Article pubs.acs.org/EF
Deep-Vacuum Fractionation of Heavy Oil and Bitumen, Part I: Apparatus and Standardized Procedure O. Castellanos Díaz,† M. C. Sánchez-Lemus,† F. F. Schoeggl,† M. A. Satyro,‡ S. D. Taylor,§ and H. W. Yarranton*,† †
Department of Chemical and Petroleum Engineering, University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada T2N 1N4 ‡ Virtual Materials Group, #222, 1829 Ranchlands Boulevard NW, Calgary, Alberta, Canada T3G 2A7 § DBR Technology Center, Schlumberger Canada Limited, 9450 17th Avenue NW, Edmonton, Alberta, Canada T6N 1M9 ABSTRACT: Distillation assays provide the most widely used characterization data for modeling crude oils in refinery processes but can only define a relatively small fraction of the boiling curve for heavy oils. One way to extend the distillation range for heavy oils is to lower the distillation pressure while still avoiding thermal cracking temperatures. A vacuum distillation apparatus (DVFA-I) that was previously used to measure the vapor pressure of heavy components was modified to fractionate heavy oils and bitumens. The modified apparatus (DVFA-II) is a batch distillation system without reflux operating at pressures down to approximately 0.01 Pa, compared with 100 Pa for a conventional vacuum distillation. With the standard procedure developed for DVFA-II, up to 50 wt % of a Western Canadian bitumen was distilled into eight cuts plus a residue. This is an improvement over the 26 wt % distilled by a spinning band vacuum apparatus. The cumulative weight percent distilled was repeatable to within 2.3% for a given boiling point range. The densities and molecular weights of the cuts from two fractionations increased monotonically versus the weight percent distilled and were repeatable to within 0.18 and 4.0%, respectively.
■
fluid characterization. Other methods are sometimes employed to supplement the characterization, including refractive index, nuclear magnetic resonance, and infrared spectroscopy.4,7,8 However, distillation is the preferred choice in the refining industry because it is practical and economical and because it provides extensive information from a single assay. Figure 1 shows a distillation curve constructed on the basis of distillation data and SARA analysis. To generate pseudocomponents, the curve is divided into boiling ranges, each with its corresponding mass fraction.
INTRODUCTION The design, optimization, and operation of processes including petroleum fluids require the characterization of the oil for modeling purposes. To predict the physico-chemical properties and phase behavior of the oil in a reliable manner, a detailed knowledge of its composition is essential. However, the task of identifying every one of the millions of components that exist in a crude oil is impractical and virtually impossible. Although considerable effort has been made to identify as many components as possible in an oil sample,1 the most common representation of crude oils is to lump groups of similar components together as pseudocomponents, each of which represents a narrow range of properties. Computational effort and the limitations of crude oil assays restrict the actual number of pseudocomponents. Typically, between five and 20 pseudocomponents are considered sufficient to properly define a crude oil.2−5 Traditionally, pseudocomponents are defined on the basis of specific gravity, normal boiling point (TBP or GC assays), and average molecular weight.6 Additional data such as PNA (paraffins, naphthenes, and aromatics) composition, SARA (saturates, aromatics, resins, and asphaltenes) composition, density, viscosity, heat capacity, and/or vapor pressure are used to further define the fractions. For each pseudocomponent, physical property correlations are used to predict the thermodynamic properties and critical constants (particularly when modeling the oil with an equation of state). Mixing rules are then used to obtain the properties of any combination of the pseudocomponents, such as the whole fluid, any phases formed from the fluid, and distillation cuts. Distillation, chromatography, and chemical-class-based separation are the most commonly used assays for experimental © 2014 American Chemical Society
Figure 1. Normal boiling point curve of a heavy oil constructed from distillation data (ASTM D-1160), a Gaussian extrapolation for the maltene boiling points, and asphaltene boiling points calculated from a Γ distribution of molecular weights. Received: February 28, 2014 Revised: April 22, 2014 Published: April 22, 2014 2857
dx.doi.org/10.1021/ef500489y | Energy Fuels 2014, 28, 2857−2865
Energy & Fuels
Article
cuts provide a basis for a thermodynamically grounded interconversion method. In Part 1 of the study, a standardized procedure for the deep-vacuum fractionation is presented with repeatability tests. The pressure-reduced distillation curve of a Canadian bitumen is presented and compared with the results of distillation assays for the same sample obtained with commercial vacuum distillation systems. The data are modeled with an equation of state to assess the consistency among the different methods. The molecular weight and density of the distillation cuts are measured and compared with expected trends. Part 2 of the study (DOI: 10.1021/ef500490h) provides an interconversion method to obtain atmospheric-equivalent boiling points from the data collected under vacuum conditions.
