- Email: [email protected]
Comparative analysis of experimental methods for quantification of small amounts of oil in water
Journal of Petroleum Science and Engineering 147 (2016) 459–467
Contents lists available at ScienceDirect
Journal of Petroleum Science and Engineering journal homepage: www.elsevier.com/locate/petrol
Comparative analysis of experimental methods for quantification of small amounts of oil in water Konstantina Katika, Mehrdad Ahkami, Philip L. Fosbøl, Amalia Y. Halim, Alexander Shapiro, Kaj Thomsen, Ioannis Xiarchos, Ida L. Fabricius n Technical University of Denmark, Denmark
art ic l e i nf o
a b s t r a c t
Article history: Received 18 September 2015 Received in revised form 18 July 2016 Accepted 4 September 2016
During core flooding experiments where water is injected into oil bearing core plugs, the produced fluids can be sampled in a fraction collector. When the core approaches residual oil saturation, the produced amount of oil is typically small (can be less than a few microliters) and the quantification of oil is then difficult. In this study, we compare four approaches to determine the volume of the collected oil fraction in core flooding effluents. The four methods are: Image analysis, UV/visible spectroscopy, liquid scintillation counting, and low-field nuclear magnetic resonance (NMR) spectrometry. The procedure followed to determine the oil fraction and a summary of advantages and disadvantages of each method are given. Our results show that all four methods are reproducible with high accuracy. The NMR method was capable of direct quantification of both oil and water fractions, without comparison to a pre-made standard curve. Image analysis, UV/visible spectroscopy, and liquid scintillation counting quantify only the oil fraction by comparing with a pre-made standard curve. The image analysis technique is reliable when more than 0.1 ml oil is present, whereas liquid scintillation counting performs well when less than 0.6 ml oil is present. Both UV/visible spectroscopy and NMR spectrometry produced high accuracy results in the entire studied range (0.006–1.1 ml). In terms of laboratory time, the liquid scintillation counting is the fastest and least user dependent, whereas the NMR spectrometry is the most time consuming. & 2016 Elsevier B.V. All rights reserved.
Keywords: Low-field NMR Image analysis UV/visible spectroscopy Liquid scintillation counting Core flooding Oil fraction
1. Introduction Core flooding experiments are used for testing enhanced oil recovery strategies (EOR) on a laboratory scale. During these experiments, the oil-bearing core plugs are, for example, subjected to water injection. As a result, both oil and water are produced from the cores and typically collected in vials (Fig. 1). Quantification of the produced oil and water can be challenging, especially when the produced oil fractions are small, near residual oil saturation. A typical core flooding equipment (Fig. 1a) consists of: i) a piston displacement pump with fluid accumulators to force fluid into a core plug, ii) a test unit (the core plug inside a heated and pressurized core holder, to simulate oil reservoir conditions) and iii) a production unit, where the produced fluids are collected. Collection of the produced fluids (effluents) may be carried out by using a two or three-phase separator (Gupta et al., 2015). However, in this case, monitoring, for example, the presence of an active EOR component or changing pH and salinity is difficult (Stoll et al., n Correspondence to: DTU Byg - Department of Civil Engineering, Section for Geotechnics and Geology, Technical University of Denmark, DK-2800 Kgs. Lyngby, Denmark. E-mail addresses: [email protected] (K. Katika), [email protected] (I.L. Fabricius).
http://dx.doi.org/10.1016/j.petrol.2016.09.009 0920-4105/& 2016 Elsevier B.V. All rights reserved.
2011). If continuous chemical analysis is required, the produced oil and water should be sampled in a fraction collector containing series of vials (Fig. 1b). Each vial collects a given amount of effluent during time and is exchanged for a new vial automatically. In a typical core flooding experiment, the amount of oil in the produced fluid gradually decreases after water breakthrough. The recovered oil is distributed among several vials, and some vials may contain only a few microliters of oil. Due to capillary phenomena, the interface of oil to air is concave and the interface of oil to water is convex (Fig. 2a). The opaqueness of the oil will thus create an illusion of more oil within the sample than the actual volume. This causes a systematic error in visual measurement of the oil of about 200 ml higher than the actual volume. The recovered oil may also be disconnected from the continuous oil phase and be attached as film or droplets to the vial walls or form drops in the water (Fig. 2b) (Tang and Morrow, 1999). Quantification of this dispersed oil cannot be achieved simply by visual observations or weight-volume measurements. In the present paper, we compare the determination of the produced oil by the different methods. The simplest method is based on directly reading high-definition photographs of the vials containing oil and water. The three other methods are based respectively on ultraviolet/visible (UV/visible) spectroscopy, low
460
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
dPi Differential pressure transducer Injection cylinders
ISCO pump (Sleeve pressure)
Pressure tapped core holder
Data logger
Back pressure regulator
Fridge
BPR
Heating Temperature jacket controller
Fraction collector
(a)
ISCO pump (Injection)
(b) Fig. 1. (a) Core flooding apparatus and (b) fraction collector with vials.
Fig. 2. (a) The curvature of oil menisci within the vials with 200 ml of oil and (b) dispersed oil on the walls of the vial.
field nuclear magnetic resonance spectrometry (NMR), and liquid scintillation counting (LSC). For each method we evaluated the range of applicability, the accuracy and the time resources involved.
2. Methods 2.1. Image analysis Image analysis has been used for crude oil monitoring by several authors (e.g. Butler et al., 2001; Berg et al., 2010; Yang, 2011; Whitby et al., 2012). The method may be applied if the fluids are sampled in vials of identical geometry. Pal and Pal, 1993, discussed the main principle and operating procedure of the image analysis technique. According to their review, in high resolution images the estimated area is of high accuracy. A picture of each vial should be taken from the same position by a high resolution camera with a low distortion lens. Special precautions must be made in order to create uniform lighting and avoid reflections. The pictures are then analyzed with the help of digital analysis software for quantification of the oil volume. For the separation between near spherical oil droplets and more irregular solid particles, a shape factor defined as 4pi*Area/Perimeter2 will be one for spheres and less than one for other shapes. Gas bubbles can be recognized and excluded because they possess different optical properties from crude oil.
2.2. UV/Visible spectroscopy UV/visible spectroscopy is applicable to quantify oil mixed in water when the oil contains asphaltene fractions, aromatic fractions or other functional groups absorbing light in the ultraviolet region (Bastow et al., 1997; Evdokimov et al., 2003a, 2003b). Evdokimov et al. (2003a), (2003b) and Yang (2011), discussed the standard operating procedures of this technique. Bastow et al. (1997) showed the reproducibility of the UV–vis method for quantifying oil in water. In addition, the authors also compared their results to the ASTM 3921 standard test method for Oil and Grease and Petroleum Hydrocarbons in Water and the results were consistent. Bastow et al. (1997), measured the absorbance at a wavelength λ ¼256 nm in order to obtain the total aromatic hydrocarbon content of a crude oil. One of the limitations of the method of Bastow et al. (1997), is that nitrate may interfere since it absorbs the wavelengths around 220 nm. Evdokimov et al. (2003a), investigated molecular aggregation of toluene solutions of crude oil and of a solid asphaltene. They used a pre-weighted Tatarstan crude oil and precipitated asphaltene from the oil and measured the adsorptivity at different wavelengths (λ). The samples showed similar absorbance at λ ¼700–750 nm. Shorter wave lengths (λ o550 nm) give systematically higher absorptivity for solutions of crude oil in toluene as compared to pure asphaltene in toluene, due to other oil constituents (e.g. aromatics) (Evdokimov et al., 2003b). The principal approach of both studies is the same: oil samples are extracted with solvent and the absorbance is measured using a UV/Visible spectrophotometer. A standard curve
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
needs to be made using a known volume of oil for calibration purposes. 2.3. Liquid scintillation counting Liquid scintillation counting (LSC) or liquid scintillation analysis (LSA) has been a very popular technique for the detection and quantitative measurement of radioactivity since the early 1950s. The technique has been most useful in studies of the life sciences and the environment, and it is also a powerful tool in the chemical and physical sciences (L'Annunziata and Kessler, 2012). L'Annunziata and Kessler (2012), summarizes the principles and practice of this technique. In the petroleum industry, radioisotopes have been used for transport properties investigation, e.g. tritiated water (3H2O) and 14C for studying permeability, preferential flow zones, and inter-well connections (Wheeler et al., 1985; Lieser, 2001). Recently 14C radioisotope and liquid scintillation counting has been applied for quantitative determination of bioethanol in gasoline (Dijs et al., 2006) and for the investigation of transport mechanisms of CO2 foam for field applications (Dugstad et al., 2011). 14C radionuclides have been used for doping crude oil in biodegradation studies (Kanaly et al., 1997). For LSC the oil fraction can be determined by mixing the sample with a scintillation liquid. The energy of the radioactive decay is then transferred to the scintillation liquid and converted to photons. By counting the photons using a photon multiplier tube (PMT), the activity of radionuclides is measured (L'Annunziata and Kessler, 2012). This is performed in a dedicated scintillation counter. In liquid scintillation counting, molecular species present in the sample interfere with the process of energy transfer generated by the beta particle by pathways that do not involve light emission (chemical quenching). Also, the sample may interfere with light transmission after the conversion of this energy into light by the scintillator if the sample fluoresces or absorbs light in the range of the wavelength emitted from the scintillator. This process is known as external or color quenching (Rogers and Moran, 1966). According to L'Annunziata and Kessler (2012) and Rogers and Moran (1966), the internal standard method is the most accurate method to compensate for quenching. The internal standard method consists of the following steps for each radioactive crude oil sample. The first step is to measure all samples and obtain a count rate value for each of them (CPM, counts per minute). Then the samples are removed from the liquid scintillation analyzer and a known radioactivity (DPM, disintegrations per minute) radionuclide standard (14C in our case) is added to each sample. After thorough mixing, samples are recounted to obtain the count rate of the sample plus the internal standard. Then the counting efficiency of the sample is determined by:
E=
Ci + s − Cs Di
(1)
where E is counting efficiency of the sample, ci þ s is the count rate after the addition of internal standard, cs is the count rate before the addition of internal standard, and Di is the disintegration rate (radioactivity, DPM) of the added internal standard. The radioactivity Ds of the sample is then calculated by:
Ds =
Cs E
(2)
The accurate determination of the radioactivity, the accuracy and reproducibility of each measurement of a given sample has been studied in L'Annunziata and Kessler (2012), with the use of the internal standard method.
