Anda di halaman 1dari 11

Dynamic and static analysis of a marine riser

C. L. KIRK and E. U. ETOK


Offshore Structures Group, Cranfield Institute of Technology, Cranfield, Bedford, MK43 OAL, UK

and M. T. COOPER
Conoco North Sea Inc., 116 Park Street, London Wl Y 4NN, UK

A frequency domain normal mode solution is presented for the dynamic response of an unbuoyed marine riser subjected to periodic excitation from a surface vessel in the direction of wave propagation. The variable tension beam-column equation is solved in terms of normal modes of free vibration of the riser and the rigid body displacement. Drag forces on the riser are represented by Morisons formula taking account of the velocity of the riser and wave-induced fluid velocity. A periodic solution for the flexural motion of the riser and the bending stress is then obtained by means of an iterative solution of the frequency response function. The drag force induced stresses arising from a linearly varying current are also determined. The results presented compare favourably with those obtained by other methods.
INTRODUCTION

A marine riser is essentially a conductor pipe connected between a fixed offshore platform, a floating platform or vessel and the wellhead at the seabed. Examples of floating structures used or proposed for use as drilling or production platforms are the semi-submersible vessel (SSV) and the tension-leg platform (TLP). The purpose of the marine riser is to provide a conductor for the conveyance of fluids and drilling hardware between the platform and the well and it may be categorized as either a drilling riser or a production riser. The floating platform will be referred to as a TLP. The present paper is concerned with the dynamic bending stresses in a marine riser, caused by direct wave forces on the riser and the excitation of the top end due to horizontal motion of the TLP. In addition to dynamic stresses the static stresses in the riser due to current forces and static offset of the TLP are determined. It should be noted that in contrast to an SSV which can perform vertical as well as horizontal and angular motions, the TLP by virtue of its stiff vertical pretensioned mooring cables is constrained to move primarily in the horizontal plane. The TLP thus removes the major drawback associated with the SSV which is that of having to disconnect the riser from the BOP stack when wave heights are of the order of 12 m. The designer of a marine riser system must take into account the expected peak dynamic bending stress in the riser, the maximum bottom angle or riser slope at the BOP (or ball joint), in addition to possible fatigue failure due to fluctuating stresses superimposed on a high mean stress. Two aspects of the influence of bending stress of importance are (1) the maximum bending stress which usually occurs at about 2/3 of water depth may give rise to local buckling, and (2) the bending stress near the top of the riser plus the tensile stress which is a maximum may be a substantial fraction of yield. To produce a reliable and safe design, a realistic assessment of the loading environment on a marine riser is essential. The major factors that should be taken into
0141 1187/79/030125Il$O2.00 @ 1979 CML Publications

consideration are: (1) wave forces associated with most severe operating sea state, (2) water depth, (3) static platform offset due to wind, current and slow drift, (4) current forces on the riser, (5) horizontal motion characteristics of the TLP, (6) excitation due to vortex shedding. Three types of riser dynamic analyses can be specified, namely (1) non-deterministic random vibration, (2) steady state or frequency domain, (3) deterministic time history. Because they are more economical in computing time (1) and (2) are often favoured over (3) when the system is linear. Method (3) is normally used for non-linear systems and when time varying parameters exist such as top tension applied to the riser. Tucker and Murtha presented a random vibration analysis of a marine riser which ignores the influence of platform motion, only taking into account wave forces acting directly on the riser assuming that the upper end is a fixed point. The influence of platform motion as pointed out by later authors is considered to be a major factor in evaluating dynamic stresses in risers. Burke* presented a steady state analysis including the excitation of the riser due to drilling barge motion. The method involves solving the variable tension, 4th order beam column equation by reducing it to a series of four first-order equations, in order to facilitate its solution by numerical integration along the riser. The influence of periodic barge horizontal motion is taken into account by including it as one of the upper end boundary conditions of the riser. The method has the advantage of being able to include abrupt changes in diameter of the riser caused by added buoyancy material. The direct wave forces are modelled by means of Morisons formula in which the drag force term is linearized using an average value of the relative fluid velocity at a given section. This velocity is selected so that the same energy is dissipated by the equivalent linear drag forces as by the quadratic drag force. The static stress in the riser due to current and mean offset is then combined with the dynamic stress. A time domain solution for a riser/drill ship system has been outlined by Sexton and Agebezuge3. The method uses an implicit finite difference solution to the tensioned

Applied Ocean Research, 1979, Vol. 1, No. 3

125

Dynamic and static analysis of a marine riser: C. L. Kirk et al.

beam column equation. Direct wave forces are calculated by Morison's equation and vessel induced motion is included in the riser boundary conditions. Random waves are represented as a superposition of a number of regular linear waves obtained by dividing the wave height spectrum into small frequency elements, each element representing a regular wave. The response of the riser due to vessel motion was found by summing the individual responses of the vessel caused by the component waves, taking account of the vessel (motion/wave height) frequency transfer function and the phase angle of the vessel motion for each wave. An identical procedure was adopted in calculating the direct wave forces on the riser. A similar type of analysis based on a finite element formulation of the equations of motion of a riser was presented by Gardner and Kotch 4. Young et al. 5 present an iterative frequency domain dynamic analysis which includes random sea states and current generated vortex induced oscillations. The wave spectrum is divided into elements each representing a regular wave and the responses are then superposed statistically. The present authors, however, believe that due to the non-linear drag forces it is not permissible to carry out linear superposition for random sea states. The correct method is to use the Monte Carlo approach in the time domain which enables the correct resultant fluid velocity to be substituted into Morison's equation for the drag force. Linear superposition of response for random seas is only permissible for linear wave forces. The frequency domain analysis given in the present paper differs from previous methods since it involves a normal mode solution of the variable tension beam column equation, taking account of rigid body motion of the riser and static offset. The method of incorporating the velocity of the riser relative to the fluid velocity in Morison's equation, accounts for the relative velocity at all sections of the riser, including rigid body and elastic modes and does not require linearization of the drag forces. TWO-DIMENSIONAL ANALYSIS FREQUENCY DOMAIN

