Ocean Management, 7 (1981) 211--230

Elsevier Scientific Publishing Company, Amsterdam -- Printed in The Netherlands

Simplified Analysis and Design of Abandonment

and Recovery of Offshore Pipelines

F. Andreuzzi and G. Maier

SNAM-progetti S.p.A., Milan (Italy)
Technical University (Politecnico), Milan (Italy)
(Accepted for publication October 13, 1980)

INTRODUCTION

During pipelaying exceptional rough sea conditions and other circum-

stances may make it necessary to abandon the pipe on the seafloor and later
to recover it. These operations are lengthy, costly and not easy to perform,
particularlF in deep waters.
From the engineering standpoint, the problem arises to devise an operative
sequence which is as simple, quick and, hence, as economical as possible,
under the basic constraint of safety against failure or damage in the pipe.
From the structural mechanics point of view, the requirements are rather
severe and conflicting. In fact, on the one hand, the system pipe-cable is,
loosely speaking, "more nonlinear" than sealines during normal laying opera-
tions (and generally more difficult to analyse at equal accuracy); on the
other hand, sequences of configurations have to be analysed along alternate
abandonment or recovery strategies, in order to choose the most favourable
one in a specific situation; moreover, it is desirable that at least approximate,
comparative analyses may be carried out in real time on the barge and,
hence, with limited computing time and storage.
This paper is intended as a contribution to meet the above requirements.
A very simple, computationally economical procedure for approximate
comparative static analysis of abandonment-recovery operations is devel-
oped. Numerical results in nondimensional variables are employed to con-
struct diagrams, which make it possible to visualize the evolution of the
main static and geometric quantities along alternative abandonment or
recovery processes and, hence, may be useful for the preliminary design of

211
the operation. The accuracy achieved, the possible corrections and substitu-
tions by means of more refined previous methods, and the use and limita-
tions of the proposed procedure as practical tools, are discussed on the basis
of specific deep-water cases.
In the recent, growing literature on offshore pipelines and cables, one
can find several more or less sophisticated approaches. These are applicable,
with suitable adjustments, to abandonment and recovery analysis and can be
used for a detailed study of the operation, after the preliminary, simplified
analysis and design which alone is of concern in what follows. These
approaches will not be considered here, as this would be b e y o n d the purpose
of the paper, but a fairly abundant bibliography on static sealine analysis is
given, though by no means aiming at completeness.

FORMULATION OF PROBLEM

In both the abandonment and recovery of submarine pipelines (Fig. 1) the

pipe extends between an unknown, variable t o u c h d o w n point B on the sea
b o t t o m and a cable connected to a winch on the barge. The pipe is gradually
laid d o w n on the seabed (abandonment) or more pipe is picked off the sea-
bed (recovery) by changing, as governing independent variables, two of the
following three parameters available and easily measurable on the barge: (1)
the cable tension T at the winch; (2) the position of the barge with respect
to the laid pipe along the laying route, a position which will be defined here
by the distance X between the stern pulley for the cable C (usually on the
stinger) and the location of the pulling head P on the seabottom after aban-
d o n m e n t or before recovery; (3) the length c of the cable from the aft pul-
ley to the pipe head.
Given two of these parameters and the fixed data which characterize the

B B B=P

Fig. 1. Schematic illustration and basic nomenclature.

