Evolution of reliable sensing systems which can survive, measure, and transfer data in the harsh environments has resulted in a tremendous increase of their application for integrity monitoring of different subsea components/structures in the past decade. Most of these sensory systems are designed for periodical and long term data measurement from their respective structures. However, they do not directly measure the degradation process. Instead, they intend to measure the structural responses which are affected by the degradation processes. Therefore, the raw measured structural responses should be further processed to identify the potential degradation either caused by aging or during an extreme event. Despite the significant evolution of the sensory systems, less attention has been paid to the data processing methodologies for the integrity monitoring. Different statistical pattern recognition methodologies have been successfully developed and used in the aerospace, automation and civil engineering industries for the purpose of Structural Health Monitoring (SHM). This study demonstrates application of a statistical pattern recognition technique to identify a potential damage in an umbilical line in Gulf of Mexico (GOM). Any potential damage in gas lift umbilical imposes serious environmental risk and shut-down period for production. The study uses sets of simulated sensory data to simulate the measured structural response of both intact and damaged umbilicals under different environmental loading conditions. In this regard, different sensor locations correspond to different Finite Element nodes along the length of the Umbilical. The damages are introduced simply by reducing the stiffness properties of the umbilical cross-section at hang off and touch down areas. The simulated strain time history data were pre-processed to potentially remove the effect of measurement noise and different loading. Damage sensitive features were extracted from the processed strain time history responses for intact and damage conditions of the umbilical under different environmental loadings. Autoregressive models were used for extraction of these damage sensitive features. The data from the intact conditions were used as the baseline data, and statistical analyses were performed to measure the variation of the damage features with respect to the intact condition of the umbilicals. The amounts of variations were investigated at the different sensor locations. The statistical measures were used as damage indices at different locations along the length of umbilical. Results show that the locations with large damage indices successfully corresponded to the areas with reduced stiffness properties. Although the proposed technique is implemented to monitor integrity of an umbilical, this technique can be used for risers and other flowlines to update near real time information about the integrity condition of these structures.
Bus real-time arrival prediction using statistical pattern recognition technique
