concerning the behavior of
mooring lines. In particular, the major factors contributing to the response of a cable are indi
cated and a simnle but comprehensive dynamic analysis is outlined. The nonlinearities are also
studied, but it is the belief of theauthors that their effect can be assessed very efficiently
by using analytic techniques with a better qualitative understanding than by employing numerical
techniques.
It should be emphasized that the present paper focuzcs on the preliminary design phase and the
results that are derived are simple but efficient solutions allowing a wide parametric search.
These results are by no means general solutions to the cable dynamic problem, since such analytic
expressions are impossible to obtain, but they are applicable within the frame of selecting the
various parameters involved so as to minimize excessive dynamic phenomena.
The results have been derived for a guyed tower to demonstrate how to obtain simple solu
tions in an application-oriented manner.
GUYED TOWER
The guyed tower is one of the concepts that is being studied as an economical alternative
to the jacket platform for water depths between 1,000 and 2,000 ft.
The has been described in detail in several Dublications (lJ (2] (3] (4J. The sys
a slender truss-type tower supporting the dec:, on which all working equipment is
i.. _alled. The lower end of the tower is connected to a snecial foundation which imposes small
rotational restraint on the structure. Proposed foundations include a spud can and ungrouted piles
extending along the full length of the tower. The lateral support is provided by a number of
cables whose upper end is secured on the deck by cable grips, while they pass around fairleads
located below the water level and then they extend radially to clump weights located on the sea
floor. The clump weights are connected through horizontal lines to the anchors. Buoyancy tanks
are provided so as to reduce the foundation and to additional lateral moment
(figure 1).
The system could be used to change the pretension level so as to handle extreme
weather conditions.
The primary objective is to obtain a compliant system whose first natural frequency, treating
the tower as a rigid body, is below the wave frequency range (typically between 0.2 and 1.5 rad/
sec). This objective places the guyed tower at the opposite end with respect to the wave frequency
range than the jacket platform, whose first natural period for shallow water is typically 2-4 sec.
The problems that arise include the dynamic behavior of all subsystems involved, especially
of the mooring lines: the structural integrity of the tower and the cables both in extreme loads
and fatigue: and the damage stability of the system.
MOORING LINES
The mooring line is capable of supporting external forces by re-adjusting its configuration,
given that it can support only positive tension which is tangent at each point to the line shape.
This fact noses limitations to the holding capacity of a mooring line in the case of dynamic loads,
because if the rate of change of the external force is fast, the cable configuration does not
change as fast, resulting in poor holding capabilities. In addition, the configuration forms waves
which may cause resonance phenomena and therefore large parasitic tensions.
The saq to length ratio is a very important parameter for cable dynamics. When the sag is
very small, we obtain essentially a taut wire which can easily support dynatlic forces along its
&47
axis, while it provides very little support against any fast varying The long
itudinal load can change fast except that its frequency range must remaln below the flrst longitu
-'.nal (i.e. stretching) natural frequency. The longitudinal natural are given as:
(c.lmi t (1)
where L is the cable length, E is Young's modulus and p the cable density, while the trans
verse natural frequencies are (assuming a constant tension T.)
(2)
with d the cable diameter. For a steel cable the ratio T./fd
2
must remain below the elastic
limit so by using some typical steel properties we obtain:
E .. 40 (3) . ml
c.l n (T 1!d
2
) n
n t 4
We conclude that the first longitudinal frequencies are much larger than the first trans
verse frequencies. This is the reason why the cable can sustain dynamic loads in the
longitudinal direction, while for normal wave and vortex shedding loads its longitudinal response
is quasi-static (i.e. the longitudinal stretching can be by a distributed spring).
One particular aspect of taut cables is the possibility for parametric resonance [51: The
axial quasi-static stretching of the cable causes an oscillatory change in the tension which con
stitutes the restoring mechanism in the transverse dynamics. As a result, the transverse dynamics
are characterized by an equation whose time dynamics are in the form of a Mathie'J equation. Under
certain combinations of wave frequency versus cable natural frequency, large transverse oscilla
tions may occur.
When the sag is large the stretching becomes unimportant for the transverse dynamics, which
are essentially the dynamics of a chain [61, [71. As outlined in [8], we can define a tangential
and a normal unit vector at any point along the static configuration and define a dynamic motion
along the normal (transverse dynamics) and a motion along the tangential (axial dynamics). The
consists of the sum of a solution that varies slowly along the cable length and is essen
ally a perturbation of the static configuration (catenary dynamics): and a solution which in
_ransverse resembles the motion of a taut string.
For moderate values of the sag (of the order of 1/8 of the length) both stretching and
catenary dynamic effects are important. This was first recognized in [9] and a solution was ob
tained which the theories for small sag (taut wire) large sag (chain dynamics).
The nonlinearities involved in the dynamics of a cable are composed of geometric nonlineari
ties (large amplitude motion): nonlinear stress-strain relations, especially for large dynamic
tension: and nonlinear fluid which can be represented in the transverse direction as
(4)
where F is the transverse force per unit length, pw the density of the fluid, d the cable
diameter and Un the relative velocity between the fluid particles and the cable in the normal
direction.
As we have already pointed out we are interested here in solutions useful for design. This
means that we must identify the mechanisms leading to large dynamic motions but we need not study
those extreme motions, which are known to cause failure. hs a result we try to find the dynamic
response with as simple a model as possible: From the three classes of nonlinearity indicated
above only the nonlinear drag has a definite effect on the r1ynamics we are interested in.
When a system is lightly damped the (small) nonlinearities can be studied by using a double
perturbation exnansion in amplitude and time, as for example in the method of multiple scales (101.
The characteristic of the response is that it consists of a forced part and a homogenous part
which may not be decaying.
On the contrary, a system with significant damping responds at the frequency of the forcing
function while all other terms are very small or fast decaying. This is true for nonlinear damp
ing as well and although the response is a nonlinear function of the amplitude of excitation there
is no need to in multiple time scales.
648
The drag term (equation 4) is a nonlinear function of the velocity Un so its effect depends
on the amplitude of the forcing function. For a cable driven at its upper end the motion is usu
ally equal to several times its diameter so the damping is important. This means that the method
. harmonic balance (11] or the similar method of equivalent linearization (11] can be used to
ain the cable dynamics. This fact explains the success of applying the method of equivalent
__ nearization on the problem of the forced response of slender cylinders in water driven at their
upper end [8}, (12].
For very small motions, the effect of the nonlinear damping is of the same order as the
effect of the structural damping, i.e. the system is rather lightly damped. As a result, cable
strumming (13] is characterized by phenomena of lightly damped systems such as frequency entrain
ment, multiple frequency response etc., and although we can not claim that we have a complete
model for vortex shedding, we can at least qualitatively describe cable strumming
as the limit cycle of a lightly damped nonlinear system (15].
This concludes the review of the dynamics of mooring lines which is based primarily on
physical argument rather than mathematics. In the sequel we develop quantitative techniques, we
we will refer though to the arguments of this section quite frequently, in order to simplfy the
equations used.
STATICS OF MOORING LINES
The static loads on the cable of a mooring line consist of the external forces, its own
weight and the drag force from the current.
A .At each point of the cable configuration we define a tangential and normal unit vector, t
and n respectively (Figure 2). Let all quantities referring to the unstretched cable have a
subscript o. If W is the weight per unit length, d the cable diameter, the angle between the
vector t and the horizontal, T the tension, U the local current velocity, s the coordinate along
the cable, Co the normal drag coefficient and Cf the (tangential) frictional coefficient, then
the governing equations are (8}
T
e
+
(5)
- U
2
(6)
T
e
., (7)
d" .. (l+e)
GS
o
(8)
.. (l+e)
iiS
o
(9)
e ..
(10)
where E is Young's modulus and z the vertical distance from the free surface, while
8 .,

