Photoacoustic analyzer for natural gas and short-chain hydrocarbon isotopologues

The objective of this work is the development of a sensor based on photoacoustic spectroscopy (PAS) which, contrary to the usual field of application of PAS, namely the detection of trace gases, is capable of measuring concentrations of the main alkanes of natural gas as well as of main methane isotopologues. Knowledge of the natural gas composition, whose main part is methane (7599% onaverage), allows a variety of conclusions to be drawn, including its origin and energy content. Furthermore, knowledge of the main methane isotopologues shares of natural gas allows conclusions to be drawn regarding the natural gas maturity and the surrounding rock. The latter can be used to optimize the drilling technology. Since the absorption spectra of the main alkanes and isotopologues strongly overlap, the determination of the optimal wavelength of the laser source used is not trivial, as far as different molecule types have to be differentiated simultaneously. Therefore, a mathematical algorithm was developed which calculates the optimal wavelength based on the theoretical absorption spectra. The heart of the new sensor is an interband cascade laser (ICL), which emits in the mid-infrared range around 3.3 µm wavelength. By applying the injectioncurrent modulation, it is possible to achieve more precise measurement results in comparison to indirect modulation techniques. The behavior of the laser in this mode was initially unknown and has been analyzed in this work with a Michelson interferometer. Calibration mixtures were produced, using a self-developed gas mixer, whose photoacoustic spectra were calculated by the implemented digital signal processor (DSP) real-time signal processing, based on their photoacoustic signals. The resulting spectra were used to train the machine-learning algorithm Partial Least Squares Regression (PLSR), which subsequently determines the concentrations of the main methane isotopologues and the main alkanes, based on their photoacoustic spectra. The evaluation of the developed sensor was carried out on the basis of synthetic gas mixtures, as well as with the help of real natural gas samples from various sources. It could be shown that PAS is suitable to determine the main methane isotopologues and to analyze natural gas with respect to the composition of its main alkanes. The new sensor determines the composition of the main components of real natural gas samples based on PAS in real time, which enables continuous monitoring of the natural gas composition.