461
2.4. Low-field nuclear magnetic resonance spectrometry Low field nuclear magnetic resonance spectrometry is commonly used to determine petrophysical properties of reservoir rocks (Kenyon, 1997). NMR spectrometry can directly measure the density of hydrogen nuclei in reservoir fluids and determine the presence and quantities of the different fluids (water, oil, and gas) (Coates, 1999) and has been described in both lab petrophysical and logging applications (Dunn et al., 2002). Low field NMR has successfully been applied for the determination of water and heavy oil in emulsions at reservoir conditions (Allsopp et al., 2001). Additionally, this technique was used for the determination of the water droplet size in water-in-oil emulsions, as they form in oil and gas pipelines (Majid et al., 2015). Low field NMR experiments conducted at the same parameters and magnetic field strength provide similar results independent of the user, equipment and software used. The results should be reproducible with high accuracy as illustrated in Allsopp et al. (2001). Here we investigate the application of NMR spectrometry to quantify small amounts of oil in effluent from flooding experiments. Low-field NMR spectrometry involves a series of magnetic manipulations of the hydrogen nuclei found in pore fluids. The procedure includes the alignment and realignment of the hydrogen nuclear spins by action of static and oscillating magnetic fields. After each NMR measurement, an inversion technique is used to convert the perturbation decay curve into a distribution of T2 (where 1/T2 is the transverse relaxation rate). In the case of fluid samples containing both water and oil, the T2 distribution will typically be different for water and oil. For fluids in a test vial, the T2 distribution is reflecting bulk relaxation, which is the intrinsic relaxation property of a fluid and is controlled by its physical properties, such as viscosity and chemical composition (Coates, 1999). Low field NMR spectrometry is able to estimate directly the volume of the oil and water within each sample provided the hydrogen index (HI) of each fluid is first measured. HI is a factor representing the number of hydrogen atoms per unit volume divided by the number of hydrogen atoms per unit volume of pure water.
3. Experimental procedure 3.1. Sample preparation North Sea crude oil with a density of 0.845 g/cm3 and synthetic seawater with a density of 1.019 g/cm3 (Table 1), were used to prepare the samples used in this study. The densities were measured using a digital density meter of high accuracy (7 0.001). Samples were divided into two categories: a) non-radioactive and b) radioactive. The non-radioactive (NO) and radioactive (SRO2) samples were prepared according to Table 2. The oil and the synthetic seawater were weighed in a 10 ml glass vial using an analytical balance. This set of samples was used in image analysis, UV/Visible spectroscopy, and NMR studies. First the samples were photographed for the image analysis method, then they were analyzed using low field NMR spectrometry and finally the amount of oil was determined from the UV/visible spectroscopy. The sequence of the techniques was selected carefully, so as not to destroy the samples before the last step of the analysis. Two extra sets of samples were prepared and used for the preparation of standard curves for image analysis and for UV/visible spectroscopy; NOa and NOb respectively (Figs. 3 and 4). For the preparation of the radioactive samples (SRO2), one liter of the aforementioned crude oil was mixed with 1.25 ml solution
462
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
Table 1 Seawater composition. Components
g/l
NaCl NaHCO3 KCl MgCl2 6H2O CaCl2 2H2O Na2SO4 TDS
18.01 0.17 0.75 9.15 1.91 3.41 33.39
Table 2 Non-radioactive (NO) and radioactive (SRO2) crude oil samples. Sample ID
Oil mass (g)
Oil vol (μl)
Sample ID
Oil mass (g)
Oil vol. (μl)
NO-1 NO-2 NO-3 NO-4 NO-5 NO-6 NO-7 NO-8 NO-9 NO-10
0.853 0.635 0.422 0.254 0.086 0.045 0.021 0.01 0.005 0
1007.4 750.2 498.2 300.1 102.1 53.3 24.6 11.2 5.6 0
SRO2-1 SRO2-2 SRO2-3 SRO2-4 SRO2-5 SRO2-6 SRO2-7 SRO2-8 SRO2-9 –
0.9645 0.7892 0.565 0.3076 0.1999 0.1753 0.126 0.0316 0 –
1141.1 933.6 668.4 363.9 236.5 207.4 149.1 37.4 0 –
Fig. 4. The standard curve used for the quantification of the oil within the NOsamples for the UV/visible spectroscopy method. The standard curve created from the measurement of the optical density of the N0b samples. The equations illustrated in the figure are used to quantify the oil fractions within the NO-samples. Table 3 Physical properties of Stearic acid (Lide and Milne, 19). Name: stearic acid [1-14C]
Formula: CH3 (CH2)1614COOH
Physical properties MP (°C) BP (°C) Density (g/ml)
68.8 350 0.9408
H2O: insoluble, EtOH, Benzene: slightly soluble, Acetone, Chloroform, CS2: very soluble
Fig. 3. The standard curve used for the quantification of the oil within the NOsamples for the image analysis technique. The standard curve was created from the measurement of the area of the N0a samples. The equation illustrated in the figure is used to quantify the oil fractions within the NO-samples. Samples containing less than 100 μl were excluded from the analysis since they were scattered.
of 14C stearic acid (Table 3) dissolved in toluene (7.4 MBq/ml from Perkin Elmer). The radioactive crude oil samples were placed in 20 ml polyethylene scintillation vials with screw cap. The total volume was calculated on the basis of known mass and density. A weight balance of high accuracy (7 0.0001 g) and an electrical density meter were used to measure the weight and density of the fluid respectively. The use of the scintillation vials is necessary for measuring the radioactive oil in the liquid scintillator. Another set of samples (labeled as SRO1) was prepared and used for the preparation of the standard curve for liquid scintillation counting (Fig. 5).
Fig. 5. The standard curve created from the measurement of the radioactivity of the SRO1 samples using the liquid scintillation method. The radioactivity data of these samples corresponds to the y- values and the -x- value is the volume of the radioactive oil. The equation illustrated in the figure is used to quantify the oil fractions within the SRO2-samples.
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
463
Fig. 6. (a) The area measured and (b) the detected volume of oil plotted against the actual oil volume within the NO-samples for the image analysis technique with the use of the equation in Fig. 3.
