Sie sind auf Seite 1von 7

Response of a Simple Tension Leg

P. D. Spanos
Professor,
Platform Model to Wave Forces
Departments of Civil Engineering
and Mechanical Engineering,
Rice University,
Calculated at Displaced Position
Houston, Tex. 77251
A simple single-degree-of-freedom model of a tension leg platform is used to assess
the reliability of the common practice of calculating wave-induced forces at the
undisplaced position of offshore structures. This assessment is conducted in con-
V. K. Agarwal junction with the Morison equation based modeling of the wave-induced forces on
Research Assistant, slender structural members. It is shown by numerically integrating the equation of
The University of Texas motion that the calculation of wave forces on the displaced position of the structure
Austin, Tex. 78712 introduces a steady offset component in the structural response. This is valid for
Assoc. Mem. ASME either deterministically or stochastically described wave fields. Several parameter
studies are conducted. Furthermore, reliable approximate analytical deterministic
and stochastic solution techniques are developed which conform to and, in fact,
predict the conclusions drawn from the results of the numerical studies.

Introduction
As offshore oil exploration moves into deeper waters, a new investigation of the effect on structural response of wave
class of structures, namely the "compliant" offshore forces calculated at the displaced position of the structure. In
structures, is likely to become increasingly important. order to isolate this effect from and to prevent it from being
Prominent new examples of this type are the Tension Leg obscured by other mathematical complexities, a simple single-
platforms (TLP), also called Tethered Buoyant platforms degree-of-freedom structural model is employed. As will be
(TBP), and Guyed Towers. The principal characteristic of seen, however, the equation of motion, even for this simple
these structures which distinguishes them from the more model, is beset with a nonlinearity which makes it
traditional fixed jacket type of structures, is that they unamenable to exact analytical solution. Therefore, it is
primarily ride with the waves rather than resist them. Hence, necessary to resort to numerical integration procedures in
the name compliant. A large number of studies, experimental order to get an acceptably accurate solution. Another ob-
as well as analytical, on the dynamic behavior of these jective of this study is to develop appropriate analytical
structures have appeared in the literature, see for example solution techniques. As far as the structural loading is con-
[1-9]. The majority of these studies indicate that compliant cerned, both deterministic and stochastic wave fields are
structures can have excursions with extreme values of the considered. Furthermore, in view of the fact that (TLP) are
order of several meters. likely to be installed in relatively deep waters, it is assumed
For many (TLP) the characteristic dimension of individual that "deep" water approximations [10] can be used in dealing
members (D) is small enough (D/L < 0.2) compared to wave with wave kinematics.
lengths (L) to justify use of the modified Morison equation
for the calculation of wave-induced forces. This approach is Problem Formulation
especially valid for severe sea states where the wave energy is
concentrated at lower frequencies corresponding to longer Equation of Motion. Shown in Fig. 1 is a schematic
waves. At present most of the dynamic analyses based on this representation of a four-column (TLP) subject to a
formulation are carried out assuming that the structural unidirectional wave field, deterministic or random,
displacement is negligibly small compared to wavelengths. propagating along its longitudinal axis. As is usual in the
This assumption appears quite reasonable for fixed struc- analysis of this class of structures, the hull is treated as a rigid
tures. However, it has been felt that it needs to be assessed for body. In order to study only the surge, (x) motion, all forces
compliant structures because of the relatively large excursions are considered as acting on the center of mass of the structure.
expected. Thus, the present study has, as its principal aim, the The structure can then be modeled as a single-degree-of-
freedom oscillator driven by wave loads. That is,
mx + cx + Kx = f(x, i) , (1)
Contributed by the Offshore Mechanics Committee of the Petroleum
Division and presented at the 3rd International Symposium on Offshore where m is structural mass, c is structural damping, if any, K
Mechanics and Arctic Engineering, ETCE, New Orleans, February 12-16, is the tether stiffness, and f(x, t) is the total wave-induced
1984, of THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS. Manuscript
received by the Petroleum Divison, August 19, 1983; revised manuscript force in the x direction. In this study, the wave force is
received December 7, 1984. calculated by using the modified Morison equation.

