Article pubs.acs.org/IECR
Measurement of Heavy Oil and Bitumen Vapor Pressure for Fluid Characterization Orlando Castellanos-Díaz,*,† Florian F. Schoeggl,† Harvey W. Yarranton,*,† and Marco A. Satyro‡ †
Department of Chemical and Petroleum Engineering, University of Calgary, AB, Canada Virtual Materials Group Ltd., Calgary, AB, Canada
‡
ABSTRACT: The prediction of heavy oil phase behavior, particularly with solvents, is sensitive to the characterization of the middle and heavy boiling point components of the oil. These components are typically characterized based on an extrapolation of distillation data. One method to test the extrapolated characterization is to model the vapor pressures of these fractions or residues containing these fractions. Unfortunately, the vapor pressures are too low to be reliably measured with conventional techniques. A new high vacuum static apparatus was designed and constructed for the measurement of vapor pressure of heavy oil and bitumen samples. The apparatus is capable of measuring pressures from 100 down to 0.1 Pa and temperatures in the range of 293.15−473.15 K. New procedures were developed to degas samples and obtain accurate vapor pressures at vacuum conditions. The apparatus was tested on n-hexadecane and naphthalene at temperatures between 303.15 and 363.15 K. The measured vapor pressures were, on average, all within 13% of the literature data. The vapor pressures of a Western Canadian bitumen sample (WC_BIT_B1) and three of its fractions were measured using the apparatus. The WC_BIT_B1 bitumen was modeled using the Advanced Peng−Robinson equation of state using a Gaussian extrapolation of its distillation curve for the maltene fraction and a Gamma molecular distribution for its asphaltene fraction. The measured vapor pressures were all predicted to within 3.5%.
1. INTRODUCTION Both the refining and thermal and solvent based recovery processes for heavy oil and bitumen require the prediction of their phase behavior. A key part of modeling the phase behavior is to characterize the crude oil; that is, to divide the fluid into pseudocomponents and assign properties, such as density, molar mass, and critical properties, to each component. Once the pseudocomponents are defined, the phase behavior of heavy oils and bitumen is typically modeled with an equation of state.1 Generally, the pseudocomponents of petroleum fluids are determined from atmospheric or vacuum distillation assays; however, the heaviest fraction (residue) of the oil is left undetermined because the components in this fraction have boiling points higher than the cracking temperature (approximately 573−623 K). For heavy oils and bitumen, up to 70% of the oil can be nondistillable.2 The characterization of this large heavy fraction is based on an extrapolation of the distillation ́ data and the associated property correlations. Castellanos-Diaz 2 et al. demonstrated that cubic equation-of-state model predictions of liquid−liquid boundaries are sensitive to this extrapolation. Hence, any data that can constrain the extrapolation are useful for constructing more accurate phase behavior models and predictions. Ideally, the vapor pressure data of the heavy fractions can be used to constrain or test the characterization, but the vapor pressures of heavy hydrocarbons and their mixtures are difficult to measure because they fall below 1 Pa at moderate temperatures (290−330 K). Note that, strictly speaking, the term “vapor pressure” applies to pure components not mixtures. However, the term is commonly used to describe the saturation pressure of petroleum fractions, and we have followed the common usage. © 2013 American Chemical Society
At present, few vapor pressure data are available in the open literature for heavy fossil fuel fractions and, when available, they are reported at high temperatures. For example, Wilson et al.3 and Gray et al.4,5 reported vapor pressures and heats of vaporization values for heavy coal liquids using static and ebulliometric methods at temperatures ranging from 320 to 790 K and vapor pressures above 2000 Pa. Schwarz et al.6 measured the vapor pressures of 12 fractions of crude oils, coal liquids, and tar sands using ebulliometric methods at temperatures up to 575 K and vapor pressures ranging from 400 to 200 000 Pa. Neither a standard procedure nor off the shelf equipment is available commercially for direct vapor pressure measurement of heavy hydrocarbons systems at moderate temperatures since such measurements require deep vacuum conditions. Several authors7,8 have designed static deep vacuum laboratory apparatuses to measure the vapor pressure of pure components. In these static apparatuses, samples are placed in a closed vessel connected to a pressure transducer, and the pressure is measured directly. This approach generally has a more stable temperature and is more repeatable than other techniques.9 However, this type of apparatus has not yet been tested on mixtures of heavy components. The objectives of this work are to (1) construct a static apparatus, similar to those designed by Fulem et al.,7,10 to measure the vapor pressure of heavy hydrocarbons and heavy oil fractions; (2) measure the vapor pressure of bitumen fractions and use the data to validate a bitumen characterization Received: Revised: Accepted: Published: 3027
December 10, 2012 January 29, 2013 February 1, 2013 February 1, 2013 dx.doi.org/10.1021/ie303397y | Ind. Eng. Chem. Res. 2013, 52, 3027−3035
Industrial & Engineering Chemistry Research
Article
