Yongchen Songa,
Min Haoab,
Yu Liua,
Yuechao Zhao*a,
Bo Sua and
Lanlan Jianga
aKey Laboratory of Ocean Energy Utilization and Energy Conservation of Ministry of Education, Dalian University of Technology, Dalian 116024, P. R. China. E-mail: zhaoyc@dlut.edu.cn; Tel: +86-411-84708015
bSchool of Energy and Power Engineering, Shenyang University of Chemical Technology, Shenyang, Liaoning 110142, P. R. China
First published on 29th September 2014
A method combining magnetic resonance imaging (MRI) and dual-chamber pressure decay (i.e., the pressure–volume–temperature method) was used to measure the diffusivity of CO2 in bulk n-hexadecane, a representative oil, at 21 °C. Images of the proton density of n-hexadecane were obtained by MRI using a high magnetic field and high resolution, and the pressure decay was recorded. Overall diffusion coefficients of CO2 were calculated from both pressure decay and MRI intensity data. The two results showed good agreement. MRI was successfully used to study the diffusivity of CO2 in n-hexadecane. The change of image density (i.e., proton density) demonstrates that the oil density decreases as CO2 diffuses into it. The finite volume method was applied to the images obtained by MRI, allowing the diffusion coefficient of CO2 to be directly obtained. MRI methods can measure the unsteady-state local diffusion coefficient and overall diffusion coefficient of such systems.
CO2-enhanced oil recovery (CO2-EOR) has proven to be effective for improving domestic oil production. Moreover, use of CO2 in oil recovery also provides considerable capacity for CO2 mitigation.1 The viscosity of oil can be lowered significantly by diffusion of CO2 into it; thus, the mobility of oil in a reservoir can be improved.2 Mass transfer between CO2 and oil, which is mainly controlled by diffusivity, is the first mechanism that occurs during the CO2-EOR process.3,4 To describe this process quantitatively, the diffusion coefficient of CO2 gas in oil needs to be determined.
Experimental methods used to measure the diffusivity of gas in oil can be broadly divided into two types: direct and indirect. The direct measurement of diffusivity is time consuming and involves measuring the change of concentration over time.5 Indirect methods involve measuring the change of a parameter related to diffusion over time, such as volume, density or pressure.6,7 The pressure decay method is a typical indirect method used to measure diffusivity. Riaziet et al.8 reported a pressure decay method using a pressure–volume–temperature (PVT) cell. In their experiment, both the position of the gas/liquid interface and gas pressure decay in the PVT cell were monitored during the diffusion process. Zhang et al.9 measured the diffusion coefficient of CO2 in heavy oils using a similar experimental method except that the position of the interface was considered constant. They also presented a nonlinear regression method that allowed the diffusion coefficient to be obtained directly from pressure decay data. Recently, non-intrusive detection methods such as X-ray computed tomography (CT) and magnetic resonance imaging (MRI) have been extended to the area of petroleum research. Liu et al.10 observed the miscible and immiscible CO2 displacement of n-decane in glass-bead packed beds using an MRI scanner under high magnetic field. Residual oil saturation was quantitatively measured by intensity analysis of magnetic resonance (MR) images. Song et al.11 determined the minimum miscible pressure of CO2 and n-decane by the MRI method, and their results agreed well with previous data obtained with traditional methods. Meanwhile, X-ray CT has been used to measure density distribution in heavy oil and hydrocarbon solvent mixing experiments.12–14 In these experiments, concentration profiles were obtained from CT images based on the relationship between bulk density and CT image intensity. Other researchers have used MRI to observe mass-transfer processes. Fisher et al.15 used MRI to observe vapor extraction processes. They performed bitumen and heavy-oil vapor extraction experiments using butane and propane as solvents in dry sand and sand with connate water. In their experiments, the MRI signal, which corresponded to the oil saturation, provided not only the position of the vapor chamber but also concentration gradients.