Journal of Energy Resources Technology DECEMBER 1984, Vol. 106/437

Copyright © 1984 by ASME


Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Specifically, the force / on the unit length of a member of acceleration a(x, t) in equation (5) is, in this case, a random
diameter D is obtained by the equation process. This can be related to another random process, the
ocean surface elevation ij, which is conveniently specified in
f(x,y,t) = p - ^ D2{Cm - 1)(« - x) + p - ^ Z>2u terms of its power spectrum. The most commonly used power
spectrum is the single-parameter Pierson-Moskowitz (P-M)
spectrum given by the equation
+ — pCd.D(w-x)l«-xl (2)
S„(«)=-Vexp(-—), (12)
In equation (2) p denotes the density of sea water, Cm and C d CO \ CO /
are respectively the inertia and drag coefficients, and u and x
represent the water particle and structural displacements, where
respectively. As usual, the dot over a variable represents a = 8.1 x 10~ 3 (Phillips constant)
differentiation with respect to time. In general, u, it and u are
functions of x, y and /. Throughout this study, the linear (12a)
(Airy) wave theory will be used for wave kinematics. Ac- B = 0.74(-^-)\
cording to this theory the y dependence of velocity is ex-
pressed by the relation In equations (12) and (12a) co denotes frequency in rad/s and
ky [/is wind speed in m/s. Once again relying upon the "deep"
u(x,y, t)=ii(x,0, t)e , (3) water approximation of linear wave kinematics, the power
where X = 2ir/L is the wave number. With ds denoting the spectrum for a(x, t) can be expressed in terms of Sm as
submerged depth of structure, the excitation f(x, t) in equation S„a(co) = g 2 ( l - e ^ ^ ) 2 S „ ( c J ) (13)
(1) can be written as
A zero-mean process specified in terms of its power spectrum
can formally be represented as [15]
Ax,t)=\ f{x,y,t)dy. (4)
J -ds
u(0 = V2\ cos(2iTcot+\b)\/S„Jco)dco (14)
The last term on the right-hand side of equation (2) is the Jo
velocity square dependent drag part of the wave force. For
usual (TLP) member sizes the contribution of this component In equation (14) S„„(co) is the one-sided power spectrum of the
is of the order of 10 percent or less of the total force [11], process v(t), and \p is a random variable uniformly distributed
Thus, it will be assumed herein that the drag effect can be in (0, 2 T ) . The foregoing pseudo-integral representation is
reasonably approximated by means of linear viscous damp- usually made meaningful by defining it to be the limit of a
ing. Alternatively, use could be made of readily available partial sum. This interpretation is frequently used [16, 17] in
linearization schemes [12-14]. Based on this simplification synthesizing a random process by superposing a large number
and equations (2)-(4), the equation of motion for the of harmonics with randomly varying phases. This approach
structure can finally be reduced to the following form leads to the following representation of the process a(x, f)
Mx+Cx + Kx = Ga{x, t). (5) N
a(x, t) = j2,Yjg(l-e-x'dsWSrirl(u)i)Aco'COs(kiX-coit+\l/i)
In equation (5) M and C incorporate both structural and /=i
hydrodynamic components,
(15)
In equation (15) N is selected such that
a(x, t) = f u(x, y, t)dy = " ^ — * "(*> °> ^ (6>
S „ ( u ) s 0 f o r a > > coN. (16)
and Using equation (15) the equation of motion for the stochastic
case can be written as
G=^-pirD2C„, (7)
N
Mx+Cx + Kx = V2gG £ ) (1 - e "x'rf^)VS„(co)Aco
;=i

Deterministic Excitation. Consider the structural motion cosQiiX-Wit+ypi). (17)


induced by a regular harmonic wave of height H, wave length It is seen that the structural equation of motion for
L, and period T. The qualities L and Tare related through the deterministic excitation, equation (10), as well as that for
following dispersion relation corresponding to "deep" waters stochastic excitation, equation (17), is nonlinear. To the best
2-7T CO2 27T of the authors' knowledge, no exact analytical solutions are
X-_ = _ ;«=-. (8) available for these equations. Thus, it has been deemed ap-
L g T propriate to conduct first purely numerical studies to gain a
The water particle acceleration at the still water level (SWL) is reasonable understanding of the structural response behavior,
given by the equation and its sensitivity to variations of relevant physical
co2H parameters.
u(x, 0 , 0 = —z- sin(Ax- ait). (9)
Numerical Studies
Substituting equation (9) into equation (5) yields
Numerical integration studies have been performed on
Mx+Cx+Kx=Fsm(cot-'Kx), (10) equations (10) and (17) using basic structural data similar to
where those of reference [7]. Specific values that have been used are
listed in Table 1. Some of the damping values like 0.2 and 0.3
F=-co2-G(^~-). (11) may, at first sight, appear rather high. But numerical studies
performed with full nonlinear drag term retained in the
Stochastic Excitation. Consider next the structural motion Morison equation indicate that these values are not
due to a still unidirectional but irregular wave field. The unrealistic.