212
situation (seabottom profile, sectional properties of pipe and wire), the static
equilibrium configuration and the stress state of the system, including the
third parameter, can be defined (see Fig. 1).
We will consider here only "normal" situations, namely such that the fol-
lowing assumptions are legitimate.
(1) The system is two-dimensional, i.e., it can be studied in the vertical
plane containing laid and suspended spans and the barge axis.
(2) The seafloor can be regarded as flat, horizontal and rigid.
(3) Dynamical effects can be allowed for merely by suitable amplification
of the safety factor.
(4) The cable has no bending stiffness and its lift-off point from the
stinger is fixed at a single pulley during the operation.
(5) The loading due to current is negligible.
From the mechanical standpoint, abandonment and recovery consist of
substantially equal operations performed in reversed sequential order. There-
fore, only abandonment will be considered henceforth; its phases are briefly
described below, with reference to a barge with a rigid stinger.
(a) The pipehead slides from the tensioner to the stern end of the stinger
while the barge moves forward. Little option is left in this phase. The stinger
geometry is given, as set for the laying operation. The pipe bending stiffness
plays an essential role. The control of stresses, both in the sagbend and over-
bend, has to be carried out by means of the same methods and software
appropriate for accurate analysis of normal pipelaying, allowing for the uni-
lateral contacts on the stinger rollers (e.g., by means of the approach and
code concisely described in Maier et al., 1979). At the end of this phase, the
pipe slope at its head P coincides with the stinger slope at its aft end. When
the latter angle is fixed, this coincidence, and the condition of vanishing
m o m e n t at pipe head, uniquely define the tension T1 at the winch and the
position X1 of the barge.
(b) The second phase starts from the situation above specified as the end
of phase (a), and ends when the pipe assumes the straight configuration at
rest on the seafloor. The length L of the suspended pipe span BP and the ver-
tical distance d from b o t t o m to pipehead P reduce to zero, the latter starting
from the distance D from b o t t o m to the head final position in phase (a), i.e.,
to-the last pulley C for the cable ( D i s usually less than the water depth).
Simultaneously the suspended cable length c varies from cl = 0 (Fig. 1).
(c) The last stage concerns the cable alone and will not be considered in
the sequel.
The second phase (b) admits many alternative strategies, i.e., ways of vary-
ing two of the above three parameters T, X, c available to the operator on
the barge; hence, it gives rise to the problem of choosing between them, i.e.,
of designing the operation. Only this phase (or abandonment " t o u t court",

213
 in the strict sense) will be of concern here, aiming at simple tools for com-
parative analyses of its alternate options, in view of deep-water applications.

SIMPLE FORMULAE AND GRAPHS FOR COMPARATIVE ANALYSIS

For the static analysis of abandonment and recovery in "normal" condi-

tions in the above-specified sense, the suspension cable can be quite rigor-
ously interpreted by a natural catenary. The pipe is a very slender beam in
the large displacement regime;its bending stiffness EI, however, affects the
equilibrium config~Jration to some extent mainly near the ends B and P, For
the present purposes, we will assume first that the pipe configuration

V+dV

__. \ M ~dM

(a)
H ds

v~ T
i.e c
q
D ¸

(b)
$"

yl

f~

~,- - H , , r / 1 , 1 , : , / I i--.
I/,'/ii/,-"/// /
Xp i Xc X
~,^~

F i g . 2. T h e d o u b l e ~ a t e n a r y model: definition of symbols.

214
 "A priori", such an hypothesis is suggested by
also b e a n a t u r a l c a t e n a r y .
the following circumstances: it appears not too rough for deep waters, where
the pipehead elevation d exceeds 50--80 m during most of the operation; it
turns out to be conservative as for the evaluation of the maximum bending
strain era; of course, it drastically simplifies the analysis. Its accuracy and
limitations will be considered after the developments which follow.
With reference to Fig. 2a the equilibrium equations for the pipe under the
above hypothesis (which implies throughout zero bending moment M and,
hence, zero shear force Q), besides H = const, are:

dV= wds Hcos 0 -- Vsin 0 = 0 (1)

As usual in structural analysis of sealines (e.g., Palmer et al., 1973; Gnone et

al., 1975), the hydrostatic load per unit length and, hence, the submerged
unit weight w are regarded as vertical, and a correction is made afterwards,
by adding - - A p ( A being the cross-section area, p the water pressure at this
section) to the axial force N calculated on this assumption. It is convenient
(Dixon and Rutledge, 1969), to define the length L, weight of which
equals the vertical force Vp at the upper end P, and which coincides with the
length of arch B P as long as shear is supposed to vanish (L = OP in Fig. 2).
The following nondimensionalizations simplify formulae and developments:

z - s/L = V/Vp h - H/wL (2)

the non-dimensional curvilinear co-ordinate z being 0, where V = 0 (at the

catenary vertex) and 1 at the upper end. Thus, simple mathematics leads
from eq. 1, account taken of the boundary conditions 0(0)--lr/2, x ( 0 ) =
y(0) = 0, to the known relations which follow *:

0 = artg h/z (3)

x = L h Arsin z / h (4)

y = Lh([1 + (zlh)2] 1/:-1) (5)

N = w L ( h 2 + z2) 1/2 = H + w y (6)
L / R = --dO / d z = h ( h 2 + z2) -1 (7)

* T h e n o t a t i o n a r t g , A r s i n e t c . f o r inverse trigonometric and hyperbolic functions is pre-

ferred here to the more usual one in English~peaking countries ( t a n -1 , s i n h -1 , e t c . ) .

