On-line multi-gas component measurement in the mud logging process based on Raman spectroscopy combined with a...

Area of Science:

• Petroleum Engineering
• Analytical Chemistry
• Spectroscopy

Background:

• Online gas analysis during mud logging is crucial for oilfield recovery, reservoir characterization, and anomaly detection.
• Current methods like Gas Chromatography (GC) and Gas Mass Spectrometry (GMS) are costly and time-consuming.
• Raman spectroscopy offers rapid, in-situ analysis but faces challenges with accuracy due to environmental factors and spectral overlap.

Purpose of the Study:

• To develop a highly reliable and sensitive gas Raman spectroscopy system for online, quantitative gas analysis in mud logging.
• To improve the accuracy and reduce detection limits compared to existing methods.
• To enable continuous monitoring of multiple gas components during oilfield operations.

Main Methods:

• Designed a gas Raman spectroscopy system incorporating a near-concentric cavity structure to enhance signal acquisition.
• Employed a quantitative model combining 1D Convolutional Neural Networks (1D-CNN) and Long Short-Term Memory (LSTM) networks.
• Integrated an attention mechanism (AM) to further optimize the performance of the CNN-LSTM model.

Main Results:

• The system successfully achieved continuous online detection of 10 hydrocarbon and non-hydrocarbon gases.
• Achieved low limits of detection (LOD) ranging from 0.0035% to 0.0223% for various gas components.
• The CNN-LSTM-AM model demonstrated high accuracy with average detection errors between 0.899% and 3.521% and maximum errors from 2.532% to 11.922%.

Conclusions:

• The proposed gas Raman spectroscopy system and CNN-LSTM-AM model provide a stable, accurate, and sensitive solution for online gas analysis in mud logging.
• This advanced method overcomes the limitations of traditional techniques, offering significant improvements in efficiency and cost-effectiveness.
• The system is well-suited for real-time monitoring in oilfield recovery, enhancing the identification of reservoir characteristics and drilling anomalies.