含硫气田集输管道腐蚀预测软件应用

Application of corrosion rate prediction software in sour gas gathering and transportation pipeline

关键词:  含硫气田  管线腐蚀预测  腐蚀速率  BP神经网络  腐蚀影响因素

Key words:  sour gas field  corrosion prediction for pipeline  corrosion rate  back propagation neural network  corrosion affecting factor

摘要:

针对川渝地区含硫气田集输管道材料腐蚀问题,结合几种经典的、应用较广的腐蚀速率预测模型进行了对比分析。根据对影响腐蚀的主要因素CO₂分压、H₂S分压、液体流速、运行温度等开展机理分析,推导建立了一种腐蚀速率预测半经验模型,并通过试验数据确定了模型中的待定系数。在此腐蚀预测模型的基础上,开发了基于BP神经网络算法的含硫气田集输管道腐蚀预测软件。应用该软件对川渝地区某气井集输管线腐蚀情况开展了预测,对预测结果与同等参数条件下的腐蚀挂片试验实测结果相比较,软件计算准确度≥90%,应用效果较好。 

Abstract:

Aiming at the material corrosion in gathering and transportation system in sour gas fields, this article analyzed contrastively several classical and widely used corrosion rate prediction models. Essential factors such as partial pressure of CO₂, partial pressure of H₂S, liquid flow rate, temperature have been taken in consideration to develop a semi-empirical prediction model for corrosion rate. The undetermined coefficients in the model are determined by the experimental data. Based on this corrosion prediction model, a prediction software has been developed by back propagation(BP) neural-network algorithm. The software is utilized to predict corrosion rate of the pipeline material in a certain gas field in Sichuan/Chongqing area. The prediction result compared with coupon test result under the same parameter condition, the prediction accuracy of software is higher than 90%, which means it has a good application effect.