F. B. Seyed
John Brown Engineers & Constructors Ltd, 20 Eastbourne Terrace, Paddington, London W2 6LE, UK
& M. H. Patel
Department of Mechanical Engineering, University College London, Torrington Place, London WC1E 7JE, UK
ABSTRACT Derivations are presented for calculation of pressure and internalflow induced forces on flexible risers and other curved pipes using a mathematically rigorous approach. Approximate and exact methods are presented for calculation of pressure forces on straight and curved pipes in two dimensions. The mathematical identity of these equations with thosefor effective tension is illustrated. The force arising from the flow of an internal fluid of constant density is then calculated and combined with those for pressure forces in derivation of the catenary equations including pressure and internal flow terms. It is shown that internal flow contributes a new term to the expression for effective tension. These governing equations are then reduced for the specific cases of simple catenary, steep-S, lazy-S, steep-wave and lazy-wave risers. In each case, the solution method has been presented and the goveming equilibrium and geometric compatibility conditions cited. Key words: flexible riser, pressure force, internal flow, slug flow, catenary equations cables, umbilicals, flowlines.
121
Marine Structures 0951-8339/92/$05.00 1992 Elsevier Science Publishers Ltd, England. Printed in Great Britain.
122
In the absence of other external and internal forces, the equilibrium of a pipe is determined by its self-weight. The profile of a suspended pipe of negligible bending stiffness and uniform self-weight is given by the classical catenary equations. The presence of additional internal and external forces considerably modifies the governing equations. The forces of concern are those due to self-weight, internal and external pressure forces (buoyancy), steady current drag and steady internal flow. The effect of internal and external pressure forces on the geometry of the pipe has, in the past, caused a great deal of confusion. The load system due to these forces has two different influences on the pipe. Firstly, it modifies the effect of pipe self-weight and, secondly, controls the pipe curvatures in a way similar to the tensile forces in the wall of the pipe. Conventionally, these forces are calculated through a buoyancy analogy where the buoyancy force given by the Archimedes principle is modified to exclude the effects of the end pressures, thereby allowing for pipe continuity. This approach is discussed in detail in Refs 1--4. This is based on the assumption of small curvatures and, therefore, reduces in accuracy with increasing pipe curvature and water depth. In such cases, maintaining the accuracy of finite element models requires a rapid reduction in element size. As a result, more accurate expressions are needed to accommodate larger element sizes and help reduce computer storage and time requirements. The influence of internal flow has, in the past, been ignored in all flexible riser analyses. The profile of a flexible pipe is determined by its effective weight and the tensile forces that are generated along the pipe to support this weight. The response of the pipe to internal and external excitation is resisted through its structural stiffness and its tensile forces which confer a geometric stiffness to the pipe. The internal flow forces are later shown to interact with the pipe through changes to its tensile force distribution. Hitherto, the consequences of this interaction have not been explored in detail. With recent advances in prediction of mechanical behaviour of flexible pipes and development of models for assessment of long term wear and fatigue of flexible pipes, it has become essential to investigate the modes of interaction of internal flow with the pipe in detail. In this paper, the governing equations of equilibrium for a flexible riser have been derived from first principles and adapted for each of the standard flexible riser configurations; namely, simple catenary, steep-S, lazy-S, steep-wave and lazy-wave. The resulting equations include pressure and internal-flow terms and provide a framework for rapid
123
assessment of different configurations for the given application. Absence of internal flow effects could lead to total misprediction of inter-layer forces a n d hence fatigue and wear of the pipe.
".,.