0 (11)
I
F .,
Cd do
n (12)
I
F
2"P Cf'lld
t o (13)
The derivation of equation (10) requires that Poisson's ratio is 0.5 which is an acceptable
assumption at this stage. We must emphasize the fact that the hydrostatic force is normal to the
cable configuration and not vertical. This leads to the formulation of the problem in terms of
the effective tension T as defined in equation (7) rather than the actual tension. Essentially,
we can say taat all staiic solutions derived for cables in the air are valid for cables in water
(in the absence of current) if the tension T is replaced by T
e

This leads to surprising results because when the depth becomes large the actual tension
may become negative, while the extension (which depends on T
e
according to equation (10) is
positive.
This is due to the squeezing- action of the external hydrostatic pressure, which causes
stretching in the longitudinal direction although the tension may be negative. This facet has
549
been overlooked sonetimes in the literature and may lead to significant errors.
If the mooring line employs a chain then the above are not applicable because each chain
link is practically entirely submerged so the buoyancy is vertical and the effective tension is
equal to the tension. i.e. for a chain equation (7) is replaced by
T T (7a)
e
The actual tension therefore in a chain is larger than the tension in a cable under the
same loading conditions.
If the current action is iqnored. since usually the current is a near surface phenomenon,
then the extensible catenary solution can be obtained:
T - __H_. JH
2
+ (V-Wl [L-s) 2 (14)
e
x H {sinh-
1
[V-Wl (L-s)1 sinh-I [v-W
l
Hs
VI H--j - ---g T AE (15)
y - 0 +;nIL-OJt
(16)
V W
tan. Ii - RI (L-s) (17)
where V is the vertical and H the horizontal force at the top end, L is the unstretched
cable length, and:
(18)
(19)
Note that i. the unstretched coordinate and the subscript 0 has been ommitted for con
venience. The stretched length becomes:
(20)
while it i. reminded that
ainh-h) In[x+J1+x
2j
if x (21)
If the force at the top is large compared to the net weight of the cable, i.e.
F /V
2
+H
2
W
l
L
then a shallow sag configuration is obtained, while elasticity effects are important. By
Ysing a perturbation expansion with W1L/F and F/AE the small we can obtain a simple
solution to the static equations. By keeping the first two orders we find
T
e
F - WI
(22)
W F s
.0 .s+ f U
2
(sl)ds
l
+ t
b
(23)
o
F
x
cost ' s + cos4i S
o o
s'
b+
ttl U
2
(.s2) dS
2
dS
l
+ tbSJ
(24)
o 0
F
+ EA s +
Y -
s'
+ b+

s
J
S
1
U
2
(s2) dS
2
dSl+tbS]
(25)
o 0
where if D is the water depth
(25)
650
while
It should be mentioned that the original equations (5) through (13) are easy to implement
.. a finite difference scheme in the computer and then solve them iteratively fully accounting for
current and elasticity effects. The analytic solutions are of value so as to check the numerical
solutions, but primarily to derive efficiently the overall statics of the multi-leg system.
DYNAMICS OF MOORING LINES
By using the normal and tangential vectors defined in figure 2 we can derive the equations
of motion in the normal and tangential directions: (The effect of the current on the added mass is
omitted).
(JI'
e
m m
-Wlsinljl + R (l+!)
o
ar-
0 t (28)
0
(mo+ao}ir
+ m u 1t
"" -Wcosljl +T
e

R
o dt n (29)
0
where m is the cable mass per unit unstretched length, a the added mass per unit length,
o
u is the tangeRtial and v the normal cable velocity and R
t
, R are the fluid forces in the t, n
n
directions respectively:
TT 2 'iN
R .. (ldo p+a ) r + F (Vn-v) Ivn-v I (30)
n o n
R .. Ft(Vt-u} Ivt-u I (31)
t
where V is the normal and V the tangential velocity of the fluid. Also, the following
compatability must be
_ v