Laboratory distillation assays are limited to temperatures below approximately 573 K to avoid thermal cracking. For conventional oils, atmospheric and vacuum distillation (TBP, ADC, ASTM D-86, D-1160, and D-2892) provide data to characterize 80−95 wt % of the oil. However, only a relatively small fraction of heavy oil or bitumen is distillable using these methods, as indicated in Figure 1. To extend the range of the fractionation, high-vacuum conditions are necessary. Figure 2 shows hypothetical distillation curves (boiling point curves) of heavy oil under three different pressure conditions:
■
EXPERIMENTAL METHODS
The materials and experimental methods are described below, except for the deep-vacuum fractionation, which is discussed in detail later. Materials. A Western Canadian bitumen sample (WC_Bit_B1) was utilized in this work; selected physical properties are provided in Table 1. The SBD assay is reported in Table 2.
Table 1. Selected Physical Properties of the Bitumen Sample (WC_Bit_B1) Figure 2. Conceptual distillation performance on a heavy oil or bitumen at different operating pressures.
atmospheric, vacuum, and high vacuum. All of the distillation curves stop at the onset of thermal cracking (approximately 573 K), since beyond this point the substance is chemically transformed and the collected distillation data are no longer related to the original fluid composition. With this limitation, atmospheric distillation methods can fractionate approximately 9−10% of a heavy oil or bitumen. If the pressure of the system is reduced, the boiling points of the oil constituents are reduced as well, and a greater amount of sample can be fractionated before the cracking temperature is reached. For example, commercial vacuum distillation assays such as ASTM D1160 and the spinning band distillation (SBD) technique, which operate at pressures close to 132 Pa (1 mmHg),4 can fractionate 20−30 wt % of a heavy oil or bitumen. The vacuum potstill method (ASTM D5236) is a high-vacuum method to distill heavy hydrocarbon mixtures and can reach pressures as low as 13.3 Pa. The conversion of the actual distillation temperature at subatmospheric pressure to atmospheric-equivalent temperature (AET) is done by using the Maxwell and Bonnell equations, which were developed for conventional oils. Techniques such as short-path distillation can generate high-boiling-point fractions with end points as high as 993 K. However, boiling points cannot be determined during the distillation because this technique is a nonequilibrium process, and therefore, post analysis of cuts by simulated distillation is required.10,11 While ASTM D-5236 can extend fractionation to higher atmospheric-equivalent boiling temperatures, there is still no standard method that encompasses the complete boiling range for heavy oils.9 The objective of this study was to develop a new apparatus and methodology to generate multiple cuts of heavy oil and bitumen samples using deep-vacuum conditions at temperatures ranging from 370 to 573 K and pressures below 10 Pa. Originally developed as a vapor pressure measurement device,14 this apparatus has been adapted for a second use as a deepvacuum distillation unit. Vapor pressure measurements on the
property
value
average molecular weight [g/mol] average specific gravity initial boiling point [K] asphaltene/solids content [wt %] wt % distilled using SBD
510 1.007 486 17 26
Table 2. Spinning Band Distillation (SBD) Assay of WC_Bit_B1 normal boiling temperature [K]
wt % distilled
511 526 541 552 563 575 587 597 613 623 631 640 648 653
3.2 4.9 6.5 8.1 9.7 11.4 13 14.6 17.9 19.5 21.1 22.7 24.3 26
Molecular Weight Measurements. Molecular weights of the bitumen and its distillation cuts were measured in toluene at 323 K using a Jupiter model 833 vapor pressure osmometer calibrated with sucrose octaacetate. Briefly, the osmometer measures the voltage difference between a thermistor contacted with solvent and a thermistor contacted with a solution of the solute of interest in the same solvent. For each sample, a series of solutions were prepared at concentrations of 0.001, 0.002, 0.005, 0.010, and 0.020 kg/L, and the voltage difference was measured for each. The voltage difference/ concentration ratios were extrapolated versus solute concentration, and the y-intercept was converted to molecular weight using a calibration constant. More details of the theory and procedure are provided elsewhere.12 The repeatability of the measurement is ±15%. 2858
dx.doi.org/10.1021/ef500489y | Energy Fuels 2014, 28, 2857−2865
Energy & Fuels
Article
a total volume of 20 cm3. Figure 3 shows a schematic of the originally developed apparatus.