124
F. B. Seyed, M. H. P a t e l
?, Density of fluid 0 Angle of normal to the centre of element from x-axis Angle a r o u n d the circumference of pipe The area of the element on the surface of the pipe is given by: dA = R d x d 0 Pressure at the axis of pipe, distance x from the pipe mid-point is: Pc = Po - yxcos0 (2) (1)
For this position, the pressure on an element of surface area is given by: P = Po - ~,xcos0 -
yR sin~ sin0
(3)
Considering external pressure first, the force on the element of area dA is given by: dF = PdA (Po - yx cos 0 - ?'R sin~ sin0)Rd~ dx
= -
(4)
The components of this force in the coordinate directions are given by: dFx = - d F sintp cos0 dFy = - d F sin~ sin0 dFz = - d F cosO (5) (6) (7)
The total force in each direction is calculated by integration of the above equations. We first have:
L/2 2n
Fx = f
,x = -L/2
f
~ 0
- d F sin~ cos0
(8)
(9)
F y = -.x= f
(10)
giving:
Fy = rtR2yL sin20
(11)
M a t h e m a t i c s o f f l e x i b l e risers
125
Finally:
L/2 ~x = - L / 2 2n = 0
The above equations are valid for external pressure. For internal pressure, a reversal of signs occurs and y, R and Po are replaced by their internal equivalents. Using subscripts i and o to denote internal and external quantities respectively, the total pressure force on the element is given by: (l'iAi - yoAo)L sin0 cos0 Fy = (r-,4i - ~'oAo)L sin20
Fx =
(13) (14)
Note that the resulting expressions are independent of fluid pressures and are hence depth invariant. Further, the expressions are frame variant since the positive y-axis has been chosen as the axis of pressure gradient. The above derivations present a mathematically rigourous derivation of the buoyancy force experienced by a straight pipe immersed in a fluid. The effects of end pressures are ignored as the pipe is assumed to be continuous in geometry. This approach is equivalent to a mathematical re-statement of Archimedes principle for a continuous pipe. These equations are of direct application to analysis of columns of semi-submersibles and members of jacket structures as well as marine pipelines.
2.2 Approximate equations for a curved pipe
Consider the element in Fig. 2 which is subjected to external hydrostatic pressure acting on its outer surface and excluding the ends. The resultant force acting on the element may be calculated through integration of pressures over the surface area of the pipe. Using the following specific notation: ds P Pb Pt 0 O Length of element measured along the arc Pressure at centroid of base Pressure along the base circumference Pressure along the top circumference Element angle measured from the x-axis Angle around the circumference of pipe
the arclength of the shaded area may be written as: ds = (r + R cosq0d0 (15)
126
Pb
t:> X
Y
Fig. 2. Curved pipe element for approximate analysis.
Using P to denote the pressure at the centre of the base, the pressure along the circumference of the base may be found using: Pb = P - yR cos0 sin0 (16)
The pressure at any point along the top edge of the element is given by:
Pt = Pb -- Y sin0
ds
(17)
using sin0 = dy/ds. The area of the shaded section of the element is given by: dA = R d 0 d s The force acting on dA is: dF = 1 (Pb + Pt) dA (18)
(19)
127
(20) (21)
dF =
(22)
Resolving this force into components and integrating them between the limits (0 ---, 2n) yields expressions for the resultant forces in the x a n d y directions. The force components in the x direction is:
217
Fx = f~ = 0 - d F sin0 sin0
Fx = - n R : [P + 1,r (cos0 - sin0 dO)] sin0 dO
(23) (24)
Fy = f - d F cos0 sin0 J# =0
Fy = nR: [P + 1,r (cos0 - sin0 d0)] cos0 dO
(25) (26)
The above results consider external pressure only. If internal pressure is also included, the final expressions for the resultant pressure forces in the x and y directions become:
Fx = [P-r,4i- PoAo + r ( 1 , ~ i - 1,oAo) (COs0 -- sin0 dO)] sin0 dO (27) Fy = - IPiAi - PoAo + r (1,iAi - 1,oAo) (cos0 - sin0 dO)] cos0 dO (28)
where subscripts i a n d o are used to denote internal and external values of fluid pressure, density or radius, respectively. The above derivation follows the approach suggested by Young et al. 5 who through algebraic errors miscalculated the equations for the vertical force component. The
128
~Po"dPOAo
Upthrust
- ),o A odS
T+dT
T+dT
Te+ dTe
t
Dry Weight
~sAsdS
T e = T " Pi A i + PoAo
Wc = ~sAs+~Ai- rA o
Fig.