3.2. Analysis of samples 3.2.1. The image analysis method High resolution images (using a 15.1 megapixel camera) were taken of standard samples and of NO samples and images were analyzed with the help of the digital analysis software ImageJ (Schneider et al., 2012; Abramoff et al., 2004). The diameter of the vial was used to calculate the size of each pixel, and then the oil area was marked manually and calculated from the number of pixels. The standard data prepared for standardization of the image analysis method indicates a linear dependency between the calculated surface area and the predetermined volume of oil for samples containing more than 0.1 ml oil (Fig. 3). Samples containing less than 0.1 ml oil were removed from the analysis. The equation on the graph (Fig. 3) was used to determine the oil volume based on the area detected for each NO-sample (Fig. 6). 3.2.1.1. Advantages and disadvantages. The image analysis method is relatively simple, cheap and fast, as it takes no more than ten minutes to take the picture and analyze the result for each vial, it does not require much preparation and elaborate equipment. Under conditions where multiple data have to be processed it may save time and usage of expensive equipment. The method is nondestructive and has high accuracy when the oil is distributed regularly and the amount is higher than 100 μl (Fig. 6). It was observed that when the oil amount is smaller than 100 μl, the method is unreliable because of the difficulty in manual indication of the regions of low oil amounts. The method cannot properly account for the irregularly dispersed oil phase (e.g. in the form of the drops or films on the vial surfaces). 3.2.2. The UV/visible spectroscopy analysis method The NO samples and samples for the standard curve were treated using the UV/visible spectroscopy as described by Evdokimov et al., 2003a, 2003b. Approximately 3 ml of toluene were weighed and added to each NO sample. The samples were then shaken in order to mix oil and toluene, and afterwards they were stored to let the liquid separate into two phases. The upper phase containing oil dissolved in toluene was transferred into a quartz glass cuvette (path length 10 mm) and the optical density was measured using a UV–VIS Spectrophotometer (UV mini 1240,
Shimadzu) at λ ¼750 nm. If the spectrophotometer showed an absorbance above 1 (corresponding to samples with more than 0.2 ml oil), the sample was further diluted in order to get an accurate measurement. Between the measurements, the glass cuvette was rinsed with toluene and ethanol followed by drying with compressed air. Only pure grade toluene and ethanol were used in this experiment. For the standard samples, the absorbance data was then plotted against the oil volume. Two linear regression equations were obtained as standard curves (Fig. 4), which were then be used to quantify the amount of oil in the NO samples (Fig. 7). 3.2.2.1. Advantages and disadvantages. The UV/visible method is a low cost, fast (less than 5 min per sample) and overall simple method. It is reproducible and can give direct analysis of the oil-inwater samples. The method can determine the amount of oil at any condition, even when the oil is dispersed in a vial in the form of drops. The method is destructive, which is a major disadvantage. The samples cannot be used for other analysis after the solvent is added. Furthermore, the standard curve needs to be made using the same oil as the samples that are going to be analyzed. Oil volumes higher than 200 μl require dilution. 3.2.3. The liquid scintillation counting method For the LSC method, the oil was doped with low-concentration, radiolabeled stearic acid [1-14C] (Table 3), which is fully dissolvable in the oil. The labeled acid is not soluble in water and no radioactivity is transferred to the water. 14C stearic acid has a known radioactivity of 222346 dpm. For standard samples and SRO2 samples, the internal standard method was selected to anticipate quenching effects due to the chemical complexity and color of the crude oil. 15 ml of scintillation liquid Ultima Gold (Edler, 2015) was added to each of the samples. All the samples were analyzed using the scintillation counter. The energy window used for analysis was 0–156 keV, which is the energy spectrum of beta emission for 14C radionuclide. The counting time was set to 30 min. For the standard samples, the radioactivity was then plotted against the oil volume and a linear regression equation was obtained as standard curve (Fig. 5), which was then used to quantify the amount of oil in the SRO2 samples (Fig. 8). A higher
464
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
Fig. 7. (a) The optical density data and (b) the detected volume of oil plotted against the actual oil volume within the NO-samples for the UV/visible spectroscopy by using the equations in Fig. 4.
deviation between detected and actual crude oil volume is observed above 0.6 ml. It might be attributed to a quenching effect. 3.2.3.1. Advantages and disadvantages. The liquid scintillation apparatus has a tray for carrying many samples. So in the present study 40 samples can be analyzed in one batch. By liquid scintillation, low levels of radioactivity and, thus low amounts of oil can be detected. The signal is proportional to the radioactive oil amount and is easy to calibrate. Disadvantages of this method are chemical and color quenching. The standardization must compensate for this. It is especially important for oil samples that absorb light in the range of the wavelength emitted from the scintillator. The liquid scintillation analysis is destructive. The samples cannot be reused after the scintillation liquid is added. According to L’Annunziata and Kessler (2012), the internal standard method is the most accurate method to correct for quenching. The risk of radioactive fluid leaks is important.
Ingestion and skin contact with the radioactive material can be hazardous. The relatively high cost of the counting apparatus and additional expenses for the disposal of radioactive effluents have to be taken into account. Contamination surveys should be performed regularly in lab facilities. The handling of radioactive samples should be done correctly. 3.2.4. The NMR method NMR measurements were made at room temperature with a GeoSpec2 NMR Core Analyzer at 2 MHz. The T2 relaxation curves were measured by using a recycle delay (repetition time) of 40 s, number of echoes 32768, CPMG inter echo spacing (τ) 50 μs and 64 scans. The π/2 and π pulses were 9.75 μs and 19.5 μs, respectively. The T2 relaxation spectra were generated using the WinDXP (Oxford Instruments, UK) software. For the NO samples, the hydrogen nuclear spin in crude oil relaxes faster than in water, hence, the T2 distribution can be
Fig. 8. (a) The radioactivity measured and (b) the detected volume of oil plotted against the actual oil volume within the SRO2-samples for the liquid scintillation analysis by using the equations in Fig. 5.
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
465
Fig. 9. Typical T2 distribution of an oil/water sample (0.05 ml oil and 3.06 ml water).
divided into two areas (Fig. 9). The peaks with T2 less than 1 s correspond to the signal from oil, whereas the peaks with higher T2 represent the water signal. The hydrogen index for the crude oil used in the current study is different from the index for the synthetic brine. The area under the peaks may therefore be translated into the volume of fluid emitting this signal after a linear correction. A 10% reduction of all oil volumes detected from the peak area was required in order to quantify the exact volumes of oil within the samples (Fig. 10). Fig. 11 represents the quantification of the water fraction. 3.2.4.1. Advantages and disadvantages. Samples in glass vials can be used in the NMR probe without any extra preparation. The volume of each fluid does not interfere with the accuracy of the measurements. The NMR technique is not destructive to the samples. The volume of oil and water can be quantified in the presence of radioactive materials. The NMR measurement is safe and poses no risk to health. During NMR, the sample is not exposed to open air or other possible sources of contamination. This is an important point which gives the opportunity to measure any internal sample change resulting from for example storage. A drawback of this technique is related to the sample containers. The method can only be used in, “low-field NMR friendly” materials. Some plastics can create noise during the measurements and reduce the Signal-to-Noise ratio (SNR). In this work we choose to enclose our fluids in glass containers. The NMR probe used in this work has a limit on the volume of the sample. Each
Fig. 11. The water volume detected from the low field NMR spectrometry.
vial has to contain fluid that corresponds to less than five centimeters height. Above that height the protons cannot be magnetized. Currently there is no automatic sample changer for the apparatus, so the method is time consuming. Analysis of one sample takes 40 min, because it has to equilibrate with the apparatus temperature before data recording.
4. Discussion We were able to obtain a good correlation of the amount of oil added in the glass containers and the amount of oil detected from all four techniques as reflected in high correlation coefficient (R) and low standard deviation (s) for the analysis of 10 samples (n) (Table 4). Advantages and drawbacks of each method are summarized in Table 5.
Fig. 10. (a) The volume detected and (b) the corrected volume of oil plotted against the actual oil volume within the NO-samples for the low field NMR spectrometry. Fig. 10b was created after applying a 10% volume correction to all data in Fig. 10a.
466
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
Table 4 The correlation coefficient (R) and the standard deviation (s) of the samples (n) analyzed in each method. The errors of each measured value are based on the reproducibility of each technique between the standard curve and the final measurements. Method
n
R2
s (μl)
Error (μl)
Image analysis UV/visible spectroscopy Liquid scintilation analysis Low field NMR
10 10 10 9
0.96 0.99 0.92 0.99
599 105 378 56
117 47 142 –
The errors for the image analysis technique were calculated based on the performance of this technique on the entire range of oil volumes; including samples with less than 0.1 ml of oil. Low field NMR and UV/visible spectroscopy were able to detect the oil of the samples within a wide range of oil volumes; from a few microliters to several milliliters. Low field NMR spectrometry was shown to be the most precise in respect to accurate quantification of the oil fractions, but UV/visible spectroscopy produced almost similar accuracy throughout the measurements, as illustrated by correlation coefficients and standard deviations (Table 4). On the other hand, the UV/visible spectroscopy is a user dependent technique that requires manual operation to add toluene to the samples and the use of standard curves which might introduce errors to the results. The standard curves for this technique must be produced with the same oil sample that is going to be analyzed. If the oil volume is higher than 0.2 ml the sample should be diluted several times and an error from the extensive manual dilution might be introduced. Finally, low field NMR provides an accurate determination of the water fraction within each sample, giving the opportunity to quantify the amount and the type of fluids within each sample. A drawback of the NMR method is that the entire volume of the oil should be within 5 cm height. Similarly to the two abovementioned techniques, liquid scintillation counting resulted in a high correlation coefficient and low residuals in the wide range of oil samples (Table 4). During the analysis, the residuals proved to be higher at bigger amounts of oil, proving this technique more accurate at smaller amounts of oil. Finally, the image analysis method showed a correlation between the added and detected amount of oil with a correlation coefficient equal to 0.99 for oil volumes above 100 μl (Fig. 6). As the amount of oil decreases within the sample, this technique provides a coefficient of less than 0.97 for the entire range of volumes and the level of residuals increases unless only samples containing oil volume higher than 0.1 ml are used (Table 4). A drawback in the image method is its dependency on oil distribution within a vial. Unlike other methods, the visual method cannot
be applied in more complicated cases such as shown in Fig. 2b or in cases that emulsification phenomena occur.
