Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
By
Copyrighl
Offshore Technology Conference on behalf of _the American Institute of Mining, Metallurgical, and
Petroleum Engineers, Inc. (Society of Mining Engineers, The Society and Society of
Petroleum Engineers), Association of Petroleum Geologists, American Institute of Chemical
Engineers, American Society of Civil Engineers, American Society Engineers, Institute
of Electrical and Electronics Engineers, Marine Technology Society, Society of Exploration
Geophysicists, and Society of Naval Architects and Marine Engineers.
This paper was prepared for presentation at the Sixth Annual Offshore Technology Conference
to be held in Houston, Tex., May 6-8, 1974. Permission to copy is restricted to an abstract of
not more than 300 words. Illustrations may not be copied. Such use of an abstract should contain
conspicuous acknowledgment of where and by whom the paper is presented.
motions, and of the wave forces on where symbols [] and {} indicate matrix and
the structure are determined by use of the vector, respectively. [M s ] represents the
equation (3). The effect of having a lumped structural masses and rotational inertias
body accelerating in water, is included by the about the vertical z-axis; [Cs] and [K]
so-called "added hydrodynamic mass" (or virtual represent the structural damping and stiffness
mass), whose coefficient in this investigation coefficients, respectively; {O(tl}, {(Htl}
is assumed to be CM -I , where CM is the co- and {u(t)} represent the structural acceler-
efficient of inertia. ations, velocities and displacements (rota-
tions), respectively; and {Pt(tl} represents
Statistical quantities of structural re- the wave forces on the structure.
sponse in the orthogonal horizontal directions
and rotation about a vertical axis are obtained. Arranging the displacement vector as fol-
Previously an equivalent theory to lows
statistics of structural response in one hori-
zontal direction using the one-dimensional
wave height spectrum has been developed by
Foster (4) and and Penzien (5-6). A (4)
summary of this theory and numerical results are
given in (7).
This paper is based on the Ph.D where {Ux (tl) and {U y (tl} are the displace-
tion of the primary author (8), a former student ment vectors in the x and y directions, re-
of the co-author. spectively; and {Ue(t* is the rotational
vector; the matrices in Eq. (3) can be written
DYNAMIC RESPONSE TO RANDOM WAVE FORCES in the following partitioned form
Equations of Motion NP
Pt x' (t) =I Px. (7)
I k =I Ik
The dynamic equations for this
model can be written in matrix form as NP
r> (tl =" p ( 8)
"ty'1 Lk=1 y'k
I
NP
Pte, (tl = L [- Y'k Px (tl + x.Ik PYik (t)] (9)
I k=1 I ik
OTC 2050 _BENT BERGE AND JOSEPH PENZIEN 175
where PX"k (t) and PY'k (t) represent the where (Tvr is the standard deviation of the
wave at node ik id the x and y direc- relative velocity. Substituting Eq.(14) into
tions, respectively; and NP is the number of Eq.(13), the linearized drag factor becomes
nodes where the wave forces are applied at level
i.
dI = .j (Tvr (15)
The wave forces are described by the
toorison equatiop (3), modified to the following
form for oscillating cylinders. Thus, only the standard deviations of the rela-
tive velocities between water particles and
structure need be determined in order to linear-
p(t) =cMP b ii(tl - (C M-I) P b u (tl ize the drag forces. This is an iterative pro-
I cedure because the relative velocity deviations
I
+2 C oP a v(tl-u(tll{idtl-u(t) (10) calculated in one step are based on the linear-
ized drag forces calculated in the previous
where CM and Co are coefficients of inertia step. Fortunately, acceptable convergence is
and drag, respectively; P is density of fluid; obtained after very few iteration
b is volume of member; a is projected area
of member perpendicular to_motion of fluid; The wave forces, Eq.(lO) , can now be ex-
ii(tl and v(tl are fluid acceleration and
pressed in the following linearized form
velocity, respectively; and U(tl and u(tl
are structural acceleration and velocity, re- p(tl =CMP bii(tl - (CM -llpbU{tl
spectively.