out and the variation of riser stress along its length is obtained by superposition of dynamic and static stresses. Consider an unbuoyed riser of uniform external diameter D and length d, subjected to a top tension T1 as shown in Fig. 1. The differential equation of motion of the riser has been derived by Huang and Dareing v and is given by:
El i ' x 4 - ( ~ ' - ' q p " x ) i ' x 2 + #p~ = x +t"{t2 = 0 ,

(l)

where p, = total mass per unit length of riser and contents including fluid added mass, gp., = effective weight per unit length of riser including buoyancy effects. I f p l , p,, are the densities of water and drilling mud, di, d o are the internal and external diameters of the riser, w is the mass per unit length of the riser and ~,,, is the height of the head of mud above mean water level, then
rr To=T, -4 .qp,,6,,d2

(2)

and
7~

p, = w + 4

(p,,d~ -

p fd~)

(3)

The 'effective' tension (To - g p , x ) in equation (1)is identical to that discussed in ref. 5 if the effect of internal diverter pressure is ignored. Consider now the excitation of the riser due to a prescribed periodic motionf(t) of the upper end, i.e. T L P motion. It is assumed that the riser is ball jointed at the sea bed and at the T L P where it is taken to coincide with the mean water level. An approximate solution to equation (1) due to Mindlin and Goodman 8 is assumed in the form:
v(x, t ) = 9 ( x ) f ( t ) + ~ sin(nzrx/d)q,(t)
n=l

(4)

The forces acting in the direction of wave propagation on a riser attached to a T L P will consist of: (1) forces due to platform motion, (2) drag forces acting directly on the riser due to wave motion, (3) drag forces arising from current. As discussed by Morgan and Peret 6 transverse periodic forces may also arise by vortex shedding due to current and wave motion. The resulting transverse motion in the case of resonance can lead to considerable increases in both the in-line and transverse drag coefficients. Such effects although of practical importance, are not included in the present paper which is concerned primarily with in-line motion due to platform excitation with the drag force being included mainly for purposes of damping and to a lesser extent as a forcing function. Young et al. s state that vortex shedding due to current is included but give no details of how the effects of transverse motion are coupled with the in-line motion. In order to present a clear description of the dynamic steady state response of the riser we will begin by considering forces due to platform motion only. The influence of the drag force arising from relative motion of the riser will then be introduced, neglecting the effect of current. A static analysis of deflection under current forces is carried

TI

Vo

= f-,-"~"~,, I
T 2-%

I j.~-..~ MWL
'

__.7

x .~ ~,. t "CrUg~;t. ~
~~

SSV/TLP
Riser
r/

Figure 1.

Schematic" of marine riser

126

Applied Ocean Research, 1979, Vol. 1, No. 3

Dynamic and static' analysis of a marine riser." C. L. Kirk et al.


where ,q(x)= 1 -(x/d) represents rigid body displacement of the riser and the summation terms represent the elastic response in the normal modes. The assumption of sin(nrcx/d) for the elastic mode shapes is reasonable for tension ratios T~>I (T~=top tension/riser weight) and for d~<500 ft. For T~<I it is found that the exact mode shapes obtained by the Frobenius and Rayleigh-Ritz methods differed from sin(nnx/d) and the solution is less accurate. Equation (4) satisfies both the rigid body and elastic mode displacements of the riser. Substituting equation (4) into equation (1) gives:
~[ [[El(mr~d) 4 +
n

It can be seen that equations of motion for the first m generalized coordinates, expressed by equation (7) are coupled through the summation term which corresponds to t?2v/c~x and ~.v/~x in equation (1). If this term is 2 neglected the equations become uncoupled and the ruth natural circular frequency is then given by equation (8). To study the influence of the coupling term on the natural frequencies and mode shapes of the riser four terms will be used in equation (4). Since the term involving E(m, n) is small compared with the F(m, n) term, it is subsequently omitted.

To(n'lr/d}2] sin (mrx/d) -

Eigenvalue analysis Writing equation (7) is matrix form for m, n = 1, 2, 3, 4 we have:

,qp~{nlr/d)2x sin (ngx/d) + .qp,(nrt/d) cos (nnx/d)]q,, +


~q "

[l

q2~
1
1

~!3~i +
,14 2

t),~/4,, sin (nrcx/d) = ii P f f - p,.q(x)f


II

(5)

Equation {5) is multiplied throughout by sin{mrcx/d) and integrated over the length of the riser, which using the integrals given in Appendix A, gives the result:

[El,m./d,4d2 + To(m./d,z d -.qp.,(m./2)2Jq..+P~ iIm 2


'qP" ~1

0c3F(2,11, 0)22, 30%F(2,3), 4~3F(2,4)1 iq2~ a3F(3,1), 2a3F(3,2), 0)32, 4a3F(3,4) / qa %F(4,1), 2e3F(4,2 ), 3~3F(4,31, 0)4 2 J q4-

n2 E(m, nlq, +,qp.~(rt/d) ~ nF(m, nlq,


n

[ cq(1) i ... ~0~1(2) ) =j{tJ)ot,(3){


(6) k a,(4) J

i~:(1)) ~ ~2(2)~ +j:(t)~2(3)i \ a2(4) ] {4.1 + [K]