above results provide corrected results a n d include a change of axes to the standard Cartesian system. The same results m a y be obtained using a global equilibrium analysis of the fluid plug surrounding the pipe. Using the equivalence of forces shown in Fig. 3, the total b u o y a n c y force acting on the pipe should exactly equal the sum of forces due to end pressures a n d those resulting from integration of fluid pressures over the surface of the pipe. In the absence of other forces, the equilibrium of the fluid plug requires that the resultant of the end pressures balances the resultant of pressures acting n o r m a l to the surface of the pipe. The resultant force in the x direction due to e n d pressures is given by:
fx = PA cos0 dO - (P - ~,r sin0 dO)A cos(0 + dO) = PA cos0 dO - (P - ~,r sin0 dO)A (cos0 - sin0 d0)
= A sin0 dO [P + r r
(29)
129
It is noted thatfx andfv are exactly equal to F.,. and F,. but act in the opposite direction which confirms the equilibrium of the pipe. Furthermore, it may readily be shown that if the pipe plane makes an angle a to the vertical, an additional cos(a) multiplier is introduced into eqns (16) and (17) which perpetuates through the equations but leaves the final form of results unchanged. If the final expressions for forces are then transformed to local coordinates in the plane of curvature, the same expressions as eqns (27) and (28) result. This confirms that the concept of effective tension remains valid in three dimensions.
2.3 Exact equations for a curved pipe The method proposed in the previous section used a linear interpolation function for calculation of fluid pressures along the arclength ofthe pipe using the exact values at its ends. Although this provides very accurate results for rigid and flexible riser applications, its accuracy diminishes with increasing curvature of the pipe. For pipes of large curvature and single sections of pipe-bend, more accurate expressions are thus needed. In this section, the previous formulation is extended to provide an exact integration of fluid pressures over the surface of the pipe. The sole assumption here is that the centre-line of the pipe is a circular arc. The following notation are specific to this derivation: P Pc Po a 0 Pressure at the surface of the element Pressure at the axis of pipe at given arclength Pressure at centre of element (centroid of half-arclength) Angle of a position on the pipe surface from centre of curvature Angle of centre of element from centre of curvature Angle around the circumference of pipe
With reference to Fig. 4 the area of the element on the surface ofthe pipe subtending an angle a from the centre of curvature is given by: dA = R (r + R sin~) dq~ da (31)
Pressure at the axis of the pipe at this position can be expressed in terms of the pressure at the centroid of the element:
Pc = P o - y r
(sina - sin0)
(32)
For the same position, pressure on an element of surface of the pipe subtending an angle ~ from the axis of the pipe is given by:
P = Po - yr
(sina - sinO) -
yR
sina sin~
(33)
130
F B. Seyed, M. H. Patel
Considering external pressure first, the force on the element of area dA is given by:
d F = P dA
= [Po - ?'r (sina - sin0) - y R sina sinq~] (r + R sin0 R dO dq~ (34) The components of this force in the coordinate directions are given by: dF.,. = - d F sin~ cosa dF,. = - d F sinq~ sina dF: = - dFcosq~ (35) (36) (37)
The total force in each direction is calculated by integration of the above equations. First: F,.= J,,f'+a/2)= (o- ao/2, fi"= o - d F sin~ cosa (38)
which after lengthy manipulation gives: F,- = - 2 n R 2 Po sin -~- + y r sin0 sin y - sin dO Similarly:
-
ao
(ao
)]
cos0
(39)
d F sinO sina
(40)
-a
(O-dO/2)
+ ~ y r (sin dO - dO)
'
l
- dFcosg~ - 0
(41)
Finally:
(0 + dO/2) 2n
F, = f
-a =
f
(O-dO/2).t~
= 0
(42)
The above derivation applies to external pressure. For internal pressure, the negative sign changes to positive a n d y, R a n d Po are replaced by their respective counterparts. Using subscripts i and o to denote internal and
131
Po ~ . d F
de
external quantities, respectively, the total pressure force on the element is given by: F,. =
2[(PiAi-PoAo)+r(),iAi-~'oAo)sinO(1-2cosd-ff)]
(43)
(44)
132
F. B. Seyed, M. H. Patel
It may readily be shown that the definition of angle 0 in Section (2.2) corresponds to that in Section (2.3) according to the relationship:
(45)
with subscripts referring to the respective sections. Making the above substitution and using the small angle assumptions for sines and cosines of(0 + d0/2), it is easily shown that eqns (43) and (44) revert to eqns (27) and (28), respectively. The above results correct the work of Chakrabarti and Frampton 6who used a similar approach but obtained erroneous results for the force resultants. This correction relates to eqn (63) in the work of Chakrabarti and Frampton 6 which corresponds to eqn (44) of this paper. The validity of the equation may be checked for a closed torus ring section in the vertical plane where the resultant vertical force reduces to the Archemedian buoyancy force given by the product ofthe volume and specific weight of water.
