Improving our ability to detect CCS related changes in rock properties from 4D seismic data
By BIGCCS PhD student Dawid Szewczyk, NTNU
Carbon capture and storage (CCS) is an operation involving capturing CO2 from large sources, transporting and depositing it into storage sites. Fluid injection during subsurface CO2 storage leads to many changes in the reservoir and its surroundings. Among many others, changes in pore pressure and stresses experienced by the formations, transitions of fluid saturations, fluid substitution, modifications of the pore-fluid distribution or temperature changes alter the properties of rocks. Those changes may lead to a number of problems connected with the performance of the storage sites and their stability. Indeed common fears connected with CCS projects are the potential rock failures and associated leakages. Therefore, monitoring of the formation during and after injection is necessary. For this purpose the 4D (Time-lapse) seismic represents the most proven and reliable method. 4D seismic involves acquisition, processing and interpretation of repeated over time seismic surveys. Interpretation of such data among others is based on rock physics models. With development of new experimental methods and increasing understanding of rock physics phenomena, the rock physics models become more and more accurate. Yet for the time being there is still a number of challenges associated with interpretation of acquired 4D seismic data that needs to be addressed before we will be able to access a full picture of the reservoir behavior during CO2 storage.
Beside previously mentioned effects connected with changes of stresses and fluid states within storage sites there are also more basic questions concerning interpretation of seismic data. One of such questions is linked with dispersion of acoustic waves in rocks i.e. frequency dependency of the acoustic wave velocity and as a consequence frequency dependent properties of rocks. Typical frequencies of sound waves used during field surveys are below 100 Hz. Taking into account the properties of the rocks, the wavelengths associated with such waves may be in order of several hundred meters. This fact makes it impossible to propagate such seismic pulse through a few centimeter long laboratory samples. It is also a reason why most of the available experimental data has been acquired within ultrasonic frequency range. Taking into account that rock physics models used in the interpretation of 4D data are mostly developed or validated based on the laboratory experiments, it becomes clear that linking properties of rocks obtained with low and high frequencies plays the key role in monitoring of CO2 storage site performance.
First attempt to access the discrepancy between low (seismic) and high (ultrasonic) frequency measurements of rock properties was done by Spencer in the early 1980’s [Spencer, 1981]. The force oscillatory technique, now known as the quasi-static or low-frequency technique, allowed determination of Young’s modulus and Poisson’s ratio at seismic frequencies. Within few places around the globe where quasi-static measurements can be made, probably the most state-of-the-art low-frequency apparatus had been built in the laboratory of SINTEF Petroleum Research. It allows for determining static, quasi-static and ultrasonic properties of tested specimens under biaxial stress conditions within a single experiment. We are utilizing this equipment to study the effects connected with modifications of fluid saturations and stresses experienced by the formation for the typical reservoir and overburden materials under a wide frequency range.
We have performed series of experiments in which dispersion in Mancos Shale and Pierre Shale (outcrop samples) was investigated for different water saturations and stress states. Water saturations have been established by exposing shale core plugs to different relative humidities (RH), ranging from 11% to 100%. In addition, for Mancos Shale, oven-dried samples were studied. In all tests performed with Mancos Shale, pore pressure was kept at atmospheric pressure while confining and axial stress were varying depending on type of performed experiment. In case of Pierre Shale pore pressure, axial stress and confining pressure were changing with different types of experiments. In addition to the traditional stress sensitivity experiments, constant mean stress scenarios were executed on Pierre Shale to simulate the behavior of the overburden during injection of CO2 into the reservoir followed by the depletion.
Obtained results were analyzed and compared with the existing physical theories which allow us to identify possible causes of the observed phenomena or if there is such need to propose new concepts. Experimental data were later used to calculate the AVO (amplitude versus offset) responses in order to check how these different test conditions affect the AVO analysis, typically used for determining rock properties based on field seismic data.
Produced data, besides addressing the fundamental questions connected with physical properties of examined materials and physical processes they are undergoing, allows us to make a link between typical field and laboratory data. They improve our understanding of various phenomena that storage sites experience during their lifetime and allow us to identify them during monitoring processes. Presented approach increases our proficiency in management of CO2 storage formations and it contributes to the advancement of public knowledge connected with CCS.
Figure 1 (a) Frequency dependence of the Young’s modulus for different water saturations of Mancos shale. The lines represent fits to the experimental data based on the Cole-Cole model (b) anisotropic AVO response of three different frequencies 1 Hz, 21 Hz and 500 kHz each at two different saturations 12% and 86%.
Spencer Jr., J. W. (1981), »Stress relaxations at low frequencies in fluid-saturated rocks: Attenuation and modulus dispersion», J. Geophys. Res., 86(B3), 1803–1812.