00457949188 $3.00 + 0.

00
0 1988 Pergamon Press plc

A SIMPLE AND EFFICIENT ALGORITHM

FOR THE STATIC AND DYNAMIC ANALYSIS
OF FLEXIBLE MARINE RISERS
RUMBOD GHADIMI
College of Aeronautics, Cranfield Institute of Technology, Cranfield, Bedfordshire MK43 OAL, U.K.

(Received 12 August 1987)

Abstract-The lumped mass discretization method is used to derive the equations of motion of flexible
risers in three dimensional space. Nodal shear forces and bending moments are included in the
formulations and fluid loading on the structure is calculated using the Morison equation. A simple model
of the sea bed contact for catenary risers is given and tested. Nonlinear equations describing the riser
motions are solved in the time domain using the tangent stiffness incremental approach combined with
the Wilson-theta numerical integration algorithm. Computational efficiency is improved via a compact
‘profile storage’ technique, used for the assembly of element matrices, given by Taylor [Computer
procedures for finite element analysis, in The Finite Element Method (Edited by O.C. Zienkiewicz), 3rd
Edn. McGraw-Hill, New York (1977)]. Consideration is also given to natural periods and mode shapes
for small oscillations about the static equilibrium position of the riser. Comparison of results with
experimental values, analytical solutions and independent results from the published literature show very
good agreement.

INTRODUCTION large displacements, small strain analysis of strings

together with the variational or Galerkin method of
The concept of employing flexible catenary risers, as solution. The inextensibility constraint in this rather
opposed to tensioned rigid risers, for use in the sophisticated approach introduces extra unknowns,
offshore oil and gas industry has necessitated the namely the Lagrange multipliers, into the equations
study of static and dynamic behaviour of flexible of motion. The method is generally more suitable for
hoses and riser systems. Catenary risers provide a the analysis of offshore pipelines, however, the con-
compliant link between the floating production facil- straint condition can be easily removed from the
ities at the free surface and the subsea well head. Due equations. McNamara et al. [8] used hybrid beam-
to its low flexural rigidity the riser can exhibit large column finite elements to include the inextensibility
three dimensional motions under the action of waves condition in their two dimensional model for flexible
and current. Nonlinearity of the problem is further risers and pipelines. McNamara and Hibbitt [9] car-
exaggerated due to the presence of quadratic damp- ried out a three dimensional finite element analysis
ing terms arising from the fluid drag loads in the using beam-column elements taking into account the
governing equations of motion. axial, torsional, bending and transverse shear strains
A number of the published works on the analysis in their formulation. They state that in situations
of cables and offshore mooring lines can be related to where the axial rigidity of a pipe is much larger than
the dynamics of flexible hose systems but not many its bending stiffness, as in the case of a flexible riser,
give methods that can be applied easily and directly numerical ill-conditioning of the coefficient matrices
to slack catenary riser systems in three dimensions. can lead to poor or unstable computations hence
The lumped mass method was used by Walton and requiring the introduction of an inextensibility con-
Polachek [l] and more recently by Nakajima et al. [2] straint condition.
to model mooring systems and cable lines in two Here a simple three dimensional lumped mass
dimensions. Peyrot and Goulois [3] and Peyrot [4] model is presented taking into account the shear
based their mathematical model on curved cable forces and bending moments at the nodal points. An
elements of zero bending stiffness and made use of the approximate method similar to that used by
elastic catenary equations to derive the equilibrium Ractliffe [lo], based on the geometry of curves in
equations at the nodal points. Their method offers the space and the classical Bernoulli-Euler theory of
advantage of using much larger elements in a static elastic rods, is employed to calculate the bending
analysis; however, its computational efficiency for a moments along the riser. All forces and displace-
three dimensional dynamic analysis becomes ques- ments are measured relative to a global coordinate
tionable. Garrett [5] presented the nonlinear equa- system thus removing the necessity for any
tions of motion of inextensible elastic rods in three local-global axis transformations. The objective is
dimensions after Love [6] and Nordgren [7] using therefore to formulate a simple and efficient algo-
541
542 RUMBOD GIUDMI