3 INTERNAL FLOW FORCES IN TWO DIMENSIONS The flow of a slug within a curved pipe results in forces whose magnitudes are primarily dependent on the nature of the flow, densities of constituent parts and the velocities with which they travel through the tube. The uncertainties associated with the flow prevent an accurate modelling of the large spectrum of flow regimes that are known to exist. The steady component of internal-flow induced force is derived here. Consider the curved element of the pipe illustrated in Fig. 5 where: Pi P2 Ut U2 0~ 02 Pressure at lower end of pipe Pressure at upper end of pipe Input velocity at lower end Output velocity at upper end Angle at lower end Angle at upper end
Using/7,. and F,' to denote horizontal and vertical forces exerted on the fluid by the pipe, the horizontal force acting on the fluid plug is given by: P1A i cos01 - P2A i cos02 + Fx The total change in fluid momentum is: /9 iA iU2 cos02 - p iA i Ui2 cos01 (47) (46)
133
Ul' PI
)
11
(48)
(49)
(50)
(51)
The forces exerted by the fluid on the pipe are equal and opposite to Fx and Fy and are given by:
(52)
(53)
It is noted that the pressure terms are inaccurate in the presence of external pressure when the net force must be calculated through integration of inner and outer pressures over the curved surface of the tube. The present derivation overlaps with the pressure integration
134
F. B. Seyed, M. H. Patel
scheme described earlier. Care must, therefore, be exercised in considering a combination of slug flow and pressure calculation using pressure integration where the preceding method is employed. Analogy with eqns (29) and (30) shows that this overlap may be removed by eliminating the pressure and weight terms from eqns (52) and (53) to give the components of the internal-flow induced force as:
Fxs = piAi (U21 cos01 - U~2 cos02)
U~2sin01 dO]
(54)
Using the small angle assumptions and with Ui denoting the average internal flow velocity. Similarly:
Fy s --" p'r4i (U]I sin01 - U~ sin02)
(55)
4 GOVERNING EQUATIONS Consider the element of pipe shown in Fig. 6. The pipe is assumed to be of uniform cross-section and weight distribution along its length. The forces acting on the pipe, using the notation on the figure, are:
Fxi Internal pressure force in the x direction
External pressure force in the x direction Internal flow force in the x direction Internal pressure force in the y direction External pressure force in the y direction Internal flow force in the y direction Bending moment Drag force component along x-axis (per unit length) Drag force component along y-axis (per unit length) Shear force Tensile force Weight per unit length of pipe element
135
Vo Ho<
x T+ dT
__ T
O ~
ds
For small values of dO, the following approximations are valid: cos(0 + 60) ,~ cos0 - sin0 60 sin(0 + 60) ~ sin0 + cos0 60 (58) (59)
Replacing small changes in quantities by their equivalent differential operators and making use of the above approximations: - ( T sin0 -'- Vcos0) dO + dTcos0 + d V s i n 0 + Fxi + Fxo + Fxs
+ Nx6s = 0 (6O)
(61)
Multiplying eqn (60) by ( - sin0) and eqn (61) by (cos0), and adding gives:
T dO - d V + (Fyi + Fyo + Fys + ivy 6s - w 6s) cos0
-
(62)
136
F.B.