Density Measurements. The densities of bitumen and the associated distillation cuts were measured with an Anton Paar MDA 46 density meter at 293 K. Since the samples were very viscous, the density was determined indirectly from solutions of the sample in toluene at concentrations from 0.001 to 0.02 kg/L. The density of the sample was determined assuming there was no excess volume of mixing.13 The repeatability of a density measurement is ±0.5 kg/m3 but the repeatability of the extrapolated density is ±7 kg/m3. Elemental Analyses. The elemental compositions of the boiling fractions were determined via combustion using a Carlo Erba EA 1108 elemental analyzer that simultaneously determines the total carbon, nitrogen, hydrogen, and sulfur. Standards fell within 0.3% of the theoretical values. Oxygen was detected using a different instrument configuration. Measurements were performed by the Chemistry Department at the University of Alberta. Spinning Band Distillation. Distillation data were measured using SBD. A 0.1 kg sample of dewatered bitumen was placed in a boiling flask, which was continuously heated until both liquid and vapor flows were observed, as indicated by one drop of liquid per second exiting the condenser. A period of 40 min was required to equilibrate the liquid and vapor flows within the system. Then the reflux valve was opened for a reflux ratio of 5:1. The temperatures of the vapor and boiling liquid sample were measured, and the AET of the vapor was calculated using the Maxwell−Bonnell interconversion method for reduced distillations.4 The desired cut temperature was set for each cut, and up to four cuts of the initial bitumen were collected in the receivers. Once a temperature of 573 K was reached in the boiling flask, the distillation was stopped, and mass balances were performed to determine the amount of bitumen distilled. The operating pressure in the SBD was 390 Pa (3 mmHg), and 573 K was set as the maximum operating temperature of the apparatus to ensure that the cracking temperature of the sample was not exceeded. Maltene Preparation and Asphaltene Content. The deepvacuum fractionation was performed on maltenes obtained by deasphalting the bitumen sample using n-pentane. A known amount of bitumen was mixed with a solvent at a ratio of 40 mL of solvent per gram of bitumen. The mixture was sonicated for 60 min and left to settle for 24 h. Then the supernatant was decanted and filtered through a 24 cm Whatman filter. Solvent was added to the sediment at an amount equivalent to 10 vol % of the initial solvent added. The mixture was sonicated for 45 min and left to settle. After 16 h of settling, the solution was filtered through the same filter used previously. The filter cake was washed with n-pentane until the liquid leaving the filter was colorless and then dried for 8 days at 333 K. The filter cake includes asphaltenes and any coprecipitated material and is here termed asphaltene/solids. The asphaltene/solids content is the mass of the filter cake divided by the mass of bitumen. Finally, all of the filtrate from both filtrations was placed in a rotary evaporator, and the solvent was evaporated to recover the maltenes. It should be noted that the bitumen used in these tests consisted primarily of components with carbon numbers above 9. Nonetheless, the deasphalting procedure may strip some light components from the heavy oil. The procedure can also leave residual n-pentane in the maltenes. These issues will be discussed later.
Figure 3. Simplified schematic of the deep-vacuum fractionation apparatus DVFA-I.
To fractionate a bitumen sample using DVFA-I, the sample temperature was raised to the initial condition at T1. The initial temperature of fractionation is the temperature at which a detectable amount of condensate is visually observed in the cold finger. To determine this initial condition, the sample was heated gradually and the cold finger was observed; a value of T1 = 423 K was found to be optimal for the WC_Bit_B1 sample. At this point, the sample was left open to pump suction by leaving all the connecting valves open. The upper section of DVFA-I, above the sample vessel, was left at a temperature 20− 30 K higher than T1 to facilitate vapor transport and avoid condensation in the inner pipe. The cut was collected by condensation in the cold finger, which was replaced when the fractionation process at T1 stopped. To determine when a complete sample was taken at a given temperature, the volume of the cut collected as a function of time was measured. If the volume of the cut did not change during 72 h, the sample was considered to be complete. Then the sample vessel was isolated and the temperature was raised to a value T2 higher than T1. Once the temperature was constant to within ±0.1 K, the fractionation process was repeated. A set of cuts was collected at temperatures between 423 and 563 K. It should be noted that the pressure of the distillation could not be measured because of the temperature limitations of the gauges. The deep-vacuum fractionations were performed on maltenes instead of the whole bitumen. Asphaltenes are known for their molecular complexity, high viscosity, and high hypothetical boiling points. Removing the asphaltenes reduces the sample viscosity and thereby enhances the mass transfer, decreases the distillation time, improves the separation, and possibly increases the amount of oil that can be distilled. Two fractionations were performed on WC_Bit_B1 in the proof-of-concept test; the results are listed in Table 3. Figure 4 shows a photograph of the five cuts collected for fractionation 1. As the boiling point of the cut increased, the color of the cut changed from light, transparent amber to dark greenish-brown, ending in an opaque black for the residue. The measured densities (Table 3) of the cuts increased from the first to the last cut. We also visually observed slower flow, consistent with higher viscosity, in going from the first to last cut. These
■
DEEP-VACUUM FRACTIONATION: PROOF OF CONCEPT Castellanos-Diaz et al.14 introduced a static apparatus for the measurement of vapor pressures of heavy oils and bitumen. The same apparatus, essentially operated as a batch distillation with no reflux and one theoretical stage, was used in the present work to prove the concept of deep-vacuum fractionation. This deep-vacuum fractionation apparatus, herein denoted as DVFAI, consists of a series of stainless steel Swagelok fittings connected with copper O-rings, a liquid-trap section (cold finger), a turbomolecular pump, and two thermocouples connected to a temperature controller. The sample vessel has 2859
dx.doi.org/10.1021/ef500489y | Energy Fuels 2014, 28, 2857−2865
Energy & Fuels
Article
doubled the distillable portion of the bitumen. The repeatability of the fractionation was estimated by fitting the data from fractionation 1 and then determining the deviation of the fractionation 2 data from the fitted curve (Figure 5). The average deviation of the distillation data was ±4.3 cumulative wt %.