(32a)
eN + u (l+e) (32b)
ds ds o\.
O O
T and e are defined again by equations (7) and (10) respectively.
e
It is important to compare the dynamics of the overall system with those of the cables, in
order to decide for the type of governing equations to be usp.d. Let us consider small amplitude
motions of the cable around an average static configuration defined by a tension To(S) and 1jI0(s).
Then if we subtract the static equations from equation (28) through (32) we obtain
(33)
(34)
(35)
(36)
where IjI , T are the time varying parts of 1jI, T while the subscript 0 has
l
been ommitted troM s again for convenience. Note thatethe nonlinear drag force has been included
in equation (33) because, according to the discussion in an earlier section, this will govern the
type of technique to be used. No fluid particle motion has been assumed in the equations (33)
(36)
Since the mooring lines are under large tension the effect of elasticity is as important as
the effect of the sag. This is the reason for including the elasticity term in equation (36).
Note that still no longitudinal elastic waves are present as explained in an earlier section. The
solution of equations (33) through (36) has been derived for a horizontal, flat catenary in [9],
while the solution for an inelastic cable has been derived in [8].
551
Using the data provided in Appendix 1 and the solution derived in Appendix 2, the natural
frequencies are derived as shown in Table 1 (i.e. by taking
It is important to determine the effect of the damping term. If is small, a multiple
technique must be applied, while for larger F
n
the method of balance is appropriate
,lce the cable will be forced to follow the motion of the tower the excursions will be large .
compared to the cable diameter, and the contribution of the damping is important. As a result the
method of harmonic balance can be used.
Assuming a single frequency excitation at the top, of amplitude a and frequency w
o
, following
the standard procedure of the harmonic balance (11], we replace the nonlinear term FnV Ivi with the
equivalent linear ter.m bY, where (8]:
(37)
while the response will be predominantly at the exciting frequency Woe
GUYED TOWER DYNAMIC
The natural frequency of the rigid body dynamics of the guyed tower is placed below the
range of wave frequencies, typically around 0.25 rad/sec. This fact combined with the results of
table 1 indicates that the dynamics of the cables and the rigid body dynamics of the system are
separated by a factor of 4.
When we consider the dynamic response of the tower at low frequencies, therefore (when it is
expected to move significantly), we can consider the cables to respond quasi-statically. There
is one important consideration though: The clump weight is sUbject to inertia and drag forces, so
that its response will be dynamic, i.e. there will be dynamic amplification and phase difference
in its response, although the cables behave essentially as (nonlinear) springs.
(a) Quasi-static Modeling
Let us first adopt a quasi-static modeling of the mooring lines. Then the analytic solu
tions for the statics can be used to derive the force-displacement relation. The first part of
the cable connects the upper end of the clump weight to the tower and its configuration is that of
a shallow sag cable so that equations (22) through (27) can be used, or, if the current can be
neglected, the elastic catenary equations (14) through (21) provide better accuracy. The distri
ed clump weight behaves clearly as a heavy inelastic catenary, so that equations (14) through
) can be used by neglecting the elasticity terms.
The third of the line is the cable connecting the lower end of the clump weight to the
anchor. Its transverse motions are small so it can be modelled essentially as an equivalent
spring.
This modeling neglects the inertia and drag forces on the clump weight, as well the
frictional forces between the weight and the bottom. It is used to provide a first rough estimate
of the dynamic behavior of the tower.
By matching the solutions for each of the three parts we obtain a relation between the ver
tical and horizontal forces at the upper end and the horizontal displacement at the top. Figure 4