in the maltenes was removed from the sample by the degassing process described later in this paper. Residues were obtained from the two maltene samples by the deep vacuum distillation using the apparatus developed in this work. The procedure was similar to the vapor pressure measurements described below except that the sample was left open to vacuum at each temperature step and the distillate was collected in a coldfinger. Two residues equivalent to 75 and 88 wt % of the maltenes, at temperatures of 423 and 453 K, respectively, were collected after distillations of the two maltenes samples. Details are provided elsewhere.14,15 2.4. Heat Capacity Measurements. The liquid heat capacity of the WC_BIT_B1 sample was measured using a differential scanning calorimeter (DSC) TA Q2000 V24.9 at the Steacie Institute for Molecular Sciences NRC-CNRC in Ottawa, ON. The DSC was calibrated to the heat capacity of indium. The liquid heat capacities obtained from the DSC have a precision of 2%. 2.5. Vapor Pressure Apparatus. The operating principle of the vapor pressure apparatus is very simple: a sample is opened to a fixed volume initially at vacuum and allowed to equilibrate at the desired temperature. The equilibration pressure is a combination of the vapor pressure of the sample and small leaks inside the apparatus. The readings are corrected based on the leak rate to obtain the vapor pressure of the sample at the given temperature (section 2.7). The design criteria were as follows: • upper operational temperature limit: 473 K (maximum rating of diaphragm pressure gauge) • lower operation temperature limit: ambient (no cooling) • upper operational pressure limit: 133 Pa (maximum rating of diaphragm pressure gauge) • lower operational pressure limit: 0.1 Pa (minimum rating of diaphragm pressure gauge) The apparatus consists of two components: a degassing or sample preparation apparatus (DA) and a vapor pressure measurement apparatus (VPMA). The DA is used to remove dissolved gases and light solvents from the sample; the VPMA is used to measure the vapor pressure of a sample or to fractionate a sample. The use of the DA prior to the VPMA ensures that the sample is clean before it is evaluated. Each component is described in detail below. Vapor Pressure Measurement Apparatus (VPMA). A schematic of the VPMA is provided in Figure 1. The apparatus consists of six different systems which are discussed below: piping, sample chamber, pressure measurement, temperature control, pumping, and cold finger. In addition, seals for the fittings are a critical component of any vacuum apparatus, and they are discussed as well. Piping. Stainless steel SS316 from Swagelok was used for the apparatus. The pipe diameter was determined based on two criteria: thermal transpiration and pump time. The former is mitigated with a larger pipe diameter whereas the latter is minimized with a smaller pipe diameter. A pipe diameter of 2.45 cm (1 in.) was selected to avoid thermal transpiration effects and yet provide an acceptable pump-down time. Caution is required at pressures below 0.1 Pa since thermal transpiration corrections to the vapor pressure may be required. Sample Chamber. The sample chamber consists of a sample vessel and a metal valve. The sample vessel is a stainless steel (SS) Swagelok full nipple, with ConFlat (CF) 133 fittings and an approximate volume of 20 cm3. The bottom part is sealed
used for phase behavior modeling. The design and testing of the apparatus are discussed below. For the second objective, vapor pressures were measured for a Western Canadian bitumen sample (WC_BIT_B1) and three of its residues. The liquid heat capacity of WC_BIT_B1 was also measured since heat capacity and vapor pressure are related through standard thermodynamic relationships.14 The WC_BIT_B1 bitumen was characterized following the procedure of ́ et al.2 and modeled using the Advanced Castellanos-Diaz Peng−Robinson equation of state.11 The model was tested on the data for the vapor pressures of the oil and residues as well as the liquid heat capacity of the oil.
2. EXPERIMENTAL METHODS 2.1. Materials. n-Eicosane (C20H42, CAS no.: 112-95-8) was used to calibrate the apparatus and naphthalene (C10H8, CAS no.: 91-20-3) and n-hexadecane (C16H34, CAS no.: 544-76-3) were selected to test the repeatability and reproducibility of the apparatus. Naphthalene has well-established vapor pressure and calorimetric data and is recommended as a reference compound for vapor pressure measurement below 1000 Pa.12,13 nHexadecane and n-eicosane are petroleum-related paraffins with molecular weights of 224 and 282 g/mol, respectively, which are similar to the molecular weight of the lightest fraction of bitumen and heavy oil and are expected to exhibit similar vapor pressures. Pure component samples were purchased from Aldrich Chem., with purity greater than 99.7 wt % for both chemicals. The bitumen sample (WC_BIT_B1) is from a Western Canadian reservoir and was provided dewatered by Shell Chemicals Americas. 2.2. Bitumen Property Measurements. Selected physical properties of WC_BIT_B1 bitumen were measured at the University of Calgary and are presented in Table 1. The Table 1. Selected Physical Properties of Bitumen Sample from the Western Canadian Bitumen (WC_BIT_B1) property
value
average molecular weight average specific gravity initial boiling point asphaltene and solids content
510 g/mol 1.007 213 °C 17 wt %
molecular weight was measured in toluene at 323 K using a Jupiter Model 833 vapor pressure osmometer, calibrated with sucrose octacetate. The bitumen density was determined from measured densities of a series of solutions of bitumen in toluene at different concentrations. The densities of the solutions were measured with an Anton Paar MDA 46 density meter. The initial boiling point and distillation curve data, Figure 7, were measured using spinning band distillation. 2.3. Fractionation of Bitumen. The maltene fraction of the WC_BIT_B1 bitumen was obtained by deasphalting the sample with n-pentane at a 40:1 w/w ratio of n-pentane:bitumen. The mixture was sonicated for 60 min and left to equilibrate for 24 h. After equilibration, the mixtures were filtered through a 24 cm Whitman filter. The filter cake was washed with n-pentane until washings were colorless. The dried filter cake is the asphaltenes plus any nonasphaltene solids that coprecipitate (typically less than 1 wt % of the filter cake). The asphaltene and solids content is reported in Table 1. The washings were added to the filtrate which was then placed in a roto-vaporator to remove the n-pentane. Any residual pentane 3028
dx.doi.org/10.1021/ie303397y | Ind. Eng. Chem. Res. 2013, 52, 3027−3035
Industrial & Engineering Chemistry Research
Article
Figure 1. Vapor pressure measurement apparatus (VPMA).