The main objective of the present study is to use MRI technology to measure gas diffusion coefficients in oils, and investigate diffusion and mass transfer processes in situ. Diffusion coefficients of CO2 in n-hexadecane are determined by measuring the pressure in a dual-chamber PVT system. MRI with high spatial resolution is used to observe the concentration distribution of CO2 during diffusion. In addition, pressure data are recorded to calculate the molar mass of CO2 diffusing into n-hexadecane. Pressure decay analysis is performed to calculate the steady-state diffusion coefficient of CO2 in n-hexadecane. This study helps us to understand the diffusion behavior of CO2 in a representative oil, so it will provide useful information for several oil recovery processes.
Images were measured using a 400 MHz NMR spectrometer (Varian Inc., Palo Alto, CA, USA) with microimaging kits including a 55 mm i.d. gradient coil, which provided a maximum gradient magnetic field of 50 Gauss per cm, and a 40 mm i.d. 400 MHz 1H Millipede vertical imaging probe. All data and images were processed with a data processor.
n-Hexadecane of 98% purity was contained in a glass vial (i.d. = 8.8 mm) placed at the bottom of the sample vessel. Gaseous CO2 of 99.99% purity was injected via a syringe pump into the gas cylinder to attain the desired pressure.
All experiments were performed at 21 °C. The glass vial was filled to a height of about 20 mm with n-hexadecane. The system was placed under vacuum to extract air. After careful tuning, shimming and setting of the pulse parameters, the oil sample was scanned to obtain an initial MR image. CO2 was then injected into the gas cylinder with the syringe pump until the desired pressure was reached. The pressure in the gas cylinder was set between 3000 and 5000 kPa. The gas valves were then turned on to allow CO2 to enter the sample vessel and diffuse into the oil. The pressure was automatically recorded at the same time. Proton density images of the longitudinal planes along the diffusion direction were obtained while the pressure was being plotted. Each experiment was stopped after the pressure maintained a constant value (equilibrium pressure) for 1 h. When the experiment had finished, the system was slowly depressurized to atmospheric pressure, dismantled and cleaned before starting a new experiment.
The steps of applying MRI into diffusion experiment imaging are as follows:
(1) Put the n-hexadecane-containing glass vials upright in the sample vessel, and keep the height consistent with the imaging probe.
(2) Debug the MRI analyzer; open Vnmrj in the image acquisition computer, and enter “tune” into the dialog box; tune by regulating the knobs “tune” and “match” below the probe; then start to shim: click “manual shim” on the “start” interface, entering the shimming interface; regulate by using the right and left mouse buttons, until the images are appropriate.
(3) Setting of parameters. All images were acquired with a standard spin-echo multi-slice (SEMS) pulse sequence, because this sequence is free from the impacts of main magnetic field inhomogeneity, and produces high-quality images. The specific imaging parameters are as follows: echo time (TE) = 4.9 ms, repetition time (TR) = 5 s, image data matrix = 96 pixels × 96 pixels, field of view (FOV) = 40 × 40 mm2, slice thickness = 2 mm.
(4) Image acquisition. After gas injection was started, the MRIs were continuously acquired, until the end of the experiment.
(1) The swelling height of the oil is negligible, that is, z0 remains constant during the experiments.
(2) The temperature remains constant.
(3) The volatilization of n-hexadecane is negligible.
(4) The diffusion coefficient, D, does not change markedly with concentration during the experiments.
(5) The gas phase is a pure (single-component) gas.