shows a plot of the horizontal force versus horizontal displacement for a single cable (data from
Appendix 1). Hhen several cables participate we can project all forces along the direction of
motion and obtain an overall force-displacement relation as shown in figure 5 for 20 symmetrically
placed cables (the graph depicts an odd function so only half of the curve is shown).
The relation thus obtained is that of a nonlinear spring: For displacements below a critical
value the relation is that of a weakly hardening spring; above the critical value the relation
changes fast to a softening spring. In physical terms the clump weight allows relatively large
restoring forces for moderate displacements, while for large displacemnts it allows large motion
so that the tension level is kept below the breaking limit.
It is to study the effect of such a nonlinear spring on the rigid body dynamics
of the tower. As we have already mentioned, damping is a very serious consideration when treating
the dynamics of nonlinear systems, so we start by considering the effect of the nonlinear damping:
We write the equation of motion of the tower by equating the moments acting on the structure with
the bottom as reference point. In order to make the analysis possible we write separate drag terms
for the tower velocity and the fluid particle velocity:
(381
where I i. the moment of inertia of the tower, I the added moment of inertia, C the non
linear damping coefficient, M6 the excitation moment, the horizontal cable force and F
y
the
vertical cable force. 0 is tHe distance from the bottom point to the cable attachment and the
angular motion of the tower. It ia convenient to define an equivalent drag diameter of the tower
io, and an equivalent inertia diameter d
1
(39)
(40)
where d
i
(i-I, 2, N) are the diameters of the N members of which the tower section con
.ists. Then
(41)
(42)
where P is the water density, 0
0
the water depth and Co the drag coefficient. Using
figure 5 we can derive a polynomial approximation to the hor1zontal force-displacement curve in
the form
+ I
3
(Xl_X'
0 (43)
where aI' a , I , l curve-fitting constants and x the clump-lifting displacement. The
approximation vel; shown in figure 5 for the spicific example considered. For the
vertical force it is sufficient to consider
Next we change from the angle t to the horizontal displacement x,
x(t) - t (t) 0
and then nondimensionalize by setting
i.e.
(45)
so as to obtain
t t
1+1a
tw
0
(46)
(47)
(48)
where
C x
C*.
J.+1a (49)
f* (u)
1
u > 1
(1+1 ) w'
a 0 (50)
while F is the nondimensional excitation moment amplitude and if 2 is the frequency of
excitation
553
._ Q
"'1 wo (51)
By definition w is the natural frequency of the tower for small displacements around the
equilibrium. Yt is interesting to use the data provided in appendix 1 and derive a nu
Lical value for the coefficients of (48):
+ 0.3566 u F*{u) + F sin(wlT)
(52)
u < 1
0
f* (u)
l 0.2289 (u-l)
+ (uJ-l) u > 1 (53)
W
o
0.2275 rad/sec
(54)
The perturbation techniques are expected to provide reliable data in this case since all
nonlinearities have relatively small coefficients. This is indeed the case as simulations have
shown and it is important to note that the damping term has a relatively large coefficient: As
simulations have verified the
The nondimensional form of the equation of motion is convenient in assessing the effect of
nonlinearities, it should be noted though that, as equation (46) indicates, the form of the
equation is correct for amplitudes of the order of x (i.p.. for u of the order of 1). For small
amplitudes we must nondimensionalize with respect toOa smaller amplitude, in which case the value
of C* will be significantly smaller, i.e. the small motions of a guyed tower are lightly damped.
Figure 7 shows the relation between nondimensional amplitude {with respect to xo)and fre
quency W for damping, as derived by using perturbation- techniques (see appendiX 3). The
spring form is clearly shown while some areas are unstable (thick lines) so
that jump phenomena can occur. For the typical drag coefficients of a guyed tower, the resulting
damping is large so that the method of harmonic balance can be used instead, i.p.. the tower re
,sponds only at the frequency of excitation.
