Grey Prediction of Corrosion Rate of Crude Oil Distillation Unit Pipelines in a Refinery

Serious erosion-corrosion was found in the crude distillation unit for proeessing middle east crude oil in a petrochemical company. The thickness of some positions like elbow was measured every month, and the thickness change was monitored. The monitored data series is quasi smooth sequence. The series after accumulated generating has quasi exponent law. The series meets the conditions for establishing gray differential equations. GM(1,1) model was established by grey system theory according to this series. The model was used for the simulated prediction of the elbow corrosion rate. The examined statistical data show that the average relative error was less than 15% and the prediction accuracy was above 85%. So, using GM(1,1) to predict the corrosion rate of crude oil distillation unit's pipelines is feasible and effective.