Artificial neural networks for burst pressure strength of corroded subsea pipelines repaired with composite fiber-reinforced polymer patches / Mohd Fakri Muda

The application of Composite Fiber-Reinforced Polymer (FRP) patches for rehabilitating corroded subsea pipelines is a burgeoning field in offshore technology. However, the current finite element analysis-based modeling is time-consuming and lacks comprehensive defect coverage. This underscores the rising demand to explore predictive models for subsea pipeline repair, particularly in the oil and gas sector, to ensure sustained and stable operations. Developing an effective prediction model that utilizes Artificial Neural Networks (ANN) to correlate with the repaired assessment method, particularly composite FRP, can potentially overcome these limitations. Hence, this study aimed to present an effective method to evaluate the strength of repaired subsea pipelines to sustain burst pressure loads and determine the suitability of Composite FRP repaired assessment to multi-level corrosion in subsea pipelines using the finite element analysis and ANN modeling. The research methodology unfolds in three pivotal phases. Phase 1 is dedicated to the meticulous analysis of historical data, employing statistical techniques that align with relevant offshore codes. Phase 2 shifts the focus towards finite element modeling, providing deep insights into structural behavior. Finally, Phase 3 marks the development of an influential ANN prediction model, leveraging essential input data. The efficacy of the suggested method was demonstrated by comparing the output of the ANN with the historical FE output. A computational model for predicting the burst pressure strength of repaired pipelines with composite FRP patches was employed using the ANN algorithm. The geometry of corrosion damage was defined by three physical parameters, namely length, width, and depth. Finally, the computational model was validated by comparing the results with refined FE method solutions. Based on the results, it was observed that the composite repaired material study was ineffective when the predicted burst pressure decreased after the repaired analysis was carried out. In contrast, composite FRP repaired method was effective for defect sizes greater than 50 mm x 50 mm at any level of corrosion. Furthermore, the published ANN models were able to predict the burst pressure of the corroded and repaired subsea pipelines. In short, the proposed method was considered useful for developing a quick procedure for the composite FRP based repair scheme of corroded subsea pipelines.