Corrosion in Wet Gas Piping: Root Cause, Mitigation, and Neural Network Prediction Modeling

摘要: This paper discusses the root causes and operational mitigations of corrosion anomalies reported for an FPSO wet gas system, crucially, proposes a neural network (NN) prediction model. The NN model involves ‘back-propagation’ processing each nodal cause mitigation to obtain value which when combined with weight then summed, provides output value. is used further adjust weights. Each correlates magnitude influence on overall rate. ability train (i.e., weight-adjustment during processing) makes it responsive adaptable, such that fresh data inputs are made in ‘forward-propagation’ mode, into large modeling database has been developed (which includes number susceptibility factors), significant increases accuracy predicting rate integrity behavior system can be achieved. identified will useful understanding internal degradation mechanisms operating systems general.