Research On Corrosion Model And Evaluation In Water Recycle Heat Exchanger System For Crude Refining Equipment

Corrosion issues have serious impact on crude refining equipment’s safe, stable, long-term operation of water recycle system in oil refinery and petrochemical enterprises. There are many corrosion types in refining enterprises, each corrosion factors are different from others and the relation between them is very complicated. It is useless to prediction and control the corrosion of heat exchanger system by normal test. Study the present situation ofrecycle heat exchanger system corrosion has great importanceof refining equipment’s stable operation and economical issue.Information fusion is an information processing technology which analyzes multi-information completely. This paper divides corrosion information into three parts:process parameters, corrosive substanceparameters and corrosion test parameters. Using information fusion to comprehensive deals with various parameters, and builds corrosion prediction model.It can reduce the data error impact on prediction precision by using neural network which based on Grubbscriterion to deal with the crude data. The models are built on three neural networks:BP neural network, RBF neural network and user-defined neural network. The input variable is corrosive substanceparameters, and the output variable is corrosion test parameters. It compares training time and prediction precision with three neural networks. The result is that BP neural network is the best method in building recycle heat exchanger corrosion model.This paper uses five methods toqualitative and quantitative analysis different corrosion factors based on corrosion prediction model. Although the results are different, turbidityand hardnessare the most important factors; and then is phosphorus content; pH, conductivity and basicity are relatively small.Corrosion prediction softwareis developed in this paper. There are three main functions: predict corrosion test parametersfast and accurate, self-optimizing corrosion prediction model when increase the data samples and optimize the control of corrosion.Information fusion provides an effective analyzing method on corrosion prediction and control. Based on corrosion prediction model and sensitivity analysis, we can provide the basis for recycle heat exchanger corrosion.