Académique Documents
Professionnel Documents
Culture Documents
C. M. Larsen
Departmentof Marine Structures, This paper describes how the design of a catenary riser can be formulated as an
Norwegian Universityof optimization problem by using riser costs as the criteria function, design requirements
Science and Technology (NTNU), in terms of maximum allowable stress and buckling capacity as constraints, and riser
7034 Trondheim, Norway dimensions as free variables. The theory has been implemented in a computer pro-
e-mail: callas@marine.ntnu.no gram that can generate an optimized riser design for given design parameters such
as water depth, diameter, pressure, and platform excursions. The developed software
consists of a conventional program for two-dimensional riser analysis and a set of
T. Hanson standard routines to minimize a nonlinear function subjected to general constraints.
Norsk Hydro a.s., A case study where design parameters and requirements have been varied is also
Research Centre, presented. The importance of buckling versus allowable equivalent stress as the most
Bergen, Norway critical constraint has been investigated for varying water depth. The conclusion of
e-mail: tor.david.hanson@nho.hydro.com
this work is that optimization is a useful tool for riser design, and that the proposed
strategy for selection of design variables and constraints will enable an engineer to
identify designs with minimum costs in an efficient way.
zation with and without nonlinear constraints is given by Walsh a Formal constraints, stating that all free variables must be
(1975). Moe et al. (1971) applied optimization on design of within a range given by predefined minimum and maximum
ship structures. A more recent application on stress joints for values
workover and production risers has been presented by Larsen a Global geometry, saying that the sum of all segment lengths
et al. (1995). Papalambros et al. (1982) has applied a more must exceed the length of a straight line between end points
analytical approach to the riser problem than what is presented of the riser. Using the notations seen from Fig. 1, we have
in this paper. L
The optimization software used in the present study has been Y, li > ~/(DIST + A P O S ) 2 + (O + H ) 2 (5)
developed by Schittkowski ( 1985, 1994) and has proved to be i= 1
Nomenclature
L =
no. of segments in riser A P O S = distance between near and far ~ra = axial stress
M =
no. of constraints position (rb = bending stress
N =
no. of free variables Ii = length of segment i (rh = hoop stress
D =
water depth 1B = length of bottom segment ~rr = radial stress
H =
upper end position above sea pi = internal pressure ~req = equivalent stress
surface Pe = external pressure ~ry = yield stress of material
DIST = horizontal distance between Ai = internal area of riser pipe E = modulus of elasticity
ends, near position Ae = external area of riser pipe u = Poisson's ratio
Journal of Offshore Mechanics and Arctic Engineering MAY 1999, Vol. 121 / 91
where A~ is the area of the circle with radius r. F r o m these 60 is a parameter to define initial ovalization
equations, one can easily derive the following formulas for hoop
and radial stress components at the outer and inner surfaces of
the pipe:
Table 2 Riser case definition
piAi + (Pi - pe)Ai - peAe
Crh,ext Riser case W~er depth(m) APOS (m)
Ae - A~
1 300 30
C7..... : --Pe (9) 2 900 80
3 1500 130
Similar equations can be found for the inner surface
0.8
Dma x - Dmi n
60 - (r5)
Dm~ + Dmin E
0,6-
0.2-
P~1- 1 - u 2 (16)
while the yield pressure p y is given by 0.0- I ' I ' i ' I '''t i
0 200 400 600 800
t D i s t a n c e f r o m b o t t o m e n d a l o n g t h e riser ( m )
Pr = 2O'y ~ (17)
Fig. 3 Buckling criteria, riser case 1
300 -
~ ~ ~/ ~
300 -
Axial
Bending
a. 200 -
.... Ho~Li_~de ~
I1. 200 -
/ -f ~- Radial, inside
I \ J ~ Max equivalent
0- 0-
-100 -100
' I ' I ' I ' ' I ' I ' I ' I '
0 200 400 600 800 500 1000 1500 2000 2500
Distance from bottom end a l o n g riser (m) Distance f r o m b o t t o m end along riser (m)
Fig. 2 Stress distribution, riser case 1 Fig. 4 Stress distribution, riser case 3
0.0
\ sensitive to vertical motions at upper end. Such motions may
easily cause a global buckling behavior, and hence excessive
500 1000 1500 2000 2500 bending stresses.
Distance from bottom end along the riser (m) Fatigue is another failure criterion that needs to be consid-
ered, which again calls for dynamic analyses. A two-dimen-
Fig. 5 Buckling criteria, riser case 3 sional frequency domain solution may be relevant to analyze
fatigue in the wave zone. However, the riser may also have a
significant fatigue damage close to the touchdown point. Here, a
DIST = horizontal distance between touchdown point and linear dynamic solution combined with the analytical boundary-
upper end at near position (cf., Fig. 1 ) layer solution published by Pesce et al. (1997) is expected to
yield results of sufficient accuracy.