I
Calling the relative velocity between wa-
+ 2 CoP a dj(V(fl-i.J(fl) (16)
ter particles and structure Yr(tl ,given by
Solution of the Linearized Equations of Motion
"r(t) = v{t) -u(t) (11)
Eq.(16) shows that the linearized forcing
the drag force in Eq.(lO) can bewritten_ functions depend on both water particle and
structure motions. Including the terms that
PD(t) Ivr(tllvr(tl (12) depend on structural motion in the mass and
damping matrices, Eq. (3) can be rewritten as
Thus, the drag force is proportional to the _
relative velocity squared with the exception of
the absolute sign, which preserves thedirec-
[M]{ii(f)}+ [c]
{U(fl} + [K]
{u(tl} = {P(fl} (17)
tion of the force. Practical solution of this where [M] represents the sums of structural
dynamic problem requires a linear set of differ- and hydrodynamic (added) masses; [C] represents
ential equations. The drag force ;is_therefore the sums of structural and hydrodynamic damping
linearized by the method of equivalent linear- coefficients; and is the forcing vector
ization (9). The mean square error introduced depending only on the water particle motion.
by replacing the nonlinear term Iilr(t) Iilr(tl
with dI ilr (t) is minimized (10). This gives The added mass matrix [MaJ and the hydro-
the following expression for the linearized dynamic damping matrix [Ch] may be partitioned
drag factor as the matrix [A] in Eq.(5). (The coupled
hydrodynamic damping coefficients for transla-
dJ =
[lv 2]
E r IV r
(13)
tion in x-direction and rotation are neglected
such that vibrations in the x-direction are not
2
E [ vr ] coupled with vibrations in the y-direction and
rotation. These damping coefficients are zero
where E [J denotes the average over the en- in the initial calculation of the structural
semble. response, and they will be very small even when
the drag forces are linearized.)
Since the input process (water particle
motion) is assumed to be a zero mean ergodic Elements ii of the diagonal submatrices are
Gaussian process ,the linearized output process
is also a zero mean ergodic Gaussian process. (1) the added mass matrix
Thus, the probability density function for the
relative velocity is NP
m axx " = m ayy = L(CM-Ilpb ik (18)
2 II Ii k= I
exp(-ilr ) (14) NP 2 2
2 (T'v 2 maaeIi =L{C -llpb'lk (Y'k +x' k ) (19)
r
k. I M I I
THRW-DTXENS1ONALSTOCHASTIC!RESFONSE
76 OF OFFSHO.RE.
TOWERS TO WAVE FORCES OTC 2050
c~yyii
= :,Byik ..... .. . ..... ...
. (22)
Modes in the x-directionare not coupled
with the coupled modes in the y-directionand
rotation. Thus, the modal matrix may be parti-
tioned as follows
CMX3..
,,$l[~xil(y;; 1 . . . . . . . . . . . ..(23)
~Yikxil(2
@------
[1 lo
_-:x
@, . . . . . . . . . . . ,.O...O . . . . .J(?4)
chy6,. = ch6yii
11
= $l@yik
, ik . . . . . . . . . . . ..(24)
[1 o ~@ye
[1
L
where
where [@xx1iS~Nx N matrix and [@ye]
is a 2N x 2N matrix; and where N is the number
.. .. 0 .(25)
@,ik=~CDp axikdfxik .0 . . 0 ***.
of levels of the dynamic model..
y. +) k~[aik (t)].............(28)
Yik(t)+pyikvyik . generalized stiffness matrix........... (38)
1
P8 t)=k~l
NP
iw
that is linearly relsked to the response in nor.
mal coordinatesas given by
3N
s
v,.
lk
~x.
I
=Lxxi,
r=l
HX)J)
zi(t) (t).