5. Conclusions We may summarize the applicability of the four different methods in regard to: a) accuracy of the detected amount of oil, b) experimental time, c) technical work power, d) post destruction of the samples, e) money, and f) final recommendation. a) With respect to accuracy, low field NMR spectrometry and UV/ visible spectroscopy proved to be able to detect the oil fraction in the entire range of oil volume. The same accuracy was achieved with the liquid scintillation technique for samples with less than 0.6 ml oil. Finally, the image analysis is only suitable for larger amounts of oil, above 0.1 ml. b) With respect to experimental time, low field NMR spectrometry can only test one sample at a time, which might be very time-consuming, especially in the case of water flooding experiments with hundreds of samples per flooding. Each measurement requires 40 min and the overall time required for analysis is unrealistic. It can be used in a small selection of samples to control the results obtained by other techniques or in cases where the studied vials do not exceed a couple of tens. UV/visible spectroscopy requires 5 min for each analysis and an additional 5 min for dilution when necessary. The liquid scintillation analysis requires 30 min for the quantification of the oil in each sample, but the tests can be performed automatically with batches of 40 samples per run. The image analysis technique is able to detect the oil volume in one sample at a time and each analysis can be done in 10 min. When using image analysis, it is an advantage to use the same type of vials for fraction collectors throughout the core flooding experiments. This can save time for the user. But even after the proper selection of vials, the amount of time spent for the detection of oil with the image analysis in hundreds of vials coming out of the fraction collector is unrealistic. c) With respect to technical work power, low-field NMR spectrometry is a very time consuming technique but does not require technical work power during equilibration time. UV/ visible spectroscopy is relatively fast but requires constant presence of a technician. The main advantage of liquid scintillation counting is that the measurements can be performed overnight and save valuable time and technical work power. Image analysis requires technical work power constantly. d) With respect to post destruction, out of the four methods, NMR spectrometry and image analysis proved not to interfere with the sample at all, since both measurements are
Table 5 Advantages and drawbacks of image analysis, UV/visible spectroscopy, liquid scintillation counting and low-field NMR spectrometry. Parameters
Image analysis
UV/visible spectroscopy Liquid scintillation counting
Low-field NMR spectrometry
Time consumption: Destructive: Limitations in volume:
Low No Volumes higher than 0.1 ml
Low Yes No
Low Yes No
Limitations in concentration: No
Yes
User dependent:
Yes
Yes
Preparation: Standard curves: Conditions of sample:
Photos Yes Cannot measure oil drops and dispersed oil No
Solvents Yes No
High amounts of oil (color quenching effects). Internal standard method (labor intensive). Use of doped oil. Yes Mixing with scintillation liquid
High No No – within probe detection range No
No No No
No
No
Yes
Measures water fraction:
No
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
conducted in closed vials. Liquid scintillation counting and UV/ visible spectroscopy require additives to each sample for analysis; therefore, the samples cannot be used further. When combining these techniques with image analysis or NMR, the destructive method should be applied last. e) In terms of saving money, image analysis has proven to be the cheapest of all techniques since it only requires photos of each sample and software for the analysis. Low field NMR requires the presence of a low field NMR spectrometer which is a relatively expensive piece of equipment designed for laboratory rock core analysis in the petroleum industry. UV/visible spectroscopy can be performed in a spectrophotometer that fulfills the criteria described above. Finally, liquid scintillation counting can be performed in laboratories certified to handle radioactive tracers. f) Final recommendation: image analysis is a relatively accurate technique when it comes to quantifying the oil fractions in volumes higher than a 0.1 ml, whereas liquid scintillation counting gives high accuracy for amounts of oil less than 0.6 ml. There is clearly a benefit of combining these two techniques to save time and increase the accuracy of the overall measurements. If a single method should be chosen, UV/visible spectroscopy is the most trustworthy. If all measurements are conducted by the same user, any errors in respect to user and pipette dependency can be diminished. Low field NMR and image analysis can be used in addition to this method, in order to certify the results.
Acknowledgments The research has been carried out in the framework of the BioRec and SmartWater projects supported by the Danish Energy Agency (EUDP-10-II 64011-0009), Innovation Fund Denmark, Mærsk Oil AS, DONG Energy AS, and Novozymes AS.
References Abramoff, M.D., Magalhaes, P.J., Ram, S.J., 2004. Image Processing with ImageJ. Biophotonics Int. 11 (7), 36–42. Allsopp, K., Wright, I., Lastockin, D., Mirotchnik, K., Kantzas, A., 2001. Determination of oil and water compositions of oil/water emulsions using low field NMR relaxometry. J. Can. Pet. Technol. 40 (7), 58–61. Bastow, T., Durnie, W.H., Jefferson, A., Pang, J., 1997. Ultraviolet spectroscopy for the analysis of oil-in-water effluent using isopropanol as co-solvent. Appl. Spectrosc. 51 (3), 318–322. Berg, S., Cense, A.W., Jansen, E., Bakker, K., 2010. Direct experimental evidence of wettability modification by low salinity. Petrophysics 51 (5), 314.
467
Butler, J.B., Lane, S.N., Chandler, J.H., 2001. Automatic extraction of grain-size data from gravel surfaces using digital image processing. Am. Soc. Civ. Eng. J. Hydraul. Div. 39, 519–529. Coates, G.R., Xiao, L., Prammer, M.G., 1999. NMR logging: principles and applications. S.l.: Gulf Professional Publishing. Dunn, K., Bergman, D.J., LaTorraca, G.A., 2002, Nuclear magnetic resonance-petrophysical and logging applications. In: Handbook of Geophysical Exploration: Seismic Exploration, Elsevier Science Ltd., Pergamon Press, New York. Dijs, I.J., Windt, E., Kaihola, L., Borg, K., 2006. Quantitative determination by 14C analysis of the biological component in fuels. Radiocarbon 48 (3), 315–323. Dugstad, O., Opel, K., Fjelde, I., 2011, Improved Understanding of CO2/Foam EOR Techniques by AID of Tracer Technology. In: Proceedings of the Society of Petroleum Engineers, SPE EUROPEC/EAGE Annual Conference and Exhibition, Vienna, Austria, 23–26 May. Edler, R., 2015. Cocktails for Liquid Scintillation Counting – Application Note. PerkinElmer, Inc. Evdokimov, I.N., Eliseev, N.Y., Akhmetov, B.R., 2003a. Assembly of asphaltene molecular aggregates as studied by near-UV/visible spectroscopy: I. Structure of the absorbance spectrum. J. Pet. Sci. Eng. 37 (3–4), 135–143. Evdokimov, I.N., Eliseev, N.Y., Akhmetov, B.R., 2003b. Assembly of asphaltene molecular aggregates as studied by near-UV/visible spectroscopy: II. Concentration dependencies of absorptivities. J. Pet. Sci. Eng. 37 (3–4), 145–152. Gupta, R., Glotzbach, R., Hehmeyer, O., 2015. A novel field-representative enhancedoil-recovery coreflood method. SPE J., 21–34. Kanaly, R., Bartha, R., Fogel, S., Findlay, M., 1997. Biodegradation of [14C]Benzo[a] pyrene added in crude oil to uncontaminated soil. Appl. Environ. Microbiol. 63 (11), 4511–4515. Kenyon, W.E., 1997. Petrophysical principles of applications of NMR logging. Log Anal. 38, 21–43. L'Annunziata, F.M., Kessler, J.M., 2012. Liquid scintillation analysis: principles and practice. In: L’Annunziata, F.M. (Ed.), Handbook of Radioactivity Analysis, third ed. Academic Press, Amsterdam, pp. 423–573, Chapter 7. Lieser, K.H., 2001. Technical and Industrial Applications of Radionuclides and Nuclear Radiation, in: Nuclear and Radiochemistry: Fundamentals and Applications, second revised ed. Wiley-VCH Verlag GmbH, Weinheim, Germany, p. 386. Majid, A.A., Saidian, M., Prasad, M., Koh, C.A., 2015. Measurement of the water droplet size in water-in-oil emulsions using low field nuclear magnetic resonance for gas hydrate slurry application. Can. J. Chem. 93, 1–7. Pal, N.R., Pal, S.K., 1993. A review on image segmentation techniques. Pattern Recognit. 26 (9), 1274–1294. Rogers, W.A., Moran, F.J., 1966. Evaluation of quench correction in liquid scintillation counting by internal automatic external, and channels' ratio standardization methods, vol. 16(2), pp. 206–219. Schneider, C.A., Rasband, W.S., Eliceiri, K.W., 2012. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671–675. Stoll, W.M., Al Shureqi, H., Finol, J., Al-Harthy Said, A.A., Oyemade Nneamaka, S., De Kruijf, A., Van Wunnik, J., Arkesteijn, F., Bouwmeester, R., Faber, M.J., 2011. Alkaline/surfactant/polymer flood: from the laboratory to the field. SPE Reserv. Eval. Eng. 14 (6), 702–712. Tang, G.Q., Morrow, N.R., 1999. Influence of brine composition and fines migration on crude oil/brine/rock interactions and oil recovery. J. Pet. Sci. Eng. 24, 99–111. Wheeler, V.J., Parsons, T.V., Conchie, S.J., 1985. The application of radioactive tracers to oil reservoir waterflood studies. In: Proceedings of the SPE 13985/1 Presented at SPE/AIME Offshore European 85 Conference, Aberdeen, Scotland, March 10–13. Whitby, C.P., Lotte, L., Lang, C., 2012. Structure of concentrated oil-in-water Pickering emulsions. Soft Matter 8 (30), 7784–7789. Yang, M., 2011. Measurement of oil in produced water. In: Lee, K., Neff, J. (Eds.), Produced Water. Springer, New York, pp. 74–75.
