← Back to Sessions

Learning Permeability from Acoustic Emission with Distributed Acoustic Sensing

Description: 

The shallow crust holds essential natural resources, including water, energy storage or extraction reservoirs, and mineral deposits, making it essential to understand how fluids flow through near surface geologic formations. Rock permeability is a very important property because it measures the capacity and ability of the formation to transmit fluids and controls the directional movement and flow rate of reservoir fluids in the formation. Establishing reliable permeability estimates of subsurface fractured rock is remarkably challenging when considering the heterogeneous variations in rock composition, subsurface stress conditions, and the spatial variations of a fracture network through a large volume of bulk material. However, if the fluid flow rate can be reasonably approximated and the pressure differential in a system can be estimated using passive surface measurements, then the bulk permeability is obtainable with a few assumptions. These types of remote measurements to reliably estimate permeability are needed to optimize the production of natural resources that rely on knowing the hydraulic properties that control fluid flow and transport in rock formations.

Distributed Acoustic Sensing (DAS) technology is proven capable of capturing fracture displacements in boreholes with the potential to map the surrounding fracture networks. In this work, a laboratory experiment was designed to test the how well acoustic emission recorded with DAS sensors can be utilized to determine the flow characteristics. The experiment is comprised of a spherical bead pack in a pipe with fluid injected at one end and collected at the other. Optical fiber is wrapped around the pipe and is interrogated with a DAS instrument. Input and output pressure and flow velocity are measured so that permeability can be calculated directly. The results show consistent scaling characteristics that correspond well to the pressure drop across the porous media. Additionally, a machine learning model is developed that is capable of inputting features derived from the continuous noise and outputting an estimate of the bulk permeability.

Session: Seismology for the Energy Transition - I

Type: Oral

Date: 4/16/2025

Presentation Time: 05:30 PM (local time)

Presenting Author: Carly

Student Presenter: No

Invited Presentation: 

Poster Number:

Authors