Energy & Fuels 2005, 19, 1319-1326
1319
Monitoring Bitumen-Solvent Interactions with Low-Field Nuclear Magnetic Resonance and X-ray Computer-Assisted Tomography† Y. W. Wen‡ and A. Kantzas* TIPM Laboratory, University of Calgary, Calgary AB T2N 1N4, Canada Received September 14, 2004. Revised Manuscript Received February 15, 2005
This work involves the detection and monitoring of solvent interactions with heavy oil and bitumen. Two nondestructive methodsslow-field nuclear magnetic resonance (NMR) and X-ray computer-assisted tomography (CAT)swere used. It is shown that low-field NMR can be a very useful tool in understanding the relationship of viscosity, density, and asphaltene precipitation in bitumen-solvent mixtures. Such mixtures are present in solvent-related heavy oil and bitumen recovery processes, such as vapor extraction (VAPEX). As a solvent comes into contact with a heavy oil or bitumen sample, the mobility of hydrogen-bearing molecules of both solvent and oil changes. These changes are detectable through changes in the NMR relaxation characteristics of both the solvent and the oil and can be correlated to mass flux and concentration changes. Based on Fick’s second law, diffusion coefficients were calculated for combinations of three oils and six solvents. X-ray CAT scanning was also used in parallel for analysis of solvent diffusion into the bitumen. As the solvent was diffusing into the bitumen, a concentration gradient was obtained. Concentration values at certain times were used to calculate diffusion coefficients, which were compared with results obtained from NMR data, using both an analytical method and a numerical method. The diffusion coefficients were considered either as constants or as functions of solvent concentration in two models that have been developed during this research. The overall diffusion coefficients calculated for several pairs of oils and solvents at different ratios, both by NMR data and X-ray tomography, were on the order of 10-6 cm2/s.
Introduction The vapor extraction (VAPEX) process was proposed by Butler and Mokrys1 for the first time as an alternative to steam-assisted gravity drainage (SAGD) for thin reservoirs where heat lost in the formation would make SAGD uneconomical. In the VAPEX process, vapor solvents, instead of steam, are injected into the reservoir. The solvents dissolve into the bitumen and dramatically reduce its viscosity. The diluted bitumen can then drain down into a producer well by gravity. Since the original paper, many valuable experimental studies have been published using Hele-Shaw cells,2-4 pore network glass micromodels,5 magnetic resonance imaging (MRI),6 and pressure-volume-temperature (PVT) experiments.7 In this paper, low-field nuclear magnetic † Presented at the 5th International Conference on Petroleum Phase Behavior and Fouling. ‡ Now with Devon Canada Corporation, Calgary, Alberta, Canada. * Author to whom correspondence should be addressed. E-mail address:
[email protected]. (1) Butler, R. M.; Mokrys, I. J. J. Can. Pet. Technol. 1991, 30 (1), 97-106. (2) Mokrys, I. J.; Butler, R. M. Presented at the Production Operations Symposium, Oklahoma City, OK, 1993, Paper No. SPE 25452. (3) Das, S. K.; Butler, R. M. Presented at the Sixth Petroleum Conference of the South Saskatchewan Section of the Petroleum Society of CIM, 1995, Paper No. CIM 95-118. (4) Butler, R. M.; Jiang, Q. J. Can. Pet. Technol. 2000, 39 (1), 4856. (5) James, L.; Chatzis, I. Presented at the Petroleum Society’s Canadian International Petroleum Conference 2003, Calgary, Alberta, Canada, 2003, Paper No. CIM 2003-205.