(b) Dynamic Modeling
In the previous section the analysis was based on a quasi-static modeling of the cables.
As shown in the section on cable dynamics, this is adequate for modeling the inclined upper part
the cable for frequencies close to the natural frequency of the tower since the first cable
.ural frequency is 4 times higher. The only part of the cable that is not adequately modelled
by a quasi-static model is the clump weight, which is bulky and heavy so that its inertia and drag
forces are significant.
This means that the motions of the clump weight will introduce dynamic amplification and
phase lag in the overall response, although the cable response can be modelled as quasi-static.
This observation can save us significant computational effort since we do not need to model the
distributed behavior of the cable which is known to cause numerical problems unless special care
is taken for the fast longitudinal dynamics.
Let us model the inclined cable connecting the clump weight to the tower as a nonlinear
spring and the cable between the clump weight and the anchor as a linear spring. We can model the
clUMP weight as a distributed mass subject to added mass and fluid drag forces. No waves develop
along the clump weight so its space configuration can be approximated by a parabola for the part
which has lifted from the bottom and a straight line lying on the bottom for the remaining part.
approximate techniques such as suggested in [17] could also be used.
The simulation scheme is significantly simpler than a general model modeling the cables by
finite difference or element techniques. Figure 8 demonstrates the dynamic amplification and
phase lag introduced by the dynamics of the clump weight.
By ommitting damping forces such as the friction between the clump weight and the bottom,
oscillations may appear in a simulation scheme which in are quickly damped by those ne
glected forces. This is a source of trouble especially if numerical models are used for the
cables, which are very sensitive to stretching oscillations.
CONCLUSIONS
The purpose of this paper is to indicate some analytical techniques that can provide sig
nificant insight in the static and dynamic behavior of moored structures and therefore assist the
designer in selecting appropriate values for the parameters involved especially at the early
design phases.
The guyed tower, like other compliant structures, has a low rigid-body natural frequency by
design, while the mooring lines are under significant tension so that the natural frequencies of
the cables are far from the frequency range where significant guyed tower motions are obtained.
As a result, we can model the cables as nonlinear springs and concentrate on the proper simulation
~ f such components as the clump weight, whose effect on the dynamics of the tower is significant.
A simple overall model is obtained providing flexibility for configuration changes and efficiency
of computation for a wide parametric search.
The dynamics of the cables are studied as sources of parasitic forces, while their struc
tural integrity must be guaranteed against excessive dynamic tension or fatique. Some efficient
solutions are derived for the linear dynamics of stretched cables, which agree with recent devel
opments in cable dynamics. These solutions are particularly useful to analyze the behavior of
multi-leg systems.
REFERENCES
1. Finn, L. 0.,1976 -A New Deepwater Offshore Platform - The Guyed Tower-, Offshore Technology
Conference, Paper No. 2688, Houston.
2. Finn, L. D., Young, X. E., "Field Test of a Guyed Tower", Offshore Technology Conference,
Paper No. 3131, Houston.
3. Mangiavacchi, A., Abbott, P. A.,. Hanna, S. Y., Suhendra, R. N., 1980, -Design Criteria of a
Pile Founded Guyed Tower, Offshore Technology Conference, Paper No. 3882, Houston.
4. Finn, L. D., Thomas, G. G., 1980, A Guyed Tower for North Sea Production, Offshore Tech
nology Conference and Exhibition, Norwegian Petroleum Society.
~ . Lubkin, S., Stoker, J. J., 1943, Stability of Columns and Strings Under Periodically Varying
Forces, Quart. Applied Math., Vol. I, No.3.
6. Saxon, D. S., Cahn, A. 5., 1953, Modes of Vibration of a Suspended Chain", Quart. Journal
of Mechanics and Applied Math., Vol. VI, pt. 3.