438/Vol. 106, DECEMBER 1984 Transactions of the ASME

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Table 1 Parameter values used in the numerical studies
H = 20rt>i Analytic, o Numerical Integration
H = 10m, Analytic, A Numerical Integration
Structure mass: 40,000 tonnes
Added mass: 28,140 tonnes (based on Cm = 2.0)
Wave forces act on: 4, 16-m-dia columns
Submerged depth: 35 m
Total tether tension: 14,000 tonnes (force)
Tether lengths: 415 m, 830 m, 1245 m, 1660 m
Total tether stiffness: 331 kN/m, 165.6 kN/m, 110,33 kN/m,
82.75 kN/m
Structural undamped period: 90s, 127 s, 156 s, 180s
Damping ratio (total): 0.05, 0.10,0.20, 0.30 1= 415m
Wave heights: 10 m, 15 m, 20 m
Wavelength: 200 m, 400 m, 600 m, 800 m, 1000 m S8
Wind speed: 40 knots-70 knots

i' 1660m
I' 1245m
(Xx-wt) X • 830m
jl • 415m

S.W.L
0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55
HARMONIC WAVE FREQUENCY W, rod/sec
Fig. 3 Response offset component X 0 ; f = 0.2

H=20mi Analytic, o Numerical Integration


H= lOmi - Analytic, A Numerical Integration

Fig. 1 System schematic


1= 1245m
H = 20m, L = 200m
1= 330m
£= 0.3, JL = 830m

w
0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55
HARMONIC WAVE FREQUENCY Cu, rod /sec
Fig. 4 Amplitude ratio X 0 /X; f = 0.2

frequency component, to be denoted by XQ henceforth, and


the other is an oscillatory component at the wave frequency o>;
the amplitude of this will be denoted by X.
The variation of X0 and X0/X with important physical
parameters can be seen in Fig. 3-6. Figure 3 shows the
variation of X0 with w for two different wave heights for a
given damping value and for four different values of tether
length /. Figure 4 displays similar information of the am-
525.0 plitude ratio X0/X. In Figs. 5 and 6 is shown the variation of
the same parameters for a fixed tether length for different
values of damping ratio. A careful study of these figures and
several more similar studies leads to the following con-
clusions:

Fig. 2 Steady-state response; equation (10) (a) The magnitude of the steady offset increases almost
linearly with increasing damping as well as with frequency
Deterministic Excitation. Numerical integration studies on ratio co/co„; co„ denotes the undamped natural frequency of
equation (10) have been conducted for almost the entire range the (TLP). Note that an increase in tether length corresponds
of parameter values listed in Table 1. Figure 2 shows a typical to a decrease in co„.
time history of steady-state response as obtained from one of (b) The amplitude of the oscillatory component is close to
these studies. It can be seen that the structural response is what would be expected for a linear SDOF oscillator driven by
essentially oscillatory but with nonzero mean. That is, it a harmonic force F sin o>t. Some similarity is also observed in
consists of two components; one is a steady offset or zero its behavior in other respects. Specifically, it is relatively

Journal of Energy Resources Technology DECEMBER 1984, Vol. 106/439

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


C • 0.30 U = 70 Knots, £ = 0.30, £=4l5m
H = 20m, Analytic, o Numerical Integration
H « lOmt Analytic, A Numerical Integration

TIME AVERAGE
E £ = O.EO
« 0.74m

H
UJ
to 100.0 150.0 200.0 250.0
UJ TIME, sec
S
C» o.io
£ • 0.30

Fig.
1 7 Response mean - equation (17); ensemble size = 500
£ • O.IO
C • 0.05
U = 70 Knots, £ = 0.30, i=4l5m
0^25 030 0.35 0.40 0!45 0!50
HARMONIC WAVE FREQUENCY W, rad/sec
Fig. 5 Behavior of offset component X 0 ;/ = 830