Seyed, M. H. Patel
Substituting from (27) and (28) for pressure forces and (54) and (55) for steady internal flow forces gives:
T dO -
d V - [PiAi - PoAo) + r(~'iai - ~/oAo)(cos0 - sin0 dO)] cos 2 0 dO - [(PiA i - PoAo) + r ( T i A i - ToAo)(cos0 - sin0 dO)] sin20 dO - piAi U~ sin20 dO - p#li U~i cos20 dO - Nx 6s sin0
+ (Ny - w)6s
cos0
(63)
Rearranging and dividing through by (ds) and noting that (r dO = ds) gives
( T + PoAo - PiAi - piAiUi:') dO ds d ~V - + ('/0,40- r,Ai)
(64)
which forms the polar version of the governing equations. Conversion to Cartesian coordinates may be carried out using the following identities: d0_ds ~--~,//d2y)[l+ (~xy)2]-3/2 (65)
coso= [1+
dV ds
m
dV~ dx ds
dV --cos0 dx
= dV dx
1+
(68)
After substitution, ignoring the ( - s i n 0 dO) term and rearranging the (w = ~%As) term eqn (64) becomes:
(T + P o A o - P i A i - P i A i U i 2 )
1 +
= 0
(69)
137
dO
= EI~
1+
(70)
dM
ds ~ EI-~-~
day
(71)
(73)
which is the Cartesian version of the governing equations for flexible pipes.
5 ANALYTICAL SOLUTIONS
5.1 Elastic catenary equations
In the absence of fluid drag and pressure forces and after division by (ds), eqn (64) becomes: dO T ds dV ds w cos0 -- 0 (74)
where d0/ds is the exact expression for curvature. The equation for the shear force and bending moment along the beam given by eqns (68) and (70) are now used to give:
T dO
ds
- EI~-
wcosO
= 0
(75)
Equilibrium of horizontal forces between points O and A in Fig. 6 yields: Ho = Tcos0 + Vsin0 (76)
138
F B. Seyed, M. H. Patel
Substituting for V and rearranging: T -- Ho sec0 - E I ds- tan0 Substituting this equation into eqn (75) gives: dO d20 dO d30 Ho sec0 ~ - EI-d~s2~ tan0 - El--d~ - wcos0 = 0 or more concisely: d [ d20-] dO E1 -~ _ sec0 d-~j - Ho seC0 ~ - + w = 0 (79) (78) (77)
which is the equation of equilibrium for a uniform beam under selfweight and includes flexural effects. If the flexural rigidity term (E/) in eqn (79) is ignored: seC0 dO w ds Ho (80)
where 0o represents the angle at the base of the pipe. The equation of equilibrium becomes: dO T d---s = w cos0 Substituting for dO/ds from eqn (80) gives:
T(s) = Ho see0
(83)
(84)
1+
~+tan0o
(85)
139
6So
= (1 + e ) 6 s
= 1+ ~-~ 6s
(86)
where so refers to the stretched length of the element. For small So the projections of the element on the axes are:
6 x = 6So COSO
(87) (88)
6y = 6So sinO
Substituting for 6s from eqn (86) and taking the limit as s --~ 0: ~ ds [ l + ( w ; + tanOo )~1 -''~ . . . .H .w
+ Ho EA sHo~
(89)
x(s)
[sinh-~ ( Hoo ws +
(90)
Similarly:
._(w,
ds ~
+ tanOo
)[[ (w,
1+ ~
+ tanO0
+~
.o]
] ws2
(91)
y(s)
= T
Silo
Ho [[ 1 + ( Hoo ws + t a n 0 )21~/2 - ( 1
tan0o
+tan20o) u2 + 2 ~
+~
(92)
Now, using eqn (86): dso _ l + H o [ (WHo )2],/2 ds ~ 1+ + tan0o which after integration gives: So = s + (93)
H2o r
,/ws
+ ws + tanOo
ws + tanOo
140
F B. Seyed, M. H. Patel
To obtain expressions for shear force and bending moment, expressions for curvature and its first derivative are required. These are given by:
dO
W [ l + ( WS
Hoo + tan0o
)2]-,/2 (95)
ds - Ho ~-~ = ~
(w)(ws
~+tan0o
)[ (ws
1+
~+tan0o
)C
(96)
At a given arclength, the angle of the pipe from the x-axis is: 0 = tan-'(W~o+tan0o ) (97)
Comparison of eqn (83) with eqn (64) shows that, in the absence of current loading and neglecting flexural effects, these are identical in form. Further, it is readily seen that using:
Taf = T + PoAo
- PiAi - piU2Ai
(98) (99)
Wef t = y s A s d- y i A i - y o A o
to replace T and w, it is possible to use all the equations derived in this section for calculation of the profile of a flexible riser subjected to selfweight, internal and external pressure forces and a steady internal flow. These considerations significantly simplify the computation of the static profile of a flexible riser and avoid lengthy numerical procedures. Tar and Waf are usually referred to as the 'effective tension' and 'effective weight' of the pipe. Once the pipe geometry has been defined, T, which represents the physical tension in the wall of the pipe, may be derived by subtraction of other terms. An interesting observation is that according to the formulation for effective tension, changes in internal and external pressures and densities or internal fluid velocity will not affect the pipe geometry but merely alter its wall tension distribution. Further, it should be noted that the pipe geometry and its stress state are independent of the internal flow direction.