7. Pugsley, A. G., 1949, On the Natural Frequencies of Suspension Chains, Quart. Journal of
Mechanics and Applied Math., Vol. II.
R. Triantafyllou, M. S., Preliminary Design of Mooring Systems, Journal of Ship Research, March
1982.
9. Irvine, H. M., 1981, Cable Structures, Cambridge, Massachusetts: MIT Press.
10. Nayfeh, A. H., Mook, D. T., 1979, Nonlinear Oscillations, New York: Wiley - Interscience.
11. Vidyasagar, R., 1978, Nonlinear Systems Analysis, New Jersey: Prentice Hall.
12. Paulling, J. R., 1979, Frequency Domain Analysis of OTEC CW Pipe and Platform Dynamics",
Offshore Technology Conference, Paper No. 3543, Houston.
13. Ramberg, S. E., Griffin, O. M., 1975, Velocity Correlation and Vortex Spacing in the Wake
of a Vibrating Cable, ASME paper 75-FE-7, Fluids Engineering Conference, Minneapolis.
14. Wilhelmy, V., Fjeld, S., Schneider,S., 1981, Non-Linear Response Analysis of Anchorage
Systems for Compliant Deep Water Platforms, Offshore Technology Conference, Paper No. 4051,
Houston.
15. Blevins, R. D., 1977, Flow-Induced Vibration, New York: Van Nostrand Reinhold.
16. Xardomateas, G., 1982, Dynamics of a Guyed Tower Offshore Platform, SM Thesis, Massachusetts
Institute of Technology, Department of Ocean Engineering, Cambridge, Massachusetts .
17. Suhara, T., Koterayama, W., Tasai, F., Hiyama, H., Sao,' X., Watanabe, X., 1981, "Dynamic
Behavior and Tension of Oscillating Mooring Chain", Offshore Technology Conference, Houston.
ACKNOWLEDGMENTS
Part of the research reported here was supported by the Sea Grant Program under project
number NA8lAAA - D - 00069.
APPENDIX 1: Data
To demonstrate the techniques described here, the following example has been used:
Water depth
Do 1,500 ft
Tower height 1,600 ft D
t
Distance from fairlead to seafloor
D 1,400 ft
Clump weight W 200 kips
c
Clump weight length
lc 150 ft
Length of unstretched cable from fairlead to clump weight c
L 3,300 ft
Total cable length (unstretched)B
L
t
4,600 ft
Initial tension at the cable upper end c
T 300 kips
Cable Young's modulus. E 4.3*10' lb/ft
2
Cable diameter d 3.5 inch
Tower weight 4*10' lbs
Tower buoyancy 3*10' lbs
Drag diameter c do K 55 ft
Inertia diameter d
I
19 ft
Drag coefficient CD c 0.7
Number of cables 20
APPENDIX 2: Cable Dynamic Solution
We will use equations (33) through (36) of the text to derive the linear dynamics of the
cable. As shown in [B], the overall dynamic response consists of a part which is slowly varying
with space (called in the sequel the slow solution) and a part which is wave-like and varies fast
. with respect to space (fast solution). In [B], the slow part was related to the catenary effects,
because the systems considered involved considerable sag.
In the present case, large tensions are involved so the elasticity effects are significant
and must be included in the slow solution.
First, we consider all dynamic quantities as varying sinusoidally in time, i.e.
iwt
u c ue (A.ll
iwt
v v e (A.2)
T
1
" T
l
e
iwt
(A.3)
iwt
(A. 4)
.1" .1 e
- iwt
e e e (A.5)
We ommit all nonlinear terms and obtain a homogeneous set of equations expressed in terms
of ~ , n instead of u, v
d ~ iwt
u .. at c i w ~ e
(A.6)
dJ\ iwt
v .. dt .. iWT} e
(A.7)
i.e. ~ , J \ are the tangential and normal displacements respectively. Then:
(A.8)
_w
2
m ,. -WI + (A.9)
o 0 1 ds
(A.IO)
(A.ll)
Let a Ids where for shallow sag cables, a will be a small quantitiy (order E) and K
similarly all chaRges with respect of s of the static quantities will be small.
Let us first derive a solution which is fast varying with space, i.e. n(s), (5). Then
must be of order E compared to n(s) so that to leading order from (A.B)
T
1
+ s [To J} (A.12)
By using (A.IO) and (A.ll) to first order, we obtain from (A.12)
}
(A.l3)
where M a m + a From equation (A.9) to first order
o o
.. 0 + OlE)
(A.14)
so that
2
- d [
Mw n + as To
0
(A.15)
By using the WKB method as in [B] we find
n- r
CI
cos[W(s)] + sin[W(s) h C
2
L J To/M (A.16)
s
W(s) = wI
ds
(A.l?)
By using (A.ll) to first order, we find
o
]
.f To/M
t [C
I
sin[W(s)] C
2
cos[W(s)] w a (A.lS)
Next, we derive a solution which is slowly varying with respect to 5, withn and t of the
same order. We note that
dE;
CIS'"
OfF.
(A.19)
so that from (A.8)
2
T
I
'" - -
I
Mw