thermocouples were not used because their ports would be a significant potential source of leaks. Values of the temperature are displayed on Waltlow SD series readouts with a resolution of 0.1 K. The same device contains an autotuned PID (proportional−integral−derivative) controller that is used to maintain the temperature of the system at a desired value to within ±0.1 K. Electric heat tapes wrapped on the apparatus provide the heat to maintain a desired temperature. These heat tapes were selected for reliability and price and provide smallequipment flexibility and solutions. The maximum working temperature of the tapes is approximately 920 K. Heating tapes are connected to a fuse box for protection and to the PID controller. Finally, the heat tapes and pipes are insulated using insulation tape made of carbon fiber. There are two sections of the VPMA that are temperaturecontrolled. The first one is the sample chamber, set to the desired temperature value at which the vapor pressure is measured. The second section is the pressure measurement system; this section is kept at 473 K to ensure stability of the measurements, facilitate vapor transport, and mitigate possible condensation. Pumping. The pumping system is comprised of a Pfeiffer pumping station model TSH 071 E., capable of reaching a final pressure of 10−5 Pa at a rate of 60 L/s, nitrogen based (manufacturer’s rating without baking-out), and final pressures of approximately 10−7 Pa after baking-out. It consists of a
with a Swagelok CF133 blank, and the upper part is connected to an all metal angle valve with a manual actuator. A volume of 20 cm3 was sufficient to obtain an entire vapor pressure curve of a heavy substance (molecular weight higher than 200 g/ mol). Pressure Measurement. The pressure in the chamber is measured with an Inficon diaphragm gauge capable of measuring pressures in the 0.1−133 Pa range. This gauge has an internal heater which sets it at 473 K for every measurement. This high temperature ensures that the readouts are stable and mitigate possible instabilities and noise from the environment. The gauge is connected directly to a computer to record the pressures on a digital file using LabView 8.6.18 Vapor pressure readings are limited on the high end by the resolution of the gauge (133 Pa) and on the lower end by gauge resolution, pump suction, and/or adsorption/desorption processes. At pressures below approximately 0.1 Pa, absorption and desorption of the molecules to and from the pipe as well as permeation of atmospheric gas molecules through the pipe lead to indeterminable pressure measurement uncertainties.10,16 Note that these processes occur at any condition; however, their effect on the total pressure readout increases significantly as the pressure is decreased. Temperature Measurement and Control. The temperature of the system is read by J-type thermocouples attached to the pipes and vessels using adjustable metal clamps. Internal 3029
dx.doi.org/10.1021/ie303397y | Ind. Eng. Chem. Res. 2013, 52, 3027−3035
Industrial & Engineering Chemistry Research
Article
2.6. Calibration of DVMA. Thermocouple Calibration. All of the thermocouples were calibrated against a resistance temperature detector (RTD) which had been calibrated against a certified high precision thermometer (HPTAutomatic Systems Laboratories F250 Precision Thermometer Res. 0.025 C). The temperature for the calibrations was controlled with a thermostatted bath (FLUKE 6330 Calibration Bath). Pressure Gauge Calibration. Vapor pressure measurements on the VPMA were taken with a temperature controlled diaphragm gauge. For this type of gauge, a linear calibration is possible at pressures above 10 Pa. However, below this threshold, a linear calibration is not representative and a logarithmic term may be required.17 The diaphragm gauges for the VPMA were first calibrated at pressures above 10 Pa where the behavior is expected to be linear. A two-step calibration was performed because the pressure ranges of the diaphragm gauges (DG) did not overlap the pressure range of the most reliable low vacuum diaphragm gauge (CG: calibration gauge). Therefore, a Cold Cathode Pirani gauge (CCP) with an intermediate pressure range was calibrated to the GC, and then, the diaphragm gauges were calibrated to the CCP gauge. The final calibration (above 10 Pa) had the form