215
Eqs. 3 to 7 define the angle 0 and, hence, the vertical c o m p o n e n t V of the
traction, abscissae x and ordinates y (i.e., the equilibrium configuration), the
axial force N and, finally, the radius of curvature R, respectively, for the sus-
pended pipe span B P , i f its bending stiffness is neglected. For the cable the
same eqs. 1 to 7 hold almost rigorously, provided that the primed symbols
shown in Fig. 2 are used, referring to the new axes x', y' with origin in the
vertex 0 of the cable natural catenary (the length L' of which includes both
the real span PC = c and its fictitious extention O P ) .
At the pipe head P, where the two catenaries meet with equal slopes (as a
consequence of M -- 0), one can write eqs. 3, 4 and 5 for both pipe and cable
(unprimed and primed quantities, respectively). Thus, the slope continuity in
P requires:

L w = (L' - - c ) w ' (8)

The identifications of abscissae via eq. 4 and of ordinates through 5, give

rise, respectively, to the relations:

H / w Arsin L w / H = Xo + H / w ' Arsin I ~ w--

' Lc' ) (9)

d = [ ( H / w ) 2 + L2] 1/2 -- H / w = Yo + [ ( H / w ' ) 2 + ( L ' - - c)2] 1/2 - - H / w ' . (10)

At the suspended span end C eqs. 4, 5, 6, written for the cable, supply the
relations:

x c - - Xo = H / w ' Arsin L ' w ' / H (11)

D = Y0 + [ ( H / w ' ) 2 + L'2] 112 - - H / w ' (12)

T = ( w ' 2 L '2 + I-t~) i n (13)

respectively, where the abscissa xc turns out to be related to the "barge posi-
tion" X as follows (see Fig. 2b):

Xc = X + L (14)

Finally, according to eq. 7, the m a x i m u m curvature in the pipe occurs at the

touch-down point B and attains the value:

(1/R)m = w / H = e m / r (15)

The second eq. 15 defines the maximunl bending strain em (r being the

216
external radius of pipe), to be kept for safety below some admissible thres-
hold earn depending on the water depth. In fact, em can be regarded as a mea-
sure of the stress peak in the sagbend, since bending is the prevailing defor-
mation (the axial force is roughly H minus the corrective term A p and,
hence, relatively small).
By eliminating x c from eq. 11 by means of eq. 14, x0 from eq. 11 by
means of eq. 9 and, finally, Y0 from eq. 10 through eq. 12, we can derive
from eqs. 8 to 15 six independent equations in the eight variables X, T, c, d,
H, era, L, L', as w, w' and D are data characterizing the specific situation. Of
these variables, the first six have a mechanical interest, whereas L and L' will
be eliminated by other substitutions, thus reducing the six equations to four.
It is convenient to make use of the nondimensional quantities which follow:

× - X/D, r = T/wD (16)

= c/D, 77 - e r n D / r , p - H/wD, 6 - d/D (17)

m - w'/w (18)

Using the above quantities, after some algebra the four equations read:

r a = i~= + ( ~ , / m + x / 6 = + 2 p 6 ) = (19)

r = 6(1 - - m ) + m + g (20)

m x + rex/6: + 2 # 8 = ~ Arsin[1/g(m~/+ \/6 2 + 2#6)]

-- (1 -- m) # Arsin[1/#x/62 + 2p6] (21)

= 1/p (22)
When the pipe head is close to the stem end of the stinger (i.e. for 7 = 0,
8 = 1), its inclination 0c clearly acquires a practical interest; since H = T sin
0c, one immediately obtains from eqs. 16, 17 and 20:

sin 0c = p(1 + p)-i (23)

In eqs. 19--23 the only parameter depending on the situation is the weight
ratio m (18). The non-dimensional barge co-ordinate × and winch tension r,
eq. 16, will be taken as independent variables; by a simple numerical proce-
dure, easy to implement and cheap to run (even on small computers), the
other four variables can be expressed as:

= ~(X, r ) , n = 1 / ~ = n(X, r ) , 6 = 8(x,r) (24)

217
 -t.40 -~.60 0,20 L .00 i ,8@ 2.8~" ,,~ )..40 -0.60 0.20 L.O0 1.80 Z.60
.... ± . . . . . L__~.J~_.~ .... ~--- .~_ 1 ~F ~ ~ ~_ _A_ .l ~ ..... i _ -L. ~ ~ ~ ..... i .. _L. _ ~ .
oi a Io ~ a !'~

