催化裂化装置烟机结垢总量表征及软测量模型研究

Characterization and soft sensor model of fouling totality in flue gas turbine of FCC unit

烟机结垢是困扰烟机安全运行和长周期运行的难点。烟机结垢不可直接观测,很难量化,增加了结垢分析的难度。为了定量分析烟机结垢总量,首先设计一个间接指标即能效因子表征烟机结垢总量,进而建立能效因子的动态软测量模型;在假设结垢速率与影响因素为线性关系的基础上,采用主元回归表达结垢量的增加速率;依据催化裂化装置的历史数据,经回归计算得到速率方程中的各个主元的系数。经检验,能效因子与主要可控可调变量的相关度小于10%,表明能效因子几乎不包含可控可调因素的信息,能效因子作为表征量是合理的;软测量模型的自检相对误差和预测相对误差均小于4%,表明动态软测量模型的准确性。该软测量模型的结垢速率相关性分析表明,烟机结垢并没有一个主导型的因素,而是多种因素综合作用的结果。

Fouling of flue gas turbine in FCC unit is a difficult problem for safety and long period operation.Fouling of flue gas turbine couldn’t be directly observed and is difficult to quantify,which makes scaling analysis more difficult.In order to quantitatively analyze fouling totality of flue gas turbine,an indirect index(energy efficiency factor)was designed to represent the fouling totality,and then a dynamic soft sensor model of energy efficiency factor was established.Based on the linear relationship between fouling rate and influencing factors,principal component regression was used to express the increasing rate of fouling totality.Based on the historical data of FCC unit,the coefficients of each principal component in the rate equation were obtained by regression calculation.Upon examination,the correlation degrees between energy efficiency factor and the main controllable variable was less than 10%,which showed that the energy efficiency factor almost did not contain the information of controllable variable.It is reasonable to use energy efficiency factor as the characteristic parameter.Self checking relative error and prediction relative error of dynamic soft sensor model were less than 4%.It shown the accuracy of dynamic soft sensor mode.According to the correlation analysis of fouling rate based on the soft sensor model,fouling was the result of multiple factors rather than a dominant factor.