Numerical Modeling of Time-Lapse Seismic Experiments to Monitor CO\u3csub\u3e2\u3c/sub\u3e Sequestration in a Layered Basalt Reservoir

Instead of allowing carbon dioxide (CO2) generated from the burning of hydrocarbons to escape into the atmosphere, CO2 can be captured and stored. For the long term mitigation of the increasing amount of CO2 emissions, its sequestration in geological formations is promising. As a measure, its short and long term monitoring is equally important for environmental and health safety issues. The seismic method is proposed as a non-invasive monitoring technique for geological sequestration of CO2. Based on the positive results obtained from reservoir monitoring during enhanced oil recovery with CO2 floods, geoscientists plan to inject CO2 in layered basalt. CO2 mineralizes rapidly while it is exposed to mafic rocks. Due to CO2 injection, the physical properties of the reservoirs, such as the elastic moduli and the density change. These changes have effects on the seismic velocities. Although basalt seems to have advantages as a reservoir from the CO2 storage point of view, it poses some considerable challenges in terms of seismic monitoring. Strong multiple scattering from the layering of the basalt and high scattering attenuation complicate surface seismic imaging. To investigate the possibilities and limitations of time-lapse seismic monitoring of CO2 injection in a basalt reservoir, a number of numerical simulations are performed. The Spectral Element Method (SEM) and its modeling package SEM2DPACK is used for synthetic data generation. A base case (unperturbed) model is generated and the seismic velocity is perturbed in the reservoir to generate post-CO2 injection models. Among surface seismic, traditional Vertical Seismic Profiling (VSP), and downhole VSP methods used for monitoring purpose, results obtained from the downhole VSP method show significant differences between the pre- and post-injection seismic sections. Coda Wave Interferometry (CWI) theory is used to quantify velocity changes observed in base case and perturbed models. For a simple homogeneous model, CWI theory is highly applicable to detect small-scale time-lapse changes. For a heterogeneous reservoir, where the velocity perturbation is a function of space, CWI alone is not enough to quantify the time-lapse changes. Due to wave interference effects, the wave propagation paths yield complicated relations between the travel time and material velocities. In the Columbia River Basalt (CRB) Group of rocks, the proposed reservoir is bound by thick basalt layers and thus has a high contrast in acoustic impedance between the sedimentary interlayerings and the bounding basalt layers. For a source receiver pair both inside the reservoir (with low acoustic impedance) bound by layered basalt (with high acoustic impedance), the waves get trapped in between the layers for some early time. In this early time, the time-lapse travel time change has a linear relation with the velocity change, according to the CWI theory. But when the waves that leaked outside and traveled in the unperturbed area re-enter the reservoir, the relation is no longer linear. Also for the receivers outside the reservoir, the relation between the travel time difference and the velocity change is more complicated. In such cases, this relation is governed by a mathematical function (a sensitivity kernel) which depends on many factors such as the region of perturbation, the amount of time the wave spent in the perturbed region, source-receiver location, geology, intensity of the waves, energy reaching the receivers, and the path the wave has travelled. These factors make the derivation of the analytical solution of this kernel a complex task. However, the qualitative shape of this kernel is obtained