Contents lists available at ScienceDirect
Journal of Petroleum Science and Engineering journal homepage: www.elsevier.com/locate/petrol
Comparative analysis of experimental methods for quantification of small amounts of oil in water Konstantina Katika, Mehrdad Ahkami, Philip L. Fosbøl, Amalia Y. Halim, Alexander Shapiro, Kaj Thomsen, Ioannis Xiarchos, Ida L. Fabricius n Technical University of Denmark, Denmark
art ic l e i nf o
a b s t r a c t
Article history: Received 18 September 2015 Received in revised form 18 July 2016 Accepted 4 September 2016
During core flooding experiments where water is injected into oil bearing core plugs, the produced fluids can be sampled in a fraction collector. When the core approaches residual oil saturation, the produced amount of oil is typically small (can be less than a few microliters) and the quantification of oil is then difficult. In this study, we compare four approaches to determine the volume of the collected oil fraction in core flooding effluents. The four methods are: Image analysis, UV/visible spectroscopy, liquid scintillation counting, and low-field nuclear magnetic resonance (NMR) spectrometry. The procedure followed to determine the oil fraction and a summary of advantages and disadvantages of each method are given. Our results show that all four methods are reproducible with high accuracy. The NMR method was capable of direct quantification of both oil and water fractions, without comparison to a pre-made standard curve. Image analysis, UV/visible spectroscopy, and liquid scintillation counting quantify only the oil fraction by comparing with a pre-made standard curve. The image analysis technique is reliable when more than 0.1 ml oil is present, whereas liquid scintillation counting performs well when less than 0.6 ml oil is present. Both UV/visible spectroscopy and NMR spectrometry produced high accuracy results in the entire studied range (0.006–1.1 ml). In terms of laboratory time, the liquid scintillation counting is the fastest and least user dependent, whereas the NMR spectrometry is the most time consuming. & 2016 Elsevier B.V. All rights reserved.
Keywords: Low-field NMR Image analysis UV/visible spectroscopy Liquid scintillation counting Core flooding Oil fraction
1. Introduction Core flooding experiments are used for testing enhanced oil recovery strategies (EOR) on a laboratory scale. During these experiments, the oil-bearing core plugs are, for example, subjected to water injection. As a result, both oil and water are produced from the cores and typically collected in vials (Fig. 1). Quantification of the produced oil and water can be challenging, especially when the produced oil fractions are small, near residual oil saturation. A typical core flooding equipment (Fig. 1a) consists of: i) a piston displacement pump with fluid accumulators to force fluid into a core plug, ii) a test unit (the core plug inside a heated and pressurized core holder, to simulate oil reservoir conditions) and iii) a production unit, where the produced fluids are collected. Collection of the produced fluids (effluents) may be carried out by using a two or three-phase separator (Gupta et al., 2015). However, in this case, monitoring, for example, the presence of an active EOR component or changing pH and salinity is difficult (Stoll et al., n Correspondence to: DTU Byg - Department of Civil Engineering, Section for Geotechnics and Geology, Technical University of Denmark, DK-2800 Kgs. Lyngby, Denmark. E-mail addresses: [email protected] (K. Katika), [email protected] (I.L. Fabricius).
http://dx.doi.org/10.1016/j.petrol.2016.09.009 0920-4105/& 2016 Elsevier B.V. All rights reserved.
2011). If continuous chemical analysis is required, the produced oil and water should be sampled in a fraction collector containing series of vials (Fig. 1b). Each vial collects a given amount of effluent during time and is exchanged for a new vial automatically. In a typical core flooding experiment, the amount of oil in the produced fluid gradually decreases after water breakthrough. The recovered oil is distributed among several vials, and some vials may contain only a few microliters of oil. Due to capillary phenomena, the interface of oil to air is concave and the interface of oil to water is convex (Fig. 2a). The opaqueness of the oil will thus create an illusion of more oil within the sample than the actual volume. This causes a systematic error in visual measurement of the oil of about 200 ml higher than the actual volume. The recovered oil may also be disconnected from the continuous oil phase and be attached as film or droplets to the vial walls or form drops in the water (Fig. 2b) (Tang and Morrow, 1999). Quantification of this dispersed oil cannot be achieved simply by visual observations or weight-volume measurements. In the present paper, we compare the determination of the produced oil by the different methods. The simplest method is based on directly reading high-definition photographs of the vials containing oil and water. The three other methods are based respectively on ultraviolet/visible (UV/visible) spectroscopy, low
460
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
dPi Differential pressure transducer Injection cylinders
ISCO pump (Sleeve pressure)
Pressure tapped core holder
Data logger
Back pressure regulator
Fridge
BPR
Heating Temperature jacket controller
Fraction collector
(a)
ISCO pump (Injection)
(b) Fig. 1. (a) Core flooding apparatus and (b) fraction collector with vials.
Fig. 2. (a) The curvature of oil menisci within the vials with 200 ml of oil and (b) dispersed oil on the walls of the vial.
field nuclear magnetic resonance spectrometry (NMR), and liquid scintillation counting (LSC). For each method we evaluated the range of applicability, the accuracy and the time resources involved.
2. Methods 2.1. Image analysis Image analysis has been used for crude oil monitoring by several authors (e.g. Butler et al., 2001; Berg et al., 2010; Yang, 2011; Whitby et al., 2012). The method may be applied if the fluids are sampled in vials of identical geometry. Pal and Pal, 1993, discussed the main principle and operating procedure of the image analysis technique. According to their review, in high resolution images the estimated area is of high accuracy. A picture of each vial should be taken from the same position by a high resolution camera with a low distortion lens. Special precautions must be made in order to create uniform lighting and avoid reflections. The pictures are then analyzed with the help of digital analysis software for quantification of the oil volume. For the separation between near spherical oil droplets and more irregular solid particles, a shape factor defined as 4pi*Area/Perimeter2 will be one for spheres and less than one for other shapes. Gas bubbles can be recognized and excluded because they possess different optical properties from crude oil.
2.2. UV/Visible spectroscopy UV/visible spectroscopy is applicable to quantify oil mixed in water when the oil contains asphaltene fractions, aromatic fractions or other functional groups absorbing light in the ultraviolet region (Bastow et al., 1997; Evdokimov et al., 2003a, 2003b). Evdokimov et al. (2003a), (2003b) and Yang (2011), discussed the standard operating procedures of this technique. Bastow et al. (1997) showed the reproducibility of the UV–vis method for quantifying oil in water. In addition, the authors also compared their results to the ASTM 3921 standard test method for Oil and Grease and Petroleum Hydrocarbons in Water and the results were consistent. Bastow et al. (1997), measured the absorbance at a wavelength λ ¼256 nm in order to obtain the total aromatic hydrocarbon content of a crude oil. One of the limitations of the method of Bastow et al. (1997), is that nitrate may interfere since it absorbs the wavelengths around 220 nm. Evdokimov et al. (2003a), investigated molecular aggregation of toluene solutions of crude oil and of a solid asphaltene. They used a pre-weighted Tatarstan crude oil and precipitated asphaltene from the oil and measured the adsorptivity at different wavelengths (λ). The samples showed similar absorbance at λ ¼700–750 nm. Shorter wave lengths (λ o550 nm) give systematically higher absorptivity for solutions of crude oil in toluene as compared to pure asphaltene in toluene, due to other oil constituents (e.g. aromatics) (Evdokimov et al., 2003b). The principal approach of both studies is the same: oil samples are extracted with solvent and the absorbance is measured using a UV/Visible spectrophotometer. A standard curve
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
needs to be made using a known volume of oil for calibration purposes. 2.3. Liquid scintillation counting Liquid scintillation counting (LSC) or liquid scintillation analysis (LSA) has been a very popular technique for the detection and quantitative measurement of radioactivity since the early 1950s. The technique has been most useful in studies of the life sciences and the environment, and it is also a powerful tool in the chemical and physical sciences (L'Annunziata and Kessler, 2012). L'Annunziata and Kessler (2012), summarizes the principles and practice of this technique. In the petroleum industry, radioisotopes have been used for transport properties investigation, e.g. tritiated water (3H2O) and 14C for studying permeability, preferential flow zones, and inter-well connections (Wheeler et al., 1985; Lieser, 2001). Recently 14C radioisotope and liquid scintillation counting has been applied for quantitative determination of bioethanol in gasoline (Dijs et al., 2006) and for the investigation of transport mechanisms of CO2 foam for field applications (Dugstad et al., 2011). 14C radionuclides have been used for doping crude oil in biodegradation studies (Kanaly et al., 1997). For LSC the oil fraction can be determined by mixing the sample with a scintillation liquid. The energy of the radioactive decay is then transferred to the scintillation liquid and converted to photons. By counting the photons using a photon multiplier tube (PMT), the activity of radionuclides is measured (L'Annunziata and Kessler, 2012). This is performed in a dedicated scintillation counter. In liquid scintillation counting, molecular species present in the sample interfere with the process of energy transfer generated by the beta particle by pathways that do not involve light emission (chemical quenching). Also, the sample may interfere with light transmission after the conversion of this energy into light by the scintillator if the sample fluoresces or absorbs light in the range of the wavelength emitted from the scintillator. This process is known as external or color quenching (Rogers and Moran, 1966). According to L'Annunziata and Kessler (2012) and Rogers and Moran (1966), the internal standard method is the most accurate method to compensate for quenching. The internal standard method consists of the following steps for each radioactive crude oil sample. The first step is to measure all samples and obtain a count rate value for each of them (CPM, counts per minute). Then the samples are removed from the liquid scintillation analyzer and a known radioactivity (DPM, disintegrations per minute) radionuclide standard (14C in our case) is added to each sample. After thorough mixing, samples are recounted to obtain the count rate of the sample plus the internal standard. Then the counting efficiency of the sample is determined by:
E=
Ci + s − Cs Di
(1)
where E is counting efficiency of the sample, ci þ s is the count rate after the addition of internal standard, cs is the count rate before the addition of internal standard, and Di is the disintegration rate (radioactivity, DPM) of the added internal standard. The radioactivity Ds of the sample is then calculated by:
Ds =
Cs E
(2)
The accurate determination of the radioactivity, the accuracy and reproducibility of each measurement of a given sample has been studied in L'Annunziata and Kessler (2012), with the use of the internal standard method.
