Feature extraction techniques for noisy distributed acoustic sensor data acquired in a wellbore.

Area of Science:

• Geophysics
• Signal Processing
• Petroleum Engineering

Background:

• Distributed Acoustic Sensing (DAS) offers real-time infrastructure monitoring capabilities.
• Field applications of DAS are challenged by significant background and environmental noise.
• Extracting specific signals, like gas signatures, from noisy DAS data is crucial for effective analysis.

Purpose of the Study:

• To develop and present computationally inexpensive techniques for real-time gas signature extraction from noisy DAS data.
• To validate these techniques on well-scale datasets representing multiphase flow conditions.
• To assess the impact of the techniques on signal quality and data size.

Main Methods:

• Implementation of a suite of computationally inexpensive denoising techniques.
• Application of these techniques to three well-scale DAS datasets with varying multiphase flow parameters.
• Analysis of gas slug signatures in the presence of high background noise.

Main Results:

• Successful optimization of gas slug signatures despite high background noise.
• Significant reduction in DAS data size achieved without compromising signal quality.
• Demonstrated effectiveness of the techniques across different gas injection volumes, fluid circulation rates, and injection methods.

Conclusions:

• The proposed denoising techniques are effective for real-time gas signature extraction from noisy DAS data.
• These methods enhance the utility of DAS for wellbore monitoring by improving signal clarity and data management.
• The approach offers a practical solution for overcoming noise challenges in field applications of DAS.