=~1Biryr ,,. ,.., ..*.. .. *
(55)
:fjr :[~irsvxikxj,+~
xj,sxikixjfl
*(61)
where the are known coefficients. Thus
Bi~s
Variances of relative velocities are ob-
the cross spectral density finction far Z1 tained by integratingcorrespondingspectral
and
i is density functions over the frequency range and
I the linearized drag factors are obtained using
s z#)=f~B.B
js S Y rY J w )
. . . . . . . . . . . . ..(56) Eq.(is).
ii r:l~=l r
STATISTICALEXPRESSIONSFOR WATER PARTICLE
and the variance of Zi is MOTIONS
co
~z .~= ff Bir
Bis~syryJw)
r :l s .1
dw
...............(57)
-a
The random ocean waves are consideredto
be a zero mean stationaryand ergodic Gaussian
process describedby the directional spectrum.
Depending on the coefficients Bir(r=l,2,...,3N), Figure 2 shows this spectrum,which specifies
u~ . m a y be the variance of displacement,
the distributionof wave energy with respect to
shdar force, twisting moment or bending moment frequency and direction of the waves. The di-
rectional spectral density, S7T(W,0) has the
Of great interest is the extreme value of property that the volume SqqAwA6 is the mean
the response. For narrow_bandresponse, the square amplitude of the waves with frequencies
mean value of the extreme is given by the rela- and directionswithin the AwAO rec%angle.
tion (11, 12) The volume enclosed by the directional spectrum
and the we-plane is equal to the mean square
E[zi(f)lm.x]=.z}- +~-
wave amplitude.
) **0*(58)
In this investigation,the directional
where T is the duration of the spectrum and the spectrum has been expressed as a product of a
frequency v is given .ly function of frequency and a function of direc-
tion, i.e.
=~e ***co*e9*0**:(5
ST7(W,8)= DWU3)ST7(W)
where D@(0)is the directional distributionof
............... (62)
.
OTC 2050 BENT BERGE AND JOSEPH PENZIEN 179
.
27T
CASE STUDIES
Stiitj (w) = tIJ2cl (9 ) Sq; (%e)
J
o Seven deep water towers are studied in
this investigation. Six of the towers are
cosh[K(- zi+d)]cosh[/c(-zj +d)]
symmetric a vertical plane, and have
about
heights of 475 (Tower 1, 5, and 6), 675 (Tower
s in h 2
[1
Kd 2), 875 (Tower 3) and 1075 feet (Tower 4),
correspondingto.water depths of 400, 600, 800
-iK (x.-x.)Cose+ (y,-Y,) sin8]}d8.. (64)
exp {[ and 1000 feet, respectively. The torsional
t J I stiffness of Tower 5 is 75$ of chatof Tower 1
(rotationalinertias and stiffnessesare scaled
where accordingly). Tower 7 is symmetric about two
vertical planes and has a height of 475 feet,
correspondingto a water depth ,of400 feet.
cos2e x-direction
cl(e) = s in 8 coSf3 x and Y directions The mass of each tower is lumped at 7
y.dj-re ~t ion
sin 28 levels. (Figure1 shows a model with 5 lev-
{ els). Numerical values are assigned to lumped
masses, rotational inertias, flexibilitycoef-
8 is the angle with the x-axis,x is the radian ficients, coordinatesof the nodal points at
wave numbersxi, yi and Ziare the coordinatesof each level, and member volumes and projected
point ~ ,d is the depth of water and areas associatedwith the submerged nodal
points. Unfortunately,due to space limita-
%#%e)= sqq*(-@)=+ SqT(@) tions structuraldata are not included. Struc-
w ~o ....(65)
tural damping ratios of 5%are assumed for all
towers vibrating in water. Six normal modes
is the directionalspectrum def~ned for both are included in the analysis. The coefficients
positive and negative frequencies. of inertia CM and drag CD are assigned values
of 2.0 and 1.4, respectively.