461
2.4. Low-field nuclear magnetic resonance spectrometry Low field nuclear magnetic resonance spectrometry is commonly used to determine petrophysical properties of reservoir rocks (Kenyon, 1997). NMR spectrometry can directly measure the density of hydrogen nuclei in reservoir fluids and determine the presence and quantities of the different fluids (water, oil, and gas) (Coates, 1999) and has been described in both lab petrophysical and logging applications (Dunn et al., 2002). Low field NMR has successfully been applied for the determination of water and heavy oil in emulsions at reservoir conditions (Allsopp et al., 2001). Additionally, this technique was used for the determination of the water droplet size in water-in-oil emulsions, as they form in oil and gas pipelines (Majid et al., 2015). Low field NMR experiments conducted at the same parameters and magnetic field strength provide similar results independent of the user, equipment and software used. The results should be reproducible with high accuracy as illustrated in Allsopp et al. (2001). Here we investigate the application of NMR spectrometry to quantify small amounts of oil in effluent from flooding experiments. Low-field NMR spectrometry involves a series of magnetic manipulations of the hydrogen nuclei found in pore fluids. The procedure includes the alignment and realignment of the hydrogen nuclear spins by action of static and oscillating magnetic fields. After each NMR measurement, an inversion technique is used to convert the perturbation decay curve into a distribution of T2 (where 1/T2 is the transverse relaxation rate). In the case of fluid samples containing both water and oil, the T2 distribution will typically be different for water and oil. For fluids in a test vial, the T2 distribution is reflecting bulk relaxation, which is the intrinsic relaxation property of a fluid and is controlled by its physical properties, such as viscosity and chemical composition (Coates, 1999). Low field NMR spectrometry is able to estimate directly the volume of the oil and water within each sample provided the hydrogen index (HI) of each fluid is first measured. HI is a factor representing the number of hydrogen atoms per unit volume divided by the number of hydrogen atoms per unit volume of pure water.
3. Experimental procedure 3.1. Sample preparation North Sea crude oil with a density of 0.845 g/cm3 and synthetic seawater with a density of 1.019 g/cm3 (Table 1), were used to prepare the samples used in this study. The densities were measured using a digital density meter of high accuracy (7 0.001). Samples were divided into two categories: a) non-radioactive and b) radioactive. The non-radioactive (NO) and radioactive (SRO2) samples were prepared according to Table 2. The oil and the synthetic seawater were weighed in a 10 ml glass vial using an analytical balance. This set of samples was used in image analysis, UV/Visible spectroscopy, and NMR studies. First the samples were photographed for the image analysis method, then they were analyzed using low field NMR spectrometry and finally the amount of oil was determined from the UV/visible spectroscopy. The sequence of the techniques was selected carefully, so as not to destroy the samples before the last step of the analysis. Two extra sets of samples were prepared and used for the preparation of standard curves for image analysis and for UV/visible spectroscopy; NOa and NOb respectively (Figs. 3 and 4). For the preparation of the radioactive samples (SRO2), one liter of the aforementioned crude oil was mixed with 1.25 ml solution
462
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
Table 1 Seawater composition. Components
g/l
NaCl NaHCO3 KCl MgCl2 6H2O CaCl2 2H2O Na2SO4 TDS
18.01 0.17 0.75 9.15 1.91 3.41 33.39
Table 2 Non-radioactive (NO) and radioactive (SRO2) crude oil samples. Sample ID
Oil mass (g)
Oil vol (μl)
Sample ID
Oil mass (g)
Oil vol. (μl)
NO-1 NO-2 NO-3 NO-4 NO-5 NO-6 NO-7 NO-8 NO-9 NO-10
0.853 0.635 0.422 0.254 0.086 0.045 0.021 0.01 0.005 0
1007.4 750.2 498.2 300.1 102.1 53.3 24.6 11.2 5.6 0
SRO2-1 SRO2-2 SRO2-3 SRO2-4 SRO2-5 SRO2-6 SRO2-7 SRO2-8 SRO2-9 –
0.9645 0.7892 0.565 0.3076 0.1999 0.1753 0.126 0.0316 0 –
1141.1 933.6 668.4 363.9 236.5 207.4 149.1 37.4 0 –
Fig. 4. The standard curve used for the quantification of the oil within the NOsamples for the UV/visible spectroscopy method. The standard curve created from the measurement of the optical density of the N0b samples. The equations illustrated in the figure are used to quantify the oil fractions within the NO-samples. Table 3 Physical properties of Stearic acid (Lide and Milne, 19). Name: stearic acid [1-14C]
Formula: CH3 (CH2)1614COOH
Physical properties MP (°C) BP (°C) Density (g/ml)
68.8 350 0.9408
H2O: insoluble, EtOH, Benzene: slightly soluble, Acetone, Chloroform, CS2: very soluble
Fig. 3. The standard curve used for the quantification of the oil within the NOsamples for the image analysis technique. The standard curve was created from the measurement of the area of the N0a samples. The equation illustrated in the figure is used to quantify the oil fractions within the NO-samples. Samples containing less than 100 μl were excluded from the analysis since they were scattered.
of 14C stearic acid (Table 3) dissolved in toluene (7.4 MBq/ml from Perkin Elmer). The radioactive crude oil samples were placed in 20 ml polyethylene scintillation vials with screw cap. The total volume was calculated on the basis of known mass and density. A weight balance of high accuracy (7 0.0001 g) and an electrical density meter were used to measure the weight and density of the fluid respectively. The use of the scintillation vials is necessary for measuring the radioactive oil in the liquid scintillator. Another set of samples (labeled as SRO1) was prepared and used for the preparation of the standard curve for liquid scintillation counting (Fig. 5).
Fig. 5. The standard curve created from the measurement of the radioactivity of the SRO1 samples using the liquid scintillation method. The radioactivity data of these samples corresponds to the y- values and the -x- value is the volume of the radioactive oil. The equation illustrated in the figure is used to quantify the oil fractions within the SRO2-samples.
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
463
Fig. 6. (a) The area measured and (b) the detected volume of oil plotted against the actual oil volume within the NO-samples for the image analysis technique with the use of the equation in Fig. 3.
