The Study On Corrosion Information Of Atmospheric Tower By Fusion Technology

Atmospheric equipment is the fist set of crude oil processing, and it determines the normal operation and yield of oil refinery. Atmospheric tower is very important equipment in oil refinery, and the most serious corrosion happen at the low temperature system of atmospheric roof. As a result of corrosion, a lot of resources have been wasted for repairing equipments or changing new ones, even shorten the production cycle. Lack of monitoring and treatment for corrosion information are important factors causing short production cycle, however there is not a method and theory to processing and analysis variety of data, short of guiding role to safety assessment of corrosion.Analyze the process of atmospheric equipment in YanChang refinery, establish a monitoring system, including confirm the production process parameter and parameters of corrosive substances and corrosion monitoring parameters, set the data collection point, select probes and choose the methods to installation those.Divid the top of the Atmospheric tower system into several subsystems, such as top of trays and wall in the tower, connecting pipes, air cooler, and reflux accumulator. Analysis the factors of production process parameter and parameters of corrosive substances how to affect the corrosion monitoring parameters, then establish database for plenty of parameters, integration of computing three kinds of parameters above.Using MATLAB neural network to fusion the parameters which were collected form the top of the Atmospheric tower system, comparison of different algorithms for network training and get the speed and accuracy. LM algorithm achieves convergence very quickly but forecast errors are large. With BP algorithm getting convergence need a long time, and forecast errors between30-50percent. Using RBF neural network to calculate, convergence speed is faster and forecast error is less than15percent. So RBF neural network was the best one to forecast corrosion.Establish corrosion forecast system for the Atmospheric tower roof, complete the interface of the human computer interaction and choose the best forecast method for Atmospheric tower of the second unite workshop in YanChang refinery, the material and type of the equipment as a precondition, input acid value of the oil, total sulfur content, temperature, pressure, content of chloride ion, content of hydrogen sulfide, pH, predicted the content of the ferrous ions. With the change of the various parameters in the fluctuations, technical staff could get the corrosion predictions quickly and accurate, then they can control the fluctuation of process and guide the dosage of the inhibitor, in order to provide the basis for corrosion safety assessment.