Tomography of a carbonate core with ultrasonic seismic sources and Laser-Doppler Vibrometer at the laboratory scale

Carbonate reservoirs have been drawing the attention of the energy industry for decades. Quantitative information on carbonate rocks remains a challenge in exploration geophysics as these porous and fractured rocks are generally heterogeneous and anisotropic at differents scales. The interest in investigating such rocks spans many applications, for example, monitoring underground water resources and geological CO₂ storage.

We aim to study carbonate rocks at the laboratory scale that can help with large-scale seismic interpretation. This is a promising experimental approach that could lead to significant progress in the development of imaging methods for the subsoil at different scales. We developed an automated experimental prototype involving a point-like pulsed-laser (PL) or a piezoelectric transducer (PZT) as seismic sources and a single-point Laser Doppler Vibrometer (LDV) as a receiver for efficient and high-resolution seismic data acquisitions on core samples. The MHz frequency-range seismic signals recorded by the LDV are used to study V_p inside the carbonate core slice through first-arrival travel-time tomography. The seismic tomographic images obtained from both the PL-LDV and the PZT-LDV datasets are compared with an X-ray CT-scan image of the carbonate core. In parallel, numerical tests on synthetic and decimated data are run to study the hyperparameters and the resolution of the tomography tool, which helped us to establish an optimal inversion strategy on real data involving a multi-grid approach. The tomography results are completed with a sensitivity analysis through spike tests.

Synthetic and spike tests have concluded that both the PL-LDV and PZT-LDV setups may correctly reconstruct the P-wave velocity, however, the tomographic V_p images from the experimental PL-LDV and PZT-LDV datasets are different even though they share some common trends and patterns. The X-ray CT-scan image shows that the V_p model retrieved from the PL-LDV dataset is in better agreement with the CT-scan image and confirms these trends and patterns in general.

We will discuss the reasons for which the two experimental results are different, with the help of tomography on both the experimental and numerical data after decimation. The point-like feature of the PL seismic source may have contributed greatly to the higher 2D tomography resolution, among other advantages of the PL source: it should have allowed to minimize the potential bias caused by 3D information.

Therefore, we validated an experimental prototype featuring a PZT source and a pulsed-laser source for high-resolution measurements in laboratory as well as a tomography workflow: we propose an original and efficient geophysical core-slice-probing configuration based on the PL-LDV set-up leading to a more accurate tomographic P-wave velocity reconstruction as compared to the PZT-LDV set-up.