52
yj2, 1
-*'j2
t

LUBPE~ BASS
yjly
RODE1
zjl

t -
J-x.
II
-

BUOr

TETilER
ll”

Fig. 1. Schematic of flexible riser system.

rithm capable of handling various flexible riser where as shown in Fig. 2(a) S,, and Sg represent the
configurations and to validate its results against both displacements at nodes jl and j2 respectively, i.e.
theoretical and experimental data.
sji=(Xji, r,,z,,, i= 1,2. (1)

BASIC THEORY Figure 2(b) shows the forces acting at the nodes of
the jth element. Thus force equilib~~ at node jl
A schematic of the flexible riser is shown in Fig. 1.
implies the force vector relationship
The riser is divided into a finite number of elements
each consisting of a linear elastic spring with the mass Fj, = r, ~j + Fi9 + F~, - Wjl + Qj 3 (2)
of the element concentrated at the nodes. Each node
has three degrees of freedom (three DOF) describing
where
the displacements in (X, Y, 2) directions giving six
DOF per element. All ~spla~ments are measured
relative to a global axis system XYZ for which the XY ‘I; = tension in jth element =
plane lies at the still water level SWL.
Consider the jth element of the riser which con- (Lj - .L;, (EA /q> (3)
nects nodesjl andj2. The displacement vector S, can
then be written as L!, Lj= initial and finai element lengths re-
spectively
EA = riser axial stiffness
 Algorithm for analysis of flexible marine risers 543

Equilibrium of forces at node jl implies that

Fj, = (m’ + mf),, ($)j, + (MS+ mz)j, (S,)jl) (7)

where (m:),, and @I:),, denote the normal and tan-

gential added mass of riser at node j 1. (s,),, and (s,g
represent normal and tangential structural acceler-
ations at j 1. These can be expressed in terms of total
structural acceleration &I as

(s,)j, = @,I ‘sj)sj = [Al S,I3 (8)

i-z
1,A1
“”
Node jl
where

1
(a) Displacements yfxJ =J xJ
.s. -2
iAl= Y,__ ?,ij (9)
“_i?
[ Symm. I
and

(R ),I = gj:jl- (S,)j, = PI SjI3 (10)

.1(111
where

(1 A;,‘,.,_ -ijzj - *
-zjx,
PI = (l-pi)_ ^*
-zjYj
Symm. “.(l - 2;)

The equations of motion at node jl can now be

written using eqns (2) (7) (8) and (lo), thus

(m’ + mZ)jl[Clgj,
(b) Forces
=~~j+Ff:+F~,-~jl+Qj_Fi4, (12)
Fig. 2. Forces and displacements at jth element.
where

1
1 0 0
!Jj = unit tangent vector = (ij, 9j, fj)
[Cl= 0 1 0 (13)
[0 0 1
= k [(q2 - A$), (52 - $17 Cz,2 - zjl )I C4)
I F,4 = position dependent added mass term
F$, Fj, = fluid induced drag and inertia force = IpljISI (14)
vectors respectively
wjr = weight of element in fluid lumped at ipljl =
node jl = (0,0,W,,)
(my- mi)j&., (my- mi),,_tjij
(my- m~)j,f,k,
w,, = (mf, - prLD2418) g (5) '(my-mf),,Pf (m;-m:)j,ij$j.
Symm. “.:(m:-m:),,ij
I
rn;, = structural mass lumped at node jl
= mass per unit length x f Lj
(15)
p = fluid density
D = external diameter of riser Similarly, for node j2 we have
g = gravitational acceleration
Qj = shear force vector (assumed to be con- (m’ + W[Cl~j*
stant along the element).
=- ~$,+F$+F;--w~~-Q~--~, (16)
The length of the jth element is given by
where

L, = [(x, - X,, )’ + ( Yfl - r;., )* + (Z,2 - Z,,)2]1'2.(6) y2 = [plj,?sj*. (17)

544 RUMBODGHADIMI

[P), can be found by substituting j2 for jl in where

[I3 (i=t,2,3,
equation (15). Assembling (I 2) and (16) the equations
of motion of the jth element are (‘) a&
‘“[K’], = . . .6) (31)