The diffusion process can be expressed by Fick's law:9
(1) |
Based on this hypothesis and combined with Fick's law, the relationship between pressure P and time t can be acquired using semi-infinite boundary conditions and the matter balance principle. When the higher-order terms were ignored, a 2-order analytical expression was obtained as follows:9
(2) |
The first exponential term in eqn (2) fits the main part of the pressure–time curve (large k) and the second term fits the initial part of the pressure–time curve during the incubation period (small k). Numerical fitting is required to obtain the values of m1, m2, k1, k2 and Peq. Because the incubation period is relatively short, the first exponential term is more suitable to calculate the integral diffusion coefficient. The diffusion coefficient D is then given by:9
(3) |
CO2 has no signal in MRI. A previous study18 confirmed that the MR signal intensity at any location is proportional to the content of oil. When CO2 diffuses into the oil, the content of oil decreases. Thus, the signal intensity of the MR images will weaken as the concentration of CO2 increases. Therefore, the distribution of dimensionless CO2 concentration can be calculated based on the measured signal intensity distribution of the MR images at each time period. According to Fick's law, the diffusion coefficients related to time and displacement can be computed based on such dimensionless concentrations. The concentration of CO2 is related to the MR signal intensity. The normalized concentration of CO2 in oil at different positions, z, and times, t, can be written as:
(4) |
xz,t = k(I0 − Iz,t) | (5a) |
x0 = 0 | (5b) |
xeq = k(I0 − Ieq) = const | (5c) |
Here, a non-iterative, finite-volume analytical technique was used to calculate the diffusion coefficient of CO2 in oil (n-hexadecane). It was assumed that there was a small composition gradient in the oil along the diffusion direction, and that the diffusion front did not move appreciably. Thus, Fick's Law (eqn (1)) can be used to relate the concentration and diffusion coefficient.
As shown in Fig. 4, the medium domain was discretized with mesh size Δz, time step Δt and grid points zi = i × Δz (i = 1, 2, …10) and tn = n × Δt (n = 0, 1, 2, …).19,20
Assuming that the concentration in boundary A is constant and there is no flux in boundary B, the discretized equations are as follows:
Internal points
(6a) |
Boundary A
(6b) |
Boundary B
(6c) |
By arranging the discretized equations within the medium domain and at the boundary surfaces, eqn (6) can be written in matrix form as Ax = b. Vector b is composed of the concentration measurement at specified grid locations along the sample in the experiment. The components of vector x are the unknown diffusion coefficients.
Fig. 5 Pressure decay during diffusion of CO2 in n-hexadecane starting with an initial pressure of 4377 kPa. |
The collected pressure–time data were numerically fitted in origin with an accuracy of 0.974–0.993. The overall diffusion coefficient of CO2 in n-hexadecane can be obtained by the fitted k* using eqn (3). The overall diffusion data obtained under different pressures are listed in Table 1. We calculated that the overall diffusion coefficient of CO2 in n-hexadecane to be about 3.21–4.30 × 10−9 m2 s−1. Even though the exact experimental conditions used in the present work have not been used before, Wang et al.22 and Grogan23 obtained diffusion coefficients of 3.32–5.71 × 10−9 m2 s−1 and 1.8–3.21 × 10−9 m2 s−1, respectively, under similar conditions. Our diffusion coefficients are of the same order of magnitude as these.
NO. | Oil height (mm) | Pressure of gas cylinder (kPa) | Initial pressure (kPa) | k* (s) | D (10−9/m−2 s−1) |
---|---|---|---|---|---|
1 | 20 | 3000 | 2568 | 50537.3 | 3.21 |
2 | 20 | 3500 | 3093 | 49201.1 | 3.71 |
3 | 20 | 4000 | 3420 | 41170.2 | 3.94 |
4 | 20 | 4600 | 3980 | 39020.2 | 4.15 |
5 | 20 | 5000 | 4377 | 37680.5 | 4.30 |
The ROIs of MR images were cropped with image processing software. As an example, Fig. 6 shows a time series of ROIs of the diffusion process of CO2 in n-hexadecane starting with an initial pressure of 4377 kPa. As CO2 diffuses into n-hexadecane, a decrease in image intensity from top to bottom can be clearly observed because CO2 is invisible in MR images. The CO2 concentration distribution can be calculated from the image intensity data.