H = 20m, ' Analytic, ° Numerical Integration


H = 10m, - Analytic, A Numerical Integration
£=0.30

TIME AVERAGE
" 5.6 m

0.0 50.0 I00.0 I50.0 200.0 250.0


TIME, sec

Fig. 8 Response standard deviation, equation (17); ensemble size


500

» £ • 0.05
4=-.r*="-—»
0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55
HARMONIC WAVE FREQUENCY CO, rad /sec
Fig. 6 Amplitude ratio X0IX; f = 830 m

insensitive to variations of damping ratio f, and frequency


ratio w/co„ and, as expected, decreases slowly with increasing
value of each.
(c) While the magnitude of oscillatory component X
increases linearly with wave height H, that of the offset
component, X0, increases quadratically in H. Thus, in some
extreme cases corresponding to steeper waves, high damping
and deep waters, the ratio Jf 0 /^ c a n approach or even exceed
1. This last fact serves to call attention to the need of assessing
the importance of calculating wave forces at the displaced
U= 7 0 Knots, C = 0 . 3 0 C = 4 l 5 n
position of compliant structures.
Stochastic Excitation. With regard to the case of
stochastically described waves, a variety of Monte Carlo
studies of equation (17) have been conducted. The value of N
has been taken equal to sixty-four (64), and a fixed increment
Ao = (co6 - oia)/N has been used, where co0 and wb are
respectively the lower and upper cut-off frequences in the P-M
1 — I
spectrum; they depend on the wind velocity. A typical record 0.00 0 25 0.50 0.75 I.DO 1.25 1.50 I.75
of the mean value and standard deviation of the structural FREQUENtiY, rad/sec

response is shown in Figs. 7 and 8. Figure 9 shows a pertinent W0 Wb

Fourier amplitude spectrum. All these figures are based on Fig. 9 Response amplitude spectrum

440/Vol. 106, DECEMBER 1984 Transactions of the ASME

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


ensemble averages of five hundred (500) records. Examining solutions for the response are available. Therefore, an ap-
Figs. 7 and 9, it is seen that a steady offset component is proximate analytic solution will be sought. To derive an
present in the structural response. In addition, Fig. 9 also approximate solution, an optimal time-invariant equivalent
shows that the magnitude of this component is considerably stiffness K* is used. That is, equation (22) is replaced by
larger than the magnitude of any other component. Note that
this component would not be seen in the response if the forces Mx+C5;+K*x = Fsmwt, (23)
were calculated at the undisplaced position of the structure. where the stiffness K* is selected according to the criterion
Another observation that can be made in Fig. 9 is that, aside
from the zero frequency component, most of the energy is e2 = (K*x -{K+Fk cos ut)x)2dt = min. (24)
contained in the frequency band bounded by o>b.
Results of some other parameter studies are shown in Figs.
10 and 11. From these figures it can be seen that the behavior Substituting equation (19) into equation (23) and integrating
of the response for a stochastic excitation is in some respects gives
similar to that for the deterministic case. For example, for a 2 A2

fixed peak wave frequency w0, the offset component X0 is


seen to increase with ratio w0/co„ as well as with the increasing
«-«b-HW^w] (25)