[J] Sj= T/Vj+F;+F~-F;+~j-F;, (18)

and is usually referred to as the tangent stiffness
where matrix. The Wilson-theta numerical integration algo-
[%I : PI rithm can be used to predict the displacements at time
&I= -----;_-___ (19) t + T where z = 8. At, At = time increment and
f? N 1.4. The acceleration at time c + 7 is then given
[ to1 : IMj*l
I
by
[Mji]=(m”+m~)ji[C], i= 1,2 (20)
@+=)S= c,As - @S - c$‘)S. (32)
[O]= null matrix
Velocity vector
vi = [Sj, -+,,r (21)
(I+r)S = c,AS - c$‘)S _ &$ (33)
F; = pi, F;]’ (22)
and displacements

F;=@,F,:]’ (23) ft+r)S = (‘IS + AS, (34)

Fj” = F; 3F;12 = [O,0, I&:.,3O,O, Wjz]’ (24) where a, = 6/t*, a, = 3/r, a2 = 617, a3 = 2, a, = t/2.
Substituting for (‘+‘)S from (32) into (29) we
8, = [Qj, -Qjl’ (25) obtain

[a0I - KTlj AS, = Bj (35)

q = R t QIT (26)
and
sj = [Sj, 9 s,21Ta (27)
pi= “‘J(S) + [J.](a$“Sj + aS”Sj) +(‘+‘)F;
In order to be able to carry out the numerical
integration of the nonlinear equations of motion (18) +ft+~)F!_(‘+~‘)FW+(f+~)~j_(‘+‘)F~, (36)
i i
the displa~ment dependent tension term is expressed
as
Note that the right hand side of (36) contains terms
which cannot be readily determined at (t + r) and
q vj =4(s). (28) hence an iterative solution is necessary. For a static
analysis the terms corresponding to cyclic wave loads
The equation of motion at time (r + z) is then written must be set to zero.
as Using eqns (31), (21) and (4) the 36 elements of the
tangent stiffness matrix can be obtained (these are
[1;]”+I) Sj =A(S + AS) + (‘+“FjD given in the Appendix).
In finding q, from the cable stretch, accuracy can
+(r+f)F!_(I+r)F; be lost due to multiplication of a large quantity
I
(EA/L,O) by a small value (Lj - L,O). This can be
+tr+r)Bj_tt+r)F; avoided by writing

(‘1T.I = (‘--1)T./ + ATI’ (37)

(j--1,2 ,..., N), (29)
Using (6) and (3) and neglecting small quantities such
where N is the total number of elements and AS are as the high powers of displacement increments, we
displacement increments. The weight vector Fy can be have
time dependent due to variations in wetted length of
elements close to SWL.
Noting that each element has six DOF, s’, s*,
s’. . . s6, and using a Taylor series expansion we can
write

h(S + AS) = “‘&(S) + ‘“[K*]ASj, (30) +“‘(Zc - Z;l )fAZ,z - A& 111 (38)
 Algorithm for analysis of flexible marine risers 545

where AX,, , AXfl, A y/, , AYn etc. represent the

increments in nodal displacements.

Fluid loading on the riser

We consider regular waves and current travelling in , SELEMENT j
arbitrary directions at angles p and y relative to the
global X axis as shown in Fig. 3. Fluid loads are
calculated at the instantan~us position of the riser
from Morison’s formula in three dimensional vec-
torial form. The fluid inertia force at node j 1 can be
expressed as

F;, = C,,,pnD* Lj[tj, - (i;, .sj) ljj]/8, (39)

‘1
where C,,, is the inertia force coefficient. The fluid
particle acclerations are obtained from linear wave Fig. 4. Change of curvature and bending moments.
theory as

given by

and

(41) in which vcj, is the current velocity vector at jl, gj,

defines the structural velocities and vi! are the wave
(42) induced fluid velocities given by

(43) vjl = l”.rvuy,uzlj’; (50)

where (0,),I= kg, cosP (51)
cash k(d + Z,,) (QI = kg, sin
hr=~~H
sinh (kd)
w P (52)