n +
o(e 2 ) (A.20)
a
By combining (A.20) and (A.ll) we find to first order
-
T ..
-Mw
2
AE dr
I a
2
AE-Mw2 CIS (A.21)
By combining (A. 21) and (A.9) we find to first order
-
AE
d [ 1
de]
=
m ds a
2
AE-Mw
2
as (A.22)
557
For a shallow catenary such as the one considered here, sUbject to a surface current, the
value of 11 is constant to first order as easily seen from equations (5) and (6) of the text: If
T W.L (in order to have a shallow catenary)
(A.23)
where T is the tension at the top and the angle between the line connecting the cable
ends and the h8rizontal. Let
m
M (A.24)
so that (A.22) becomes
- 1
t
p2
(A.25)
or
(A.26)
and by using (A. 11) and (A.2l)
PS -ps a m
n (C
3
e - e ) p K C
4 (A.27)
In order to find the natural frequencies we add the slow and fast solutions, i.e.
Ps Ps (A. 28)
t(s) l1[C sin!w(s}] - C + C
3
e + C e
l 2 4
and apply the boundary conditions
t (0) t (L) 0
(A.30)
nCO) neLl 0
(A.3l)
The resulting equation is:
sinh(PL) sin (W
o
) [1 - ]+
+ [COSh (Pl.) cos (W ) -1
1
{!.- +
o ] P k
l
0
2
where
L w ds
W
o
.. jT(S)!M
(A.33)
w w
(A.34)
Note that if we consider a horizontal catenary, then xl k k. If the elasticity
/T(o)/M
2
then P+l1 and equation (A.32) becomes to first order in 11 :
xL { xL kL}
tan - = 0 (A.35)
which provides the symmetric and antisymmetric natural frequencies of a shallow sag chain
(9].
If we let a-+O then:
w
P "i
lE'T'P - ill
and equation (1..32) becomes:
sin(kL). Bin (IlL)- 0 (A.36)
which provides the natural frequencies (transverse and longitudinal) of a taut wire.
APPENDIX 3: Weakly Nonlinear Dynamics
Equations (48) and (50) of the text describe the dynamics of the guyed tower when the moor
ing system is modelled quasi-statically. Since the coefficients of the nonlinear terms are small
relative to 1 (order ) we can use the method of multiple scales to derive the response up to first
order (101. Essentially the technique is based on the remark that the nonlinearity changes the
amplitude of the response, but also the period (or equivalently the phase). As a result, both the
amplitude and the phase are medified by expanding the amplitude in a power series in c and by using
two time scales t and t.
In the case of (i.e. w ) we expect resonance, i.e. a single frequency response
which is excited by relatively small f8rcing amplitude (order E). To first order the method of
multiple scale predicts
(B.l)
where a and yare slowly varying functions of time. By using the next order equation de- .
rived from (48) we find a set of equations that a and y must satisfy, otherwise they produce
secular terms [101.
da F
.. - a - cosy (B.2)
\.L+w )
l
dY F
.. we (a) - + sin y (B.3)