~q., = '

= I F(t),

1 (11)

-'qP~m~[1 - ( - 1)m]j- ~ j:

or
i)l,, + 0)2.,,q,.+ ~ [o~3nF(m' n) - o~4n2E(m, n)]qn
n

or
['-I~] {F(t)} (12) To obtain the eigenvalues from equation (12), writing F{t) = 0 and assuming a solution q,, = a~' gives the matrix equation:

= ~lf+ ~X2):
where

(7)
0)2, I-q--.] ,am,'' = [K] {a.', t13)

d 2 0)" = p~d El{mrctd)') 2 + T(mxtd)2 2 -'qP"(mxt2) (8)

22[

rn=l, 2, 3, 4
the solution of which yields the four natural frequencies 0),, and corresponding column vectors {al, a2, aa, a4} ~ The values of co,, can then be compared with the results of equation (8). An exact solution for the natural frequencies of a riser using a power series solution to equation (1) (the method of Frobenius) has been given in ref. 7, hence the degree of error involved in the approximate solutions can be assessed. It should be noted that if the exact orthogonal mode shapes had been used in equation (4) the modes would be uncoupled. The approximate mode shapes are, however, non-orthogonai and lead to coupling.

cq = 29p~ [1 - ( - 1)m] ~

% = -2/mrt
(9)

~3 =29(pJP,)d~

~4 =2q I p , / d 3

E(m.

--2\~]

(m~{l-(-1)m+"]
1

(m-n) 2'1 - ( - 1).... } and

(I0)

F(m, n)=

2 m { 1 - ( - 1 ) m+"~ m2 _ n 2 ,

m 4: n

Steady state response neylectin9 interactive fluid damping In view of the difficulties anticipated as a result of introducing non-linear fluid damping, it was decided not to attempt a solution to equation (11) directly. An alternative frequency domain solution to equation (7) is taken in the formf(t) = Ae i'' and q,,(t) = a,,e ~(''+''-),where A is the amplitude of horizontal motion of the platform at

Applied Ocean Research, 1979, Vol. 1, No. 3 127

Dynamic-and static analysis of a marine riser: C. L. Kirk et al. wave 'frequency o~, a,, is the required amplitude of response of the riser in the mth mode and 7,, is a phase angle. The solution of equation (7) then yields: A(al - ~20"~2)~3 E
ii

and the drag coefficient Co is taken to be 1.10. The velocity of the riser is denoted by t')t'/tnt and the fluid velocity is given by: u = u(x)ei'"'e i~" (21)

nF(m, n)a,e i:.,

ame'7. =
=M+iN where

where

(O 2 - - 0.) 2

(14)
u(x) = o~ (15)
=--COS 0 ) I ,

H cosh [k(d + x)] 2 sinh (kd)

(22)

and q/denotes the phase angle between u and .f(t) and e i.... Writing the mth mode as sin ( m n x / d ) = (pro(x), the generalized drag force corresponding to equation (7) is given by:

A ( ~ l - - g 2 0 ~ ) 2 } - - g 3 ' ~ ) l F ( m , tl)a,, c o s 7.,

M. . . . . . .

U) m - - 0 ) 2

"----

(16)

.l},,,,(t) = p,d
--%~nF[m. n)a.

2;
0

C ( u - ?U?t)(pm(x)dx

(23)

sin

7,, (17)

N -Thus

u)2 _ o)z

Substituting equation (23) in equation (7) gives:


d

am=(M2 +N2) 1/2 and

?,,=tan

I(N/M)

(18)

2 #"+,,d

f Ci'zq'"'(x)d-~+'"'q"+%Y"F(n" c,, =
N

'')q''

An iterative solution to equation (14) can be carried out as follows. Starting with m = n = 1 on the right hand side, al and ),~ are calculated using equations (16), (17) and (18). Substituting these values into the right-hand side of equation (14), a 2 and 72 are calculated which then enable more accurate values of a~ and 7~ to be obtained. The procedure is repeated in a cyclic manner for a 3, ~3, ~4, 74 until satisfactory convergence is achieved. The foregoing analysis could be used when wave frequency 0 and the various natural frequencies tom are different, to give the steady state response of the riser neglecting riser/fluid interaction effects. In the more serious case of resonant excitation of the riser it is essential to incorporate nonlinear hydrodynamic damping. Steady state response including non-linear damping In calculating the drag force on the riser it is necessary to know the phase difference between the fluid velocity u and the platform motionf(t). This phase angle can easily be assessed because the frequencies of the dominant wave forces on a semi-submersible or tension-leg platform are considerably higher than the natural frequency of surge motion. Thus since the exciting forces are primarily dependent on fluid acceleration fi it follows thatf(t) and ti are approximately 180' out of phase, hencef(t) and u will be approximately 90 out of phase Fluid inertia force on the riser is considered negligible in comparison with the drag force. The drag force per unit length of the riser at depth x, due to the horizontal component of fluid velocity is:

= p,d2

fcu,am(x)dx+(,l+%{.
,
0

(241

F r o m equation (4) the velocity of the riser can be written:


~U= y I x ) i .+ & L
n= i

.i (p,,(x)q,,=q{x)Atu~e .... + ~/ P,,(xki,,


11: I

(25) Making use of the orthogonal property


d

f (p,,(x)cp,(x)dx = 0
0

for m 4: n and substituting equations (20), (21) and (25) into equation (23), writing fl = Copy D/p,d we have:
d

i51,,+ fl f ]u{x)e ii~ + ' ' ' - g(x)Aime i''' - q,dP,,lx)16](xlAm)e i,,,, " ,, .
0

+O.,p.,(x)l q..tx)dx + O)rnqm 2


d

F D(X, t)=CopfD(u - i)v/?t)tu - ?c/t3tI


= C(u - & / & ) where C =Copy Dlu - 8v/&l (20) (19)

= 13 [ lu(x)e i(* +'''~ -,q(x)Aitoe i''' 0 q..~%,(x)lutx)e iq* p,,(x)e it,Jr d x +

Aei"'(cq - ~2r, 2) - 0% ~ nF(m, n)q, o


n

(26)

128

Applied Ocean Research, 1979, Vol. 1, No. 3

Dynamic and static" analysis of a marine riser: C. L. Kirk et al. A steady state solution for qm is now written: q" = a"e i~''~'+ 7"~ (27) It may also be required to determine the shear reactions due to bending at the upper and lower joints of the riser, using the expression: a~(x)= E l ~ = El ~ am(mn/d)a cos(mnx/d)e ~;'- (34) t?zv ab(x) = Er (~xz = Er Z am(ran~d) 2 sin (mnx/d)e';'. (33)

which on substituting in equation (26) and cancelling e i''~t and taking Icos ~otl =2/~ gives
d

(~2 - to2)ameiV,,,W2~flf lu(x )eiO - .q(x)Aito - ot,,rit~ei:,q~"(x)l . . .


0

. . . {g(x)Aico + ct=i~oeiT.,~o,.(x)}tp,.(x)dx
d t~

STATIC ANALYSIS OF RISER

7 | lu(x )ei* - ,q(x )A ito -7,,,itoei;,,,tp,,(x)lu(x)e i* tp"(x)dx +


0

The static bending and shear stresses in the riser caused by current drag forces and platform offset, as shown in Fig. i are now considered. The equation of static equilibrium of the riser is: &4v , &2v 8v El ~ - (TO- gp,xj ~x z + gPs ~x =L(x) where the current drag force is given by: f~(x) = l p l CoD V~(x)2 The current profile is assumed to take the form: (36)

A(0q --

0~2D 2) --

o%~')lF(m, n)an el;'''


II

(28)

(35)

The modulus term on both sides of equation (28) is written in the form la + ibl, where: a = u(x) cos ~b+ a,. sin Y,,. tp"(x) b = u(x) sin ~ - 9(x)Aco - a"to cos 7"- ~o"(x) Equation (28) is then written:
d

(29)

2~ ix/a2 + h2 [u(x)ei,q)"(x)_g(x)Aitoq),,(x)}d x + A(~tl _ct2tnz)_~3~nF(m ' n)a,,


. n

a"ei;,, =

(30)

t ~ .2 - ~o2) + 2fl~ J x(' a '25 + b 2 q~2(x)dx , Ti /


0

It can be seen that the unknown quantities am and 7,, appear on both sides of equation (30) hence an iterative solution is required. It is noted that the imaginary part of the denominator in equation (30) represents the equivalent linear hydrodynamic damping. The influence of the term in F(m, n), equation (30) was found to be small. Equation (30) is expressed in the complex form: A+iB a,.e i''," :: - C + iD where / ~ a,.=4~-D~ BC-AD tany"-AC+B~ D (32)

V~(x)= Vo(1 - x/d)

(37)

from which it can be seen that f~(x) varies parabolically with depth. As in the case of the dynamic analysis we take an approximate solution to equation (35) in the form: v(x) =9(x)v o + ~ A. sin (mtx/d)
n

(31)

(38)

and

where vo = platform offset and the A, are undetermined coefficients. Substituting equation (38) into equation (35), multiplying by sin (mnx/d) and integrating over the length of the riser gives, omitting the term in E(m, n) gives: 2 A,.[{EI(mrc/d)4+ To(mn/d) 2} d ~-(mlt/2)gp~] +

in which A, B represent respectively the real and imaginary parts of the numerator of equation (30) and C, D correspond to the denominator. The iterative method of solving equation (31) is identical to that described for equation (14). Dynamic riser stress The external bending stress in the riser is obtained from:

9p~(rr/d) ~, nA,F(m, n)
n d

= ff.(x) sin (mgx/d)dx-yp~(vo/mrO[ ( - 1 ) " - 1] (39)


0

Applied Ocean Research, 1979, Vol. 1, No. 3

129

Dynamic and static analysis of a marine riser." C. L. Kirk et al.


20
WATER DEPTH 520ft [D] r ~ = ~Aj ~ ~ ~Cj
or

Inverting the D matrix gives:


15
J

~A,

[D]

'

IC I

(43)

,=,,
(/3

O
ne n _.1

which on multiplying out the matrix product will yield the deflections coefficients Aj, A 2, A 3, A 4. The static deflection of the riser is then obtained by substituting these coefficients in equation (38). The bending and shear stresses are then calculated by replacing a,, by A,, in equations 133) and (34) and writing 7,,=0. The tensile or compressive axial stress is given by:

"~ nI--

5.0

MODE 2 MODE 3 MODE 4

o tIx)= ( L - w x )
4 (d,, - d i )
2 2

(44)

<

NUMERICAL ANALYSIS A drilling riser was selected for study, the details of which follow. The case considered refers to a value of top tension for which the tension ratio T, (top tension/riser weight) was equal to 1.95, where

0 i-.