(53)
si~~(~+Zj,)
h*=+H (45)
sinh (kd) Similar expressions can be used for node j2. We also
note that if Z,, ) 0 then node jl will lie outside the
g~=co@k(~,sin~ +X/,cos~)-mot] (46) fluid region and hence the fluid loading on the jth
element must be adjusted. For partially submerged
g, = sin[k( Yj, sin p + Xj, cos IL) - of]; (47) elements near the free surface the length Lj is replaced
by the actual wetted length with respect to SWL.
o = wave frequency, H = wave height, k = wave Fluid drag and inertia force for the jth element in
number and d = water depth. (36) can be written as
Fluid drag force at node jl is given by

F$ = PCDDV, IVrl LjI4, (48)

where C, is the drag coefficient of the riser and V, is

the relative velocity vector normal to the element
Shear forces and bending moments
Y
An approximate method of allowing for shear
force and bending moment is considered by reference
to Fig. 4. The bending moment at node m and
adjacent elements j and k is obtained from beam
theory as

I&,, = EI R,b,,,, (56)

where El is the flexural stiffness of the riser and

Fig. 3. Wave and current directions. curvature R,,, is estimated from the change of slope
546 RUMBODGHADIMI

between the jth and kth elements thus: where C, is the inertia coefficient for the buoy and
0, = fluid acceleration at buoy:
Rm=2(sk--j(/(L,+Lj). (57)
f7*= [tibcos y, ziibsin y, db]* (70)
The unit binormal vector b, is given by
rib, db = horizontal and vertical fluid acceler-
b, = s, x n, (58) ations at buoy respectively.

where skj is the unit tangent vector at node m The fluid loading on each buoy, their structural
approximated by a weighted average as follows: mass, added mass and the buoyancy forces must be
added to the relevant elements in the assembled
Skj = (L,Sj + Lj,S,)/(Lk + Lj) (59) global arrays.

Sea bed contact for catenary risers

and the unit normal vector n is given by
Variable contact length of the riser segment
n=(S,-Sj)/(S,-Sj(. (60) between the buoy and the well head is depicted in
Fig. 1. Neglecting friction forces the ground contact
Substituting from (58), (59), (60) and (4) in (56) we can be modelled by the application of an upward
obtain thrust at nodes near the sea bed. The upthrust must
overcome the weight of the riser and other dynamic
M, = 2EZ R,/(Lj + L,)2, (61) loads once a node reaches the bottom. A sudden
application of force must however be avoided to
where

iCLk9j + LjPk)(ik - 4) - CLk2j + L~ik)(Pk- i;>l

R, = [(L& + Lj,i,)(& - 2,) - (L,Zj + r;,a,)(i, - ij)] . (62)

[CLki,+ Ljak)($k-iJ - CLki;+ Lj9k)(2k
- ij)l

The shear force at element-j can be obtained from the

change of bending moment along the element, i.e. prevent numerical instabilities during time domain
simulation. Thus the force can be applied gradually
Ql= sj X CM,*
- M,II/L,. (63) in the manner shown in Fig. 5 where upthrust is zero
for nodes far from the bottom. As a node reaches the
Fluid loading on buoy(s) sea bed at 2 = -md the force will increase smoothly
Assuming a spherical buoy of diameter Db the fluid to the value required at the bottom where the
upthrust equals the downward loads of magnitude,
drag force is calculated from
say nw, where n is a constant and w = weight lumped
at the node. The variation of upthrust is expressed as
F: = G&‘; V,/V,lP~ (64)

where C,, = drag coefficient of buoy and V, is the F, = nw[l - tanh(jZ + tt)1/2, (71)
relative velocity vector found from
where B and q are found by imposing some conditions
V,, = [VI%,~yb?~z51r (65) such as F,,=nw at Z = -d and F,=O.O5nw at
Z = -md as shown in Fig. 5.
v,,=(v,cosr +ubcos~ -a,) (66)
Natural periods and mode shapes
Vyb= (V,, sin y + u, sin p - J&,) (67)
Natural periods are of some value in flexible riser
vrb = (Vb- -a, (68) analysis where the possibility of vortex induced oscil-
lations, say due to current, exists. They are also useful
in which in interpreting the simulation results if wave periods
and a natural period are close.
Veb= current velocity at buoy
The natural undamped modes and frequencies are
I+, vb = horizontal and vertical wave particle obtained by neglecting external and position de-
velocities at buoy, respectively pendent terms (except tension) on the right hand side
&,, JJ, i, = structural velocity components of buoy. of eqn (18). This gives the nonlinear equation of
motion as
The fluid inertia force is given by

