折半搜索法预测二元混合物的共沸点

Binary search method for predicting azeotropic temperature of binary mixture

在化工分离过程设计中,共沸点预测的作用十分重要,目前常用的方法有牛顿迭代和牛顿同伦等,都需要求解大型非线性方程组,且牛顿迭代法易发散,本文提出修正UNIFAC模型的逐次代入法同时与折半搜索联合的算法,既克服牛顿迭代计算时因取初值不合适时而容易发散的缺点,又不需要求解大型非线性方程组,且计算速度快,计算中对逐次代入法进行改进,使温度初值的取法更简捷,且无发散现象,通过验证乙醇,苯等10多种二元混合物,计算过程均可在1MS以内完成,计算所得共沸点与文献所载实验值比较,平均误差<1%,共沸点组成与文献所载实验值比较,平均误差<2%,证明该法不但可用于二元混合物共沸点预测,又可在相应大型数据库中查找可产生共沸效果的混合物。

To predict Azeotropic temperature is very useful in separating operation.The methods to predict Azeotropic temperature usually are Newton iteration and homotopy-Newton hybrid method.These two methods need to solve the system of nonlinear equations.This article provides a method of predicting azeotropic temperature.This method is a complex method union with successive substitution method and binary search based on modified model of UNIFAC.This method not only overcome the defect of divergence of Newton iteration,but also can calculate rapidly,and it needn＇t to solve the system of nonlinear equations.In the process of calculating,not only the initial data of temperature is get easily and in effect but also overcome the defect of divergence.Through calculating more than ten binary mixtures such as ethanol-benzene etc.,they can be finished within 1 MS.The result sets out that the method can be used to search Azeotropic mixture based on database.