resonance (NMR) and X-ray computer-assisted tomography (CAT) scanning have been used to measure the physical properties of mixtures of heavy oil and bitumen samples with kerosene, hexane, naphtha, heptane, pentane, and toluene at several ratios at room temperature and pressure. Low-field NMR has potential as a tool for measuring the properties of a reservoir fluid.8 NMR measures proton density and relaxation characteristics of a given fluid. A spectrum of signal amplitudes at different relaxation times is the final result of a measurement. The amplitude is related to mass or volume, and the relaxation times provide a measurement of the fluid mobility in space. NMR logging and benchtop tools were designed with water as the target fluid. When fluids other than water are used and quantitative measurements are required, the amplitude must be compared to the amplitude of water of the same mass or volume. This relative amplitude is called relative hydrogen index (RHI), and it is a characteristic property of any hydrogenbearing fluid. NMR measurements are simple and nondestructive but capable of yielding an incredible wealth of informa(6) Fisher, D. B.; Singhal, A. K.; Das, S. K.; Goldman, J.; Jackson, C. Presented at the 2000 SPE/DOE Improved Oil Recovery Symposium, Tulsa, OK, 2000, Paper No. SPE 59330. (7) Dauba, C.; Quettier, L.; Christensen, J.; Le Goff, C.; Cordelier, P. Presented at the SPE Annual Technical Conference and Exhibition, San Antonio, TX, 2002, Paper No. SPE 77459. (8) Coates, G. R.; Xiao, L. Z. H.; Prammer, M. G. NMR Logging Principles and Applications; Butterworth-Heinemann: Oxford, 1999.
10.1021/ef049764g CCC: $30.25 © 2005 American Chemical Society Published on Web 05/18/2005
1320
Energy & Fuels, Vol. 19, No. 4, 2005
Wen and Kantzas
tion about the reservoir fluid under investigation in a particular sample.9,10 The relaxation time of a fluid in the bulk state is inversely proportional to its viscosity.10 When a solvent is added to a heavy oil or bitumen sample, the mixture viscosity decreases dramatically as the solvent content increases. NMR spectra reflect these changes. A viscosity model was developed for oilsolvent mixtures using different oils and solvents over a wide composition range.11 Asphaltenes are the n-pentane- or n-heptane-insoluble fractions of crude oil that remain in solution under reservoir temperature and pressure conditions. They are destabilized and start to precipitate with decreasing pressure and temperature, and/or with compositional changes.12 During the experiments presented in this work, when the ratio of the paraffinic solvents mixed with bitumen increases, asphaltenes begin to be observed at the bottom of the samples. As the asphaltenes precipitate, the proton density in the fluid hydrocarbonsolvent mixture changes. Thus, the content of asphaltenes could be estimated through the RHI change during an NMR test.11 This paper will also demonstrate that NMR can be used to measure changes of oil-solvent properties as a function of time. From such “transient” data, one can then predict diffusion coefficients. Because NMR measurements are “bulk” measurements, without any spatial variability considerations, whereas mass-transfer models are based on spatial and temporal variability, CAT scanning is also used in parallel experiments to monitor concentration profiles and recover diffusion coefficients. The CAT-scanner-derived diffusion coefficients and the NMR-derived diffusion coefficients are then compared. Diffusion coefficients in the solventbitumen systems were considered both as a constant and as a function of concentration. Diffusivity is an important parameter, both in the VAPEX process and in solvent injection processes, because it determines how fast the solvent dilutes the oil. Diffusivity of the solvent in the oil correlates strongly with the properties of the solvent and the oil, such as the size of the molecules and the viscosity.13 Oballa and Butler13 have studied the diffusion of toluene in the Peace River bitumen system in quasisteady-state conditions (diaphragm cell method) and non-steady-state conditions (free diffusion method) using a vertical cell, which consisted of two closely spaced glass plates. The overall diffusion coefficient was calculated as a function of concentration. It was found to vary by 1 order of magnitude over the concentration range (0%-100%).