F’B= C,,,, pnD; VJ6, (69) [Z]s = f(S) - F”. (72)

 Algorithm for analysis of flexible marine risers 547

UPTHRUST, Fu

Fig. 5. Sea bed contact of catenary riser.

For small amplitude oscillations about the static in allocated memory space and in a reduction in
equilibrium position we can write computing time.
Algebraic equations involving increments of nodal

[I
displacements are solved using Crout’s method
[I] (S + AS) =f(S) + ; AS - F, (73) (see [I 3]), which is a variation of the Gauss elimi-
nation technique. In this method the decomposition
which simplifies to of coefficient arrays into upper and lower triangular
forms can be effectively carried out using a vector dot
[I] AS - ]Kr] AS = 0, (74) product subroutine.
Time domain simulation is started by assuming an
in which 0 is the null vector. initial stationary position for the riser (i.e. Sj = 0,
Assuming simple harmonic motions (74) becomes j= I,2 , . . . , IV). Initial riser tension is also set to zero.
the standard eigenvalue problem The incremental displacements are then computed
from eqn (35) after which nodal coordinates, veloci-
[[K*] + [1]w;] AS = 0, (75) ties and accelerations at time (t + 7) are estimated.
These are substituted in the right hand side of (36) and
which enables the mode shapes and natural fre- increments of nodal displacements are recalculated
quencies o, to be computed. until convergence is achieved. Note that the tangent
stiffness matrix must be updated regularly.

COMPUTATIONAL APPROACH NUMERICAL RESULTS

(a) samurai periods and mode shapes of a suspended

Assembly of element matrices into global arrays is
carried out using a compact ‘profile storage’ tech- chain
nique given by Taylor [13] which takes advantage of Simulation results were validated against the
storing only the non-zero terms of a banded matrix theoretical solutions of Saxon and Cahn [I I] for free
in a vector array. This results in a considerable saving oscillations of suspended chains. Figure 6 shows the

u kY,a
-___---
I
MODE MODE
2
-----STATIC HJUIlIGRIUM
POSITION
10 ftfMfNTS

Fig. 6. Suspended chain model.

548 RUMBOD
GHADIMI

Table 1. Natural periods of a suspended chain mathematical model consisting of 10 elements, the
Natural period (set) static equilibrium position from computer results and
In-plane two of the in-plane mode shapes. Saxon and Cahn
Mode No. Saxon and Cahn [I l] Present study give the natural periods in terms of chain length and
1 9.55 10.07 the angle that end points make with the horizontal
2 5.57 5.80 which was found to be about 66” from simulation
3 3.93 4.24 results.

YESSEt
2

L -I
I
\ SW1
6 -
X
9
Q
Q
t Q
4
a a
a a
;\ RISER \.

1 SEABED

- -- - STATIC EQUlLlSllUM POSITION

MODE I MODE 2

Fig. 7. In-plane mode shapes for a steep S riser configuration.

EA= l.53Sw10° N
El = 2.086~10’ NM*
MASS = 10.7 KS/M
LENQTN-350 M

20 ELEMENTS

Fig. 8. Hanging riser configuration.