141
flexible riser applications. These equations are obtained through substitution of a fLxed seabed angle into the elastic catenary equations developed earlier. Two modifications are introduced at this stage. First, the effects of axial elongations are ignored; this is permitted due to its small magnitude in cases of practical interest. Secondly, it will be assumed, hereafter, that the seabed slope is zero. This simplification is justified since seabed plains have typical slopes of the order of 5 with higher slopes restricted to hilly regions for which the use of catenary equations is not justified. For a zero seabed slope (0b = 0) in Fig. 7, eqn (90) simplifies to:
(100)
(101)
y(s) = - ~ 2 [ c o s h ( ~ o o ) - l ]
(102)
Squaring eqns (101) and (102), subtracting the former from the latter, and reducing we get: Ho _ s 2 - y 2 w 2y (103)
Y
Ho
I~
X
:',
142
F. B. Seyed, M. H. Patel
For completeness, it is noted that for a non-zero seabed slope, the above expression becomes: Ho s z _ y2 w -- 2(y s e C O b - - S tanOb) (104)
These expressions are of use in determining the top tension requirements in terms of the water depth and suspended pipe length. Equation (104) is equally applicable to flexible riser cases where the base angle is known. In pipeline design, the horizontal vessel pull is restricted by the vessel power and usually assigned a maximum which is within a safe margin of the maximum vessel capacity. Situations also arise where a damaged section of the pipe needs to be retrieved and replaced or other cases where adverse weather conditions force the operator to halt laying and lower the pipe end onto the seabed. The above two cases are normally referred to as pipeline pick-up and lay-down. Both these operations require the suspended length o f the pipe to be determined. Equation (103) provides this on rearrangement. The criterion during lay-down and pick-up is that the bending stress or equivalently the curvature at the seabed is kept constant. Equation (95) shows that this implies a constant Ho. For example, in the case of pipeline lay-down, having capped the pipe end on the vessel, a cable is attached to this section which is paid out keeping a constant horizontal tension. This corresponds to the design tension for laying the pipe. The vessel movements should clearly duplicate the reverse of those used in laying. Additionally, the total length of cable is equal to the suspended length of pipe given by eqn (92) which provides: Ho _ y cosOt w 1 - cosOt (105)
This form of the equation is the most suitable for pipe laying as it allows the determination of all variables in terms of quantities at the vessel attachment point. In a typical situation, the water depth is known and the required vessel pull is a constant design parameter which allows the required stinger angle to be determined from eqn (105). This may also be calculated from eqn (97). Given a length of pipe and its required horizontal and vertical offsets, the problem is now one of determining the length of pipe resting on the seabed. In Fig. 7, the seabed is assumed to be of constant slope 0b and the pipe of total length, s, which is the sum of the length resting on the seabed, Sb and the free length, s f. As illustrated earlier, for a given seabed slope, the pipe shape is a function of water depth only. This is because the
143
imposition of a constant 0b removes one of the variables required in solution. The horizontal top tension is easily determined using eqn (104) and allows the computation of the total horizontal offset generated. This may then be compared to the required value and Sb altered until the total generated offset is within the required tolerance of the given value. The mathematical statement of geometric compatibility may then be written as the following system of simultaneous equations: SbCOS0b + - ~
-x
= 0 (106)
S--Sb--Sf = 0 Ho _
W
(107)
sf 2 - - (Y 2(y s e c 0
- - Sb s i n 0 b ) 2 b -- sftan0b)
(108)
Note that eqn (108) is an alternative representation of the vertical geometric compatibility. Although the above equations may be reduced into one by substitution, the resulting expression does not have a closed form solution and must be solved iteratively. The iteration is exceedingly simple and requires incremental increases Of Sb until eqn (106) changes in sign to indicate that the solution has been overshot. Starting with the quantities from the previous step, Sb is now incremented in smaller steps and the procedure repeated until convergence.