Table 3. WC_Bit_B1 Maltene Cuts Obtained Using DVFA-I trial
cut
T [K]
fractionation 1
1 2 3 4 5 residue 1 2 3 4 5 6 7 8 residue
423 463 493 533 563 >563 423 443 463 483 503 523 543 573 >573
fractionation 2
cumulative wt % of bitumen distilled
density [kg/m3]
10.4 22.8 29.7 46.4 56.9
933.9 952.0 982.3 998.6 1018.5 1055.1 918.1 958.1 959.1 999.1 1001.5 1005.0 1016.3 1013.1 1050
5.51 17.72 27.79 34.53 41.29 45.43 50.00 56.08
■
STANDARDIZED DEEP-VACUUM FRACTIONATION PROCEDURE The results of the proof-of-concept tests were encouraging. However, DVFA-I posed significant challenges such as a very long experimental time per cut, potential sample entrainment, and insufficient amounts of cut sample collected for further analysis. To overcome these challenges and standardize the DVFA, a second-stage apparatus, DVFA-II, was developed utilizing the same basis of design as DVFA-I. Modification of the Original Deep-Vacuum Apparatus. Originally DVFA-I had an initial sample volume of 20 cm3 with a 5 cm3 volume capacity for condensate collection. As eight cuts were desired for the oil characterization and at least 7 cm3 of each cut was required for property measurements, the volume of the initial configuration was insufficient. In addition, because of the small diameter of the sample vessel and high viscosity of the fluid, many days were required for each cut to diffuse through the deasphalted heavy oil, making the complete procedure very time-consuming (2 months per fractionation). Simply increasing the sample volume to obtain larger cuts would increase the height of the fluid and the corresponding mass transfer resistance, hence increasing the distillation time per cut. Another issue was the proximity of the sample vessel to the collector device and the absence of a physical barrier to avoid entrainment. Coloring of the light cuts was observed (Figure 4), indicating overlapping and splashing of the sample. An important constraint was that additional fittings and seals would increase the apparatus leak rate, which would increase the pressure of the system and therefore decrease the ability of the DVFA to extend the distillation of heavy oils. To solve these issues without adding significantly more fittings, the original cylindrical sample vessel was replaced with a tee fitting and two additional concentric reducers. The new configuration uses Swagelok fittings and the same metal gasket seals to minimize leaks. It provides a larger sample volume (110 cm3) to obtain the desired cut volumes as well as a larger sample surface area in order to reduce the mass transfer resistance and therefore produce a corresponding increase in the mass transfer rate. The reducers above the tee-shaped sample vessel are required to couple the tee diameter with the diameter of valve 1. Because of their conical shape, the reducers also act as a barrier for the entrained liquid when vapor is produced during the distillation (Figure 6). When assembled and in the absence of fluid, the system is able to achieve pressures as low as 1 × 10−7 Pa. Figure 7 presents a comparison of the times required to collect the same volume of a cut with a boiling range between 443 and 463 K from the same initial amount of oil (0.11 kg) using the original and modified sample vessels. With this change, the total time for the distillation of WC_Bit_B1 was halved. Standardization of Fractionation Using DVFA-II. The next step was to define an initial sample volume, cut-point temperatures, and number of cuts as part of a standard procedure. The finalized distillation procedure can be divided
Figure 4. Photograph of the five different cuts and the residue of WC_Bit_B1 maltenes from the fractionation 1 of the DVFA-I proofof-concept test.
