Grey analysis of corrosion on the overhead condensing system

Abstract

Corrosion is a serious problem in atmospheric distillation unit when the plant refines high-acid crude oil from Shengli Oil Field. How to measure and predict the corrosion and related factors is a significant task in refinery. The GM (1, 1) model can only predict the sequence of exponential law, scribe the monotonic change event. We take acid values as a sample, use the data from the first to 6th month, and establish GM (2, 1) model and Verhulst model to predict the data from 7th to 9th month. For the GM (2, 1) model, the maximum relative error is 11.8%, the mean relative error is 7.2%, and the mean relative error of prediction data is 13.2%. For the Verhulst model, the maximum relative error is 7.0%, the mean relative error is 3.1%, and the mean relative error of prediction data is 10.6%. The conclusion is that the Verhulst model is better than DGM model and the Verhulst model can be used for corrosion pate prediction.