No abstract

Over the last few years, lateral buckling has gone from a secondary issue to one of the major concerns in pipeline design. Not many years ago pipelines were thought of as flexible elements, which could absorb virtually any kind of displacement, but due to the trend to increase fluid temperatures and after a few major environmental accidents, it seems that lateral buckling became the major design issue. Burial was then assumed to be the only safe solution. This is the normal reaction to every traumatic experience, but now that there is alertness and that accidents are being prevented, it is once again time to re-evaluate and see where caution has become excessive and what cheaper alternatives can be used. Several papers have been written over the last 3 or 4 years addressing this issue, [10–13] for instance, and, in general, one could say that there is a consensus regarding the need to know more about how pipelines move, when they buckle laterally, and to what extent they should be allowed to do so. Still another issue, which has been discussed along with this one, is related to how cyclic motions (due to cycles of heating and cooling) can aggravate the problem. Attention is drawn to the fact that the buckling analyses are usually performed based on models conceived to simplify the design, while, on the other hand, construction and installation pay a penalty, because of unnecessary conservatism, which could be avoided if a bit more effort was put into the design. Just to illustrate what is being said, let us consider a typical lateral buckling problem and how the practice has decided it should normally be treated: 1. Determine the pipeline embedment length (that for which the pipeline will build up sufficient axial friction to anchor the axial force due to the temperature variation). 2. Build a model twice that size, embedded at both ends; using a program, which can model axial and lateral friction (this is a nonlinear analysis). 3. Build a prop type lateral installation deviation at the center of the model. 4. Analyze the pipeline assuming that the expansion from both sides will build into that deviation, thus causing the pipeline to buckle at that section. In spite of having become a traditional design approach, it is associated with a conservative model. The bottom is assumed flat, the soil model is a simple elasto-plastic spring and, also, that there is only one lateral imperfection, which will concentrate all the axial expansion. The object of this paper is to show that it can be very advantageous to spend more time and money, on the design, using more detailed analysis models, in order to save much more on the construction and installation. This will be done by sharing the experience gained based on the use of such models, during several recent pipeline projects, handling over 300km of pipes, with diameters varying between 6 and 34 inches in water depths ranging from 0 to 2000m.

The design of rigid submarine pipelines has been the object of extensive research work over the last few years, where the most relevant issues include upheaval and lateral buckling problems. Both of these problems systematically associate temperature and pressure loads, where the treatment of the first is obvious, while the latter have always been a matter of discussion. In 1974 Palmer and Baldry [1] presented a theoretical-experimental contribution, in which they have set a pattern that has been followed ever since. Another similar and well known paper was published by Sparks in 1983 [7], who only present a physical interpretation of this same theory. Most of the present day industry codes define an effective axial force, according to which, fixed end pipelines will be under compression due to internal pressure. The starting point of the discussion presented in [1] was that internal pressure produces a lateral force, which is numerically equal to the pressure times internal cross-sectional area times the pipeline curvature: q=p.Ai.d2y/dx2(1) This equation is demonstrated further ahead in this paper. Palmer and Baldry then based their arguments on the traditional equation of the pinned column buckling problem, studied by Euler [2]: EId4y/dx4+Pd2y/dx2=0(2) for which the well known solution is: P=π2EI/L2(3) and on the associated problem studied by Timoshenko [3], which adds a distributed lateral load q to the same problem: EId4y/dx4+Pd2y/dx2=q(4) Replacing q with the lateral pressure given above, they were able to have their own problem fall back onto the Euler solution: EId4y/dx4+Pd2y/dx2=p.Ai.d2y/dx2P-pAi=π2EI/L2(5) After correcting for the Poisson effect they were able to determine the new critical axial force caused by the pressure. Unfortunately, however, the arguments set forth in [1] have been misunderstood. The fact that both axial force and lateral force multiply curvature does not make them forces of the same nature. Being able to add them has solved a mathematical equation, but still hasn’t converted the lateral force to axial. The authors wish to prove that [1] presents no more than a tool, which can be used in the analysis of global buckling problems of pipelines subject to both temperature and pressure. It will be shown, however, that this pressure will not produce an axial force, as now-a-days prescribed conservatively in many pipeline codes, which is even used for stress checking.

Galgoul et al. (2004) have written a previous paper in which they have pointed out the conservatism of the latest recommendations for pipeline freespan evaluations, associated to the way the axial force is considered in the determination of the pipeline natural frequency. First because it fails to consider the fact, that the axial force of a sagging pipe, subject to temperature expansion, is much smaller than that of a straight pipe. Second because the effective axial force caused by internal pressure should not be used to determine the pipeline natural frequency. Fyrileiv and Collberg (2005) also discussed this aspect. In order to back up their previous arguments the authors decided to perform some tests an axially restrained pipeline at both ends, which was pressurized in order to justify their claims that these pipelines are not only under tension (and not compression), but also that their natural frequencies increase instead of reducing, although they do bend out because of the pressure, reaching a point of instability. The authors understand the effective axial force concept and the enormous simplifications, which it brings to an otherwise cumbersome problem, but wish to emphasize that these advantages are not unlimited and that this is one of these restrictions. To back up the text results a finite element model has been produced, in which the internal pressure is taken into account as it actually is (and not as an axial force) to show that the pipe wall stresses can only be obtained correctly in this manner.

No abstract