severity of sea-stated. Several similar studies not presented Then, considering the steady solutions of equation (23), the
here due to space limitations indicate that X0 increases almost following equations are derived
linearly with damping ratio, which is again similar to the FCw
behavior observed in the deterministic case. However, it is A = (26)
emphasized that an increase of the peak frequency OJ0 results (K*-o)2M)2+(Co,)2'
in an increase of the wave spectral width. Thus, the qualitative and
trends of the deterministic response should not be expected F(K* - oi2M)
unreservedly for the stochastic response as well. B (27)
(K*-u2M)2 + (Cu)2'
A comment about the results in Figs. 7-8 may be in order.
The mean and standard deviation obtained through a Monte
Carlo study are themselves random variables whose variances Considering simultaneously equations (25) through (27), it is
are inversely proportional to the ensemble size n. Specifically, recognized that they can be solved to determine the unknown
ffmean s 1/Vrt, while ffstddev s 1/V2«. This should explain K*, A, and B. After determining A, B and K*, the offset
why estimates for < x > are in general much rougher than component X0 is found by ensuring that the solution given by
those for ax for a given n. Pertinent discussions may be found equation (19) satisfies equation (22) on the average. This
in references such as [19]. condition leads to the following equation
Fk A
Approximate Analytical Solution Xn = (28)
1 ~Y
Based on the results of foregoing numerical studies, it The procedure based on equations (25)-(28) has been used
appears reasonable to conclude that the structural response to estimate the response amplitudes for all the cases studied
can be written approximately as the sum of a steady or time earlier using numerical integration. The results obtained
invariant component plus a time-dependent component. That through the foregoing solution procedure are shown alongside
the results of numerical studies in Figs. 3-6. It can be seen that
x(t)=X0+x(f) (18) there is a good agreement between solutions obtained through
Deterministic Excitation. For the case of a deterministic numerical integration and the analytic estimates in the entire
wave of frequency co, the component x(f) is expressed in the range of cases considered. The numerical value as well as
following form trends are in a close agreement with each other. In this regard,
attention is also drawn to the fact that equations (26)-(28)
x(t)=A cos wt + Bsmut, (19) conform to, and in fact can predict, the conclusions drawn
where A and B are constants to be determined. from the results of the numerical studies.
In most situations pertaining to offshore structures, the
structural excursion, though not negligible, is expected to be Stochastic Excitation. For the case of stochastically
quite small compared to the wavelength. This is also borne described waves, it is more convenient to start from equation
out by the results of the numerical studies. Therefore, it could (5). Expanding a(x, t) in a Taylor series about x = 0 and
be argued that second and higher order terms in the quantity retaining only the first two terms, equation (5) can be written
Xx = 2-KX/L may be neglected. Here, to first order, one may as
invoke the approximations
MX+CX+\K-G— I ]x=Ga(0,t). (29)
sin Xx = Xx (20) L ax U=o J
and Define
cos Xx s 1. (21) G a(0, t) -Fi<f) (30)
Using these approximations in equation (10) gives (31)
G
^ x=0
Mx + Cx + (K+Fk cos wt)x = F sin oit (22)
Then, equation (29) can dx
be written more compactly as
Equation (22) is linear with periodic coefficients. There are
important questions related to stability and parametric Mx + Cx + {K-F2(t))x=Fi(t). (32)
resonance for homogeneous and inhomogeneous systems of Assume that
this type. A large body of literature devoted to the study of
these problems already exists, for example [18], and the x(t)=X0+x(t), (33)
regions of parametric resonance are well known. Thus, it is \\7n £>t*ei
Wild C
reasonable to assume that any well-designed (TLP) would be
designed to operate away from the zones of parametric £(40) =x0 Eim)=0; (34)
resonance. However, it appears that no general closed form the symbol E[»] denotes mathematical expectation. The aim

Journal of Energy Resources Technology DECEMBER 1984, Vol. 106/441

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


now is to obtain response statistics; the mean X0 and the
h(.T)E{Fx(t-T)F2(t)dr, (47)
standard deviation ax = ax. For this purpose the time-varying
stiffness in equation (32) is replaced by an "equivalent" time-
invariant stiffness. That is, the quantity and
f{x) = [K-F2(.0\x~[K-F2m&o +x) (35)
is replaced by K*x. Accordingly, equation (32) is replaced by
the equation E\x2F2) = £ [ J " / I ( T ) F , ( ? - T)rfrjo h{T')Fx{t- T')C?T'F 2 (O]
Mx + Cx + K*x=Fi(t). (36)
The minimization criterion for selection of K* is similar to = \X\ "dTdT'h(T)h{T')E[Fl (t - T)F, (t - r')F2(t) ].
equation (24), except that in this case ensemble averaging is Jo Jo
used in place of temporal averaging. This approach yields (48)
E{xf(x)} Equations (43) and (44) can now be substituted into equations
K* (37)
E{x2 (47) and (48) and the required expectations can be evaluated
The procedure of evaluation of various expectations in- by integrating. After some tedious manipulations and making
volved in equation (37) is as follows. Equation (36) yields for use of the fact that for a uniformly distributed phase \p
the spectra Sx and SF the following relationship £{sin(« + \W) =£{cos(' + iW)=0, (49)
Si(«)= \H(w)\2SFl(u), (38) one obtains
2
where \H(u) I is the transmittancy function of the linear E{xF2} =
system described by equation (36) and is given by the equation
2
2G fco„ f»6 OJ3S„0(CO)
2
doi (50)
m 2 (39) M Jo [((»d - o)) + (fw„ ) 2 ][(«d+«) 2 +(f U J I ) 2 ]
^ =^-^HCf
The acceleration process a(x, t) and, therefore, both Fx and F2
are zero-mean processes. Thus, the response x in equation (36)
is also a zero-mean process with variance and
E{x2F2}=0
o2x=E[x2 S Oo