3.2. Analysis of samples 3.2.1. The image analysis method High resolution images (using a 15.1 megapixel camera) were taken of standard samples and of NO samples and images were analyzed with the help of the digital analysis software ImageJ (Schneider et al., 2012; Abramoff et al., 2004). The diameter of the vial was used to calculate the size of each pixel, and then the oil area was marked manually and calculated from the number of pixels. The standard data prepared for standardization of the image analysis method indicates a linear dependency between the calculated surface area and the predetermined volume of oil for samples containing more than 0.1 ml oil (Fig. 3). Samples containing less than 0.1 ml oil were removed from the analysis. The equation on the graph (Fig. 3) was used to determine the oil volume based on the area detected for each NO-sample (Fig. 6). 3.2.1.1. Advantages and disadvantages. The image analysis method is relatively simple, cheap and fast, as it takes no more than ten minutes to take the picture and analyze the result for each vial, it does not require much preparation and elaborate equipment. Under conditions where multiple data have to be processed it may save time and usage of expensive equipment. The method is nondestructive and has high accuracy when the oil is distributed regularly and the amount is higher than 100 μl (Fig. 6). It was observed that when the oil amount is smaller than 100 μl, the method is unreliable because of the difficulty in manual indication of the regions of low oil amounts. The method cannot properly account for the irregularly dispersed oil phase (e.g. in the form of the drops or films on the vial surfaces). 3.2.2. The UV/visible spectroscopy analysis method The NO samples and samples for the standard curve were treated using the UV/visible spectroscopy as described by Evdokimov et al., 2003a, 2003b. Approximately 3 ml of toluene were weighed and added to each NO sample. The samples were then shaken in order to mix oil and toluene, and afterwards they were stored to let the liquid separate into two phases. The upper phase containing oil dissolved in toluene was transferred into a quartz glass cuvette (path length 10 mm) and the optical density was measured using a UV–VIS Spectrophotometer (UV mini 1240,
Shimadzu) at λ ¼750 nm. If the spectrophotometer showed an absorbance above 1 (corresponding to samples with more than 0.2 ml oil), the sample was further diluted in order to get an accurate measurement. Between the measurements, the glass cuvette was rinsed with toluene and ethanol followed by drying with compressed air. Only pure grade toluene and ethanol were used in this experiment. For the standard samples, the absorbance data was then plotted against the oil volume. Two linear regression equations were obtained as standard curves (Fig. 4), which were then be used to quantify the amount of oil in the NO samples (Fig. 7). 3.2.2.1. Advantages and disadvantages. The UV/visible method is a low cost, fast (less than 5 min per sample) and overall simple method. It is reproducible and can give direct analysis of the oil-inwater samples. The method can determine the amount of oil at any condition, even when the oil is dispersed in a vial in the form of drops. The method is destructive, which is a major disadvantage. The samples cannot be used for other analysis after the solvent is added. Furthermore, the standard curve needs to be made using the same oil as the samples that are going to be analyzed. Oil volumes higher than 200 μl require dilution. 3.2.3. The liquid scintillation counting method For the LSC method, the oil was doped with low-concentration, radiolabeled stearic acid [1-14C] (Table 3), which is fully dissolvable in the oil. The labeled acid is not soluble in water and no radioactivity is transferred to the water. 14C stearic acid has a known radioactivity of 222346 dpm. For standard samples and SRO2 samples, the internal standard method was selected to anticipate quenching effects due to the chemical complexity and color of the crude oil. 15 ml of scintillation liquid Ultima Gold (Edler, 2015) was added to each of the samples. All the samples were analyzed using the scintillation counter. The energy window used for analysis was 0–156 keV, which is the energy spectrum of beta emission for 14C radionuclide. The counting time was set to 30 min. For the standard samples, the radioactivity was then plotted against the oil volume and a linear regression equation was obtained as standard curve (Fig. 5), which was then used to quantify the amount of oil in the SRO2 samples (Fig. 8). A higher
464
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
Fig. 7. (a) The optical density data and (b) the detected volume of oil plotted against the actual oil volume within the NO-samples for the UV/visible spectroscopy by using the equations in Fig. 4.
deviation between detected and actual crude oil volume is observed above 0.6 ml. It might be attributed to a quenching effect. 3.2.3.1. Advantages and disadvantages. The liquid scintillation apparatus has a tray for carrying many samples. So in the present study 40 samples can be analyzed in one batch. By liquid scintillation, low levels of radioactivity and, thus low amounts of oil can be detected. The signal is proportional to the radioactive oil amount and is easy to calibrate. Disadvantages of this method are chemical and color quenching. The standardization must compensate for this. It is especially important for oil samples that absorb light in the range of the wavelength emitted from the scintillator. The liquid scintillation analysis is destructive. The samples cannot be reused after the scintillation liquid is added. According to L’Annunziata and Kessler (2012), the internal standard method is the most accurate method to correct for quenching. The risk of radioactive fluid leaks is important.
Ingestion and skin contact with the radioactive material can be hazardous. The relatively high cost of the counting apparatus and additional expenses for the disposal of radioactive effluents have to be taken into account. Contamination surveys should be performed regularly in lab facilities. The handling of radioactive samples should be done correctly. 3.2.4. The NMR method NMR measurements were made at room temperature with a GeoSpec2 NMR Core Analyzer at 2 MHz. The T2 relaxation curves were measured by using a recycle delay (repetition time) of 40 s, number of echoes 32768, CPMG inter echo spacing (τ) 50 μs and 64 scans. The π/2 and π pulses were 9.75 μs and 19.5 μs, respectively. The T2 relaxation spectra were generated using the WinDXP (Oxford Instruments, UK) software. For the NO samples, the hydrogen nuclear spin in crude oil relaxes faster than in water, hence, the T2 distribution can be
Fig. 8. (a) The radioactivity measured and (b) the detected volume of oil plotted against the actual oil volume within the SRO2-samples for the liquid scintillation analysis by using the equations in Fig. 5.
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
465
Fig. 9. Typical T2 distribution of an oil/water sample (0.05 ml oil and 3.06 ml water).
divided into two areas (Fig. 9). The peaks with T2 less than 1 s correspond to the signal from oil, whereas the peaks with higher T2 represent the water signal. The hydrogen index for the crude oil used in the current study is different from the index for the synthetic brine. The area under the peaks may therefore be translated into the volume of fluid emitting this signal after a linear correction. A 10% reduction of all oil volumes detected from the peak area was required in order to quantify the exact volumes of oil within the samples (Fig. 10). Fig. 11 represents the quantification of the water fraction. 3.2.4.1. Advantages and disadvantages. Samples in glass vials can be used in the NMR probe without any extra preparation. The volume of each fluid does not interfere with the accuracy of the measurements. The NMR technique is not destructive to the samples. The volume of oil and water can be quantified in the presence of radioactive materials. The NMR measurement is safe and poses no risk to health. During NMR, the sample is not exposed to open air or other possible sources of contamination. This is an important point which gives the opportunity to measure any internal sample change resulting from for example storage. A drawback of this technique is related to the sample containers. The method can only be used in, “low-field NMR friendly” materials. Some plastics can create noise during the measurements and reduce the Signal-to-Noise ratio (SNR). In this work we choose to enclose our fluids in glass containers. The NMR probe used in this work has a limit on the volume of the sample. Each
Fig. 11. The water volume detected from the low field NMR spectrometry.
vial has to contain fluid that corresponds to less than five centimeters height. Above that height the protons cannot be magnetized. Currently there is no automatic sample changer for the apparatus, so the method is time consuming. Analysis of one sample takes 40 min, because it has to equilibrate with the apparatus temperature before data recording.
4. Discussion We were able to obtain a good correlation of the amount of oil added in the glass containers and the amount of oil detected from all four techniques as reflected in high correlation coefficient (R) and low standard deviation (s) for the analysis of 10 samples (n) (Table 4). Advantages and drawbacks of each method are summarized in Table 5.
Fig. 10. (a) The volume detected and (b) the corrected volume of oil plotted against the actual oil volume within the NO-samples for the low field NMR spectrometry. Fig. 10b was created after applying a 10% volume correction to all data in Fig. 10a.
466
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
Table 4 The correlation coefficient (R) and the standard deviation (s) of the samples (n) analyzed in each method. The errors of each measured value are based on the reproducibility of each technique between the standard curve and the final measurements. Method
n
R2
s (μl)
Error (μl)
Image analysis UV/visible spectroscopy Liquid scintilation analysis Low field NMR
10 10 10 9
0.96 0.99 0.92 0.99
599 105 378 56
117 47 142 –
The errors for the image analysis technique were calculated based on the performance of this technique on the entire range of oil volumes; including samples with less than 0.1 ml of oil. Low field NMR and UV/visible spectroscopy were able to detect the oil of the samples within a wide range of oil volumes; from a few microliters to several milliliters. Low field NMR spectrometry was shown to be the most precise in respect to accurate quantification of the oil fractions, but UV/visible spectroscopy produced almost similar accuracy throughout the measurements, as illustrated by correlation coefficients and standard deviations (Table 4). On the other hand, the UV/visible spectroscopy is a user dependent technique that requires manual operation to add toluene to the samples and the use of standard curves which might introduce errors to the results. The standard curves for this technique must be produced with the same oil sample that is going to be analyzed. If the oil volume is higher than 0.2 ml the sample should be diluted several times and an error from the extensive manual dilution might be introduced. Finally, low field NMR provides an accurate determination of the water fraction within each sample, giving the opportunity to quantify the amount and the type of fluids within each sample. A drawback of the NMR method is that the entire volume of the oil should be within 5 cm height. Similarly to the two abovementioned techniques, liquid scintillation counting resulted in a high correlation coefficient and low residuals in the wide range of oil samples (Table 4). During the analysis, the residuals proved to be higher at bigger amounts of oil, proving this technique more accurate at smaller amounts of oil. Finally, the image analysis method showed a correlation between the added and detected amount of oil with a correlation coefficient equal to 0.99 for oil volumes above 100 μl (Fig. 6). As the amount of oil decreases within the sample, this technique provides a coefficient of less than 0.97 for the entire range of volumes and the level of residuals increases unless only samples containing oil volume higher than 0.1 ml are used (Table 4). A drawback in the image method is its dependency on oil distribution within a vial. Unlike other methods, the visual method cannot
be applied in more complicated cases such as shown in Fig. 2b or in cases that emulsification phenomena occur.
