Correlation of PVR, Octane Numbers and Distillation Curve of Gasoline with Data from a Thermal Wave Interferometer

Abstract

The ASTM standard methods for experimental determination of gasoline properties, such as octane numbers and Reid vapour pressure (PVR) are, in general, expensive, time consuming and cumbersome. Therefore, the study and development of faster and cheaper methods is of great industrial and scientific interest, particularly when one thinks of large numbers of gasoline samples to be tested. However, gasoline is a mixture of hydrocarbons and presents very complex chemical properties that require the use of several analytical techniques, and so the development of any new method will need to be validated through a series of statistical techniques. This contribution is aimed at the validation of a Thermal Wave Interferometer (IOT), conceived and developed by Vargas et al. at the Norte Fluminense State University — UENF, for the determination of octane numbers (MON and RON), distillation curve and PVR of gasoline and gasohol. The IOT determines these properties through a mathematical correlation of the thermal diffusivity of the components present in the gasoline, or gasohol, sample. This IOT validation study involved the use of a multivariate regression technique, namely SIMCA (Soft Independent Modelling of Class Analogy) for the classification of sample data and identification of outliers and the development of new correlation models for the prediction of gasoline octane numbers. In this study 97 samples of various types of commercial gasoline and gasohol, the latter containing up to 30% v/v of ethanol, were prepared and had their corresponding octane numbers (MON and RON) determined through the CFR motor method and their thermal diffusivities determined through the IOT. These experimental data were then correlated and a PLS (partial least squares) model, based on 89 thermal diffusivity variables, was developed for the prediction of the AKI, a parameter which is the mean value of the octane numbers (MON and RON) of a sample, i.e. AKI = (MON+RON)/2. Better results were obtained when the model was built for distillation curve points (Initial boiling point, 10% evaporated) and PVR data, and the thermal diffusivity data (30-870s) were centred at the mean value and smoothed, giving an RMSEC =1.832 and an RMSEV=2.270. These results indicate that the IOT is a promising, fast and cheap method for the prediction of AKI, provided the user takes into account the inherent errors of the experimental method. Thus, the IOT may be very useful as a tool for undertaking screening analysis, i.e., selection of gasoline, or gasohol, samples that must be submitted to the more expensive and time consuming CFR motor method.