0"8

1.0

1"2 TENSION

1./~

1"6

1"6

2"0

RATIO. Tr

Figure 2. Natural periods of riser

Tr=(Tl/,q-4)[d,,(p,-pf)+d2(p.,-p.,)]
7[ 2

Equation (39) can be written:

The basic data given in Table 1 are used to model the drilling riser given in Case 500-20-1, A P I Bulletin 9, but the wave height and period are varied.

Natural J?equencies AmG,, + gpsOz/d) ~ nF(m, n)A, = C,,


n

(40)

where

Gm=[{EI(mTz/d)4 To(mzt/d) + 2}
and

~-(mg/2)9psl
d 2 -

The natural frequencies of the first four modes were calculated by four methods for the sake of comparison. The approximate methods of equations (8) and (13) gave almost identical results indicating that the term yps~vfi3x in equation (1) has little effect for the risers considered. The exact solution for the frequencies was determined using the method of Frobenius and an additional Rayleigh-Ritz solution was also developed (see Appendix
Table 1. Data spec!fications Jor example riser analysis 9

C,. = pf CoD V2(d/2m~) [1 + (mn)2 { ( - 1)m- 11] 2 gps(Vo/mrt)[( -- 1)" -- 1]


(41)

To illustrate the solution to equation (40), it is assumed that four terms are sufficient to describe the deflection of the riser, but more terms can be easily be added if required for greater accuracy. Writing ~=gps(~/d) and taking m = 1, 2, 3, 4, equation (40) can be written:

2,
fiF(3,1), t~F(4,1), 2~F(3,2), 28F(4,2),

3 t2,3,,
G3, 3~F(4,3),

4,1,41iA) It11 4 12,4,| A1 c2


4~F(3,4) 1
G4 .J A3 A4 C3 C4 (42)

Top tension, T1= 1.927 105 lb Water depth, d Riser length, d Density of steel, p~ Density of water, Pl Density of mud, p,, Outer diameter of riser, d,, Inner diameter of riser, d~ Cross-sectional area oJ"riser, A, Weight of riser in air Modulus of elasticity, E Drag coefficient, Co (including kill and choke lines) Equivalent drag diameter (including kill and choke lines), D Static offset, Vo Surface current speed, I/,,, (for static analysis only) Wave height, H Wave period, T Vessel surge (peak to peak), 2A Vessel surge phase angle, q'

520.0 fl 520.0 ft 490 Ib/ft3 60.4 lb/ft3 89.8 lb/ft3 16.0 in 14.75 in 30.18 in2 172.42 lh/ft 4.32 lO9 lb/ft2 1.10 26.0 in 15.0 ft 0.84 ft/sec 20.0 ft 9.0 sec 4.0 ft 75.0 deg

130 Applied Ocean Research, 1979, Vol. 1, No. 3

Dynamic and static analysis of a marine riser: C. L. Kirk et al.


Table 2. Natural period,s o]' marine risers./br T, = 1.
Approximate natural period [see) Mode 1 2 3 4 I 2 3 4 1 2 3 4 I 2 3 4 (eqn 81 14.45 6.35 3.53 2.21 15.81 7.28 4.22 2.73 17.26 8.27 5.00 3.34 18.55 9.11 5.65 3.89 (eqn B9) 16.0 6.6 3.6 2.24 17.83 7.71 4.36 2.78 19.86 8.95 5.26 3.45 21.65 10.04 6.08 4.09

Mode shapes
Exact natural period (Frobenius) 16.1 6.5 3.56 2.22 18.06 7.60 4.29 2.75 20.29 8.81 5.90 3.38 22.32 10.08 5.90 3.97

Water depth (It) 520

600

700

For d = 5 2 0 ft and T,> 1 the assumed mode shape sin (mnx/d) was found to agree closely with the Frobenius and Rayleigh-Ritz solutions especially so for the case T, = 1.9. The latter mode shapes were almost indistinguishable for T~> 1.0 and water depths up to d = 800 ft but a difference was observed for n = 1 and d = 1000 ft. For the sake of simplicity the present analysis is restricted to sin (mnx/d). It is, however, recommended that for T~< 1.0 and d > 520 ft, the more accurate mode shapes be used otherwise the bending stress in the riser will be underestimated. This is because the simple sin (mnx/d) function does not adequately yield the correct curvature of the riser near to the bottom end.

800

DISCUSSION OF RESULTS

Dynamic analysis
The transfer function assumed for the T L P surge m o t i o n , f (t) is shown in Fig. 3. For a tension ratio T, = 1.95 the riser was excited at the first and second natural periods (T 1 =9.13 sec, T2 =4.3 sec) for wave steepness ratios H/L = 1/15, 1/20 respectively, the corresponding wave heights being H = 28.5, 21.4 ft for mode 1 and 6.33, 4.75 ft for mode 2. For comparison with the API results for various riser programmes 9 the riser was also excited approximately mid-way between the first two natural periods at T= 7 sec for H/L= 1/15 ( H = 16.67 ft). The results are shown in Figures 4, 5, and 6 for a drag coefficient Co = 1.10.

