Chemometric prediction of swelling potentials of aviation fuels towards elastomers using various spectroscopic data

Abstract

This work presents the development of chemometric models, based on spectroscopic data, predicting the swelling potentials of aviation fuels towards three ready for use acrylonitrile-butadiene-rubber (NBR) types exhibiting acrylonitrile contents of 18, 28 and 39%. For this purpose quaternary model substance mixtures, containing compounds, which are typical for aviation fuels were prepared. As a second set, exclusively aliphatic synthetic aviation fuels were blended with varying amounts of an aromatic fraction. After immersion in the prepared fluids, the extent of the equilibrium mass uptakes by the elastomers was determined and interpreted in terms of different mixtures’ formulations. UV/Vis- and infrared (IR) spectra were collected of all liquids. Additionally, gas chromatography/mass spectrometry (GC/MS) analyses were carried out. Based on the obtained data, sum mass spectra, which are basically a merging of all recorded mass spectra of a chromatographic run, were generated. Spectroscopic data and swelling characteristics were correlated using partial least squares regression (PLS). Here two different models, based on both training sets were established for each spectroscopic method. Obtained regression coefficients underwent a detailed evaluation to get a better understanding of factors influencing the prediction of swelling potentials. It was found, that they reflect well-known relations between the composition of a liquid, represented by specific spectral information and their corresponding swelling potential. The models were validated by estimating swelling potentials of two commercial jet fuels Jet A-1 and ReadiJet. Infrared spectra pose the most reliable base for the chemometric prediction of swelling characteristics of NBR containing 18% acrylonitrile with prediction errors of 5 and 2% for Jet A-1 and ReadiJet, respectively. Also sum mass spectra allow a relatively accurate estimation.