with viscosities of 670 000, 130 000, 14 000, and 6000 mPa s, respectively, at 25 °C. Kerosene, toluene, naphtha, heptane, hexane, and pentane were added to the oils at several predefined mass fractions: 100%, 99%, 96%, 93%, 90%, 85%, 80%, 70%, 50%, 30%, and 0% oil (100% solvent). The oil samples were heated to 40 °C, and solvent was added while stirring. The resulting solvent-oil mixtures were subsequently cooled. NMR spectra were measured at 25 °C (room temperature) using an Ecotek FTB benchtop relaxometer. The viscosities of the mixtures were measured at the same temperature using both a Haake rheometer and a Wells-Brookfield cone/plate viscometer. The densities were measured using a DMA 45 digital density meter under the same conditions. As 70% heptane, hexane, and pentane were added to 30% Cold Lake oil, precipitation was observed at the bottom of the cells. The NMR spectra of the precipitates (without any oil) were also measured. Diffusion Measurements by Nuclear Magnetic Resonance. For diffusion experiments of solvent in bitumen, the solvent was placed on top of the oil, and then the samples were sealed; thus, diffusion occurred vertically. NMR spectra were gathered over a period of 4 weeks and the changes in the spectra were related to the changing oil properties as solvent diffused into the heavy oil. The diffusion measurements were static, with time t ) 0 set as the time when solvent was placed in contact with the oil. There was no mechanical stirring or displacement to aid the mixing. As a result, mixing was due only to the concentration gradients in the two fluids. Delay of mixing due to the density contrast was neglected. All diffusion experiments were performed in small vials (5 cm high and 2.5 cm in diameter). The samples were measured in the Corespec 1000 low-field NMR relaxometer at a controlled temperature of 30 °C and ambient pressure. Heptane was added to the Cold Lake bitumen in amounts of 5, 10, 15, 20, 30, and 50 wt %. Diffusion Measurements by Computer-Assisted Tomography. Complementary to the NMR diffusion experiments, heptane diffusion into Cold Lake bitumen was conducted in the CAT scanner, using a sealed beaker 10 cm high and 5 cm in diameter. As with the NMR tests, heptane was placed on top of the Cold Lake bitumen, and CAT scanning began at t ) 0. The experiment was conducted at 30 °C. The samples were placed in a constant-temperature bath between scans. A model GE9800 CAT scanner was used. Cross-sectional images were obtained at regular intervals along the length of the beaker during the diffusion process. The two-dimensional CAT slices show the diffusion process along the length of the beaker. The images showed a homogeneous profile in the horizontal direction; thus, a onedimensional diffusion process was considered in this experiment. An average computed tomography (CT) number in each horizontal direction was calculated, and then the changes in the average CT number,14,15 relative to vertical distance, were obtained.
Experimental Procedures
Solvent and Heavy Oil/Bitumen Mixture. As solvent was added into the heavy oil/bitumen, the NMR spectra showed very distinct changes. Figure 1 represents the NMR spectra change as heptane was mixed with Peace River oil in seven different ratios. The oil transverse relaxation time (T2) becomes longer as the viscosity decreases.8 There is more than one peak in some mixtures of heavy oil/bitumen solvent, as shown in Figure 1. At high
Nuclear Magnetic Resonance of Oil-Solvent Mixtures. The four oils used in the solvent experiments were from the Peace River, Cold Lake, Edam, and Atlee Buffalo regions, (9) Allsopp, K.; Wright, I.; Lastockin, D.; Mirotchnik, K.; Kantzas, A. J. Can. Pet. Technol. 2001, 40 (7), 58-61. (10) Mirotchnik, K.; Kantzas, A.; Aikman, M.; Starosud, A. J. Can. Pet. Technol. 2001, 40 (7), 38-44. (11) Wen, Y.; Kantzas, A. Presented at the 55th Annual Technical Meeting of the Petroleum Society, Calgary, Alberta, Canada, 2004, Paper No. CIM 2004-065. (12) Hayduk, W.; Minhas, B. S. Can. J. Chem. Eng. 1982, 60. (13) Oballa, V.; Butler, R. M. J. Can. Pet. Technol. 1989, 28 (2), 6369.
Experimental Results and Discussion
(14) Kantzas, A. In Situ 1990, 14, 77-132. (15) Wen, Y. Characterization of Bitumen-Solvent Interactions Using Low Field NMR, M.Sc. Thesis, University of Calgary, Calgary, Canada, 2004.
Monitoring Bitumen-Solvent Interactions
Energy & Fuels, Vol. 19, No. 4, 2005 1321 Table 1. NMR Parameters for Asphaltene Experiments
echo spacing, TE number of echoes time to next train number of trains
Figure 1. Changes in the NMR spectra, relative to the amount of solvent.