Using the factorizationgivenby Eq.(62),
Eq.(64) may be integrated over the angular range The ocean waves are described by several
by expandingthe terms that depend on the angle directional spectra, given by one-dimensional
9 in Fourier series to yield the following spectral densities and directionaldistribu-
result tions at Xl_discrete frequencies,corresponding
to the factorizationin Eq.(62). For most
s~~(~)z~ U2S+U)
examples, spectral densities are calculated
lj
by the Pierson-Moskowitz(13) formula, namely
cosh[~(-zi+d)]
cosh[K(-zj+d)]
0.0081g2
S77$CIJ) =
~5 ldQ+Jl ~~o
[1
sinh2
Kd
. ..(71)
where g is the accelerationof gravity and w
~Cn exP~n(~ij+&)]Jn(KAij) . . . . . . . . ..(66) is the mean wind speed at a height of 64 feet
n=-CO
above the sea surface. This equation is used
where to calculate spectral densities for wave spec-
27T
traA (50 ft/see), B (75 ft/see) andC (100 ft/
cn=L Spectral densities for wave spectrum D
27T J CI(8) D@(L?) exp(in8)d8
. . . . ..(67) eC)
are selected to investigatethe effect of two
o storms coming from different directions. The
-1 Yi-Yj directionaldistributionsare described by the
a,.= tan . . . . . . . . . . . . . . . . . . . . . . ..(68) circular normal (14), namely
II x.-x. I
H 1 J DW(9) = exp a cos(8-a) ... . . ..(72)
2~IO(a) [ 1
Aij = (xi- xj)2+(yi- y~2- ...*.. . . ..0.(69) where lo(a)
is a modified Bessel function of
zero order and a is a measure of the concen-
and Jn() is a Bessel function of first kind tration about the mean angle u . The higher
and order n. Expressions for water particle the value of a the more concentratedis the
velocity-accelerationand accelerationsare distributionabout the mean angle. A circular
readily obtained, since normal with a=6 is used for the directional
distr~buttonof the wave energy spectral
. . . . . . . . . . . . ..(70) density in Figure 2.
Svv =i~svi=-iw S ~u =W2 S t t i
THREE-DIMENSIOii.KL
STOCHASTIC RESFONSE
180 OF OFFSHORE TOWERS TO WA~ FORCES nrr!
----- 70<0
,-
Directional distributionA and D use a=6 Thus, the largest rotational response 5s ob-
the two lowest frequencieswhere wave spectral tained when the mean direction of both storms
densities are given, a=5.6 for the two next is perpendicularto the plane of symmetry. Com-
frequencies,etc., and a =4 for the highest paring Curve G (Tower 5, directionaldistri-
frequency (frequencyNo. 11). Directional_ bution A) with Curve A> the effect of 25$
distributionB and E use a=10 and distribu- reduction in the torsional stiffness is observed
tion C uses a=100. Curve H (Tower 6, directionaldistribution)
shows the response when the leg spacings are
Calculationof the response of a 7 leyel doubled. The increasedtorsional stiffness
tower (21 degrees of freedom), symmetric about causes smaller rotationalresponse than for
a vertical plane, requires approximately1 Tower 1, Curve A, except when the angle between
minute central processor time on a CDC 64OO the mea,ndirection of the wave advance and the
computer when 11 discrete frequenciesare used plane of symmetry is between O and 45 degrees.
in the numerical integrationof the spectral This is due to increased effect of the leg
density functions. spacings on the rotationalexitationwhen the
waves are nearly parallel to the plafleof sym-
Results metry. Lines I and J (Tower 1 and 7, respec-
tively, directionaldistributionD) represent
Figure ~ shows the lowest mode shapes..for rotational responseswhen the angle between
Tower 1 and6. (The lowest modes for the-other the mean direction of wave advance and the plane
towers have similar shape, except tower 7 where of symmetry varies between 45 and 90 degrees foY
translationaland rotational free vibrations the waves of different frequencies.