turbomolecular pump, a dry diaphragm backing pump, and a display. The pressure at pump suction is measured by a Cold Cathode-Pirani pressure gauge. The pump cannot handle liquid, and hence, condensates are removed before the suction with the coldfinger system. Cold Finger. For sample collection and pump protection, a modified centrifuge tube with a tee-relieve system was designed and constructed, which is used as a cold finger; Figure 1. The vapor exiting the sample vessel enters the cold finger from the top of the cross into the inner tube. Subsequently, the vapor follows a winding path to the bottom of the centrifuge tube. Ideally, the cold finger and the winding path provide enough contact time for the vapor to condense at the bottom of the Pyrex tube and be collected. The cold finger is cooled by means of an oil or dry ice bath. The noncondensable gas (mainly air) is liberated through the outer section of the tee-relieve system and goes into the pump. Seals. The seals are the main source of leaks. To minimize these leaks, CF flanges with copper gaskets were used, capable of holding vacuum down to 10−8 Pa. Note that a metal gasket is only useable once and must be replaced every time the flanges are disconnected. Degassing Apparatus (DA). The degassing apparatus (DA) is similar to the VPMA. This apparatus is used to prepare the sample for vapor pressure measurement. A schematic of the degassing apparatus is provided in Figure 2. The pressure is
PcalH = c1PDG + c 2
(1)
where PcalH is the calibrated pressure above 10 Pa, PDG is the gauge reading, and c1 and c2 are calibration constants. Subsequently, the DG gauges were calibrated for pressures below 10 Pa. In this case, the calibration was performed by comparing the measured vapor pressure of n-eicosane with literature data.19 Note n-eicosane was chosen since it has available experimental data, it covers the calibration pressure range of interest and is a representative molecule present on heavy oil and bitumen; the literature data of n-eicosane were fitted with a Cox equation to facilitate the calibration where there were gaps in the data. The fitting with the Cox equation is discussed later. A temperature dependent calibration of the following form was required to match the literature data: ⎛T ⎞ PcalL = PcalH exp −c3⎜ NL − 1⎟ ⎝ T ⎠
Figure 2. Simplified DA (degassing apparatus) schematics.
(2)
where PcalL is the calibrated pressure below 10 Pa, c3 is a calibration constant, T is the temperature in kelvin, and TNL is the gauge temperature at which the vapor pressure equates to 10 Pa in kelvin. The calibrated vapor pressures of n-eicosane are compared with the extrapolated literature data in Figure 3. The calibration was verified against vapor pressure data for nhexadecane and naphthalene as discussed later. 2.7. Vapor Pressure Measurement Procedure. Preparation of Apparatus. Before the apparatus is tested and used for vapor pressure measurement, it must be baked out to remove water and remnants of oil from the inner pipe surface. The bake out procedure prevents outgassing of volatile substances during the vapor pressure measurement, which would invalidate the measurements.16 The entire apparatus is heated at 573 K for 3 days open to pump suction. Note that the pressure gauge must be removed before baking to protect its electronics. Hydraulic Testing. The system is tested hydraulically to ensure that the leaks within the pressure and temperature operable ranges have been minimized. Note, it was not possible to completely eliminate leaks at high vacuum. The apparatus is
measured by a temperature-controlled MKS Baratron diaphragm gauge. The steady measurement temperature is 423.15 K. This gauge has a pressure range from 1 to 1000 Pa. Note that this gauge can be used as the parallel gauge on the VPMA. Moreover, using the DA for cleaning and preparing the sample for the VPMA means that the DA handles more volatiles and impurities than the VPMA. These impurities need to be condensed before they reach the pump (otherwise, they will condense at the pump and suction will be lost). An 8 L LACO cold trap with NW25 connections is used as a condenser instead of the cold finger in Figure 1. Dry ice or water ice can be used to keep the cold trap at temperatures below zero, assuring complete condensation of volatiles. Dry ice is preferred since it does not leave traces to be cleaned. Finally, unlike the VPMA, KF flanges are sealed with Viton rubber gaskets. Rubber gaskets can hold vacuum down to 0.01 Pa and are subject to elastomeric deformation due to temperature cycles in the equipment, which, in turn, creates leak sources. Note that the DA and rubber gaskets are a previous prototype of the VPMA and the metal gaskets. 3030
dx.doi.org/10.1021/ie303397y | Ind. Eng. Chem. Res. 2013, 52, 3027−3035
Industrial & Engineering Chemistry Research
Article
is recorded, and a cycle is completed. The degassing of a sample consists of several cycles until at least four to five subsequent cycles repeat. At this point, the sample is sufficiently degassed to be transferred to the VPMA. Figure 5 shows a complete run of cycles for the degassing of n-hexadecane at 298 K. The data can be divided into three
Figure 3. Literature and measured vapor pressure data of eicosane with linear calibration and with linear + nonlinear calibration. The dotted 45° line represents a perfect calibration. Figure 5. Complete cycle of hexadecane degassing at 298 K.