' L:'~

fi : ,,,; , , / , ~cn

°~1 ":"; " " " " k z o~! , , " , I ' ' , ,
~:-~ ," ," ' ," ," ," '" ," ,' . . . . . . u] ~! ,' ,' ' '"',' " " , " , " ,' ' ""
Z [ ~' / ~' " ,' / / / ,/ " I~ -~4 I , , lI ~ , i / i / / -#

~- ///l/Ill .// , //

-~ .~o -o.~o o.~o ~.oo t'.~o ~'.~o -~ .~o -o.~o o.~o ~.oo d.,o ~'.~o
BRRDE POSITION CHI BRROE POSrT~ON CHI

b ~]b
Mu

,.9o
c~ 0.67 ~ L,70
-- 1.50
t 90 ~-- ' 0"81 ~ ~ - - - - - 1,30 i z

0.90 ] :2:
"----------- O,70 F
- ° \ ~ ~ o.~o ~- ~'~ ~ o . ~ o

O.LO

-t ,4o -o.~o o.~o ~ .oo t,~o ~,6o ~- .4o -~.eo o~.zo -1-
BIqROE POS]'T/ON CHI BRROE POSITION CHI

.,*o= -~.so %.zo L lj.oo Lieo J zi6° ~, -t.,*o -~,so otzo ttoo ~l.eo__L_ zj.eo ¢o

c ~ o ~AMMA I~ ~I J ~ ~
¢ u: c=
• ¢m ¢~

//
Z
0¢ o o ~| ~o

o /
i o

~ it
I ---I~--
-;.40 I -~.eo' 0'.2o ~ t'.oo
-o.so '
o~2o t.eo
x~,oo l,eo T z~eo2.60 ~1,40
BQROE P O S I T I O N CHI
BQRGE POSITION CHI
Fig. 3. Constant-level curves of: (a) p i p e h e a d - b o t t o m distance 5, (b) m a x i m u m strain 17 in
the pipe and horizontal thrust/~; (c) cable length ~f, as functions o f barge position X and
winch tension T, for weight ratio m = 0.2.

Fig. 4. Same as in Fig. 3, but for weight ratio rn = 0.8.

218
 ¸,2.0 I -f.,o - F '° , o~oo , o~4o i oiso , ti~,o , t:so = ;:too , zf,,o ,

0.40 c~
=o CABLE LENGTH
c~

t¢

t~
---"----- 2 .$0
iI
/

I
I
I
CI: ,-F---- ! .so
F- t
t
&z
u)
t
Cz --'----- ~..70

-o
(/3 "-------- 1.60
Z
~D
c'-
t-- •- - - - - - t .SO /
s
-}
1. t O t , "

~.6o

rRL FORCE

1 I I l
~..6o z'.oo 2'.4o

Fig. 5. Comparative analysis o f four a b a n d o n m e n t strategies for a situation w i t h m =

0.4 2 6.

and the constant value curves of these functions can be plotted, thus gener'
ating, for any fixed parameter m, three curve families like those shown in
Fig. 3 (for m = 0.2, a low value corresponding, say, to flooded pipeline) and
Fig. 4 (for m = 0.8, corresponding to heavy cable and light pipe). Fig. 5
shows the three curve arrays for m = 0.426 (as in the example considered
below), superposed in a single diagram.

PRELIMINARY ANALYSIS

Diagrams as that in Fig. 5 represent a synopsis of behaviours of the pipe-

cable system for any possible abandonment or recovery strategy, defined by

219
a time history of barge position and winch tension (×, r), or of tension and
cable length (T, 7), or of cable length and barge position (7, ×).
Let us consider as an illustrative example a situation characterized by the
data: D = 615 m, w = 0.46 kN m -1, w' = 0.20 kN m -1 (hence m = 0.426), r =
0.325 m, E I = 197 000 kN m s. At the beginning of abandonment phase (b),
let the tension be T~ = 817 kN. Thus, the starting point in the nondimen-
sional graph of Fig. 5 is represented by a on the 7 = 0 line. According to the
present analysis, this point turns out to correspond to the inclination 0c =
41 ° of the pipe head with respect to the vertical direction.
Four alternative abandonment strategies are depicted in Fig. 5.
(1) First, the barge moves forwards keeping c o n s t a n t t h e w i n c h t e n s i o n
and releasing cable; second, it moves backwards, decreasing the tension,

T
( r~,)
'i 800

/'
1000 5
600

4
®

3
50O X

I i 1
I
100 : i
. . . T'----. ...... -.7 ...... . ........ . - - T . - - ~ .......... . - - - ~ c ,
0 50 100 150 lO0 250 300 350 400 450 " ~ §SO 800f615

® @

tx
I

Fig. 6. Constant-tension--constant-cable abandonment (dotted lines in Fig. 5). Above,

evolution of the main quantities: double-catenary model (solid lines) vs. "exact" analysis
(dashed lines);below, configuration sequence.