144
F. B. Seyed, M. H. Patel
r/'//////////////////////////////////g
H-
TI sin0 = 0
(109)
where T~ refers to the tensile force at the top of the tether. Note that at the seabed, the tensile force becomes To (-- Tn - w l ; where l is the tether length as shown in Fig. 8) to allow for the weight of the suspended pipe. Vertical force equilibrium gives: Fb -- V - T1 cosO = 0 The inclination of the tether to the vertical is: 0 = tan -I F b H =V ) (111) (110)
For a given separation (x,y) between the buoy and the vessel and a constant length s, calculation of H and Vrequires solution of the catenary equations using an iterative procedure. This solution algorithm must be combined with the algorithm for the overall equilibrium equations which must also be solved iteratively. A typical algorithm for solution of the catenary equations is given by Peyrot and Goulois 7 and may be used here. An adjustment must be made to the pipe slopes at the buoy interface to avoid a discontinuity. Design should ensure that for expected motions of the buoy, the pipe is allowed to freely adjust its slope at the buoy interface away from the area where the riser is clamped to the buoy. It must be noted that solution of the overall equilibrium equations for steep-S riser geometries may equally be attempted using equations like eqns (90) and (92) directly in the overall system equations. However, the use of such a method would require the use of a general purpose
145
mathematical algorithm for solution of systems of simultaneous nonlinear equations. The suggested alternative offers greater simplicity and relies on the use of a standard algorithm for solution of the catenary equations, examples of which are widely available. 5.4 Lazy-S risers Lazy-S risers are a modified form of the steep-S variety where the lower section of the riser, instead of being vertical, forms a simple catenary to the bed. Lazy-S risers are used where the horizontal offset between the vessel and the well-head is too large to be accommodated by steep-S type risers. A lazy-S riser is illustrated in Fig. 9. The equations of horizontal equilibrium at the buoy may be written as: H i - H 2 + Tl sin0 = 0 (112)
where H! and H2 are the horizontal components of riser tension at the buoy and T~ is the tether tension at the top. Using the equations for vertical equilibrium:
Fb --
T1 cos0 - Vl - //2 = 0
(113)
where VI and V2 refer to the vertical component of tension due to adjacent riser parts. The tether inclination may then be calculated by combining the above equations to give: 0 = tan-I (FbHI ~ Z V2)
-/-t2
(114)
146
F. B. Seyed, M. H. Patel
HI,H2, V! and 1:2 must all be determined from the solution of the catenary
equations to either side of the buoy. The global equilibrium is determined iteratively. Starting values of variables are determined for zero tether inclincation. Equilibrium of forces at the buoy is checked and the required tether movement to balance this force calculated. For the new buoy position, the equilibrium is checked again and the procedure repeated until convergence is established. An algorithm for adjustment of riser slope at the buoy interface is required as with the steep-S risers.
x i -X
= 0
(115)
Ho
i=1
Y3
s2 e3 s1
s3 Y2 Yl
el
I_Xl_l_ t_ X2 _1_ X X3 I
147
Z
i=l where:
Yi-
Y = 0
(116)
x; Yi = Wi
'.-o + ' a n ; ) -
(117)
(118)
(119)
where (n = 3) for the pipe shown in Fig. 10. The focal consideration is the constancy of rio which is merely a statement of horizontal equilibrium. This arises from the lack of all external horizontal force resultants. It is noted that these equations may be expressed in terms of the two variables Ho and 01, all other quantities being related. An iterative procedure is required for solution of this class of equations. Many standard mathematical algorithms are available for this purpose. For this work, the NAG (Nottingham Algorithms Group) routine C05NBF was utilised.