 Algorithm for analysis of flexible marine risers 549

Table 2. Comparison of static deflection coordinates for riser shown

in Fig. 8
X(m) Y(m)
Present Present
Node study Ref. [8] Theory study Ref. [8] Theory

1 0.0 0.0 0.0 0.0 0.0 0.0

2 1.62 1.62 1.63 -4.62 -4.73 -4.73
3 5.09 5.10 5.14 - 13.77 -14.11 - 14.09
4 13.20 13.27 13.38 -31.58 - 32.36 - 32.30
5 23.55 23.87 23.96 -48.19 -49.32 -49.25
6 37.29 38.20 38.09 -62.13 -63.27 -63.30
7 55.29 56.88 56.48 - 69.80 -70.40 -70.56
8 74.99 76.56 75.85 - 66.47 - 66.84 -66.79
9 90.83 92.58 91.45 - 54.27 - 54.73 - 54.45
10 102.61 104.20 103.07 -38.10 -38.52 -38.22
11 111.64 113.06 112.00 - 20.25 -20.59 -20.34
12 118.87 120.09 119.15 -1.61 -1.87 - 1.66
13 124.87 125.89 125.09 17.47 17.27 17.43
14 129.98 130.81 130.15 36.81 36.33 36.78
15 134.42 135.08 134.55 56.31 56.20 56.29
16 138.35 138.86 138.45 75.92 75.84 75.90
17 141.87 142.23 141.94 95.61 95.55 95.60
18 145.06 145.28 145.10 115.35 115.32 115.35
19 147.97 148.07 147.99 135.14 135.12 135.14
20 149.34 149.37 149.34 145.04 145.04 145.04
21 150.00 150.00 150.00 150.00 150.00 150.00

Table 1 summarizes the results obtained. For better (d) Comparisons with model tests
agreement, especially at higher modes, the number of The experimental results of Hartnup [12] for two
elements in the model must-be increased. steep S model risers at l/SO scale were used for
Figure 7 shows the results of the programme for comparisons with computer simulations:
two of the inplane mode shapes for a ‘steep S’ flexible (i) Riser 1 (with fixed top end). Diameter = 6 mm,
riser configuration. total length = 3.97 m, structural mass = 0.0836 kg/m,
EA = 3.012 x 103N, EI = 9.88 x 10-5Nm2; nominal
(b) Static analysis of a hanging riser
dimensions (see Fig. 12) a = 1.48 m, b = 2.58 m, c =
Comparisons were made with the results of 0.73 m, buoy diameter = 0.1 m, buoy mass = 0.1 kg,
McNamara et al. [8] and with theoretical values wave height = 255 mm, wave period = 1.69 set, water
(accurately calculated using standard catenary equa- depth = 2.13 m.
tions) for a riser, details of which are given in Fig. 8. (ii) Riser 2 (with top motions). Diameter = 7 mm,
Results for the nodal coordinates, riser tensions and total length = 3.68 m, structural mass = 0.087 kg/m,
support reactions are presented in Tables 2 and 3. EA = 3.6 x lo3 N, EZ = 1.61 x 10m4 Nm’; nominal
Figure 9 shows the results for distribution of bending dimensions (see Fig. 12) a = 1.59 m, b = 2.33 m,
moments along the riser together with the plots c = 0.72 m, buoy as in (i), wave height = 323 mm,
from two other finite element programmes mentioned wave period = 1.98 set, water depth = 2.13 m, vessel
earlier in the introductory section. surge (in X direction) amplitude = 65 mm, vessel
heave amplitude = 53 mm.
(c) Static and dynamic analysis of a freely hanging
catenary riser touching the sea bed
A freely suspended catenary riser, given in Fig. 10,
Table 3. Comparison of support tensions and reactions for
was modelled and the simulation results for the static
riser shown in Fie. 8
case were compared with the standard catenary re-
sults. These are presented in Tables 4 and 5. Present
Variable study Ref. [8] Theory
The approximate model for sea bed contact was
tested for a case with dynamic loads induced by a Tension at
node 1 (kN) 34.81 37.76 35.92
storm wave of height 29m and period 17 set plus
Tension at
prescribed horizontal and vertical motions of a vessel node 21 &N) 86.65 92.17 87.83
at the top node. The results for vertical displacements Horizontal reaction
of nodes near the sea bed are given in Fig. 11, at node 1 (kN) 11.49 11.57 11.44
showing the take up and lay down of the riser at the Horizontal reaction
at node 21 (kN) 11.51 11.92 11.44
bottom.
550 RUMBODGHADIMI

0 200 400 600

BENDING MOMENT (Nm)

Fig. 9. Static bending moments for hanging riser.

