KAUST


Farooq's group for Advanced Sensing Technology & Energy Research 
logo100(FASTER)

Conventional methods of gas chromatography and isotope-ratio mass spectrometry require significant time for analysis and do not support real-time, high-resolution decision-making in the field. Our laser-based system addresses this gap by providing a rapid, in-situ, and non-intrusive method for natural gas analysis.

Optical setup and CNN model flow chart

The sensor uses a mid-infrared (MIR) DFB-ICL near 3.3 µm to target the vibrational bands of C1-C5 species, arising from C-H stretching motion. This range allows for comprehensive coverage of all alkanes' absorption features. CNN auto-encoders are employed to distinguish overlapping absorbance spectra of the different alkanes. The system's accuracy was validated in laboratory tests, achieving a 5% relative error, making it a robust, calibration-free, non-intrusive solution for natural gas analysis.

(a) Mirrored comparison between measured and simulated absorbance of a mixture comprising 30% methane and 3%
ethane at different pressures, illustrating the agreement between measured and simulated data. (b) Evaluation of CNN model performance on both the remaining 30% of the split and 7 experimental mixtures.

Impacts:
  • Operational Efficiency: Enables real-time decision-making for natural gas production.
  • Cost Reduction: Minimizes the need for recurrent calibration and delays associated with laboratory analysis.
  • Field Applicability: Robust and non-intrusive, designed to operate under varying pressure conditions, making it suitable for on-site monitoring.

EXPEC Advanced Research Center, Saudi Aramco