@ Original Contribution NMR IMAGING OF FLUID DYNAMICS IN RESERVOIR CORE B, A. BALDWIN AND W. S, YAMANASHI Phillips Petroleum Co., Bartlesville, Oklahoma and City of Faith Medical and Research Center, Tulsa, Oklahoma 74004, USA ‘A medical NMR imaging instrument has been moditied to image water and oil in reservoir rocks by the construc tion of a new receiving col. Both oll and water inside the core produced readily detectable proton NMR signals, ‘while the rock matrix produced no signal. Because of similar 7; NMR relaxation times, the water was doped with a paramagnetic fon, Mn-+2, to reduce its 7; relaxation time. This procedure enhanced the separation between the oil and water phases in the resulting images. Sequential measurements, as water imbibed into one led from the other end of a core plug, produced a series of images which showed the ‘dynamics of the fluids. For water-wet Berea Sandstone a flood front was readily observed, but some of the oil ‘was apparently left behind in smal, solated pockets which were larger than individual pores. After several addi tional pore volumes of water flowed through the plug the NMR image indicated a homogeneous distribution of oll. The amount of residual oil, as determined from the ratio of NMR intensities, closely approximated the resid- tual oil saturation of fully flooded Berea samples measured by Dean-Stark extraction. A Berea sandstone core treated to make it partially oil-wet, did not show a definitive flood front, but appeared to channel the water around the perimeter of the core plug. The relative ease with which these images were made indicates that NMR imaging can be a useful technique to Follow the dynamics of oil and water through a core plug For a variety of production processes. Keywords: Porous medi INTRODUCTION ‘The petroleum industry has used average fluid satura- tions to determine the amount of oil originally in place and potentially producible by various produc- tion processes from the beginning of scientific reser= voir analysis. These average saturation measurements do not define the distribution of the fluids inside the core plug or reservoir. To maximize the efficieney of oil recovery and/or increase the accuracy of calculat- ing recoverable reserves one should know how the flu- ids are distributed inside the reservoir rock. For the design and evaluation of alternate production meth- ods the change in distribution with various treatments is important. Simple mathematical and experimental ‘models can be used to predict that fluid flow through a reservoir can occur either in a heterogeneous or homogeneous fashion. The former would leave mul- tiple pores of oil surrounded by water, while the lat- Reservoir core; Fluid movement, ter would leave oil as microscopic globules inside individual pores.'? Channeling between injection and producing wells would appear to represent the heter- ‘geneous case; however, inside the channel the oil may be homogeneously distributed. It is probable that the mechanism changes with time and from one reser- voir to another, In the early 1950s attempts were made, using radio- tracers and gamma-ray absorption,** to locate the water movement through a core. Recent improve- ‘ments, primarily in data acquisition, have caused a reexamination of these techniques.” They can locate the Flood front by rapid changes in saturations, but due to resolution and averaging over the cross section of the core cannot locate residual pockets or channels through the original fluid filled porous media. A number of ingenious laboratory tests have been de- vised to follow fluid dynamies; however, these require artificial systems such as glass bead packs, glass plates RECEIVED 6/10/87; AccePreD 2/1/88, Acknowledgment ~ The authors wish to acknowledge C. Crandall, C. MePherson and A. Sy (City of Faith) who helped operate the instrument and collect the data and E. 495 “Todd (Phillips Petroleum Co.) who performed the sample preparations and helped with the coil construction. ‘Address correspondence to B.A. Baldwin, Phillips Petro: leum Co., 136-GB, Bartlesville, OK 74006.494 M te. and cannot be applied to real reservoir rocks. By placing a dye in the flooding fluid one can determine a 3-dimensional image of the flood front. However, since the core must be cut for analysis this technique precludes further testing of an individual core. X-ray computer aided tomography" has been able to re~ solve oil and water phases in cores; however, compen- sation for the rock