SWAY - ( B E A M
0.6

FORDRILLING VESSEL/ (Ref 2 ) ~

WAVES)

0.5 o 0-4 0.3 0-2

=
w o m

/ SURGE-(HEADJ / ON WAVES). / YI / T L P / S / FOR3LEG

0.0
80"0

AMPLITUDE OF RISER MOTION ( f t ) 0-0 2-0 4.0 6-0 MWL

0"1
160-0
I I I I

5 10 15 20 WAVE PERIOD (sec)

25
~-

240.0

Fiyure 3. Surge and sway response Jbr TLP/SV V

.=J

B). The results are given in Table 2 for a tension ratio T, = 1. Figure 2 shows the variation of natural period with T,, calculated from equation (8). In the approximate solutions for natural frequencies it is not possible to take account of the effect of riser offset whereas offset can be included in the Frobenius method. It was found that an offset of 15 ft increased the odd mode periods and decreased the even mode periods by about6%. Figure 2 shows that an increase in the tension ratio T~ has most effect on decreasing the first mode natural period and as expected has less effect on the higher modes. F r o m Table 2 it can be seen that the natural periods of the first three modes are in the range of typical wave periods, hence excitation at or near to resonance is a distinct possibility.

:z
0 r~ 14. LU r,.;

320-0

z
O3 (:3
I.-

400.0

480.0

520.0 0.0

1.0

2.0KSI3.0

4-0

5-0

Figure 4. Dynamic bending stress for resonance in first mode (TLP)

Applied Ocean Research, 1979, Vol. 1, No. 3

131

Dynamic and static analysis of a marine riser." C. L. Kirk et al.


AMPLITUDE OF RISER MOTION (ft) -1"0 0-0 1-0 MW L
,,

0"0,

TLP SURGE +-0"16 ft


80"0 - Tr = 1"95

resonant response of the riser is governed by non-linear hydrodynamic damping. Referring to the denominator in equation (30), the equivalent linear damping is obtained from the imaginary term as:

AND * 0-12 ft

H=I " ~ 20
L

fi,, = ~t2/D- f x/a~+ b 2 qo~(x)dx


lt p ~tu,,,d . o

(45)

160.0

(H= 4-75ft)

240.0

BENDINGX
STRESS
._I

>/

H 1 T'=~

I'~ (H= 633 ft,


HEAD ON WAVE S

320"0
o
ne h

LU U Z

400-0 R

I ._."

I--U3

The variation of fie with frequency and steepness ratio is shown in Fig. 7a for the first four modes of the riser. It can be seen that []~ is a maximum for mode 1, having a value of 6.2~,,, of critical for H/L= 1/15 at 09=0.688 rad/sec (T =9.13 sec). For the higher modes the damping is much smaller, particularly at higher frequencies, which explains why the response in the second mode is significant. In this case fl~=0.5% for H/L= 1/20 and 0.7~,, for H/L= 1/15. The magnitudes of the maximum bending stresses, although small when considered from the point of yield or local buckling of the riser are important in relation to the fatigue life and integrity of the riser screwed joints. A comparison of the exciting forces in the numerator of equation (30)due to vessel motion and drag, indicates that vessel motion is the major cause of dynamic riser bending stress.

480.0 I-

"<

Static analysis
-

-1-0

-0.5

0.0 KSI

0-5

1.0

The variation of bending stress due to current drag forces was calculated for T~= 1.95, the maximum value being + 1000 lb/in 2 at 77% water depth.
AMPLITUDE OF RISER MOTION (ft) 0'0 1-0 2-0 3"0 MWL 0'0

Fidure 5. Dynamic bending stress]or resonance in second mode (TLP)


The maximum dynamic bending stress arises for resonance in the first mode with H/L = 1/15, the value being + 3066 lb/in 2, the location being at 65~o of water depth. A secondary local maximum bending stress of _ 2587 lb/in 2 occurs at 20% of water depth. By examining the curves of bending stress in Fig. 4 it is evident that contributions to the bending stress occur due to response in the higher modes, although excitation takes place at the first natural period. The presence of higher mode deflections is however scarcely discernible in Fig. 4. For the steeper wave, H/L = 1/15, the bending stress is about 15~,, higher than for H/L = 1/20, for both resonant responses. It is noted that the T L P response at H/L = 1/15 is 33% higher than for H/L = 1/20 (i.e. 20/15). The reason for the smaller percentage increase in bending stress at the higher level of excitation is clearly because of non-linear hydrodynamic damping. The maximum stress for resonance in the second mode is _+ 1160 lb/in 2 as shown in Fig. 5. In this case the amplitudes of T L P motion are reduced by a factor of about 1/16. The maximum elastic deflection in mode 2 is about 1/7 of the deflection in mode 1 and the bending stress ratio is 1/3. Thus it can be seen that although the excitation at a period, T=4.3 sec is 1/16 of that at T=9.13 sec, the level of response and bending stress at the lower period is significant. The response and bending moment in the riser for excitation in between the first two modes is shown in Fig. 6 with the maximum value of _+ 1270 lb/in 2 occurring at 21% of water depth. The results shown in Fig. 4 compare favourably with the API results 9 for T~= 1.95. It has already been stated that the amplitude of

60"0

160'0

wv

._1 240-0
:r O 320'0 nh UJ U Z Ul

400.0

480.0 520.0 0.0

1-0

2.0KS I 3"0

4"0

5-0

Fiyure 6. Dynamic bendin9 stress]or excitation between first and ,second modes (TLP)

132

Applied Ocean Research, 1979, Vol. 1, No. 3

Dynamic and static analysis of a marine riser: C. L. Kirk et al.


