Determination of the composition of a mixture of gases by infrared analysis and chemometric methods
Résumé
We present the results of a chemometric analysis in the mid and near-infrared range for a gas mixture, with a composition close to that of natural gas. Different multivariate regression methods, i.e. principal component regression (PCR) and partial least squares (PLS), were used both with reference data from chromatographic analysis and with spectroscopic data, to find the best experimental spectroscopic conditions in order to achieve reasonable estimates of the composition gas mixture. The results shows that the PLS analysis in the selected near-infrared spectral intervals containing the first and second harmonics for the C-H stretching frequencies as well as in the spectral intervals corresponding to the frequencies for combinations of stretching and bending, gives an estimation in good agreement with those from chromatographic analysis. In the second part, we used the PLS calibration results to find the composition of natural gas samples. These preliminary results show that it is possible to determine the composition of the gaseous mixture by infrared analysis.