5. Conclusions We may summarize the applicability of the four different methods in regard to: a) accuracy of the detected amount of oil, b) experimental time, c) technical work power, d) post destruction of the samples, e) money, and f) final recommendation. a) With respect to accuracy, low field NMR spectrometry and UV/ visible spectroscopy proved to be able to detect the oil fraction in the entire range of oil volume. The same accuracy was achieved with the liquid scintillation technique for samples with less than 0.6 ml oil. Finally, the image analysis is only suitable for larger amounts of oil, above 0.1 ml. b) With respect to experimental time, low field NMR spectrometry can only test one sample at a time, which might be very time-consuming, especially in the case of water flooding experiments with hundreds of samples per flooding. Each measurement requires 40 min and the overall time required for analysis is unrealistic. It can be used in a small selection of samples to control the results obtained by other techniques or in cases where the studied vials do not exceed a couple of tens. UV/visible spectroscopy requires 5 min for each analysis and an additional 5 min for dilution when necessary. The liquid scintillation analysis requires 30 min for the quantification of the oil in each sample, but the tests can be performed automatically with batches of 40 samples per run. The image analysis technique is able to detect the oil volume in one sample at a time and each analysis can be done in 10 min. When using image analysis, it is an advantage to use the same type of vials for fraction collectors throughout the core flooding experiments. This can save time for the user. But even after the proper selection of vials, the amount of time spent for the detection of oil with the image analysis in hundreds of vials coming out of the fraction collector is unrealistic. c) With respect to technical work power, low-field NMR spectrometry is a very time consuming technique but does not require technical work power during equilibration time. UV/ visible spectroscopy is relatively fast but requires constant presence of a technician. The main advantage of liquid scintillation counting is that the measurements can be performed overnight and save valuable time and technical work power. Image analysis requires technical work power constantly. d) With respect to post destruction, out of the four methods, NMR spectrometry and image analysis proved not to interfere with the sample at all, since both measurements are
Table 5 Advantages and drawbacks of image analysis, UV/visible spectroscopy, liquid scintillation counting and low-field NMR spectrometry. Parameters
Image analysis
UV/visible spectroscopy Liquid scintillation counting
Low-field NMR spectrometry
Time consumption: Destructive: Limitations in volume:
Low No Volumes higher than 0.1 ml
Low Yes No
Low Yes No
Limitations in concentration: No
Yes
User dependent:
Yes
Yes
Preparation: Standard curves: Conditions of sample:
Photos Yes Cannot measure oil drops and dispersed oil No
Solvents Yes No
High amounts of oil (color quenching effects). Internal standard method (labor intensive). Use of doped oil. Yes Mixing with scintillation liquid
High No No – within probe detection range No
No No No
No
No
Yes
Measures water fraction:
No
K. Katika et al. / Journal of Petroleum Science and Engineering 147 (2016) 459–467
conducted in closed vials. Liquid scintillation counting and UV/ visible spectroscopy require additives to each sample for analysis; therefore, the samples cannot be used further. When combining these techniques with image analysis or NMR, the destructive method should be applied last. e) In terms of saving money, image analysis has proven to be the cheapest of all techniques since it only requires photos of each sample and software for the analysis. Low field NMR requires the presence of a low field NMR spectrometer which is a relatively expensive piece of equipment designed for laboratory rock core analysis in the petroleum industry. UV/visible spectroscopy can be performed in a spectrophotometer that fulfills the criteria described above. Finally, liquid scintillation counting can be performed in laboratories certified to handle radioactive tracers. f) Final recommendation: image analysis is a relatively accurate technique when it comes to quantifying the oil fractions in volumes higher than a 0.1 ml, whereas liquid scintillation counting gives high accuracy for amounts of oil less than 0.6 ml. There is clearly a benefit of combining these two techniques to save time and increase the accuracy of the overall measurements. If a single method should be chosen, UV/visible spectroscopy is the most trustworthy. If all measurements are conducted by the same user, any errors in respect to user and pipette dependency can be diminished. Low field NMR and image analysis can be used in addition to this method, in order to certify the results.
Acknowledgments The research has been carried out in the framework of the BioRec and SmartWater projects supported by the Danish Energy Agency (EUDP-10-II 64011-0009), Innovation Fund Denmark, Mærsk Oil AS, DONG Energy AS, and Novozymes AS.
References Abramoff, M.D., Magalhaes, P.J., Ram, S.J., 2004. Image Processing with ImageJ. Biophotonics Int. 11 (7), 36–42. Allsopp, K., Wright, I., Lastockin, D., Mirotchnik, K., Kantzas, A., 2001. Determination of oil and water compositions of oil/water emulsions using low field NMR relaxometry. J. Can. Pet. Technol. 40 (7), 58–61. Bastow, T., Durnie, W.H., Jefferson, A., Pang, J., 1997. Ultraviolet spectroscopy for the analysis of oil-in-water effluent using isopropanol as co-solvent. Appl. Spectrosc. 51 (3), 318–322. Berg, S., Cense, A.W., Jansen, E., Bakker, K., 2010. Direct experimental evidence of wettability modification by low salinity. Petrophysics 51 (5), 314.
467
Butler, J.B., Lane, S.N., Chandler, J.H., 2001. Automatic extraction of grain-size data from gravel surfaces using digital image processing. Am. Soc. Civ. Eng. J. Hydraul. Div. 39, 519–529. Coates, G.R., Xiao, L., Prammer, M.G., 1999. NMR logging: principles and applications. S.l.: Gulf Professional Publishing. Dunn, K., Bergman, D.J., LaTorraca, G.A., 2002, Nuclear magnetic resonance-petrophysical and logging applications. In: Handbook of Geophysical Exploration: Seismic Exploration, Elsevier Science Ltd., Pergamon Press, New York. Dijs, I.J., Windt, E., Kaihola, L., Borg, K., 2006. Quantitative determination by 14C analysis of the biological component in fuels. Radiocarbon 48 (3), 315–323. Dugstad, O., Opel, K., Fjelde, I., 2011, Improved Understanding of CO2/Foam EOR Techniques by AID of Tracer Technology. In: Proceedings of the Society of Petroleum Engineers, SPE EUROPEC/EAGE Annual Conference and Exhibition, Vienna, Austria, 23–26 May. Edler, R., 2015. Cocktails for Liquid Scintillation Counting – Application Note. PerkinElmer, Inc. Evdokimov, I.N., Eliseev, N.Y., Akhmetov, B.R., 2003a. Assembly of asphaltene molecular aggregates as studied by near-UV/visible spectroscopy: I. Structure of the absorbance spectrum. J. Pet. Sci. Eng. 37 (3–4), 135–143. Evdokimov, I.N., Eliseev, N.Y., Akhmetov, B.R., 2003b. Assembly of asphaltene molecular aggregates as studied by near-UV/visible spectroscopy: II. Concentration dependencies of absorptivities. J. Pet. Sci. Eng. 37 (3–4), 145–152. Gupta, R., Glotzbach, R., Hehmeyer, O., 2015. A novel field-representative enhancedoil-recovery coreflood method. SPE J., 21–34. Kanaly, R., Bartha, R., Fogel, S., Findlay, M., 1997. Biodegradation of [14C]Benzo[a] pyrene added in crude oil to uncontaminated soil. Appl. Environ. Microbiol. 63 (11), 4511–4515. Kenyon, W.E., 1997. Petrophysical principles of applications of NMR logging. Log Anal. 38, 21–43. L'Annunziata, F.M., Kessler, J.M., 2012. Liquid scintillation analysis: principles and practice. In: L’Annunziata, F.M. (Ed.), Handbook of Radioactivity Analysis, third ed. Academic Press, Amsterdam, pp. 423–573, Chapter 7. Lieser, K.H., 2001. Technical and Industrial Applications of Radionuclides and Nuclear Radiation, in: Nuclear and Radiochemistry: Fundamentals and Applications, second revised ed. Wiley-VCH Verlag GmbH, Weinheim, Germany, p. 386. Majid, A.A., Saidian, M., Prasad, M., Koh, C.A., 2015. Measurement of the water droplet size in water-in-oil emulsions using low field nuclear magnetic resonance for gas hydrate slurry application. Can. J. Chem. 93, 1–7. Pal, N.R., Pal, S.K., 1993. A review on image segmentation techniques. Pattern Recognit. 26 (9), 1274–1294. Rogers, W.A., Moran, F.J., 1966. Evaluation of quench correction in liquid scintillation counting by internal automatic external, and channels' ratio standardization methods, vol. 16(2), pp. 206–219. Schneider, C.A., Rasband, W.S., Eliceiri, K.W., 2012. NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671–675. Stoll, W.M., Al Shureqi, H., Finol, J., Al-Harthy Said, A.A., Oyemade Nneamaka, S., De Kruijf, A., Van Wunnik, J., Arkesteijn, F., Bouwmeester, R., Faber, M.J., 2011. Alkaline/surfactant/polymer flood: from the laboratory to the field. SPE Reserv. Eval. Eng. 14 (6), 702–712. Tang, G.Q., Morrow, N.R., 1999. Influence of brine composition and fines migration on crude oil/brine/rock interactions and oil recovery. J. Pet. Sci. Eng. 24, 99–111. Wheeler, V.J., Parsons, T.V., Conchie, S.J., 1985. The application of radioactive tracers to oil reservoir waterflood studies. In: Proceedings of the SPE 13985/1 Presented at SPE/AIME Offshore European 85 Conference, Aberdeen, Scotland, March 10–13. Whitby, C.P., Lotte, L., Lang, C., 2012. Structure of concentrated oil-in-water Pickering emulsions. Soft Matter 8 (30), 7784–7789. Yang, M., 2011. Measurement of oil in produced water. In: Lee, K., Neff, J. (Eds.), Produced Water. Springer, New York, pp. 74–75.