0
Sy((J)d(j). (40)
Thus, with all the expectations needed in equation (37),
given by equations (40), (42), (50) and (51), an iterative
(51)

The requirement that equation (33) satisfy equation (32) on scheme based on equations (37), (38) and (41) can be used to
the average yields determine X0 and ax.
E[xF2) The foregoing procedure has been used to estimate response
Xn (41) statistics for several cases, and some of these results are
K shown in Figs. 10 and 11 along with the results of numerical
From equation (35), studies. Comparison of the two sets of results shows them to
E[xf(x)} = E{x[(K-F2)(X0+x)]} be reasonably close. Even though the approximate analytic
procedure seems to somewhat underpredict response
= KE{x2}-X0E{xF2)-E{ x2F2) (42) statistics, in almost all cases the analytical solutions are within
2
The two expectations E{xFx) and E{x F2] are obtained by 10 percent of the digital simulations. More importantly, they
starting with the harmonic superposition representation for do seem to exhibit the right trends.
the random processes F^t) and F2{t). Use of equation (15) in
equations (30) and (31) yields
F , ( 0 = Ga(0, t) = yf2GEj[Sa(u}j)Au]W2 costy - ujt) (43) Concluding Remarks

da It has been shown that the calculation of wave forces on the


F2{t) m G displaced position of a structure is necessary to discern the
dx presence of a steady offset component in the response which
= -T/2GEj[Sa(uj)Aw]l/2\j sinWj-ajt). (44) would otherwise remain undetected. It is worth noting that
this effect is entirely different from the usual drift forces
In the time domain the solution of equation (36) can be associated with the nonlinear fluid dynamic effects. This is
written as evident from the fact that the present formulation does not
involve any velocity squared type of terms, and yet an offset
component is found to be present.
x(t)=\"h(T)Fl(t-T)dr (45) The significance of the offset component in determining the
total structural excursion is a function of the sea-state as well
where the impulse response function h(t) is given by the as of structural parameters. Some qualitative conclusions in
equation this regard can be drawn from the results of deterministic
studies. Thus, it is seen that, for extreme waves as well as for
h{t)= —— e-K'sirKovO. (46) extreme water depths, the offset component can constitute
nearly half or more of the total response. It must be noted,
and co„ = \f(K/M), f = C/2Mu„, and cod = u„ V(l - f2) are, however, that the present study does not include the drift
respectively, the natural frequency, the damping ratio and the effects of nonlinear drag and, in cases where these are
damped natural frequency of the structure. Using equation significant, they must be evaluated separately. In estimating
(46) one can write structural response due to random wave fields, the inclusion
of the offset component should help refine the prediction of
maximum possible excursions under extreme sea-states and
E{xF2] = E^'hlTWdt-ddT > F 2 (o] the confidence intervals associated with these. The effect is,

442/Vol. 106, DECEMBER 1984 Transactions of the ASME

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Analytic Approximation
0
Analytic Approximation Monte-Carlo Simulation
Monte-Carlo Simulation C • 0.30
0.30

z
o

t
X O 9
<
§ CVJ
o

> 4l5m

0.20 0.25 0.30 0.35 0.40 0.45


RANDOM WAVE PEAK FREQUENCY WQ, r a d / s t c
0.20 0.25 0.30 0.35 0.40 0.45 0.50 Fig. 11 Stationary response standard deviation versus wave spectrum
RANDOM WAVE PEAK FREQUENCY w 0 , r a d / s e c peak frequency