220
holding the cable length unaltered and decreasing the tension (dotted lines).
(2) A stage with the barge moving forwards at constant tension and
increasing wire length is followed by a stage with decreasing tension and
moderately increasing cable length, while the barge keeps its position (dash-
dot lines);
(3) First, the tension is lowered and the cable released at fixed barge posi-
tion, while the peak stress in the sagbend increases; then this stress is kept
fixed at a suitably chosen admissible level (higher than during normal pipe-
laying; here, specifically, em= 0.625~), by varying simultaneously position
and tension (dash-double dot lines);
(4) Multistage operation, with repeated constant position--constant ten-
sion sequences where the maximum stress is kept close to the admissible
leveh

c I × T j [m
( : ) ~(m) (kN), (~/~)

1\

~ - - i 400.
-ot ;,-"
I 200

loo! c,,,
,

0 50 100 1S0 200 ;~SO 300 3~0 400~"~-,~ 450 500 5~50 600 61S

®®®
t¢¢

.' , , .

Fig. 7. Constant-tension--fixed barge abandonment (dash-dot lines in Fig. 5): above,

evolution of the main quantities: double-catenary model (solid lines) and corrections from
second-order beam theory (dashed lines); below, configuration sequence.

221
 In each case the evolutions of all mechanically meaningful quantities can
be easily read on the graphs (or tabulated by the computer) and checked
with respect to strain limits (structural safety) and to operative limitations
(such as maximum available winch tension and cable length, mooring capac-
ity to provide horizontal thrust). This can help in adjusting the strategies
envisaged (e.g., by modifying their turning points) and/or in selecting the
most favourable one, account being taken of time consumption and simplic-
ity. The variations of quantities and configurations indicated in Fig. 5 are
further illustrated (in solid lines) in Figs. 6, 7 and 8 for the first three of the
above abandonment strategies, respectively (for 3 the condition em =
0.625%o is imposed up to the end, to show the consequent exceedingly
high requirements of cable length).
Our purpose here is not to assess the merits of alternative strategies, but
merely to briefly appraise the reliability of the above simple "double cate-
nary" analysis and its possible corrections.

c x T~
@ ® (kN t

1000
® 60¢ .
.4

llO0.
j/"
Y ! 40¢ "3'

'2

100~
l/ . , I. . . . D .
0 50 100 ,~ 2;0 ~o ~o ~io ~oo ~so ~ sso 6oole,~

//
///

X
1

Fig. 8. As in Fig. 7, for a b a n d o n m e n t w i t h c o n s t a n t m a x i m u m strain b y d o u b l e ~ a t e n a r y

analysis.

222
 SIMPLI FI ED VERSUS "EXACT" METHODS

The evolutions of tension T, cable length c, barge co-ordinate ×, horizon-

tal force H and maximum bending strain era, along the abandonment opera-
tion (1) constant tension--constant cable, are depicted in Fig. 6a as func-
tions of the distance d from seabottom to pipehead. While the solid lines
represent the results obtained here on the basis of the above "double-cate:
nary" model, the dashed lines show the results supplied by an "exact"
method, for the same constant values T, = 817 kN in the first phase, c~ =
700 m in the second phase. This method is "exact" in the sense that it solves
the uniform beam equation in large displacement (see later eq. 26) by means
of successive integrations supplemented by suitable boundary conditions
and, hence, is affected only by the errors imbedded in the Gauss numerical
integration adopted (in the present case 17 Gauss points have been used
along the pipe) besides, of course, round-off errors. This method, imple-
mented in the computer code SOLAB of SNAM-Progetti, has been described
by Matteelli et al. (1976), and appears comparable, so far as the accuracy
achieved and generality are concerned, to other published methods (see
Datta and Basu, 1977; Gnone et al., 1975; Palmer et al., 1973), though
founded on different approaches. Convergence occurs for every ratio d/D.
Comparative remarks between exact and double-catenary analysis, corro-
borated by this and other numerical computational experience, are as fol-
lows.
(a) As for the tension, cable length and barge position, the agreement
between the two analyses is fairly satisfactory everywhere for the engineer-
ing purposes.
(b) The maximum bending strain provided by the simple model is under-
standably affected by an error, generally on the conservative side, which
become significant for d less than about 60 m and exceedingly large as d
tends to zero.
(c) ComputationaUy, the results plotted in Fig. 6 required 50 sec. CPU
time (on an IBM 370/168 computer) and 150 K storage occupation when
obtained by the "exact" method, in contrast to negligible computing time
and 2 K when obtained by the double~catenary method.
The bending stiffness of the pipeline is expected to affect the slope at the
pipehead P (and to make it different from the slope of the adjacent cable
section). With reference again to the specific deep-water situation considered
so far, but for varying tension T, we have compared double-catenary results,
eq. 23, to SOLAB results for the pipehead inclination ¢Oc with respect to
the horizontal when ~ = 1, i.e., at the abandonment beginning. The error of
the simple model is practically acceptable for all tension T, as shown in
Fig. 9.