Table 4. Comparison of static coordinates for catenary riser shown in

Fig. 10

X(m) Z(m)
Node Present study Theory Present study Theory

1 255.0 256.9 25.0 24.1

2 253.5 255.2 10.0 9.2
3 251.5 252.9 -9.8 - 10.6
4 250.4 251.6 - 19.8 -20.5
5 249.3 250.4 -29.7 - 30.4
6 247.1 247.7 -49.6 -50.3
7 244.7 244.9 -69.5 -70.1
8 240.8 240.4 -99.2 -99.7
9 235.6 234.5 - 133.9 - 134.2
10 228.6 226.8 - 173.2 - 173.5
11 217.2 214.9 -221.9 - 222.0
12 201.1 199.3 -269.3 - 269.5
13 177.4 176.5 -313.3 -313.9
14 156.8 156.6 -335.2 - 336.2
15 139.5 139.5 -345.2 -346.5
16 129.9 129.9 -347.9 -349.2
17 124.9 125.0 -349.0 - 349.9
18-24 lie on the sea bed

Table 5. Comparison of results for riser shown in Fig. 10

Variable Present study Theory

Top tension (kN) 165 167

Length of riser
lying on sea bed (m) 120 121.5
Top angle relative to
horizontal (degrees) 84.6 83.7
 25yr255 m -,rd

15
6 RISER LENGTH 540 METRES
EA= 3.27x10* W
El= 3.4 x IO4 NM2
MASS= 102 KS/M
DIAMETER= 0.2766 M
WATER OEPTtl=350 M
23 ELEMENTS

SEA BE0 NOOE 24 7

23

Fig. 10. Freely suspended catenary riser.

___-_- -____-_-_c__---- _m----

rg;___I_;:---
ul

-380-
N
X-360.
S
= -340 -
-320-

-3807
Rseo.
G40.
:
-320+ I 1 I ?
TIME
L- 6 SNJ
Fig. 11. Vertical displacements near the sea bed for riser shown in Fig. IO.

551
552 RUMBOD GHADIMI

‘-=-

- EXPERIMENT

.-
-

I- b
STATION
\n
3, (LIUOY

F-
L BASE I J---J----J---u--J

Fig. 12. Model riser configuration and the static equilibrium position of Riser 2.

s
w
-8O-

H -SIMUlRTION ----EXPERIMfNl 1121

80-

-8O-
I

Fig. 13. Horizontal and vertical motions of a point near SWL for model Riser I.
 Algorithm for analysis of flexible marine risers 553

-SlMUlbTlON ----fXPERIMtN1 [I23

0.06

I I
I
L 1 set -J TIME

Fig. 14. Variations of dynamic top tension for model Riser 1.

Due to relatively low values of Reynolds number motions were found to be relatively small and an
for the fluid flow (in the order of 1000-2000 near accurate comparison without scaling errors was not
SWL) a C, = 1.5 was chosen for both risers. C,,, was possible. The dynamic base tension for Riser 2 was
taken as 2.0. not measured in the experimental arrangement.
Figure 12 also shows the static equilibrium position
of Riser 2 given by simulation and experimental
models. CONCLUSIONS
Comparison of results for the horizontal and verti-
cal displacements of a node near SWL for Riser 1 is A simple and efficient algorithm for the three
given in Fig. 13 and variations of dynamic tension at dimensional analysis of flexible risers has been
the top and at the base of the riser can be seen in presented. The algorithm is capable of handling
Figs 14 and 15 respectively. various riser confi~ations including those in which
Figures 16-18 show the time histories of nodal catenary segments come into contact with the sea
displacements for Riser 2 at the stations indicated in bed.
Fig. 12. Amplitudes of the experimental and simu- Comparison of results with analytical solutions,
lated dynamic top tension for this case were found to data from other finite element programmes and a set
be 0.269 and 0.295 N respectively. Vertical buoy of experimental results showed very good agreement.