Fig. 6 Series of ROI images at different times during the CO2 diffusion process for an initial pressure of 4377 kPa. |
The normalized concentration distribution of CO2 along the diffusion direction at different times calculated by eqn (4) for an initial pressure of 4377 kPa is shown in Fig. 7. Along the diffusion direction, the concentration of CO2 decreases gradually. For a given diffusion distance, the concentration of CO2 increases with time, which is consistent with the experimental results of Islas et al.24
Fig. 7 Normalized CO2 concentration distributions calculated from MR images with an initial pressure of 4377 kPa. |
The diffusion coefficients calculated from eqn (6) for initial pressures of 2568, 3093, 3420, 3980 and 4377 kPa are depicted in Fig. 8. All of the diffusion coefficients decreased as the diffusion distance increased. Moreover, the diffusion coefficients decreased dramatically near the gas/oil interface and gradually decreased to a stable value as the diffusion distance increased. The maximum diffusion coefficient was observed near the interface, indicating that concentration and density gradients exist in n-hexadecane during dissolution and diffusion of CO2. In addition, because the dissolution of CO2 is much higher at the gas/oil interface, the concentration gradient at this interface is relatively high, which causes convection and results in increased diffusion. Furthermore, Fig. 8 also reveals that the diffusion coefficient in a given region decreases over time, which suggests that the influence of the density gradient on convection and the incubation period decreases. In addition, diffusion coefficients increase with increasing initial pressure, which is consistent with the findings of the pressure decay method.
Fig. 8 Diffusion coefficients measured at different diffusion distances and times for initial pressures of (a) 2568 kPa, (b) 3093 kPa, (c) 3420 kPa, (d) 3980 kPa and (e) 4377 kPa. |
MRI can be used to measure the overall diffusion coefficient of CO2 in n-hexadecane. The overall diffusion coefficient is the average of diffusion coefficients measured during the time it takes for diffusion to reach equilibrium. First, the computational time step was determined, and then the distribution of dimensionless CO2 concentration was calculated using eqn (4). The position dependence of diffusion coefficients was computed using eqn (6). According to the relationship between diffusion coefficient and position, the trapezoidal rule was used to obtain the averages that represent time-dependent diffusion coefficients. Likewise, considering the relationship between diffusion coefficient and time, the trapezoidal rule was then used to average values to calculate the overall diffusion coefficient for each initial pressure. Fig. 9 shows the overall diffusion coefficients for different initial pressures calculated from the experimental MRI data and by the PVT method. According to eqn (3), the error between the MRI and PVT is about 1.6–4.3%. The two results showed good agreement. MRI can measure both unsteady-state local diffusion coefficients and overall diffusion coefficients. In addition, according to the experimental results, the overall diffusion coefficients increase along with the increase of the initial pressure.
(1) The concentration of CO2 decreases along the diffusion direction and is higher during the later stages of the reaction than during the initial stages.
(2) The diffusion coefficient of CO2 is a function of both diffusion time and diffusion distance. The diffusion coefficient decreases as diffusion distance and time increase until an equilibrium state is reached.
(3) The initial pressure has an obvious effect on the diffusion coefficient of CO2, with pressure being directly proportional to diffusion coefficient.
(4) The influence of the incubation period decreases with increasing diffusion time. Mass transfer after the incubation period is mainly governed by molecular diffusion.
D | Diffusion coefficient (m2 s−1) |
P(t), Peq | Measured and equilibrium pressure, respectively (kPa) |
t | Time (s) |
xeq(P) | Oil/gas interface molar concentration at a given time (mol cm−3) |
z0 | Height of the oil in the vessel (cm) |
xz,t | CO2 concentration at position z and time t from MRI images |
xeq | CO2 concentration at equilibrium state |
I0 | Initial image intensity at time t = 0 |
Iz,t | Image intensity at position z and time t |
Ieq | Image intensity at equilibrium state |
x | Molar concentration of CO2 gas |
This journal is © The Royal Society of Chemistry 2014 |