223
(deg

90

60

4oj
ii
1.5 2 2.S 3 3.5 4
2
300 400 500 600 700 800 900 1000 1100 T (kN)

Fig. 9. Pipehead slope vs. tension at the beginning of abandonment: double-catenary

(solid line) vs. "exact" (dashed line) analysis.

Clearly, all the above plots generated by "exact" analysis depend on the
pipe bending stiffness E I , the incidence of which can be measured (Dixon
and Rutledge, 1969) by the nondimensional parameter:

c~2 = E I / w L 3 (25)

In general, it can be said that within the parameter ranges of practical

interest for deep-water sealines, the double-catenary model becomes unac-
ceptable and misleading (and does need corrections) o n l y when the pipehead
approaches the seabottom (say for d < 6 0 - 80 m) and when the cable
length c is close to the water depth (i.e., 7 -* 1) and, hence, the horizontal
force H becomes small (say p < 1 - 2). It is worth noting that the latter con-
dition is responsible for the divergence (to a finite value) of the em solid line
as d tends to zero in Fig. 6 (this divergence is attenuated when longer cable
lengths are employed, as in Figs. 7 and 8). In fact, obviously, em becomes
infinite when the pipe conceived as deprived of bending stiffness is lifted by
a vertical cable shorter than D.

CORRECTIONS

In order to compensate for its serious localized deficiencies observed in

the foregoing, the simplified analysis developed can be supplemented by the

224
following corrective analyses, still preserving its peculiar simplicity.
The "stiffened catenary" method (Dixon and Rutledge, 1969; Plunkett,
1967) proved to be of remarkable practical value for pipelaying analysis. It
provides approximate solutions to the equation:

~2 d20/dz 2 + h cos 0 -- z sin 0 = 0 (26)

which generalizes eq. 3 when E I is different from zero and, hence:

M = --EI dO/(is, Q = dM/ds (27)

are not zero (see Fig. 2a); then, the characteristic length L is smaller than, no
longer equal to, the suspended pipe length, so that the nondimensional
co-ordinate acquires an (unknown) negative value z0 = --~h -in in B and is
z = 1 in C. The stiffened catenary solution in terms of inclination consists of
the natural catenary term (3) and of two corrective addends which allow for
flexural stiffness effects at the ends. Since each addend dies off at increasing
distance from the relevant end, only one addend is needed for the present
purpose (correction of the max strain, which will occur near, but no longer
at, the touchdown point B). Therefore the solution to be used here reads
(Plunkett, 1967):
Z

O(z) = artg h / z - - c~h -s/2 exp[--1/~ / (h 2 + ~2)1/4 d~] (28)

z0

/- ..... j

;
I

o 10 ~o ~ 4b 5'0 go 7'0 io go 1~o .'o 1~o I d~

Fig. 10. Maximum bending strain at small d/D: a, double caten~; b, "exact"; c, stiffened
cat~nazy; d, second-order, ]ineaz beam theory.

225
where:

Zo . . . . ah -x/2 (29)

In eq. 28 ~ is z when it acquires the role of integration variable. By means of

analytical developments here omitted for brevity (similar, however, to those
expounded in Dixon and Rutledge, 1969), one can easily generate, using eq.
28, two equations which define L and the co-ordinate Zm of the max curva-
ture section. After the numerical solution of these (non-linear, transcen-
dental) equations, the max. strain em is calculated straightforwardly. The
results for the example in point are plotted in Fig. 10 (dash-dot line). By com-
paring them to those from the double-catenary (solid line), it can be ob-
served that the error is significantly reduced, except for very tow distances d
(<30 + 40 m). This is expected, since the stiffened catenary method must
reflect the trend of the double-catenary results for 6 -* .0 through the first
term on the right-hand side eq. 28; moreover, it is known (Dixon and Rut-
ledge, 1969; Plunkett, 1967) that the method tends to diverge for large a
(i.e., for large E I and/or small L and, hence, d).
A correction which is far easier and probably more effective in practice,
can be derived for small pipehead-bottom distances from "moderately large"
displacement, linear beam theory. Considering only second-order geometric
effects, the equation governing the pipe span becomes linear and reads {see
Fig. 2b):