connected to a Varian 797 leak detector with a helium mass spectrometer. Leak tests show that the average leak rates are 6.6 and 10.1 Pa·cm3/s for the VPMA and DA, respectively. This means, for instance, that a flow of 6.5 × 10−5 cm3/s of air enters the VPMA at normal atmospheric conditions externally. These results show good sealing in the fittings as compared to other vacuum systems.16 Degassing. Samples to be characterized often have light impurities, such as water and light solvents, that affect the vapor pressure measurement since their partial pressures are significant in comparison with the expected vapor pressure of the sample. For example, heavy oil contains traces of light oils, or solvents such as naphtha and toluene, coming from the oil extraction processing as well as residual water. In order to measure vapor pressure correctly, the sample must be degassed. The DA is run in cycles to monitor the degassing performance. First, a baseline pressure for the system is reached by pumping the system from valve V1 onward with the sample chamber connected (see Figure 2). The baseline pressure depends on the speed of the pump (Sp), the capacitance of the pipe (C), the leak rate (Ql), and the outgassing rate (Qo). Then, valve V2 is closed and V1 is opened, simultaneously. A step change increase in the pressure is recorded, Figure 4. The magnitude of the increase depends on the pressure exerted by the sample (Qv) and the leak rate (Ql). Finally, Valve V1 is closed and V2 is opened simultaneously. A rapid change in pressure back to the baseline
major sections. The first one, labeled “air”, refers to the degassing of gases trapped on the sample and from the interior of the sample chamber. The second section, labeled “water + solvents”, refers to the degassing of water and light solvents or impurities present on the sample and within the walls of the chamber. The third section, labeled “steady state measurement”, is reached when the degassing is finished and a constant vapor pressure is recorded. Note that the three different sections are not always clearly differentiated and some overlap does occur. Vapor Pressure Measurement. Once the sample and the sample chamber have been degassed, the sample chamber is isolated from the DA by closing V6 in Figure 2. The chamber is disconnected and reconnected to the vapor pressure apparatus; Figure 1. The latter must have been already baked out, cleaned, and vacuumed at a pressure lower that whatever baseline was used on the degassing apparatus. Subsequently, the vapor pressure apparatus is run in cycles in the same manner as the degassing apparatus; Figure 4. A lower leak rate is obtained with the metal gaskets in the VPMA (rather than the rubber gaskets in the DA), so that a more accurate vapor pressure can be determined. The pressure for any given cycle is obtained by extrapolating the pressure trend caused by the leak rate back to the start time of the cycle.
3. WC_BIT_B1 VAPOR PRESSURE MODEL The vapor pressure data of the WC_BIT_B1 bitumen and fractions were modeled with the advanced Peng−Robinson equation of state (APR EOS). The APR is the Peng−Robinson equation of state with the volume translation suggested by Mathias et al.20 as well as specially fitted alpha functions for polar compounds of interest for natural gas processing such as alcohols and glycols. Details can be found in the work of ́ et al.2 and the VMGSim User’s manual.11 Castellanos-Diaz The bitumen was divided into pseudocomponents for the EOS model based on the spinning band distillation data using ́ et al.2 Briefly, the procedure recommended by Castellanos-Diaz the normal boiling point data (NBP) for WC_BIT_B1 was extrapolated as shown in Figure 6. The maltenes NBP extrapolation was performed using a Gaussian probability distribution whereas the asphaltene NBP was predicted using
Figure 4. Typical degassing cycle. 3031
dx.doi.org/10.1021/ie303397y | Ind. Eng. Chem. Res. 2013, 52, 3027−3035
Industrial & Engineering Chemistry Research
Article
4. RESULTS AND DISCUSSION 4.1. Validation of VPMA with Pure Components Vapor Pressures. Vapor pressure and heat capacity data for naphthalene and n-hexadecane were obtained from Lemmon’s recommended database.19 These data have 95% confidence intervals of ±0.01% and ±3%, respectively. To facilitate comparison with measurements with the new apparatus, the data were fitted with the Cox equation: ⎛ T ⎞ ln PV = ln Pref + ⎜1 − ref ⎟ exp[aV + bV T + cV T 2] ⎝ T ⎠
Note the same procedure was used for n-eicosane for the calibration of the pressure gauges. The fitted Cox equation constants and the AARD of the fitted equation are provided for all three components in Table 5. The fitted vapor pressure curves for n-hexadecane and naphthalene are plotted in Figure 7. The vapor pressures of naphthalene and n-hexadecane were then measured using the VPMA at temperature values ranging from 303 to 363 K; Table 6. Figure 7 compares the new data with the fitted Cox equations. The AARD between these new data and the Cox equations are 13% for naphthalene and 4% for n-hexadecane.
Figure 6. Spinning band distillation data and extrapolated normal boiling point of WC_BIT_B1 bitumen.
