精馏塔顶馏分油终馏点软仪表建模探讨

Study on soft sensor prediction for naphtha end point of hydrotreating fractionator

在对炼油装置中精馏塔顶气相穹室内气相物质状态分析的基础上,依据理想气体状态方程、道尔顿分压定律、石油馏分平均相对分子质量与馏分油中平均沸点的关系,建立了精馏塔塔顶馏分油终馏点软仪表的一种三输入模型并确定了各模型系数的正负。该模型可实现石脑油或粗汽油终馏点的在线实时预测,为常压塔、裂化分馏塔先进控制技术应用提供了关键控制指标的实时在线测量。通过对某公司4套装置的石脑油或粗汽油终馏点软仪表预测结果及预测偏差柱状图分析可知,预测曲线的趋势与化验数值变化趋势一致,大多数样本的预测偏差能控制在2.5℃以内。目前还存在在线分析仪存在投资较大、故障率较高、维护强度大的缺点,应不断地探索建立更加准确、可靠的软仪表模型。

Through the analysis of the status of gas phase material in the overhead gas phase chamber of distillation column and on the basis of gas phase equation,Dalton＇s law,the relationship between average molecular weights and average boiling point of oil fractions,a model is obtained for the end point of overhead fraction of fractionators,and the positive and negative of model coefficients are determined. By using this model,the end point of naphtha or crude gasoline can be predicted on real time,which provides real-time on-line measurements of critical items for the application of advanced control technology for ADU,VDU and cracking fractionators. It is known from analysis of the predictions of end points of naphtha or crude gasoline from 4 process units of a company by software instruments,the prediction curves are consistent with laboratory data.The prediction deviations of most samples can be controlled within 2. 5 ℃. At present,the on-line analyzer is big in investment,high in accident rates and high in maintenance,the more accurate reliable instrument models should be established for application.