This paper reviews the current methodology for assessing pipeline embedment against field data, where embedment levels were under-predicted leading to an unconservative design. An updated model is provided for predicting pipeline embedment. One important principle behind this new approach is to define the operative soil strength, based on reconsolidation of the remoulded soil beneath the weight of the empty pipe during the intervening period between laying and flooding. This approach is based on pipeline embedment data from the field, which has demonstrated that soil around the pipe can become fully remoulded during installation. Therefore, embedment in the flooded condition should be assessed using an operative strength from reconsolidation, rather than the intact strength, which is current practice. This paper also provides current equations that improve the prediction of embedments over one-half diameter, by proposing methods to take account of the increasing buoyancy with depth, the reducing influence of heave mounds and modified penetration resistance at deeper embedments.
Pipeline embedment is a fundamental input to the assessment of pipe-soil interaction, which is the largest uncertainty faced in the design of pipelines subject to lateral buckling and walking phenomena. However, pipeline embedment is notoriously difficult to predict due to the inherent uncertainty of the installation process. Improving the prediction of pipeline embedment is critical to reducing the range of predicted pipe-soil responses in design.
A current project has compared high quality observations of pipeline embedment from an earlier phase of the same field development, against predicted embedments using more recent, high quality soils data from parallel pipeline routes, including fully remoulded soil strength from cyclic penetrometer testing. While current design approaches were under-predicting embedment levels, leading to a potentially unconservative design; this updated methodology provided a much-improved match to actual embedment data. The revised model was successfully verified against measured levels of pipeline embedment and is now being used for the design of future pipelines in the area, with improved certainty and easing of the design challenge.
Under-predicting pipeline embedment could be critical to design integrity. The methodology presented in this paper has the potential to improve assessments of embedment and avoid unconservative pipe-soil responses on future developments.