Near-infrared multicomponent analysis in the spectral and Fourier domains: energy content of high-pressure natural gas

Abstract

Four multicomponent analysis methods have been evaluated: inverse Beer's law with optimal frequency selection, principal component regression, partial least squares, and principal component regression where the spectra have been preprocessed with a Fourier transform. The methods were applied to a set of near-infrared spectra of mixtures of methane, ethane, and propane, measured at pressures of 100, 250, 500, 750, and 1000 psi, to assess the feasibility of monitoring the energy content of natural gas spectroscopically. There are two approaches to the energy monitoring problem. With a standard multicomponent procedure, the first step is to determine the contribution each component makes to the total; then the contribution are summed to yield the total energy. Alternatively, the energy content may be found directly by using a P-matrix formulation of the problem. For either approach, superior energy values were obtained, with average percent errors less than 0.5%, when the spectra were Fourier transformed prior to processing.