Mean Load Impact on Mooring Chain Fatigue Capacity: Lessons Learned From Full Scale Fatigue Testing of Used Chains

Fatigue of mooring chain is for many floating offshore installations a limiting factor in design. With aging installations and the need for field life extension beyond the original design life, questions on mooring chain endurance are raised. Current SN curves utilized in fatigue limit state (FLS) calculation are based on full scale testing of new chain, performed at a high mean load level (20% of the chains minimum breaking load (MBL)). The high mean load level in the tests do not correspond to the conditions for many chains in operation, as mean load in fatigue relevant seastates are often significantly less than mean load used in the new chain fatigue tests. Mooring chains in operation also experience different degree of corrosion, both general corrosion and pitting. Surface roughness and corrosion pits contribute to crack initiations, and thus reduce fatigue capacity. Fatigue tests with new chain condition cannot be assumed representative for corroded chains. As part of mooring integrity programs, Equinor has been replacing mooring chains since year 2000. To assess actual fatigue capacity, many chain segments have been full scale fatigue tested. First tests started in 2011, and the tests cover different degrees of corrosion. The tests have been performed at typical mean load levels relevant for operation of the installations, which for most cases are less than 20%MBL. From these tests it is observed that fatigue capacity in some cases are better than expected for new chain, even for chain segments with significant corrosion. Fatigue test results show a large effect of the mean load. For test cases with significant corrosion and high mean load (20%MBL), a significant reduction in fatigue capacity compared to new chains is found. This paper presents some of the fatigue test results on used chain, highlighting the effect of the mean load for the given chain conditions. Effect of corrosion at mean load of 20%MBL is also included. The paper discusses some of the underlaying causes for the mean load dependency.