matrix contribution must be made. A study using NMR chemical shifts has also been able to locate oil and water in artificial cores." The pre ent work was undertaken to determine the feasibili of using a tool developed for medical applications, NMR imaging," to follow the dynamics of fluids in reservoir rocks. EXPERIMENTAL A diagram of the coil and core holder is shown in Fig. 1. The excitation coils are located in the large body cavity. The modified receiving coil is located in the center of the cavity with its axis perpendicular to the static axis of the field. The sample is placed in the center of the receiving coil. All materials other than the coils must be non-metallic and non-magnetic. The sample containers for the preliminary experiments were polyethylene, glass or styrofoam, all of which are invisible in the final image. For the flow experi- ‘ments, a sample container (see Fig. 2) was constructed using polymethylmethacrylate platens at either end of the core; the assembly was constrained inside a Viton sleeve by Teflon shrink tubing. Water was introduced through vinyl tubing. A water-wet ceramic frit was placed in the water inlet to ensure proper mixing at Fig. 1. Diagram of sample holder, coils and water inlet netic Resonance Imaging @ Volume 6, Number 5, 1988 the face of the core and inhibit oil flow back into the water line. The amount of water imbibed was mea- sured by the change in the height of the water col- umn. The maximum pressure exerted by the water column was approximately 10 em of water. Thus the experiments described here are largely imbibition water floods. Imbibition is the spontaneous uptake of water by a porous media through capillary action, The homogeneity of the signal, with the modified receiving coil, is shown by the image of distilled water in a glass vial, Fig. 3. Intensity profiles, both axially and radially across the sample, show very constant intensity. The small variation at each end of the sam- ple could be edge artifacts or even variations in the bottle dimensions. This homogeneity means that the final image will depend only on variations of signal from the sample. Preliminary experiments to determine the feasibil- ity of NMR imaging, see Fig. 4-5, were performed on a Picker International Model MR VISTA 2055. This ‘TEFLON SHRINK ‘TUBING. ceramic / FIT [ poten ron \ wre S Fig. 2. Diagram of core holder. Fig. 3. NMR image of distilled water in a glass vial. The intensity along the line connecting the two crosses is shown, below the imageNMR imaging of Maid dynamics @ B, A, Batbwis ab W. S, Yawanasi 495 Fig. 4. Images of core samples containing various concentrations of fluids. Description given in Table 1 c D to distilled water as a funetion of delay time and concentration. Concentration, in & re 0.0, 0.0020, 0.0080, 0,032 and 0.068. adding MnCi2 +, clockwise from lower left a496 Mag machine had a magnetic field strength of 0.255 Tesla and operated at 10.6 MHz using the head coil. The flow experiments (Fig. 6-8) were made on the same ‘machine after equipment modification which resulted in a field strength of 0.5 Tesla at an excitation fre- quency of 21.3 MHz. For the flow experiments a sole- noid receiving coil with an optimized filling factor, was constructed to enhance both intensity and homogeneity of the signal over the test area, see above and Fig. 3. The images were made with a single spin echo pulse sequence with TE = 40 ms and TR = 1000 ms. The image matrix was 256 x 256 and the viewer matrix was 512 x $12. The number of images aver- aged was 2, the field of view was 30 cm and slice thickness was 5 mm. Berea sandstone ‘ores were used, as received, for c ctic Resonance Imaging @ Volume 6, Number 5, 1988 water-wet samples. Partially oil-wet cores were pre- pared by flooding Berea sandstone with a mixture of toluene and Dri-Film 104, The chalk cores were obtained from the North Sea and cleaned with tolu- ene. The cores were initially saturated using the cen: trifuge method. Fluid saturations at the end of a test were determined by a Dean-Stark vapor phase extrac tion with toluene. RESULTS AND DISCUSSION ‘The first experiments consisted of imaging several cores saturated with various combinations of oil and water and even one with no fluids. These images, Fig. 4, were a single slice across the center of the cylindri- cal core which produced the image of a disk. The list D Fig. 6. Flow of water into water-wet Berea from the left. (A) 0% PV, (B) 30% PV, (C) 49% PV and (D) 77% PY