Rock Berms as Axial Restraint for Flowlines on Soft Soil: Improving Design Through Centrifuge Modelling

Abstract Rock berms are used to restrain flowlines from moving axially by adding vertical load to enhance resistance. Should the contact pressure be reduced at the flowline-berm interface, such as in response to ‘arching’ of the rock berm due to pipe settlement, this resistance may be substantially reduced. This study shows how the development of axial restraint forces is complex, with shear (‘friction’) at the pipe-berm and pipe-soil interfaces influenced by system settlement. We examine the restraint provided by rock berms through three phases of centrifuge model testing. Phase 1 isolates the pipe-berm interaction through ‘trapdoor’ tests on a false (rigid) seabed using actual berm profiles and scaled rock. Phase 2 models a slightly overconsolidated clay seabed with load from simulated berms held constant, thereby isolating the effects of pipe settlement and cyclic hardening at the pipe-soil interface. Finally, Phase 3 - which most closely represents the actual field behaviour - models the berm/pipeline/seabed system to investigate the combined effect of pipe settlement, arching and frictional response. The trapdoor tests identify that berm arching can occur in situations where simulated settlement of the flowline leads to redistribution of the rock berm weight, such that the berm weight acting on the flowline falls towards zero. Axial cycling is shown to help recover the berm's effectiveness, but is dependent on the amount of settlement - with greater settlements generating less axial restraint recovery. The constant vertical load tests confirm that axial resistance developed along the pipe-soil interface is strongly influenced by the vertical load delivered by the rock berm. Testing showed that axial resistances increase with consolidation and hardening, and also that settlements were modest and observed to be a function of vertical load. Axial resistances developed in the final phase of testing were the highest of all, as the tests include all restraining actions. These tests suggest that arching does occur under ‘realistic’ conditions - but that the effect is modest and largely eliminated by ongoing cycling. The findings reduce current design uncertainties and have already been incorporated in an offshore project where rock berms are being used to mitigate axial movements of a flowline. In particular, this novel centrifuge modelling confirms the potential for berm arching and loss of restraint, but also shows that arch collapse leads to recovery of flowline restraint - enabling the potential to reduce rock berm volume compared to the case of assuming arching leads to permanent loss of resistance.