0.,6.MOOE ,{ o ~ \ ~ 0~z ~ r-,
"lODE 2

"~ ,~
MODE 1, w 1 " 0 " 6 8 8 rad/sec

For a 16 in. riser with a tension ratio T, = 1.58 and d = 400 ft, Burke 2 gives a maximum bending stress of about +8000 lb/in 2 for H - - 3 0 ft, T = 2 0 sec. The bottom angle due to offset, waves and current was 4.6 . The results of the present analysis are given in Figs. 8, 9, 10, and can be summarized as follows:

DO
0"011~.

,o
i ~ ~t ~ x

,.O.WL
m

/sec

0o o

zooo
02

8o.o
0"4 0'6 0-8 1.0 1'2 1"4

.,I

,oo

('~/~-

MODE 1. w 1 =0"688 rcad/sec

~"

/~!H-21'4ft:

L
I
:
/
(,~ 0'000 1 0.2

/,Y\
%..\
~
l l'Z ,.,a (RAO/SEC) ~ ).4 I (~

//'A/
I /./ / ..~"BENO,NO
48o.o1- / ) ' ~ . / "
l 1.0

i.ooE, <~,~..,~z..~..~-....._~ ~
b T

=~
I 0.4 I 0.6 II 08 WAVE FREQUENCY,

'520~
0'0 2"0

4"0KSI6.0

8'0

10"0

Fiyure 7. Equivalent linear dampin9 ratio for sur9e motion of(a) T L P in head-on waves and (b) drill vessel in a beam sea Total maximum stress For the T L P combining the maximum dynamic bending stress in mode 1, Fig. 4, with the bending stress due to current and the axial stress at the 67% level gives a resultant stress of (4400+4000) lb/in 2. It must be emphasized, however, that resonant excitation in mode 1 with a wave height of 28.5 ft is considered to be an extreme case. Bottom angle The total bottom angle due to riser deflection is also of interest. For mode 1 the angle due to dynamic response was 1.43 and the angle due to current was less than 0.1 . The maximum bottom riser angle is obtained as the sum of the dynamic angle and the static angle due to current and vessel offset. Comparison with other published work The results given in the present paper can be compared with those presented by Burke 2 for a 16 in. riser. The transfer function for sway motion of the drilling barge is shown in Fig. 3 where the incident waves are broadside to the vessel. In the case of a SVV or T L P the transfer functions for sway and surge are similar but for a typical surface vessel such as a drill barge, the sway transfer function for periods between 5 and 10 sec is about four times as great as for surge.

Figure 8. Dynamic bendin9 stress for resonance in first mode (drill vessel)

o.c

AMPLITUDE OF RISER MOTION (ft) -2.0 -I-0 0-0 1"0 2.0 MWL ICTION

80.

160"~

E j 240"

0 re 320"
14. tU C.) Z

I-~n
Q

400,

480.

520-q

-4.0

-2.0

0.0 KSI

2.0

4.0

Figure 9. Dynamic bendin9 stress for resonance in second mode (drill vessel)

Applied Ocean Research, 1979, Vol. 1, No. 3

133

Dynamic and static analysis of a marine riser: C. L. Kirk et al.


AMPLITUDE OF RISER MOTION (It) 0"0 2"0 4'0 6-0 8-0 M W L

oo

SURGE = -*3.53f
80.0 - r r = 1.95

BEAM WAVES
160.0

"J

240"0

BENDING STRESS _H=L t 15 (H= 16"/*/* ft)

dimensional resonant response of a marine riser subject to vessel motion in regular waves. The non-linear hydrodynamic damping for the riser was found to increase considerably as the amplitude of vessel motion increased, thereby providing a natural means of limiting resonant riser deflections. The numerical results obtained were found to be comparable with those obtained by other methods but they obviously depend strongly on the value assumed for Co. In this respect there is an urgent need for data from full scale tests on oscillating risers including the effects of current. Comparison of the exciting forces due to vessel motion and drag force show that vessel motion is the dominant source of riser bending stress.

0 r,. u. tu o z
b-

320"0

u~

400.0

REFERENCES
1 Tucker, T. C. and Murtha, J. P. Nondeterministic analysis of a marine riser, Proc. 5th A. Offshore Technol. Con./'. Houston, 1973, Paper No. OTC 1770 2 BUrke,B. G. An analysis of marine risers for deep water, Proc. 5th A. Offshore Techno/. Conj'. Houston, 1973, Paper No. OTC 1771 3 Sexton, R. M. and Abzeguge, L. K. Random wave and vessel motion effects on drilling riser dynamics, Proc. 8th A. O[l;~hore Technol. Conl~ Houston, 1976, Paper OTC 2650 4 Gardner, T. N. and Kotch, M. A. Dynamic analysis of risers and caissons by the element method, Proe. 8th A. O[]/shore Technol. Cot!If Houston, 1976, Paper OTC 2651 5 Young, R. D. et al. Dynamic analysis as an aid to the design of marine risers, Trans. ASME, 1978, 100, 200 6 Morgan, G. W. and Peret, J. W. Applied Mechanics o] Marine Riser Systems. Petroleum Engineer Publishing Co., Dallas, 1976 7 Huang, T. and Dareing, D. W. Buckling and lateral vibration of drill pipe, J. Eng. Ind. Trans. ASME, 1968, 90, 613 8 Mindlin, R. D. and Goodman, L. E. Beam vibrations with time dependent boundary conditions, J. Appl. Mech. Trans. ASME, 1950, 72, 377 9 Comparison of marine drilling riser analyses, Am. Petrol. Inst. Bull. 2J, January 1977