Fig. 10 Stationary response mane versus wave spectrum peak


frequency

however, not expected to be very important for long-term Loads and Structural Response of a Tension Leg Platform," ASME, AMD -
Vol. 37, Computational Methods for Ofjshore Structures, ASME, 1980.
fatigue studies. 7 Gie, T. S., and de Boom, W. C , "The Wave Induced Motions of a
It is hoped that the approximate solution procedures Tension Leg Platform in Deep Water," Proceedings of OTC, Vol. 3, 1981, pp.
presented herein can be used to advantage in performing 89-98.
engineering studies. Clearly, these should enable one to 8 Lundgren, J., and Berg, A., "Wave Induced Motion of a Four-Column
quickly assess the importance of calculating wave forces at the Semi-submersible Obtained from Model Tests," Proceedings of OTC, 1982.
9 Jefferys.E. R., andPatel, M. H., "Dynamic Analysis Models of Tension
displaced position of the structure. In this regard, it is worth Leg Platform," ASME JOURNAL OF ENERGY RESOURCES TECHNOLOGY, Vol.
noting that the computation time required for a Monte-Carlo 104, Sept. 1982, pp. 217-223.
simulation using five hundred (500) records is over three 10 Sarpkaya, T., and Isaacson, M., Mechanics of Wave Forces on Offshore
orders of magnitude greater than that required for the ap- Structures, Van-Nostrand Reinhold Co., New York, 1981.
11 Bea, R. G., and Lai, N. W., "Hydrodynamic Loading on Offshore
proximate analytic solution. Platform," Proceedings of OTC, Vol. 1,1978, pp. 155-168.
12 Dao, B. V., and Penzien, J., "Comparison of Treatments of Non-Linear
Drag Forces Acting on Fixed Offshore Platforms," Applied Ocean Research,
Vol. 4, No. 2, 1982, pp. 66-72.
13 Spanos, P-T. D., and Chen, T. W., "Response of Dynamic System to
Flow-Induced Load," International Journal of Non-Linear Mechanics, Vol.
15, 1980, pp.115-126.
References 14 Spanos, P.-T. D., and Chen, T. W., "Random Response to Flow-
1 Angelides, D. C , Chen, C-Y., and Will, S. A., "Dynamic Response of Induced Forces," ASCE Journal of Engineering Mechanics Division, Vol. 107,
Tension Leg Platform," Proceedings of Boss, Vol. 2, 1982, pp. 100-122. EM6, Dec. 1981, pp. 1173-1190.
2 Kirk, C. L., and Etok, E. U., "Dynamic Response of Tethered 15 Kinsman, B., Wind Waves, Prentice Hall, Englewood Cliffs, N. J., 1965.
Production Platform in a Random Sea-State," Proceedings of BOSS, 1979, pp. 16 Borgman, L. E., "Ocean Wave Simulation for Engineering Design,"
139-164. ASCE, Journal of the Waterways & Harbors Division, Nov. 1969, pp. 557-583.
3 Isaacson, M., "Response of Compliant Structures to Steep Waves," 17 Shinozuka, M., Fang, S-L. S., and Nishitani, A., "Time-Domain
Proceedings of OTC, Vol. 1,1982, pp. 281-289. Structural Response Simulation in a Short-Crested Sea," ASME JOURNAL OF
ENERGY RESOURCES TECHNOLOGY, Vol. 101, Dec. 1979, pp. 270-275.
4 Yoneya, T., and Yoshida, K., "The Dynamics of Tension Leg Platforms
18 Ibrahim, R. A., "Parametric Vibrations," in 6 parts, I & II with Barr, A.
in Waves," ASME JOURNAL OF ENERGY RESOURCES TECHNOLOGY, Vol. 104,
D. S., Pt. V with Roberts, J. D., The Shock and Vibration Digest, Jan.
Mar. 1982.
1978-May 1978 and Sept. 1981.
5 Albrecht, H. G., et. al., "Non-Linear Dynamic Analysis of Tension Leg
19 Spanos, P-T. D., "Monte Carlo Simulations of Responses of Non-
Platforms for Medium and Greater Depths," Proceedings OTC, Vol. 1, 1978,
Symmetric Dynamic Systems to Random Excitation," Journal of Computers
pp.7-15. and Structures, Vol. 13, 1981, pp. 371-376.
6 Liu, D., et al., "Integrated Computational Procedure for Hydrodynamic

Journal of Energy Resources Technology DECEMBER 1984, Vol. 106/443

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Das könnte Ihnen auch gefallen