changes in physical properties indicate that a qualitatively successful fractionation was achieved. The data from Table 3 are plotted as a boiling curve (at the apparatus pressure) in Figure 5. The cumulative weight percent of the maltenes distilled is 67−68 wt % (55−56 wt % based on bitumen). As shown in Table 1, SBD fractionated 26 wt % of the bitumen. Hence, the deep-vacuum method approximately
Figure 5. Boiling point profiles for the WC_Bit_B1 bitumen sample at the DVFA-I apparatus pressure. Symbols are data, and the line is a fit to the fractionation 1 data. 2860
dx.doi.org/10.1021/ef500489y | Energy Fuels 2014, 28, 2857−2865
Energy & Fuels
Article
All of the fittings after valve 1 are set 30 K higher than the sample temperature in order to avoid condensation in the pipes and enhance the mobility of the produced vapor toward the collector device. Once the DVFA-II apparatus is reassembled, valves 1 and 2 are both opened and the turbomolecular pump is switched on. Degassing of the n-pentane from the oil sample is then achieved by holding the sample vessel at 323 K for 6 h. After this time, the cold finger is replaced once again with a new, preweighed cold finger. The sample vessel is disconnected and weighed to determine the remaining mass. The temperature of 323 K was selected to avoid losses of light ends from the oil but was sufficient to remove the residual n-pentane from the deasphalting procedure used to prepare the maltenes. At the end of the degassing process, approximately 0.01−0.22 wt % of the initial maltenes were collected in the cold finger. Gas chromatography (GC) assays confirmed that this distilled material was almost entirely n-pentane. GC assays of the subsequent lightest cuts confirmed that all of the pentane had been removed. Potential losses of light ends will be discussed later. Fractionation. Following degassing of the sample, the sample vessel is reconnected, and fractionation is begun by increasing the temperature at a rate of 10 K/h until the desired cut temperature is achieved. The temperature is then held at the cut temperature until the volume in the cold finger does not change for 24 h. Cuts are collected at temperature intervals of 20 K. For heavy oils, the first cut occurs at approximately 403 K. The fractionation produces eight cuts plus a residue. It should be noted that 20 K was selected as the temperature interval between cuts since it balanced the required sample volume for each cut with a reasonable number of cuts to use in the oil characterization. Also, two or three of the lightest cuts had to overlap with the region distillable by SBD in order to validate the method. The following three steps are repeated for every cut collected. (i) set the cut temperature; (ii) run the distillation for the cut until the volume in the cold finger does not change for 24 h; (iii) close the valves, turn off the pump, disconnect the cold finger, weigh it, and replace it with a new, preweighed cold finger. The maximum cut temperature is 563 K, thus avoiding thermal cracking of the sample. At the end of the distillation, the pump is switched off, all of the fittings are disassembled, and the mass of every fitting is recorded. Any change in the mass of a fitting is likely caused by condensation of the heavier cuts and is used in the determination of losses during the procedure, as will be discussed later. The cleaning procedure is repeated, and the collected cuts are refrigerated and stored in glass vials with nitrogen caps.
Figure 6. Simplified schematic of the modified deep-vacuum fractionation apparatus DVFA-II.
Figure 7. Effect of the modified sample vessel design on the distillation time for the 443−463 K cut.
in three general steps: preparation of the apparatus, degassing, and fractionation. Each step is discussed below. Preparation of the Apparatus. Before the start of the fractionation, all of the fittings are washed with toluene and left in a vacuum oven for 24 h at 373 K and 8000 Pa. After they are cleaned, all of the parts are weighed individually and then assembled together and connected to the pump as shown in Figure 6. The empty DVFA-II is baked out at 573 K for 1 day to avoid contamination of the sample and cuts collected; contaminants are collected in the cold finger. The main purpose is to clean the surfaces of the pipes by removing remaining amounts of contaminants and water vapor by desorption. Degassing. Degassing of the sample is begun by disconnecting the sample vessel, weighing it once more to check for any major change in mass due to desorption of contaminants, and then carefully adding 0.11 kg of deasphalted bitumen into the sample vessel. The sample vessel is then reconnected to DVFA-II, and two thermocouples are placed on its outside wall. Next, the cold finger used to collect the contaminants is removed and replaced with a new, preweighed cold finger. Heating and insulation tape is then uniformly wrapped over the entire DVFA-II apparatus except for the cold finger. This provides for temperature control for all of the noncold-finger parts of DVFA-II. A bath of oil and dry ice is placed in the cold finger section to maintain its temperature at 253 K.
■
RESULTS AND VALIDATION OF THE STANDARDIZED PROCEDURE Two fractionations were performed for the WC_Bit_B1 sample following the standardized procedure. Table 4 presents the distillation data collected for the maltenes as well as the density and molecular weight of each cut. The repeatability of the DVFA-II fractionations was within an average deviation of ±1.5 wt % (Figure 8). The distillation curves were fitted using cubic splines to avoid discontinuities for the sole purpose of 2861
dx.doi.org/10.1021/ef500489y | Energy Fuels 2014, 28, 2857−2865
Energy & Fuels
Article
Entrainment. One of the major issues with the initial apparatus was significant entrainment of liquid droplets in the boiling vapor. To assess the impact of the modified apparatus on entrainment, the distillation curves from the standardized fractionations are compared with the proof-of-concept fractionations in Figure 8. It should be noted that the amount of bitumen distilled in the standardized procedure is 7 wt % less than that obtained in the proof-of-concept apparatus. While a lower recovery does not sound like an improvement, it is in fact evidence of a cleaner separation and a reduction in entrainment. Consider the early stage of the distillation. Figure 9 is a photograph of the cuts of the WC_Bit_B1 maltene cuts from fractionation 4. Upon comparison of Figures 4 and 9, it is possible to observe a clearer color for the first cuts in Figure 9, where the standardized procedure was followed. Also, the difference between the weight percent distilled for the first cuts in fractionations 1 and 2 in the proof-of-concept tests was 4.9 wt %, whereas the difference between the first cuts in fractionations 3 and 4 with the standardized procedure was only 2.1 wt %. The progressive changes in color and viscosity shown in Figure 9 were similar to those in the initial fractionations, but there are clearly sharper distinctions between the colors of the cuts, confirming that entrainment has been reduced or eliminated. Finally, the GC assays of the initial cuts from fractionation 4 are shown in Figure 10. The cuts show a nearly
Table 4. Distillation Data for WC_Bit_B1 Obtained Using DVFA-II
trial
cut
T [K]
fractionation 3
0 1 2 3 4 5 6 7 residue 0 1 2 3 4 5 6 7 residue
403 423 443 463 483 513 533 563 >563 403 430 450 468 493 513 533 563 >563
fractionation 4
cumulative wt % of bitumen distilled 3.7 15.6 18.0 25.7 32.1 38.4 44.3 49.5 5.8 16.3 21.4 28.8 35.0 40.5 45.9 51.7
molecular weight [g/mol]
density [kg/m3]
209 242 283 293 331 380 460 496 1023 213 247 272 327 352 362 480 490 1010
890.1 920.4 954.8 970.9 977.9 987.5 998.5 1001.1 1028.5 893.3 920.6 961.9 973.2 982.2 992.0 999.6 1016.8 1035.0
Figure 10. Carbon number distributions for the first two boiling cuts from fractionation 4 and the whole WC_Bit_B1 bitumen (free of C30+ compounds).