-SlMUtAllON ---- ~~P~Rl~~Nl

WI
0.3-

c 1
t I
TIME
- lsec I-I

Fig. 15. Variations of dynamic base tension for model Riser 1.

 DISPLACEMENTS AT STATION 1 (mm)
DISPLACEMENTS AT STATION 2 (mm)
VERTICAL HORIZONTAL VERTICAL HORIZONTAL
crl
I.
qs
s
 Algorithm for analysis of flexible marine risers 555

I I

150. -slMUl~TIon ---- IIPERIMENT

[12]
mm

:
-150-
, I
TIME
- 1 set -

Fig. 18. Horizontal motions at buoy for model Riser 2.

Acknowledgemenr-The author wishes to thank Mr G. C. analysis. In The Finite Element Method (Edited by 0. C.
Hartnup of the Heriot-Watt University for his kind permis- Zienkiewicz), 3rd Edn. McGraw-Hill, New York (1977).
sion to use the results of his model tests on flexible risers.
APPENDIX: ELEMENTS OF THE
TANGENT STIFFNESS MATRIX

REFERENCES With reference to eqns (31) and (35) the elements of the
tangent stiffness matrix [K’], are given by:
1. T. S. Walton and H. Polachek, Calculation of nonlinear
transient motion of cables. J. Math. Comput. 14, (Kl),= -(n,+EIL,)+@,?-X,,)21L:
(1960).
2. T. Nakajima, S. Motora and M. Fujino, On the dy- (KT, )I = c Cx,r - X,1)( yj2 - yji)lL:

namic analysis of multicomponent mooring lines. OTC (KL), = t($2 - x,l)(z,2- zjl)lL:
Paper 4309, Proc. 14th Annual Offshore Technology
(G),= -(G),
Conference, Houston, TX (1982).
3. A. H. Peyrot and A. M. Goulois, Analysis of cable (KL)j= -(KG),
structures. Comput. Struct. 10, (1979).
(KT,)j= -(KL),
4. A. H. Peyrot, Marine cable structures. J. Strucf. Div.,
ASCE 106, 2391-2404 (1980). (K~),=-(~,+~lLj)+~(Y,2-Y,,)2/L:
5. D. L. Garrett, Dynamic analysis of slender rods. J.
Energy Resources Technol., ASME 104,302-306 (1982). (G)jzt(Y,2- ~l)(z,2-z,l)lL~

6. A. E. H. Love, A Treatise on the Mathematical Theory (K&)j= (KL)j= -(KL),

of Elasticity, 4th Edn. Dover, New York (1944).
KS),= -(G),
7. R. P. Nordgren, On computation of the motion of
elastic rods. J. appl. Mech., ASME 777-780 (1974). (G), = -(G),
8. J. F. McNamara, P. J. O’Brien and S. G. Gilroy,
(KZ),= -(Aj+tlL,) +L(zj2-z,I)2/L:
Nonlinear analysis of flexible risers using hybrid finite
elements. Proc. 5th International Offshore Mechanics (G),= -(fG),
and Arctic Engineering Symp. Vol. 3 (1986).
(KT,), = -(KL)j
9. J. F. McNamara and H. D. Hibbitt, Numerical analysis
of flexible pipes and risers in offshore applications. Proc. (KT,)j= -(KT,),
1st Offshore Mechanics and Arctic Engineering Specialty
Symp. on Offshore and Arctic Frontiers, New Orleans (KL),=(KL)j

(1986). (G), = (G),

10. A. T. Ractliffe, Dynamic response of flexible catenary
risers. Int. Symp. on Developments in Floating Pro- (G)j = (KT,)j

duction Systems (1984). (G), = (G),

11 D. S. Saxon and A. S. Cahn, Modes of vibration of a
suspended chain. Q. J. Mech. Appl. Math. 6, (1953). (G), = (G),

12 G. C. Hartnup, Model basin tests on marine risers. (G), = (G),

Reports on flexible riser tests, Dept of Mechanical
(KL),=(Kk), for m #n
Engineering, Heriot-Watt University, Edinburgh
(1986). hi = EA/Lj
13. R. L. Taylor, Computer procedures for finite element E= -EA.