Sb~d+ ~. xi-X
i=1
n
= 0
(120)
= 0
(121)
148
Y3
s2
s3
Ix1
-. . . . . x2 x
x3
Fig. 11. L a z y - w a v e p r o f i l e .
where
xi
_
Howi
[Lsinh_, \{wiSiHo+
1 + \ Ho +
(122)
Yi = -~i
tan2Oi)V2
(124)
(w,s,
subject to the boundary condition 01 = 0. Since all the loads acting on the system are vertical, the horizontal force equilibrium implies that Ho remains constant throughout the length of the pipe. This allows expression of all the above equations in terms of s~ and Ho only. Due to the nonlinearity of these equations, a standard mathematical procedure must be employed. For the purpose of this work, the NAG algorithm C05NBF which was also used in the solution of the steep-S problem was used.
6 CONCLUSIONS Expressions for pressure forces on curved pipe sections provide a mathematical derivation of the effective tension term. The resulting
149
expressions are of use in calculations relating to bends in high pressured rigid pipes where the pressure-curvature effect may only be modelled correctly using special b e n d elements. The formulation for straight elements is of particular use in the calculation of buoyancy forces on large tubulars found in semi-submersibles and jacket structures. Forces arising from a steady internal flow have been quantified and shown to contribute a new term to the effective tension expression. This becomes a new consideration in the analysis of flexible pipes. The new term acts in the same way as the internal pressures, altering the physical tension in the pipe cross-section. It is immediately seen that the absence of the internal-flow term could result in large errors in prediction of the local mechanical behaviour of the pipe. Changes to tensile force will directly influence the inter-layer forces within the pipe cross-section which control its bending as well as determining the extent of local frictional forces and wear of the pipe. Long term wear calculations excluding internal flow effects could therefore lead to grossly misleading results. The governing equations of equilibrium for a flexible riser including pressure and internal-flow effects provide very fast and accurate means of determining the static profile of flexible risers. These can be used to obtain solutions in a fraction of the time required by the finite element procedures involving incremental shifting of the pipe to equilibrium. Since numerical methods do not necessarily provide greater accuracy for calculation of the initial static profile, it should be possible to replace the static mesh generation routine in numerical analysis programs with catenary based algorithms. This could lead to significant gains in the overall run-times. REFERENCES 1. Sparks, C. P., Mechanical behaviour of marine risers mode of influence of principal parameters. Proc. Winter Annual Meeting of the Amer. Soc. Mech. Engrs., 2-7 Dec. 1979. OED Vol. 7, New York, p. 59. 2. Sparks, C. P., Cabillic, J. P. & Schawann, J. C. Longitudinal resonant behaviour of very deep water risers. Offshore Technology Conference, OTC 4317, Houston, TX, 1982. 3. McIver, D. B. & Olson, R. J., Riser effective tension -- Now you see it, now you don't! Risers, Arctic Design Criteria Equipment, Reliability in Hydrocarbon Processing, 37th Petroleum Mechanical Engineering Workshop and Conference, Dallas, Texas, Sept. 1981. 4. McIver, D. B., The linear stiffness of mooring catenaries. Paper presented at a short course on Dynamics of Compliant Structures, Centre for Advanced Engineering Studies, Southampton, 20-22 March 1985.
150
F. B. Seyed, M. H. Patel
5. Young, R. D., Fowler, J. R., Fisher, E. A. & Luke, R. R., Dynamic analysis as an aid to the design of marine risers. Trans. ASME, 100 (May 1978) 200. 6. Chakrabarti, S. K. & Frampton, R. E. Review of riser analysis techniques. Applied Ocean Research; 4(2) (1982) 73. 7. Peyrot, A. H. & Goulois, A. M., Analysis of cable structures. Computers & Structures, 10 (1978) 805-13.