A probabilistic approach to pipeline start-up structure installations

Abstract

Subsea pipeline are extensively used for the transport of hydrocarbons from offshore wells, to platforms, pump stations and to onshore facilities. Because the installation of pipelines is time consuming it is responsible for a significant amount of the total costs of a project. Thus the workability of the installation is of great importance.When installing a subsea pipeline one always begins with a start-up structure, a FLET (flowline end termination) or PLET (pipeline end termination). The start-up structure is lowered through the moonpool via the pipelay tower until it reaches the seafloor. When it’s close to the sea floor the start-up rigging is coupled to the start-up pile with the use of a remote operated vehicle (ROV). Often the moment the start-up structure transitions from a vertical to a horizontal position with respect to the sea floor the loads on the stem pipe become critical with regards to the structural integrity of the pipe. And as such dictates the workability limits of the start-up structure installation. Pipe integrity is maintained via the use of a unity check equation which is described by the design standard DNVGL-ST-F101 issued by Det Norske Veritas Germanischer Loyd (DNVGL). In this equation, the combined loading criterion, the combination of the effective axial tension, the bending moment load and the water depth is evaluated for the structural integrity of the pipe string. The purpose of this thesis is to decrease the conservatism of the equation by probabilistic modelling of the resistance parameters – yield strength, ultimate tensile stress, wall thickness, outer diameter & ovality – instead of using deterministic nominal values and in the end allowing for higher sea states to operate in which in most situations increases the workability. For start-up structure installations DNVGL aims for a target probability of failure of 10-3¬ ¬.To achieve this first a well-documented load case was found in the Ichthys project. In HMC’s pipeline database the 18” Ichthys pipeline project offered 1106 geometrical and material strength pipe line data points. This data set was filtered analysed and used two describe the (bivariate) probability distributions of the resistance parameters. Analysing the data set it was found that the wall thickness and outer diameter and the yield strength and ultimate tensile strength showed a significance correlation. Dependence models have been defined by the use of copula’s. A performed sensitivity analysis showed that in the shallow water case, which the Ichthys project is, modelling the ovality as a stochastic variable has no significance influence on the outcome of the unity check. To assess the benefits of probabilistic modelling of the resistance parameters in a more general sense the base case Ichthys situation is altered to four different load scenario’s. Two shallow water cases and two deep water cases. For shallow and deep water, one case with the original sea state, in which the unity check is below 1. And one case in which the significant wave height is increased to push the unity check value to its limit of 1. After the input, the (bi-variate) probability distributions, and the test cases were defined the sample size for the Monte Carlo was determined to be 3*106 samples to guarantee the accuracy similar to what is used in current installation analyses. Performing the Monte Carlo simulations the results showed the expected conservatism in the current method. Where DNVGL aims for a probability of failure of 10-3¬, the probability of failure in the base case was calculated to be 10-5. Which allowed for finetuning and decreasing the safety class resistance factor used in the equation by 3% in the shallow water case and 4% in the deep water case. Which makes it possible to operate in heavier sea states and thus increases the workability of a start-up structure installation in certain situations.