480.0
520.0 V

DO

11o

21o

KSI

J-o

so

Figure 10. Dynamic bending stress for excitation between first and second modes (drill vessel) Resonant response in mode 1 Max. bending stress, + 8000 lb/in 2 at 20~o and 67~o levels. Max. elastic deflection, 8 ft. Bottom angle, 4.5T~. Resonant response in mode 2 Max. bending stress, _+4600 lb/in 2 at 25% and 75% levels. Max. elastic deflection, 1.6 ft. Bottom angle, 0.86 . Excitation between modes 1 and 2 Max. bending stress, _+3800 lb/in 2 at 23% level. Max. elastic deflection, 3.5 ft. Bottom angle, 1.43. Comparing the results of the head sea case with those for the beam sea, it can be seen that although the level of excitation is about four times greater for beam sea motion of the drill barge, the maximum riser bending stress is only about 2.6 times as great. This result can be explained by reference to Fig. 7b which shows the equivalent linear damping ratio for beam seas. For mode 1 the damping has increased (for H/L = 1/15) from 6.2~ to 16~ whereas for mode 2 the damping has increased from 0.7~o to 2.5%. The considerable increase in damping thus explains why the relative increase in bending stress is smaller than the increase in the excitation amplitude. One interesting aspect of the bending stress due to excitation at the first natural period is the contribution from the second mode. This is because the triangular distribution of inertial exciting force due to platform motion and the antisymmetric second mode shape gives rise to a significant generalized force in the second mode. CONCLUSIONS The analysis described provides a simple and computationally efficient method of evaluating the two-

APPENDIX A

Integrals The following integrals are used in deriving equation (7)


d

f x sin (mTtx/d) sin{n~x/d)dx


0

1 {d~Z[

11-(-1) m+I (m_.)2

= E(m, n)
d

m 4: n

f sin (mnx/d) sin (nnx/d)dx = O,


o

m=/:n

134

Applied Ocean Research, 1979, Vol. 1, No. 3

Dynamic and static analysis of a marine riser: C. L. Kirk et al.


d

where
rn = n

f sin (mnx/d) cos (nnx/d)dx = O,


0 d

:~. = 2 o d(nrc/d) 2 - gp~(nrt/2) 2

f s i n ( m . x / d ) c o s ( n m ~ / d ) d x = d 7r m ~ lnfl___i,~ j '''+" =Flm, n), [ -(-1) ~


0

=_To d(mT~/d) 2 -- ~jp~(mrt/2) 2 2 "

(B4)

m# n
d

0%.

=gpsT[l+

(n + m) 2 I

-1

f x sin2(mTtx/d)dx
0 d

d2 4'

m = n

Total strain enerfly:


2 2

v= vb+
f (l - x/d) sin Omrx/d)dx = d mzt
0

+-q2

+
(B5)

where ft. = El(nn)4/2d 3 and ft., = El(mrt)a/2d 3 Kinetic energy:


d

f sin(mztx/d)dx= 'J [1 - ( - 1 ) ' ] mrt


o d

T= Pt J f/~dx = p~d (32 + t~2, f 2


0

(B6)

f sin2(m~zx/d)d x = d 2
0

Applying Lagrange's equation to equations (B5) and (B6) gives: (p,d/2)q. + (~. + fl.)q. + a,..q,. = 0 (p,d/2)ii.. + (a., + tim)q.. + c~,..q. = 0 (B7)

APPENDIX B
Rayleigh-Ritz s'olution .lot natural.fi'equencies and mode .shapes of riser An approximate deflection shape for the rth mode of the riser is taken in the form: G(x. t)=qmsin(mTtx/d)+q.sin(nnx/d) (BI)

The steady state solution to equations (B7) is obtained assuming q.... = ~,... sin ogt which gives the following equation for the natural circular frequencies: (P,d/2) 2m'* - )2(p,d/2)(~. + O~m ft. + tim) + + (~. + fl,)(~,. + E.) - ~ . ~, , - 0 ,
-

where m = r. n = r + 1 and q,... denote mode participation parameters which are determined by means of Lagrange's equation. The strain energies due to bending and tension of the riser are determined as follows:

(B8)

For the rth mode the natural frequency is obtained as the lowest root of equation (B8) given by:

Strain energy due to bending:


d

x/(cq,+ % + fi,, + fl.,12-4(~,, + fl,,11% + [3.,)+4~,,] (B9) (9 tl/aX ) dx = ~ - - [q2(nn/d)4 + q2(m~/d)] The ratio of the mode participation parameter is obtained from: q,./qn (~. + f t . ) - mZ(p,d/2)
O~mn

(B2) Strain energy due to tension."

(B10)

Vt=2

1;
d

(To-gpsx)(~rlfi)x)Zdx

For the first three modes, r = l, 2, 3 the following values or m and n are used in pairs, r = l(m= 1, n=2), r = 2 ( m = 2 , n=3),

which on substituting equation (B1) and evaluating the various integrals yields: O~n 2 -- %,, q,. + O~m.q.qm v, =-2 q. +-2 2 (B3)

r = 3 ( m = 3 , n=4) The rth mode shape is obtained by substituting from equation (BIO) in equation (B1).

Applied Ocean Research. 1979. Vol. 1. No. 3

135

Anda mungkin juga menyukai