The safety of deepwater risers is essential for sustainable operation of offshore platforms. The structural health monitoring (SHM) system for deepwater risers is important to detect damage and perform repairs before failure occurs. Two main sources of damage are fatigue and corrosion. Failure of the riser would not only be an economical and environmental disaster, but also have far reaching consequences affecting communities. Combining global and local monitoring can greatly increase the accuracy of damage detection and fatigue estimation. Local inspection using robotic Magnetic Flux leakage (MFL) sensors is efficient and provides high resolution estimate of wall thickness changes due to corrosion or damage, while proposed vibration-based system identification can estimate global damage locations and fatigue life.
A new SHM system for deepwater risers was recently developed to monitor damage to the risers with both global and local monitoring methods proposed in this paper. The global monitoring is achieved by wavelet transform (WT) and second order blind identification (SOBI) method, from which, the likely location of fatigue damage is estimated. Once the location of the damage is identified by the proposed Wavelets/SOBI global identification method, local monitoring is performed using a robotic crawler with MFL sensors to further estimate the extent of the damage. Local monitoring with MFL sensors is verified by experimental results. Wavelets/SOBI global identification method is verified using Gulfstream test data.
Possible applications for the proposed SHM systems are for deepwater risers and deepwater platforms. A robotic MFL crawler can be used for in-line inspection for various pipelines. The proposed damage detection method and fatigue estimation can be adapted to other offshore structures, both fixed and floating. The proposed global method can also be used to analyze Tensioned Leg Platforms (TLP). To demonstrate the applicability of the proposed Wavelets/SOBI method to risers and floating platforms, verification using Gulfstream riser field data and TLP model data is presented in the paper.
A more efficient second order blind identification method for separation of uncorrelated stationary time series
