A prediction model based on Group Contribution Method(GCM)was established for the prediction of flash points for pure-component organic liquid.According to the structure of chemical substances,the model predefined 42 function groups,and the contribution value(or weight)of each group to the flash point prediction model had been determined.The normal boiling point(NBP)of chemical substances and the amount of each function group were utilized as input parameters,as well as flash point value as output results in the prediction model.144 organic liquids of 10 classes in DIPPR 801 database were selected as experimental samples,Ordinary Least Square(OLS)method was used to fit the model.120 of samples were randomly selected as training sets and the other 24 as prediction sets.The results showed that the predicted flash point value of pure-component organic liquid was in good agreement with the experimental value.The Average Absolute Deviation(AAD),Average Absolute Relative error(AARE)and correlation coefficient were 3.67,1.18%and 0.983,respectively.The proposed model could provide technical guidance for the evaluation of flammable and dangerous properties of liquid chemicals at port site.
关键词
基团贡献法/有机液体/单一成分/最小二乘法/闪点预测
Key words
group contribution method/organic liquid/pure-component/ordinary least square(OLS)method/prediction of flash point
维普
万方数据