the Gamma distribution function with values of α = 2.0 and η = 750. The average physical properties of the WC_BIT_B1 bitumen used in this work are provided in Table 2. Table 2. Physical Properties of WC_BIT_B1 Maltenes and Asphaltenes property
maltenes
asphaltenes
bitumen
average molecular weight average specific gravity boiling point range number of pseudocomponents
450 g/mol
1800 g/mol
510 g/mol
1.005
1.105
1.010
480−949 K 10
949−1023 K 5
480−1023 K 15
5. WC_BIT_B1 VAPOR PRESSURES The vapor pressure of the WC_BIT_B1 bitumen and fractions measured with the VPMA at a temperature range of 298 to 453 K are provided in Tables 7 and 8, respectively. The 95% confidence interval for the experimental data, based on the pressure variation from cycle to cycle, is ±3%. Table 9 shows the liquid heat capacity of WC_BIT_B1 maltenes measured with a differential scanning calorimeter. The differences between the vapor pressure of the heavy oil and the maltenes fraction are 5.0% on average, which is close to the uncertainty of the measurements. Recall that the only different between maltenes and the whole bitumen is the removal of asphaltenes. Asphaltenes have a high molecular weight and low vapor pressure, and therefore, they contribute a negligible amount of partial pressure to the total pressure of the system. Hence, the vapor pressure of the maltenes is expected to be almost the same as the bitumen. Significant differences were observed with the maltene residues because in this case the light components were removed relative to the maltenes or whole bitumen. Figure 8 shows the experimental and predicted data for the WC_BIT_B1 bitumen oil and fractions. The AARD between experimental data and model was 2.2% for the bitumen and maltenes and 11% for the fractions. This higher error for the fractions falls within the error range from mass losses and mass entrainment in the apparatus during the fractionation process. Figure 10 shows the model predictions of the experimental data for the 88 wt % residue where the mass fraction is relaxed to the limits of the mass losses found in the apparatus (3%). With this relaxation, the error between the model and the data is reduced from 11% to 3.6%. There is room to improve the fractionation process to minimize losses and this issue will be addressed in future work. Figure 9 shows the experimental and predicted liquid heat capacity of the heavy oil maltenes with an AARD of 2.1%. The predicted trend does not exactly follow the experimental data, mainly due to the predicted ideal gas heat capacity; however,
The pseudocomponents comprising the bitumen were obtained by cutting the extrapolated NBP curve in VMGSim 6.0.38.11 The maltenes were represented with 10 pseudocomponents, and the asphaltenes, by 5 pseudocomponents. Physical properties of the pseudocomponents were calculated using the correlations listed in Table 3 and are summarized in Table 4. Table 3. Physical Property Correlations Used in WC_BIT_B1 Bitumen Modeling property
correlation
molecular weight maltenes specific gravity asphaltenes specific gravity asphaltenes NBP critical temperature critical pressure critical volume acentric factor ideal gas heat capacity
Lee−Kesler Katz−Firoozabadi Yarranton Soreide Lee−Kesler Lee−Kesler Twu Lee−Kesler−Lee Laštovka−Shaw
(3)
Once the bitumen has been characterized, the advanced Peng−Robinson equation of state (APR-EoS)11,21 was used to simultaneously calculate the vapor pressure and the heat capacity of the WC_BIT_B1 bitumen and fractions. The classic van der Waals mixing rules were used with adjustable interaction parameters, kij, estimated using the Gao et al. correlation.22 3032
dx.doi.org/10.1021/ie303397y | Ind. Eng. Chem. Res. 2013, 52, 3027−3035
Industrial & Engineering Chemistry Research
Article
Table 4. Pseudocomponent Physical Properties of WC_BIT_B1 Bitumen pseudo
X
w
SG
Tc [K]
Pc [kPa]
Vc [kmol/m3]
ω
NBP [K]
Malt01 Malt02 Malt03 Malt04 Malt05 Malt06 Malt07 Malt08 Malt09 Malt10 Asph01 Asph02 Asph03 Asph04 Asph05
0.057 0.051 0.062 0.055 0.068 0.063 0.067 0.223 0.188 0.108 0.003 0.012 0.020 0.017 0.006
0.023 0.023 0.032 0.031 0.042 0.044 0.051 0.206 0.222 0.157 0.005 0.025 0.052 0.061 0.026
0.858 0.872 0.886 0.899 0.911 0.924 0.937 0.965 0.989 1.007 1.026 1.039 1.054 1.075 1.088
691.183 718.501 745.645 772.087 796.984 823.315 848.648 903.273 975.343 1040.41 1039.38 1071.83 1108.36 1149.25 1180.15
2458.660 2251.800 2059.790 1887.860 1736.870 1596.910 1471.490 1241.660 919.946 673.174 849.228 791.605 746.436 726.730 716.733
0.589 0.651 0.718 0.787 0.856 0.929 1.003 1.160 1.454 1.777 1.524 1.595 1.658 1.703 1.737
0.502 0.559 0.619 0.681 0.743 0.808 0.874 1.018 1.270 1.540 1.369 1.437 1.495 1.524 1.536
494.836 523.243 552.215 580.999 608.656 637.962 666.611 728.837 819.911 907.231 884.066 919.621 957.687 996.26 1024.61
Table 5. Cox Equation Constants for Naphthalene and n-Hexadecane Vapor Pressurea a1
substance naphthalene n-hexadecane n-eicosane a
3.272 3.024 3.950
a2
a3 −4
−9
−4.347 × 10 1.794 × 10−6 6.861 × 10−7
−2.659 × 10 −1.831 × 10−3 −1.201 × 10−3
To [K]
Po [Pa]
AARDb [%]
353.7 560.15 311.16
0.9935 101.325 1.22 × 10−5
0.01 0.43 5.8
AARD are for fit to literature data. bAARD = (1/n)∑ni=1|(ln Pi,exp − ln Pi,calc)/ln Pi,exp|
Table 6. Vapor Pressure of Pure Components Measured Using the VPMAa hexadecane
naphthalene
T [K]
PCorr [Pa]
T [K]
PCorr [Pa]
298.2 298.3 303.2 303.7 305.2 307.3 308.9 311.8 317.6 318.2 319.4 323.0 323.2 330.1 332.6 333.2 343.2 343.2 352.4 363.1 363.1 363.2
0.68 0.50 0.90 0.79 0.79 1.13 1.23 1.52 2.53 2.46 3.39 4.45 4.38 6.78 8.66 9.87 30.1 21.9 27.3 69.9 82.2 64.2
303.0 303.2 313.3 323.3 323.3 323.3 343.1
25.4 15.5 21.1 130 142 152 426
Figure 7. Experimental vapor pressure data for n-hexadecane and naphthalene. Literature data shown is regressed with the Cox equation; Table 4.