Figure 8. Boiling point profiles for WC_Bit_B1 bitumen in DVFA-II at the apparatus pressure (fractionations 3 and 4) compared with the profiles from DVFA-I (fractionations 1 and 2). Symbols are data, and the line is a fit to the fractionation 3 data.
Gaussian distribution of carbon numbers with no suggestion of entrainment of heavy components. The distribution of density and molecular weights will be discussed later but also show smooth trends and no evidence of entrainment effects. Hence,
providing an as exact as possible comparison of curves where the data points do not fall on the same x or y coordinate. Potential sources of error are considered below.
Figure 9. Photograph of the eight different cuts of WC_Bit_B1 maltenes from fractionation 4 using DVFA-II. 2862
dx.doi.org/10.1021/ef500489y | Energy Fuels 2014, 28, 2857−2865
Energy & Fuels
Article
Processing of Cut Density Data. The densities of the cuts obtained from the standardized procedure (fractionations 3 and 4) are considered here. The densities of the first five cuts were measured directly, and the remaining cuts and the residue were diluted in toluene since these cuts were too viscous to inject into the density meter. The densities of the samples that were diluted with toluene were calculated from the mixture densities assuming an excess volume mixing rule,15 given by
we conclude that entrainment has been reduced to negligible levels. Residual Pentane from Deasphalting of Maltenes. The GC assays for the first cuts (cut 0 and cut 1) from fractionation 4 show no traces of pentane (Figure 10). The chromatographic results also show the expected overlapping of compounds as a consequence of the compositional complexity of the oil and the absence of reflux in the batch distillation. Finally, when compared to the assay for the whole bitumen, it appears that the small amount of material in the C10−C11 range may have been partially stripped off during the asphaltene precipitation procedure. However, these losses are negligible, corresponding to approximately 0.006% of the total bitumen sample. This mass fraction was calculated as the difference between the amounts of lightest components (C9−C11) reported in the GC analysis of the whole bitumen and the corresponding components in the GC analyses of the initial two cuts. Losses. During the fractionation there were two sources of losses of distilled sample: (1) escape of light components through the pump during the initial stages of the fractionation and (2) condensation of heavy boiling cuts in the lines during the final stages of the fractionation. The latter losses were determined from the mass changes of the DVFA-II fittings. The former were determined from the difference between the initial mass of the sample and the sum of the masses of the residue, cuts, and condensed material on the fittings. The total losses correspond to 3% of the initial sample mass. Leak Rate. In all vacuum systems, air enters the apparatus through imperfect seals and limits the ultimate vacuum pressure that can be achieved. However, this pressure cannot be monitored during the fractionation because of temperature limitations in the pressure gauges and may be a source of systematic error. Since the leak rate varies from run to run (as the tightness of the seals varies), the distillation pressure also varies, and therefore, the distilled amounts are different. The comparison of fractionations 3 and 4 in Figure 8 indicates that there is a small positive deviation at higher temperatures. It is suspected that the error is random, but multiple repeat fractionations would be required to verify this assumption. Any differences in the leak rates are detectable only at the higher temperatures, where the leak rate is largest. Despite the leak rate error, the initial testing indicates the average repeatability of the procedure was a very acceptable 2% of the cumulative weight percent collected for a given boiling point range. The final test will be to compare the results from the standardized fractionation with results from a conventional vacuum distillation. This comparison will be performed in part 2 of the study, after the low-pressure boiling points are converted to atmospheric-equivalent boiling points.
