A METHODOLOGY TO INTEGRATE MAGNETIC RESONANCE AND ACOUSTIC MEASUREMENTS FOR RESERVOIR CHARACTERIZATION

The work reported herein represents the third year of development efforts on a methodology to interpret magnetic resonance and acoustic measurements for reservoir characterization. In this last phase of the project we characterize a vuggy carbonate aquifer in the Hillsboro Basin, Palm Beach County, South Florida, using two data sets--the first generated by velocity tomography and the second generated by reflection tomography. First, we integrate optical macroscopic (OM), scanning electron microscope (SEM) and x-ray computed tomography (CT) images, as well as petrography, as a first step in characterizing the aquifer pore system. This pore scale integration provides information with which to evaluate nuclear magnetic resonance (NMR) well log signatures for NMR well log calibration, interpret ultrasonic data, and characterize flow units at the field scale between two wells in the aquifer. Saturated and desaturated NMR core measurements estimate the irreducible water in the rock and the variable T{sub 2} cut-offs for the NMR well log calibration. These measurements establish empirical equations to extract permeability from NMR well logs. Velocity and NMR-derived permeability and porosity relationships integrated with velocity tomography (based on crosswell seismic measurements recorded between two wells 100 m apart) capture two flow units that are supported with pore scale integration results. Next, we establish a more detailed picture of the complex aquifer pore structures and the critical role they play in water movement, which aids in our ability to characterize not only carbonate aquifers, but reservoirs in general. We analyze petrography and cores to reveal relationships between the rock physical properties that control the compressional and shear wave velocities of the formation. A digital thin section analysis provides the pore size distributions of the rock matrix, which allows us to relate pore structure to permeability and to characterize flow units at the core and borehole scales. Vp, density, porosity, and permeability logs are integrated with crosswell reflection data to produce impedance, permeability, and porosity images. These images capture three flow units that are characterized at the pore and borehole scales. The upper flow units are thin, continuous beds, and the deeper flow unit is thicker and heterogeneous. NMR well log calibration data and thin section analysis demonstrate that interwell region permeability is controlled mainly by micropores and macropores, which represent the flow unit matrices of the confined aquifer. Reflection image-derived impedance provides lateral detail and the depth of the deeper confining unit. The permeable regions identified in both parts of this phase of the study are consistent with the hydrological results of high water production being monitored between two wells in the South Florida aquifer. Finally, we describe the two major methodologies developed to support the aquifer characterization efforts--(1) a method to estimate frequency-dependent scattering attenuation based on the volume fraction and typical size of vugs or karsts, and (2) a method to more accurately interpret NMR well logs by taking into account the diffusion of magnetization between large and small pores. For the first method, we take the exact vug structure from x-ray CT scans of two carbonate cores and use 3-D finite difference modeling to determine the P-wave scattering attenuation in these cores at ultrasonic frequencies. In spite of the sharp contrast in medium properties between cavity and rock and the violation of the small perturbation assumption, the computed scattering attenuation is roughly comparable to that predicted by various random medium scattering theories. For the second method, we investigate how the diffusion of magnetization between macropores and micropores influences NMR log interpretation through 2D simulation of magnetization diffusion in realistic macropore geometries derived from digital images of thin sections. In most cases, our simulations show that the resulting simulated magnetization decay rate and corresponding T{sub 2} spectrum fit the well log and core NMR results better when inter-pore diffusion is included in the simulation. Interpore diffusion moves some of the magnetized fluid from large pores to small pores, so part of the T{sub 2} distribution is shifted to smaller decay times. The shift is strongest when the rock contains small macropores that are large enough to cause bulk relaxation to dominate over surface relaxation, but small enough that the diffusion transport time scale is faster than the bulk relaxation time scale. The simulated T{sub 2} spectra are consistent with known facies characteristics