Challenges in Seismic Monitoring of Flow through Fractured Carbonates

Laura J. Pyrak-Nolte¹^,²^,³, Weiwei Li¹, and Christopher Petrovitch¹
¹Department of Physics
²Department of Earth and Atmospheric Sciences
³School of Civil Engineering, Purdue University, West Lafayette, Indiana USA

Carbonate reservoirs pose a scientific and engineering challenge to geophysical prediction and monitoring of fluid flow in the subsurface. Difficulties in interpreting hydrological, reservoir and other data arise because carbonates are composed of a hierarchy of geological structures, constituents and processes that span a wide spectrum of length and time scales. What makes this problem particularly challenging is that length scales associated with physical structure and processes are often not discrete, but overlap, preventing the definition of discrete elements at one scale to become the building blocks of the next scale. This is particularly true for carbonates where complicated depositional environments, subsequent post-deposition diagenesis and geochemical interactions result in pores that vary in scale from submicron to centimeters to fractures, variation in fabric composition with fossils, minerals and cement, as well as variations in structural features (e.g., oriented inter- and intra layered - interlaced bedding and/or discontinuous rock units). In addition, this complexity is altered by natural and anthropogenic processes such as changes in stress, fluid content, reactive fluid flow, etc. Thus an accurate geophysical assessment of the flow behavior of carbonate reservoirs requires a fundamental understanding of the interplay of textural and structural features subjected to physical processes that affect and occur on various length and time scales.

An overview of several laboratory investigations on Austin Chalk will be given to illustrate the complexity of interpreting the seismic and hydraulic properties of carbonates as well scientific challenges. On the laboratory-scale, Austin chalk is a complicated carbonate rock. Thin section analysis showed that our samples contained a weakly directed fabric that was composed mainly (99%) calcite. X-ray tomographic imaging of several samples showed that the samples contained layers, layer thickness varied among the samples and within a sample, and also showed that the density varied along and among the layers in a sample. Thin sections showed that the sources of the density variation within the sample arise from the amount of cement in each layer, the number of pellets and the type of porosity. These small-scale features played a significant role in the observed seismic anisotropy of these Austin Chalk samples. An important finding was that the axis of seismic symmetry was not perpendicular to the layers by was oriented 15 degrees from the normal to the layering. This suggests that oriented micro-cracks or pores that formed during depositional or diagenetic or tectonic processes contributed to the seismic anisotropy of the sample. This hypothesis is supported by the seismic data obtained on several samples as a function of stress. Only when a sample was loaded 15 degrees to the normal of the layers were stress-dependent arrival times and amplitudes observed. A question is whether observed seismic anisotropy is connected to the hydraulic response of the layered medium.

The structural layering of the samples affected the flow of fluids with in the rock as well as the seismic attenuation and velocity. Time-lapse seismic imaging during fluid invasion indicated that fluids would preferentially flow either up-dip or down-dip (parallel to the layering) rather then across the layers. When a sample contained a fracture, there were preferential flow paths along the fracture plane as well in the matrix of the rock. In addition, the fracture plane was geochemically altered during fluid invasions. Measurements of the aperture distribution of a portion of the fracture prior to and after fluid invasion determined that the average aperture size was reduced by roughly 400 microns and the reduction was caused by the presence of precipitation. The effect of precipitation was captured during time-lapse seismic measurements. After more than a day, the seismic amplitude of the signal was observed to increase, i.e. indicated an increase in fracture specific stiffness. Thus, geochemical interactions affected the flow paths through the fracture as well as the seismic response of the fracture.

As mentioned, a source of the seismic anisotropy in these carbonate rocks was the variation in density and porosity among the layers. An interesting affect of fluid invasion into this layered system was the effect of fluid saturation on velocity dispersion. When water invaded a sample, the impedance contrast among the layers was reduced. For samples with a layer thickness greater than a wavelength, water-saturation of the sample eliminated or reduced the dispersion observed for the sample when it was dry. For samples with a layer thickness less than a wavelength, no dispersion in the compressional wave velocity was observed as expected. Simulation of the effect of layer thickness on compressional wave velocity dispersion when the layer thickness and the wavelength are equal showed that fluid saturation can reduced velocity dispersion by decreasing the impedance contrast among the layers which affects the generation of internal multiple reflections among the layers.

An important finding from our research is the complications that arise in interpreting fracture specific stiffness when a fracture occurs in a layered medium. The displacement discontinuity theory describes a fracture as a low-pass filter, i.e. a fracture preferentially attenuates the high frequency components of a signal relative to the intact portions of the rock. However, from our experiments, a fracture in a layered structure (such as Austin Chalk) caused either an increase or decrease in the dominant frequency of the signals relative to that for the intact sample. From simulations that used a combination of the transfer matrix approach and the displacement discontinuity theory, the increase or decrease in the dominant frequency of the signal depended on layer thickness relative to a wavelength and the location of the fracture within a layered medium. If a wave is propagated through a layered medium and then across a fracture, a single fracture would attenuate the high frequency components of the signal and result in a lower dominant frequency compared to the intact case. However, when the fracture is located within a layered medium, the fracture interrupts the original wave interference that existed in an intact medium and is a source of additional new multiple reflections (i.e. constructive and destructive interference). This interruption and/or creation of new internal reflections, results in either a slight decrease in the dominant frequency (but not as much if the fracture occurred after the layering) or slight increase in frequency caused by two constructively interfering waves (direct & multiple). Thus any interpretation of fracture specific stiffness must account for the layering in the medium and the location of the fracture within the layered medium.

Finally, the anisotropy and the effect of fluid substitution on seismic anisotropy in a transversely anisotropic medium were studied. The Thomsen parameters for these Austin Chalk samples were less than 1 indicating that this type of carbonate rock is weakly anisotropic. Fluid substitution was observed to affect the anisotropy of the sample. Both low frequency limit (Backus Averaging & Gassmann’s Equation) and high frequency limit (Wylie Time Averaging) approaches were taken for studying the observed effects of fluid substitution. Both the approaches predicted the general trend of the velocity behavior upon saturation as observed in the experiments, such as a decrease in P-wave anisotropy upon saturation. However, neither approach captured the complete velocity surfaces for SV or SH shear waves. The deviations between the experimental data and the two theoretical approaches arise because of the complexity of carbonate rock, i.e., the layering, density variation, pore structure, geochemical weakening, fluid strengthening, etc. The wavelength of both shear waves was roughly 2 mm, which is close to the thickness of the layers in the sample (0.5 mm – 1mm). This places these carbonate rock systems neither in the low or high frequency limit for a frequency of 1 MHz (λ/d ~ 4 – 2) but in the transition zone between ray theory and effective medium theory. Given the complexity of carbonates on all scales, from the laboratory to the field, future work should focus on seismic wave behavior in (a) layered media in the transition zone, (b) for layered systems with two competing anisotropic sources and (c) in layered fractured systems that are geochemically altered over time.

Acknowledgments

The author wishes to acknowledge support of this work by the Geosciences Research Program, Office of Basic Energy Sciences US Department of Energy (DE-FG02-09ER16022), and the GeoMathematical Imaging Group at Purdue University.

AAPG Search and Discovery Article #120034©2012 AAPG Hedberg Conference Fundamental Controls on Flow in Carbonates, Saint-Cyr Sur Mer, Provence, France, July 8-13, 2012