L2c~2 d 4 y / d x 4 -- h d2y/dx 2 = --1/L (30)

The boundary conditions associated to these equations are:

y ( x = O) = 0 , ~xx(X = O) = 0 , y ( x = L) = d (31)

d2Y" d2Y(x = L) = 0 (32)

dx~(X = O) = 0 , dx ~

It is worth noting that setting the abscissa of the pipehead x = L (L being the
unknown suspended pipe length) is a further approximation, consistent with
the "small displacement" assumption underlying eq. 30. The static condition
32 correspond to the fact that eq. 30 has been obtained from the equilibri-
um equation (between internal and external moment) by taking the second
derivative in order to avoid explicit reference to the end shear force.
The general solution of eq. 30 is straightforward; for given elevation d,

226
 -} .20 i -~).lOj -~),40 I 0~00 I 0~40 I 0I$0 I iiZO A i[llO I ZI'O0 J

20t:
O, 40 CRBLE LENOTH

- I .Zo -O.IO -0,60 O.OB 0.40 0.80 I .ZO | .80 Z.O0 Z-40
BQROE POSITION CHI
Fig. 11. Graph of Fig. 5 supplemented b y corrective curves (dotted lines) derived from
second-order linear beam theory for EI = 197,000 kNm 2.

eqs. 31 and 32 readily define the four integration constants and the length
L, this being the solution to be achieved numerically of a single nonlinear
equation.
The curve (d) of Fig. 10 has been obtained in this way for varying d by
matching in P this solution with the cable catenary. This plot, if assumed as a
substitute for curve (a) in the range of relatively small elevations d, turns out
to represent a significant improvement in the simplified quantitative predic-
tions.
In the present linearized theory centered on eqs. 30--32, the location Xm
and entity em of the maximum bending strain can be, clearly, identified as
follows:

day. . d2y. .
dx3tXm) = 0 , e m = r-d~x2tXm) (33)

By matching in P second order beam solutions for the pipe and natural

227
catenary solutions for the cable, mathematical developments {here omitted
for brevity) almost as simple as those employed in order to generate eqs.
19--22, generate the counterparts of these equations, in the same nondi-
mensional variables of eqs. 16--18.
Still analogous though slightly more laborious, numerical work leads sub-
sequently to the functions of the type of eq. 24 or to their constant level
curves. Of course, besides the weight ratio m, also the bending stiffness E I
must be taken as parameter of the curve families, moreover the simple corre-
lation of eq. 22 is no longer valid.
Fig. 11 shows (dotted lines) only the constant max. strain curves thus
generated for m = 0.426 and E I = 197,000 k N m 2 and only in the small 5
range, where the linearized beam theory is acceptable and provides results
drastically different from those of the double catenary model. The other
curves exhibit discrepancies practically insignificant for preliminary com-
parative analysis. Thus Fig. 11 represents the graph of Fig. 5 supplemented
by necessary substantial correction, which basically preserves the simplicity
and e c o n o m y of the previous approximate analysis.
The corresponding corrections are also indicated (dashed lines) in Figs. 7 and
8.
In Fig. 11 the constant max. strain abandonment (3) is illustrated by the
dash-double dot line modified according to the new, corrective curves. The
lines marked by ~ and ~ define all main mechanical features of operation of
types (1) and (2), respectively, when these are subjected to the constrain
that the strain em= 0 . 6 2 5 ~ imposed for (3) be not exceeded. The line
represents an operation of type (3) along which the max. strain at the start-
ing situation (a) is never exceeded.
In general terms, the linearized beam t h e o r y is acceptable for the suspended
pipe as long as its m a x i m u m slope is small ( d y / d x < < 1); on the other hand,
the catenary interpretation of the pipeline is n o t meaningless when E I w -1
L -3 is relatively small, L being the "a priori" unknown characteristic length
(Plunkett, 1967). The validity ranges of both approximate approaches are
not susceptible to quantitative assessment in general and may overlap or
exhibit gaps. Therefore it must be emphasized in closing that " e x a c t " meth-
ods are needed for suitable checks of the validity ranges of the simplified
approaches to comparative analysis, not only for accurate analysis of the
selected strategy.