w
l a (l+w )
l
where (16):
1
211
o (a) .. J (B.4)

o
2
w (a) = I - -!. fTr f (B.5)
e Tra 0
In the case of a steady state response a=O,y .. 0 so equations (B.2) and (B.3) provide
WI .. J :/ + 4 (0 (a)
By using equation (B.6) we can draw the curve relating the amplitude and the frequency of
excitation WI' Note that in order for (B.6) to provide real WI the radical must be positive so
the maximum amplitude, is the solution of the equation
i.e. strongly dependent on the damping coefficient C as expected.
When the frequency of excitation w is far from 1 the system is responding with a signifi
cant amplitude only if the excitation is large (order 1). The response consists of a term of
frequency WI and a term of frequency W =1 (natural frequency of the system) which may not decay
depending on the relation between wl aRd W (possibility of subharMOnics and superharmonics).
o
559
TABLE 1
First Natural Frequencies of the Cable in (rad/sec)
Symmetric Modes
1.18
loll
101 1. 59
2
101 2.27
3
Antil>ymmetric Modes
loll
. 0.89
101 1.80
2
101
3
K 2.70
Figure 1: Guyed Tower Figure 2: Forces on a cable element
8Be.S CABLE: STATICS
1
3
Cable to
anchor
Cable tG
tower 7Be.1I
Figure 3
TOWER orrSE:TCrT.)
Figure 4




~ fJI' CML.E8
.......
.._..
.._..
_._..
,.
r
..
:II'
...
,.
f .._ ..
:II'
.,
i-'-"
...
iaa._.. 1.1._..
;
i
...._..
;.a._..
I
i.
.._...
~ ._"1
._"1
..
._"1



.... N
p

~
!i !" ~ ~ !"
TONEJt 0F'nET(n ) Clf'F'8ET(" )
Figure 6
Figure 5
NON.INEM I'.-cHSE
n .
48
l?S
Dynamic
Unstable IS
t:

Static
-
.
,
.,
t:
.
11.58
I
- n . e* 0.014
..
1
a
g
7
n . xl
10 ft.
..
~
.. IS
I .
~
~

I.sa
I
-I."

;
&II .. ..
. . .
..
. . !" ~ !" !"

..
-
..
.. i
..
..
I
..

JlRIVlNlii ~ ( ~ C ) TIP(BEC)
Figure B
Figure 7
661
Approximation
Actual