Validation of Soil Models for Wellhead Fatigue Analysis

Abstract

The focus on wellhead fatigue has increased over the last decade, both in terms of consequences of failure and methods for prediction. Wellhead Fatigue is a well integrity concern when drilling subsea wells, especially with exposure to harsh environments and extreme environmental loads. The concern increases with the use of deep water drilling rigs in shallow water. As a result, full-scale measurement has been employed in several projects to document the actual load levels experienced by the subsea wellheads during drilling. Input data uncertainty has always been a challenge when using global analysis to estimate wellhead fatigue. Instrumentation opens new possibilities to validate the global analysis results. In several measurement campaigns, it is observed that the response below the lower flex joint of the drilling riser is overestimated in global analyses. It has been suggested by some that this is an indication that global riser analyses are highly conservative. However, as suggested in previous papers (i.e. Russo et.al, ref.[11]), this discrepancy could also be explained by non-appropriate modelling of the conductor lateral soil resistance for small displacements, leading to underestimation of the soil stiffness. The soil spring model also called p-y curves are usually built following the API recommended methods that are established for foundation piles. Piles are designed for ultimate limit state focusing on displacement conditions that are not optimal for fatigue analyses, as a large part of the total fatigue damage actually occurs for small displacements. A literature review is conducted, to review the basis for the API springs, and alternative p-y-curves with increased initial stiffness have been suggested. Based on the available information four alternative soil models have been proposed. The work performed by BP on p-y curves modelling for laterally loaded conductors (ref. [2]) has been an important input for this paper. In order to illustrate the effect of initial soil stiffness in the global analysis, the present study focuses on conductors installed in homogenous and normally consolidated to slightly overconsolidated clays. This limits somewhat the number of available sites with relevant conditions for full-scale measurements, at least on the Norwegian Continental Shelf, where it is common to find layers of sand interspersed between the clay layers. However, Statoil have conducted one campaign with full-scale measurements at a location with corresponding clay conditions. In this paper, the API formula for “soft clay” and four alternative soil models, have been used as input to a global riser analysis, and the results are validated against measurements. It is the response of the lower stack, in terms of rotations and displacements of BOP, LMRP and LRS, that has been investigated. In addition, the load, in terms of wellhead bending moment has been compared. Results shows that for this given case, the Matlock-API formulation overestimates the lower stack response, compared with full-scale measurements. Comparing the proposed soil models shows that the global response is affected by selection of soil model. The soil formulations outlined by Jeanjean (2009) and Zakeri et.al (2015) give the best match with full-scale measurements for this case.