ρmix
⎤−1 ⎡w ⎛1 wT 1⎞ ⎥ F ⎢ ⎟⎟β = + − wFwT⎜⎜ + ρT ρT ⎠ 12 ⎥⎦ ⎢⎣ ρF ⎝ ρF
(1)
where wF and wT are the mass fractions of the cut (F) and the toluene solvent (T), respectively, and β12 is the binary interaction parameter of the cut and toluene. The densities of the first five cuts were also measured indirectly from solutions in toluene in order to evaluate the excess volume mixing rule. Figure 11 shows the specific volumes (inverses of the densities) of solutions of cut 1 from fractionation 3 in toluene.
Figure 11. Specific volumes of solutions of WC_Bit_B1 cut 1 from fractionation 3 in toluene at 293 K.
If the solutions were regular, then the data would follow a linear trend. However, there are clearly small excess volumes of mixing. From eq 1, the binary interaction parameters were calculated for cuts 1−5, for which the cut densities were measured directly. For cuts 6 and 7 and the residue, the same excess volume mixing rule was used to estimate both the density and the interaction parameter. The interaction parameters are negative for the lightest cuts (indicating shrinkage), negligible for the heavier cuts (indicating nearly ideal mixtures), and positive for the residue. Table 5 summarizes the specific gravity and interaction parameters for a set of cuts. The reported average absolute relative deviations (AARDs) correspond to the relative deviations between the measured specific gravities and the values calculated using eq 1. Property Distributions. The experimental molecular weights and densities of the cuts obtained from the WC_Bit_B1 sample are given in Table 4. The molecular weight and density data are plotted versus the cumulative weight percent distilled in Figures 12 and 13, respectively. The average value of the property for each cut or the residue is plotted at the cumulative weight percent distilled at the end of each cut. The data for molecular weight follow a monotonic trend. In general, the data trends are consistent with those
■
PHYSICAL PROPERTIES OF THE CUTS DVFA-II provides an opportunity to collect physical property data for heavy distillation cuts that were previously unattainable. Hence, measurement-based property distributions can now be extended to 50 wt % of the bitumen. The most commonly used properties used to supplement distillationbased characterizations are molecular weight and density. Molecular weights were determined as discussed in the Experimental Methods. However, since the densities of the heavier cuts were determined from mixtures of the cut in toluene, some additional data processing was required. 2863
dx.doi.org/10.1021/ef500489y | Energy Fuels 2014, 28, 2857−2865
Energy & Fuels
Article
4 (approximately 1.7 for cuts 0 and 1 and 1.5 for the higher cuts; Figure 13) confirm this interpretation. Since the data points for each run fall on different cumulative weight percent values, the repeatability was assessed by comparing the data with curve fits of the molecular weight (exponential equation) and density (cubic spline equation). The fits are not correlations and were used only to determine the repeatability. The AARDs were 4.0% and 0.2% for the molecular weight and density, respectively. The deviations and therefore the repeatability are within the error of the measurements.
Table 5. Directly Measured and Extrapolated Values of Specific Gravity (SG) and the Interaction Parameter (βij) for the Different WC_Bit_B1 Maltene Cuts from Fractionation 3
a
cut
SG
β12
% AARD
1 2 3 4 5 6 7 residue
0.9220 0.9565 0.9727 0.9796 0.9893 1.0003a 1.0030a 1.0300a
−0.00428 −0.00088 −0.00072 +0.00035 −0.00081 −0.00016 +0.00003 +0.00733
0.017 0.012 0.003 0.001 0.000 − − −
■
CONCLUSIONS The deep-vacuum fractionation apparatus (DVFA) developed to measure vapor pressures was tested as a batch distillation technique on a Western Canadian bitumen. The apparatus was modified to reduce liquid entrainment and distill larger volumes in less time. A standardized procedure was developed for the modified apparatus. The apparatus achieved distillation pressures of approximately 0.01 Pa, compared with 390 Pa for spinning band distillation (SBD). With the reduced pressure, the DVFA distilled 51 wt % of the bitumen, significantly more than the 26 wt % distilled by SBD. The DVFA distillation with the standardized procedure was repeatable within 2.3%. The standard procedure produced eight distillation cuts plus a residue. The density and molecular weight of the cuts were measured. Both properties exhibited smooth, monotonically increasing trends with the cumulative weight percent distilled. To quantify the repeatability of the data, the trends were curve-fitted. The average deviations were 0.2% and 4.0% for the density and molecular weight, respectively. The development of an interconversion method to determine atmospheric-equivalent boiling points is addressed in part 2 of this study.
Extrapolated values.
Figure 12. Plot of molecular weight of WC_Bit_B1 maltene cuts vs cumulative mass distilled. The data are plotted at the midpoint cumulative distillation for each cut. Symbols are data, and the line is a curve fit.
■
AUTHOR INFORMATION
Corresponding Author
*Telephone: (403) 220-6529. Fax: (403) 282-3945. E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors are grateful to the sponsors of the NSERC Industrial Research Chair in Heavy Oil Properties and Processing: Petrobras, Shell, Schlumberger, and Virtual Materials Group.
