| The petrochemical industry is the basic industry of the national economy.Hydrogenated reactor effluent air cooler is an important cooling equipment in petrochemical industry.In recent years,the high sulfur,high nitrogen and high acid crude oil processing makes it quite hard to avoid unexpected safety accidents in air cooler.The research results showed that the ammonium salt crystal deposition and multiphase flow-induced corrosion in the air cooler are the important causes of tube bundle failure.Because of lacking corresponding flow corrosion risk assessment methods and characterization prediction models,there are limitations in the status monitoring,fault diagnosis,operation prevention and control of flow corrosion.As a result,the hydrogenated reactor effluent air cooler is always in high-risk service.Hence,it is the key to ensure the safe operation of air coolers to establish a data-driven model for predicting the characteristic parameters of flow-corrosion based on the research of flow-corrosion characteristics of air coolers.This paper takes a petrochemical hydrogenation reactor effluent air cooler as the research object.the risk assessment of the flow-corrosion state of the REAC system is carried out to establish the characteristic parameters corresponding to the failure model;based on the flow corrosion characteristic experiment,the influence of temperature,angle of attack,flow rate and other parameters on the corrosion rate of the pipe wall is study to build the flow corrosion rate database;the data-driven method is employed to establish a prediction model of the flow-corrosion characteristic parameters;the flow-corrosion monitoring system based on.NET for air cooler is developed to realize the real-time monitoring.The main research contents and innovations(1)Process flow of heat exchange and phase transformation of hydrogenation reaction effluent section is analyzed.According to the operation condition,the process simulation model is established.During the investigation,the air cooler has the combined failure risk of NH4Cl crystallization and flow-induced corrosion.Then the main characteristic parameters are determined,including NH4Cl crystallization temperature,corrosion rate,flow rate and others(2)A flow-corrosion characteristics system is designed.the corrosion rate of carbon steel under different experimental conditions is investigated by electrochemical workstation.The influence law of corrosion rate is described by curve fitting.The coefficient of determination value of the fitted curve is 0.94.Hence,it is reasonable to use this curve to construct the corrosion rate database(3)A data-driven prediction model of flow-corrosion characteristic parameters based on feedforward neural network is established.The fast pruning algorithm is used to optimize the number of nodes in the hidden layer of the model,and the training process of the model is improved based on the particle swarm algorithm.Finally,a data-driven model with 48 hidden layer nodes and an error of less than 5%is obtained.(4)The.Net framework is used to develop a flow-corrosion state monitoring system.Simultaneously,the corrosion rate database and data-driven model are embedded.The innovation of this paper contains:1)to use a self-designed experimental equipment of flow corrosion characteristics to investigate the influence of different temperatures,flow rates and NH4Cl concentration on the characteristics of flow-corrosion characteristics,and a database of flow-corrosion rates is constructed;a data-driven prediction model of the flow-corrosion characteristic parameters of the air cooler is constructed based on the feedforward neural network,and the prediction model is optimized by combining pruning algorithm and particle swarm optimization... |
PDF Full Text Request