Comments. The wall thickness is seen to have an insig- Another improvement would be to apply established design
nificant variation for the riser on 300 m water depth. The reason guidelines as constraints in a user-friendly way as an alternative
for this is easy to understand from the stress distribution shown
to user-specified limits for stress and buckling usage factors.
in Fig. 2. Hoop stresses from internal overpressure (load case Other design requirements and load conditions may also be
2) is seen to dominate the stress condition even at the touch-
introduced.
down area. External pressure at bottom is slightly above 10
percent of the internal pressure, and maximum bending stress
for this load condition is less than 50 percent of the average Acknowledgment
hoop stress. Load case 1 (empty riser at near position) will The authors would like to thank Norsk Hydro A / S for permis-
give higher bending stress, but is too low to become critical. sion to publish this work, and also colleagues who contributed
As seen from Fig. 3, buckling is far from critical for this case, to the development of the applied software.
illustrated by the fact that maximum buckling interaction sum
is approximately 0.3, while maximum allowable value is 0.8. References
Hence, load case 2 will govern the design along the total length Lamr, G., and Clapeyron, B. P. E., 1833, "Mrmoire sur l'rquilibre intrrieur
of the riser for this case. des corps solides homogrnes," Mdm. divers sarans, Vol. 4.
The only result of interest for this case is the needed length Larsen, C. M., and van Hoeken, A., 1995, "Optimisation of Steel and Titanium
Stress joints for Marine Risers," Proceedings, 5th International Offshore and
of the riser, which is found by considering the far position only. Polar Engineering Conference (ISOPE), den Haas.
At this position the riser must have a sufficiently low bending Moe, J., and Gisvold, K., eds., 1971, "Optimization and Automated Design of
stress to give allowable equivalent stress in combination with Structures," Division of Ship Structures, The Norwegian Institute of Technology,
hoop and other stress components. Trondheim, Norway.
Papalambros, P., and Bernitsas, M. M., 1982, "Monotonicity Analysis in Opti-
The situation for the riser on 1500 m water depth is seen to mum Design of Marine Risers," Journal of Mechanical Design, Vol. 104, pp.
be totally different. Distributions for stress components and the 849-854.
buckling interaction sum are shown in Figs. 4 and 5. Pesce, C. P., Aranha, J. A. P., Martins, C. A., Ricardo, O. G. S., and Silva, S.,
From Table 3 it is found that the wall thickness at upper end 1997, "Dynamic Curvature in Catenary Risers at the Touch Down Point: An
Experimental Study and the Analytical Boundary-Layer Solution," Proceedings,
is slightly higher for case 3 than for case 1, obviously caused Seventh International Offshore and Polar Engineering Conference, Honolulu, HI.
by increased riser tension. At lower end, however, the wall Schittkowski, K., 1985, "NLPQL: A FORTRAN Subroutine Solving Con-
thickness is significantly increased, which is caused by the fact strained Nonlinear Programming Problems," J. C. Baltzer AG, Scientific Publish-
that the buckling constraint is active in this case. ing Company.
Schittkowski, K., 1994, Non-Linear Programming Software, Department of
Figure 4 shows that the equivalent stress criterion will govern mathematics, The University of Bayreuth, Bayreuth, Lebanon.
the design for the upper segment only, while local buckling will SINTEF, Snamprogetti, and Det Norske Veritas, 1996, "Limit State Design
control the wall thickness for the other three suspended riser Guidelines for Offshore Pipelines," The SUPERB Project, Draft Version, SINTEF
segments. Still, a reduction of thickness at an intermediate sec- Civil and Environmental Engineering, Trondheim, Norway.
Sparks, C. P., 1980, "The Influence of Tension, Pressure and Weight on Pipe
tion is possible, but the combination of high external pressure and Riser Deformations and Stresses," ASME Journal of Energy Resources
and high bending stresses requires a large wall thickness at the Technology, Dec.
lower segment. S~evik, S., 1990, "Pipeline Analysis System," Vol. 3, Theory Manual, Rev.
The wall thickness for the riser on 900 m water depth is 1.0, Reinertsen Engineering, Trondheim, Norway.
Timoshenko, S. P., and Googier, J. N., 1970, Theory of Elasticity, 3rd Edition,
seen to exhibit some variation along its length. However, the McGraw-Hill Book Company, New York, NY.
difference between maximum and minimum wall thickness is Walsh, G.R., 1975, Methods of Optimization, John Wiley & Sons, London,
only slightly above 1 mm, which is considered insignificant. U.K.