are not coupled.) Each coupled mode shape for
vibrations in y-directionand rotation is Deck rotations versus the mean direction
drawn as two separate curves; one for the com- of wave advance are shown for Towers 1 through
ponent in the y-directionand.the other fQy the 4 in Fibmre 5. The curves show that the rota-
comp.onentin .zota&iQ.n%TQe_K0t=a230na$<QU&Qn tional response for towers of various heights
nents are 40 and 80 times their real magnitude depend similarly on the mean direction of wave
relative to the translationalcomponents for advance. Deck displacementsfor these towers
Tower 1 and 6, respective.QJ.The natural_fre- a~e plotted in Figure 6 versus nean direction
quencies for the 6 modes that are included in of wave advance.
the dynamic analysis, are given in Table ~~__
Figure 7 shows d.isjjlacementsin the y-
Figure 4 sho~isstandard deviations of deck direction ati rotations plotted for three
rotation versus mean direction of wave advance different storm intensities. It is interesting
for Tower 1, 5, 6 and 7. One-d;imensional wave to notice the hi~h ratio of rotation to transla.
spectrum B is used for all curves except curve tion for 50 ft/sec wind speed walative to the
F, where wave spectrum D is used. Curves A, same ~atio for ~~ and 100 ftlsecwind speeds.
B, and C (Tower 1, directionaldistributionA, The reason for this is that the phase diffl=enc(
B and C, respectively)show that the largest of the wave forces over the horizontal extensio
rotational response is obtained when the mean of.fie structurehas decreasing effect on trans
direction of wave advance is nearly perpengi- lation while it often miEht have an increasing
cular to the plane of symmetry,and that the. ef~ect on rotation. This phase difference de-
response incrsases as the dirett~onal.di.strib- creases with increasiri~ wave Ien@hs, and the
ution becomes more narrow. When the mean wave lengths increases17iththe severity of the.
direction of flow Ls pa,rallelto the plane of storm.
symmetry the rotational response increases
with increasing directional spread, which is Ii@re 8 shows the standard deviations of
expected since.wavgsparallel to the plane of leg displacementsin the y-direction for Tower
symmetry do.not exicatethe structure in tcJr- 1. This figure gives an indication of the
sion. Curves A and C are not syrmmetric.almut effect o:fthe rotational re sponse on theQVe~all
a perpendicularcothe plane of symmetry be- structural response. If rotacionis not in-
cause inertia and drag forces are not i.nphase. eluded in the analysis Legs L and 3 would have
Curve D and E (Tower 7, cltrectional distribu- te same displacementsas Legs 2 and 4.
tion A and B) show that the rotational re-
sponse of a tower symmetric.abouttwo vertical The data discusse~ so far have been stan-
planes is very small. -Curve-F (lower1> wave dard deviations of response. While these are
spectrum D, directionaldistributionE) shows important quantities,theclesigneris more
the e?fect of-two storr.wcoming from two dis- interested in the naximunrresponse that might
ferent directions. T%e rfiean direction of the Oc c t t i durinE a storm, Fig~~re~ sho~~sh~ the
storm with the low frequencywaves is perpe~- mean peak deck displacer!ents and iotations
dicular to the plane of symmetry,while the varies with duration of the storm for ~ di.f-
response is shown versus the m.eg.n.
d_i_rect_i_o&.of?e~entwind velocities.
the storm containingthe high frequency waves.
Xrc 2050 BENT B-E _AIIO
JOSEPH_PENZIEN 181
Twisting moment, shear force and bending NOMENCLATURE
moment distributionsare shown in Figures 10
through 12, respectively. Standard levitations aX 1ayik projected areas of lumped
and mean peak values for.o.?~->s.>_a_E@
10 &.&?-U!!.. ik members
storms of these quantitiesare included.
b volumes of lumped members
CONCLUSIONS ik
d linearizeddrag factors
Based on this investigation,the fol.lowine IxikdJyik -
conclusionsand recommendationshave been de- h(t) unit impulse response function
duced:
i subscript:level i
(1) A stochasticanalysis should be car-
ried out to determine the overall ik subscript: node k at level i
dynamic response o; offshore towers.