ization used in this work. The same characterization and model were used to describe the phase behavior of bitumen−solvent systems.2 In other words, the characterization is robust for both bitumen cuts and mixtures containing bitumen and therefore is applicable for a wide range of upstream and downstream processes.
6. CONCLUSIONS A high vacuum apparatus was designed and constructed for the measurement of vapor pressure of bitumen and its fractions. The apparatus and procedures were validated against literature data for the vapor pressure of pure substances. The 95% confidence interval of the vapor pressure data was ±3%, and the measurements were within 6% of the literature data. The apparatus significantly increases the range for the vapor pressure of heavy components to be measured. For example, vapor pressures were measured for bitumen, maltenes, and
Here error is the difference of the data from the previously fitted Cox equation. Experimental uncertainty in temperature is 0.1 °C. The 95% confidence interval of the vapor pressure is ±3%. a
the error is low enough to permit heat exchanger design calculations. The predictions of vapor pressure and heat capacity in Figures 8 and 9 are essentially within the accuracy of the data and are an independent validation of the bitumen character3033
dx.doi.org/10.1021/ie303397y | Ind. Eng. Chem. Res. 2013, 52, 3027−3035
Industrial & Engineering Chemistry Research
Article
Table 7. Experimental Vapor Pressure of WC_BIT_B1 Bitumena WC_BIT_B1 Bitumen
WC_BIT_B1Maltenes Fraction
T [K]
PCorr [Pa]
T [K]
PCorr [Pa]
40.9 60.6 79.5
5.34 20.8 62.7
303.4 313.1 313.2 319.2 332.8 333.2 334.1 352.5 353.1 353.5 373.1
1.79 4.06 4.41 6.83 15.6 16.6 16.8 49.5 50.1 36.8 138
Figure 8. Experimental and predicted vapor pressure of WC_BIT_B1 bitumen and fractions using the APR EoS.
a
The variation in the temperature was within 0.1 K. The 95% confidence interval of the vapor pressure is ±3%.
Table 8. Experimental Vapor Pressure of WC_BIT_B1 Residuesa WC_BIT_B1 88 wt % residue
WC_BIT_B1 75 wt % residue
T [K]
PCorr [Pa]
T [K]
PCorr [Pa]
314.0 333.6 353.3
0.16 0.54 1.92
333.1 353.1 368.1 383.2 398.2 432.9
0.24 0.95 2.08 4.44 9.10 41.8
The variation in the temperature was within 0.1 °C. The 95% confidence interval of the vapor pressure is ±3%.
a
Table 9. Liquid Heat Capacity of WC_BIT_B1 Maltenesa T [K]
Cp [J K−1 mol−1]
T [K]
Cp [J K−1 mol−1]
273.1 277.1 281.1 285.1 289.1 293.1 297.1
767.2 772.9 779.6 787.4 799.7 811.8 825.0
305.1 309.1 313.1 317.1 321.1 325.1 329.1
849.0 861.6 874.6 889.3 901.8 914.7 929.7
a
T [K]
Cp [J K−1 mol−1]
337.1 341.1 345.1 349.1 353.1
959.4 973.8 988.2 1002.0 1016.3
Figure 9. Experimental and predicted liquid heat capacity of WC_BIT_B1 bitumen and maltene fraction using the APR EoS and the Lastovka−Shaw correlation.
Molecular weights are reported in Table 2.
maltene residues at temperatures as low as 293 K and pressures as low as 0.1 Pa. To date, experimental data for bitumen and fractions are extremely scarce and only available at temperatures near the cracking point of 573 K. The data were used to test a bitumen characterization for the advanced Peng−Robinson equation of state which Castellanoś et al.2 developed to model to fit vapor−liquid and liquid− Diaz liquid phase boundaries of mixtures of Athabasca bitumen and solvents. In this work, the same model was employed to predict the vapor pressure of WC_BIT_B1 bitumen and fractions as well as liquid heat capacity of the bitumen maltene fraction to within 11% for the vapor pressure and 3% for the heat capacity. The successful application of the same characterization to two independent data sets validates both the characterization and the vapor pressure data.