Figure 2. Changes in the NMR spectra properties, relative to heptane addition in bitumen and heavy oils.
solvent concentrations, the solvent peak maintains its integrity and the bitumen/heavy oil peak shifts to longer relaxation times. However, there are cases where the solvent peak disappears and the bitumen spectrum splits into multiple peaks. There is evidence that this is due to the solvent treatment, causing the heavy oil and bitumen to separate into components with different relaxation characteristics. Asphaltenes have the shortest relaxation range, followed by resins and saturates. The longest relaxation-range results represent those of aromatics.10 The NMR spectra characteristics of bitumen-solvent pairs were tested in three mixture groups, based on the availability of the different samples and an effort to demonstrate potential differences of paraffinic and nonparaffinic solvents with bitumen and “lighter” heavy oils. The first group included four oils mixed with heptane. The second group included Cold Lake oil mixed with kerosene, toluene, naphtha, heptane, hexane, and pentane. The third group consisted of Edam oil mixed with kerosene, naphtha, heptane, and hexane. NMRderived parameters were plotted against bitumen (or solvent) content. The results of all samples show similar trends over a wide range of oil concentrations. Figure 2 shows how the product of the geometric mean of T2 (T2gm) for the mixture and the amplitude per unit mass (amplitude index, AI) changes, relative to oil or bitumen content. As the solvent content increases, the inverse of the product of AI × T2gm monotonically decreases. The resulting curves are slightly different for each solvent-bitumen pair.11 The reason for using the parameter 1/(AI × T2gm) as the choice parameter to present the results is because this term is an NMR-derived “viscosity” term and, thus, is consistent with other work in the literature.11,16-18 (16) Bryan, J. L.; Mirotchnik, K.; Kantzas, A. J. Can. Pet. Technol. 2003, 42 (7), 29-34.
NMR parameter set I
NMR parameter set II
0.3 5000 1000 100
0.3 5000 15000 25
Figure 2 also shows that, as heptane was mixed with Peace River, Cold Lake, Atlee Buffalo, and Edam oils at concentrations of 85%.18 All the results indicated that solvents change the properties of more-viscous oils more distinctly than less-viscous oils. Thus, the experimental data, such as those shown in Figure 2, suggest that correlations of bitumen content with NMR-derived properties can be created and they could subsequently be used for component fraction determination in heavy oil/bitumen-solvent mixtures. Asphaltene Precipitation. Asphaltene precipitates were observed on the bottom of the test vials as Cold Lake bitumen (30 wt %) was mixed with heptane, hexane, or pentane. The largest amount of precipitates was obtained with pentane, and the least amount was obtained with heptane. Samples of oil-free asphaltenes that precipitated from the mixture of Cold Lake bitumen mixed with heptane (6.6 g) or pentane (6.0 g), were tested by NMR. Table 1 shows the NMR parameter sets for asphaltene measurements. When NMR parameter set I was used during the experiment, a small signal could be measured. When NMR parameter set II was used, which is suitable to detect both bitumen and solvent, it was found that asphaltenes relax too fast to be measured. The entire spectrum was noise, as shown in Figure 3. If we consider that asphaltenes alone could not be measured by NMR (using the parameters set for bitumen-solvent mixtures), we can then assign an RHI value of zero to asphaltenes. Thus, the RHI that can be calculated from amplitudes and masses in the mixture of oils with solvents does not include the asphaltenes fraction. It will be a combined signal of the solvent and the de-asphalted oil. Figure 4 shows the RHI change of Cold Lake bitumen-solvent mixtures with bitumen content. As the solvent content increases, the RHI increases. In the case of toluene, the RHI reaches a plateau very quickly. However, when paraffinic solvents are used, the RHI values increase substantially, and, often, a plateau is not reached. In all solvent-oil (17) Bryan, J. L.; Manalo, F. P.; Wen, Y.; Kantzas, A. Presented at SPE/CIM/CHOA Resource 2 Reserves 2 Results Conference and Exhibition, Calgary, Alberta, Canada, 2002, Paper No. SPE 78970. (18) Wen, Y.; Bryan, J.; Kantzas, A. J. Can. Pet. Technol. 2005, 44 (4), 22-28.
1322
Energy & Fuels, Vol. 19, No. 4, 2005
Wen and Kantzas
Figure 3. NMR decay curve of asphaltene of Cold Lake bitumen precipitated by pentane. Figure 5. Asphaltene determination from NMR. Table 2. Comparison of Asphaltene Content Asphaltene Content (%)
Figure 4. Changes in the relative hydrogen index (RHI), relative to Cold Lake bitumen content.
mixtures in which the bitumen fraction is >85% and the RHI is