■
Figure 13. Plots of density and H/C ratio of WC_Bit_B1 maltene cuts as functions of cumulative mass distilled. The data are plotted at the midpoint cumulative distillation for each cut. Symbols are data, and the line is a curve fit of the density data.
REFERENCES
(1) Smith, D. F.; Rahimi, P.; Teclemariam, A.; Rodgers, R. P.; Marshall, A. G. Characterization of Athabasca Bitumen Heavy Vacuum Gas Oil Distillation Cuts by Negative/Positive Electrospray Ionization and Automated Liquid Injection Field Desorption Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Energy Fuels 2008, 22, 3118−3125. (2) Huang, S. H.; Radosz, M. Phase Behavior of Reservoir Fluids III: Molecular Lumping and Characterization. Fluid Phase Equilib. 1991, 66, 1−21. (3) Whitson, C. H.; Brulé, M. R. Phase Behavior; Henry L. Doherty Series, Monograph Volume 20; Society of Petroleum Engineers: Richardson, TX, 2000. (4) Riazi, M. R. Characterization and Properties of Petroleum Fractions, 1st ed.; ASTM International: West Conshohocken, PA, 2005.
observed for other crude oils.16 The relatively high molecular weight of the residues may be due to self-association of the heaviest molecules in the residue, that is, the heaviest resins. The density data also increase monotonically, but there appears to be an inflection point at approximately 15 wt % distilled (cut 1). There is likely a compositional change in going from cut 1 to cut 2, with the relatively volatile paraffinic components concentrated in cuts 0 and 1 while the higher cuts are more aromatic. The H/C ratios for the cuts obtained in fractionation 2864
dx.doi.org/10.1021/ef500489y | Energy Fuels 2014, 28, 2857−2865
Energy & Fuels
Article
(5) Castellanos Diaz, O.; Modaresghazani, J.; Satyro, M. A.; Yarranton, H. W. Modeling the Phase Behavior of Heavy Oil and Solvent Mixtures. Fluid Phase Equilib. 2011, 304, 74−85. (6) Gray, R. D., Jr.; Heidman, J. L.; Springer, R. D.; Tsonopoulos, C. Characterization and Property Prediction for Heavy Petroleum and Synthetic Liquids. Fluid Phase Equilib. 1989, 53, 355−376. (7) Leelavanichkul, P.; Deo, M. D.; Hanson, F. V. Crude Oil Characterization and Regular Solution Approach to Thermodynamic Modeling of Solid Precipitation at Low Pressure. Pet. Sci. Technol. 2004, 22, 973−990. (8) Merdrignac, L.; Espinat, D. Physicochemical Characterization of Petroleum Fractions: The State of the Art. Oil Gas Sci. Technol. 2007, 62, 7−32. (9) Batistella, C. B.; Sbaite, P.; Wolf Maciel, M. R.; Maciel Filho, R.; Winter, A.; Gomes, A.; Medina, L.; Kunert, R. Heavy Petroleum Fractions Characterization: A New Approach Through Molecular Distillation. In Second Mercosur Congress on Chemical Engineering and Fourth Mercosur Congress on Process Systems Engineering, 2005; pp 1− 10. (10) Speight, J. G.; Handbook of Petroleum Analysis; Chemical Analysis: A Series of Monographs on Analytical Chemistry and Its Applications, Vol. 158; Wiley: New York, 2001; pp 132−136. (11) Altgelt, K. H.; Boduszynski, M. Composition and Analysis of Heavy Petroleum Fractions; Chemical Industries: A Series of Reference Books and Textbooks, Vol. 54; Marcel Dekker: New York, 1994; pp 44−51. (12) Yarranton, H. W.; Alboudwarej, H.; Jakher, R. Investigation of Asphaltene Association with Vapor Pressure Osmometry and Interfacial Tension Measurements. Ind. Eng. Chem. Res. 2000, 39, 2916−2924. (13) Barrera, D. M.; Ortiz, D. P.; Yarranton, H. W. Molecular Weight and Density Distributions of Asphaltenes from Crude Oils. Energy Fuels 2013, 27, 2474−2478. (14) Castellanos-Díaz, O.; Schoeggl, F. F.; Yarranton, H. W.; Satyro, M. A. Measurement of Heavy Oil and Bitumen Vapor Pressure for Fluid Characterization. Ind. Eng. Chem. Res. 2013, 52, 3027−3035. (15) Saryazdi, F.; Motahhari, H.; Schoeggl, F.; Taylor, S. D.; Yarranton, H. W. Density of Hydrocarbon Mixtures and Bitumen Diluted with Solvents and Dissolved Gases. Energy Fuels 2013, 27, 3666−3678. (16) Boduszynski, M. M. Composition of heavy petroleums. 1. Molecular weight, hydrogen deficiency, and heteroatom concentration as a function of atmospheric equivalent boiling point up to 1400 °F (760 °C). Energy Fuels 1987, 1, 2−11.
2865
dx.doi.org/10.1021/ef500489y | Energy Fuels 2014, 28, 2857−2865