CONCLUSIONS

The conclusions of this paper can be summarized as follows:

(1) Comparative analysis and preliminary design of abandonment and

228
recovery operations can be conveniently p e r f o r m e d using a synopsis o f t he
correlations am ong quantities of engineering relevance.
(2) Th e e x t r e m e l y simple double-catenary model o f t h e system exhibits
sufficient accuracy for preliminary comparative analyses in most ranges o f
practical interest in deep-water pipelaying.
(3) Partial but pr oba bl y adequate corrections can be easily obtained by
second-order, linear beam t h e o r y , f or the drastic (excess) errors which affect,
in the double-catenary analysis, the m a x i m u m bending strain when t he pipe-
head approaches the s e a b o t t o m with small horizontal thrust.

ACKNOWLEDGMENTS

Th e authors are indebted t o V. Giardinieri o f SAIPEM and R. MatteeUi of

SNAM-Progetti for i nf or m at i on and advice and to SNAM-Progetti for their
generous support.

REFERENCES

Brando, P. and Sebastiani, G., 1971. Determination of sealine elastic curves and stresses
to be expected duringlaying operations. Proc. III O.T.C. Houston, Pap., 1354.
Clauss, G., Kruppa, C., Wolf, E. and Stature, K., 1977. Parameterstudie fiber das Verlegen
yon Pipelines in grSsseren Meerestiefen, Meerestechnik, 8 (3): 83--88.
Dareign, D.W. and Neathery, 1970. Marine pipeline analysis based on Newton's method
with an arctic application. Trans. ASME, J. Eng. Ind. 92: 827--833.
Datta, T.K. and Basu, A.K., 1977. Stress analysis of submarine pipelines. Proc. Inst. Cir.
Eng., 2: 833--841.
Dixon, D.A. and Rutledge, D.R., 1969. Stiffened catenary calculations in pipeline laying
problem. Trans. ASME, J. Eng. Ind., Febr., pp. 153--160.
Gnone, E., Signorelli, P. and Giuliano, V., 1975. Three-dimensional static and dynamic
analysis of deep-water sealines and risers. Proc. VII O.T.C. Houston, Pap., 2326.
Larsen, C.M. and Kavlie, D., 1978. Nonlinear analysis of oil pipelines by potential mini-
mization. Comp. Struct., 8: 733--743.
Leonard, J.W., 1972. Curved finite~lement approximation to nonlinear cables. Proc. IV
O.T.C., Houston, Pap., 1533.
Maier, G., Andreuzzi, F., Giannessi, F., Jurina, L. and Taddei, F., 1979. Unilateral con-
tact, elastoplasticity and complementarity with reference to offshore pipeline design.
Comp. Methods Appl. Mech. Eng., 17/18: 469--495.
Marichal, D. and Clement, A., 1977. I~tude tridimensionelle de la forme et des tensions de
c~bles en ~quilibre dans un currant non uniforme. J. M~c. Appl., 1 (3): 209--223.
Matteelli, R., Andreuzzi, F. and Taddei, F., 1976. Sea-line laying in deep waters. Proc.
O.T.C., 8th Houston, Pap. 2677.
Nielsen, R. and Pendered, J.W., 1978. Some aspects of marine pipeline analysis. In:
Numerical Methods in Offshore Structures. Wiley, Chichester.
Ovunc, B. and Mallareddy, 1970. Stress analysis of offshore pipelines. Proc. II O.T.C.
Houston, Pap., 1222.
Palmer, A.C., Hutchinson, G. and Ells, J.W., 1973. Configuration of submarine pipelines

229
 during laying operations. Trans. ASME, Pap., 73-WA (Oct. 4).
Pedersen, P.T., 1975. Equilibrium of offshore cables and pipelines during laying. Dan.
Cent. Appl. Math. Mech., Rep., 80.
Plunkett, R., 1967. Static bending stresses in catenaries and drill strings. Trans. ASME, J.
Eng. Ind., Febr., pp. 31--36.
Powers, J.R. and Finn, L.D., 1969. Stress analysis of offshore pipelines during installa-
tion, Proc. I O.T.C. Houston, Pap., 1071.
Webster, R.W., 1975. Nonlinear static and dynamic response of underwater cable struc-
tures using the finite element method. Proc. VII O.T.C., Houston, Pap., 2322.
Wilhoit, J.C. and Merwin, J.E., 1967. Pipe stresses induced in laying offshore pipelines.
Trans. ASME, J. Eng. Ind., 89: 37--43.

230