r,s subscripts:mode r or s
(2) Both inertia and drag forces should
be included in the analysis. time
t
(~) Rotational response has small effect
on the total response; thus for many Ux ,U displacementsat level i
Yi -
structures a two-dimensionalanalysis i
[1
K stiffness matrix
THRZZ-DIMEWSIOIML STOCidASTICRESFONSZ
182 OF CEFSHORTOWERS TO WAVE FORCES OTC2050
.
- 2er,ey.ali.Zed stiffness matrix
[1
K %
M mass matrix (includes added The authors wish to express their sincere
[1 thariksand appreciationto the ztandard Oil
hydrodynamicmass)
Company of California for their financial sup-
. genera~ize~ mass matrix po?t, which made this investigationpossfble.
[1
M*
1 2 1 2 3 4
1 2 1 2 1 2
I 77)77
i
z
Fig.1 ~.Idealized
tower.
+fLl
e
\
\
\ , /
/ \
\
\.
\.
\\
\
\
\
\ I /
/
/
s
1/
1.,
w
\ \ \ \./
;
/ \
\ \ , M oD4 S I N
y- DI RECT I ON
\ 4 N D /?b7A TIOM
\
COU PL60
\
\ /?~E
,/\
-\ ~\ ?IODE I
\ MODE 2
\ . /VODE3
\ i, --- IV60EJ+
i, ~,i
\ \ /~
/y -
p
DlREc 7 /o
//
/f/ Y - DIf.Fc7/oN
J_. R0 7 4 7 1 0 W
4
?
o I I I I I J
o 30 60 90 120 [s0 /80
4N~~..S
6E7 WEEN MEAN DI.?EC TION OF WA VE
ADvANCE AAO X -AX I S, DE6 h3 EES
Fig.4- Standard
deviation
of rotation
of deckviidirection
of
waveadvance.
LV .4 V E SFECTl?Utl B
W4VE sP.2c7,eut7 &
DI RECT I ON AL D/S TK/6UT/OM A
2.2
2 .0
f. I f
L 6
/
/
1.4 Ll /
1.2 0.6
[0
0.8 0.4
/
/
0.6 /
.-
0.4 C?z
0.2 _/
0 o I I I [ I I 1
o Is 30 45 ho 75 90 0 /5 30 4s 60 75 90
~/V~~~&E TWELN P?E/+/V D/ RECTIOW OF ~~c~~ BET WEE d M EA N D/@6 C T /O/v OF W..? VC
i
A,oO .4,$f5 4 ,9 0 c, 9 0
/
/
/ / 1
1,
I
/
1,~
I 1 1 I [ I 1 1
I I
./ .2 .3 .4 .5 0 .0 5 ./ 0 . !5 .20
DJSPLACEIYfNT, FT D15 f LA CE IYENT, F T
..
.
..
//
F
.,$+
+-+-
., , ,_[
i
1
I I
I
1 I I I
I
1 I I I
I I I
I I
I I
1 I
I I I
I I
I
H< >
$
I I
I I
I
I
I I
I
I
I
\I
I I
I I I r I
11 . --/.5 ,/
400
~ h- 1.i
g.iskw /0 //
&
300 4-77
I :!
I
I I
I ii
200
104
.
I!i
0 1 I I I
o / 2 3 +
SAWF? f%rcz / /03 K/P$
Fig. 11 - Shear force distributionin Y-direction for Tower 1.
475 ~E/q/v PEAK VALUES
.. 0, i? #@Q@ S T o@~
. 1.5 4 18
400 gM!L . . 10 /!
\ /
\\\\.
300
\ 1.
\ \ \
\ \
\ \
\ \
\ %.
6200
\-\
106 \ \
\\
\
\ \.
d I 1 1 \ I
- .=. .
!