Correlations Between Octane Number And Autocorrelation Topological Index Of Hydrocarbon Structure In Gasoline

With the rapid development of national economy and the increase of the people's living standard, the use of automobile is more and more popular and the number of automobile is also ever-increasing. However, the tail gas of the cars becomes the chief pollution source of atmospheric pollution and has great influence on people's health and living environment. The increasing compression ratio of automobile engine is one of the effective measures to improve engine efficiency and protect the atmospheric environment, but improving the antiknock properties of gasoline fuel is the premise of the improvement of compression ratio. Octane number(ON) provides a sensitive indication of the anti-knocking behavior of the gasoline fuel, the higher the octane number, the lower the tendency for a gasoline to produce a knocking sound in an automobile engine. In the past several decades, people have attempted to thoroughly reveal and understand the intrinsic relationship between the structures of hydrocarbon and their octane numbers, and predict the octane number of gasoline. In this paper, the relationships between topological structure and octane numbers of hydrocarbon (alkane, cycloalkane and aromatic) are analyzed in detail by using the autocorrelation topological indexes and multivariate linear regression method, which provides theoretic foundation for the further exploration the relationship between structure and anti-knock properties.Relationships between the autocorrelation topological indexes and octane numbers of 29 alkanes are researched by multivariate linear regression method. Among four methods of C_node-C connection and two methods of C_node-H connection, there are quite significant correlations between octane numbes and the topological indexes calculated with the square root of carbon atom number or hydrocarbon atom number connected with node. Neglecting the high-order terms of topological indexes, namely F(5) and F(6) terms, and adding the number of carbon atom which reflecting the size of molecule, the correlation coefficients of the fitted equations change little. When the autocorrelation topological indexes of C_node-C connection and C_node-H connection are considered at the same time, the fitting results are further improved, the correlation coefficients are 0.99 and the average errors are less than 2.5. When taking into account 43 molecules, the correlation coefficients of fitting between the octane numbers and topological indxes calculating with the square root of atom number connected with node.For 34 cycloalkanes, the eigenfunction obtained from the square root method is suitable to describe the topological structure of cycloalkane molecules. The correlation coefficients of multivariate linear regression between the autocorrelation topological indexes and octane numbers are 0.9354(MON), 0.9464(MON), 0.9324(RON),0.9126(RON). The neglect of the high-order terms and the consideration of square root of the number of carbon atom have little effect on the fitting results. When the two topological structures of C_node-C connection and C_node-H connection are considered at the same time in multivariate linear regressions, the correlation coefficients obtain further improvement. To study the aromatic molecules of having experimental octane number, the fittings of the autocorrelation topological indexes of the inverse of carbon-carbon connection to the motor octane numbers present high correlation coefficients, and it is 0.9537. For the olefin molecules, the correlation between the octane numbers and topological structure reflected with autocorrelation topological indexes is not good. Inferring form the bond energy of molecules, the correlation coefficients between the octane numbers and topological indexes calculating with the square root of bond gibbs free energy are less than the results of calculating with the number of atom connecting with node. When the topological indexes are computed with the square root of the product of C-C bond gibbs free energy and the number of atom connecting with node, the fitting results are similar to the outcomes of considering the number of atom connecting with node or bond gibbs free energy, respectively.