Figure 10. Experimental and predicted vapor pressure of WC_BIT_B1 88 wt % fraction using the APR EoS. Compositions have been relaxed to within experimental error.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected] (O.C.-D); hyarrant@ ucalgary.ca (H.W.Y.). 3034
dx.doi.org/10.1021/ie303397y | Ind. Eng. Chem. Res. 2013, 52, 3027−3035
Industrial & Engineering Chemistry Research
Article
Notes
(16) Roth, A. Vacuum Technology, 3rd ed.; North-Holland: The Netherlands, 1990. (17) Fulem, M. Private Communication, University of Calgary: Calgary, 2011. (18) National Instruments. LabView 8.6, User Manual; National Instruments Corporation: Texas, 2008. (19) Lemmon, E. W.; Goodwin, A. R. H. Critical Properties and Vapor Pressure Equation for Alkanes CnH2n+2: Normal Alkanes with n ≤ 36 and Isomers for n = 4 through n = 9. J. Phys. Chem. Ref. Data 2000, 29 (1), 1−39. (20) Mathias, P. M.; Naheiri, T.; Oh, E. M. A Density Correction for the Peng-Robinson Equation of State. Fluid Phase Equilib. 1989, 47, 77. (21) Peng, D. Y.; Robinson, D. B. A New Two-Constant Equation of State. Ind. Eng. Chem. Fund. 1976, 15 (1), 59−64. (22) Gao, G.; Didiron, J. L.; Saint-Guirons, H.; Xans, P.; Montel, F. A. A Simple Correlation to Evaluate Binary Interaction Parameters of the Peng-Robinson Equation of State: Binary Light Hydrocarbon Systems. Fluid Phase Equilib. 1992, 74−85.
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors are grateful to everyone involved in the design and construction of the apparatus including Bernie Then, Mike Grigg, and Colin Branner at the University of Calgary and Michal Fulem at the Institute of Chemical Technology, Prague. We thank Catalina Sanchez at the University of Calgary for collecting part of the experimental data. We appreciate the financial support from the sponsors of the NSERC Industrial Research Chair in Heavy Oil Properties and Processing: NSERC, Petrobras, Royal Dutch Shell, and Schlumberger. We thank Virtual Materials Group for the use of the VMGSim software.
■
REFERENCES
(1) Whitson, C. H.; Brulé, M. R. Phase Behavior; Monograph Vol. 20, Henry L. Doherty Series; Society of Petroleum Engineers Inc.: USA, 2000. (2) Castellanos-Díaz, O.; Modaresghazani, J.; Yarranton, H. W.; Satyro, M. A. Modeling the Phase Behavior of Heavy Oil and Solvent Mixtures. Fluid Phase Equilib. 2011, 304, 74−85. (3) Wilson, G. M.; Johnston, R. H.; Hwang, S. C.; Tsonopoulos, C. Volatility of Coal Liquid at High Temperatures and Pressures. Ind. Eng. Chem. Process Des. Dev. 1981, 20 (1), 94−104. (4) Gray, J. A.; Brady, C. J.; Cunningham, J. R.; Freeman, J. R.; Wilson, G. M. Thermophysical Properties of Coal Liquids. 1. Selected Physical, Chemical, and Thermodynamic Properties of Narrow Boiling Range Coal Liquids. Ind. Eng. Chem. Process Des. Dev. 1983, 22 (3), 410−424. (5) Gray, J. A.; Holder, G. D.; Brady, C. J.; Cunningham, J. R.; Freeman, J. R.; Wilson, G. M. Thermophysical Properties of Coal Liquids. 3. Vapor Pressure and Heat of Vaporization of Narrow Boiling Coal Liquid Fractions. Ind. Eng. Chem. Process Des. Dev. 1985, 24 (1), 97−107. (6) Schwarz, B. J.; Wilhelm, J. A.; Prausnitz, J. M. Vapor Pressures and Saturated-Liquid Densities of Heavy Fossil-Fuel Fractions. Ind. Eng. Chem. Res. 1987, 26, 2353−2360. (7) Fulem, M.; Růzǐ čka, K.; Růzǐ čka, V.; Simecek, T.; Hulicius, E.; Pangrac, J. Vapour Pressure and Heat Capacities of Metal Organic Precursors, Y(thd)3 and Zr(thd)4. J. Crystal Growth. 2004, 264, 192− 200. (8) Monte, J. S. M.; Santos, L. M.; Fulem, M.; Fonseca, J. M.; Sousa, C. New Static Apparatus and Vapor Pressure of Reference Materials: Naphthalene, Benzoic Acid, Benzophenone, and Ferrocene. J. Chem. Eng. Data 2006, 51, 757−766. (9) Weir, R. D.; de Loos, Th. W. Measurement of the Thermodynamic Properties of Multiple Phases; IUPAC, Physical Chemistry Division, Commission on Thermodynamics, Elsevier: The Netherlands, 2005. (10) Fulem, M.; Růzǐ čka, V. Personal Communication, Institute of Chemical Technology, Prague, 2009. (11) Virtual Materials Group Inc VMG. VMGSim Version 5.0.5, VMGSim User’s Manual; Calgary, Alberta, 2010. (12) Sinke, G. C. A Method for Measurement of Vapour Pressures of Organic Compounds Below 0.1 Torr. Naphthalene as a Reference Substance. J. Chem. Thermodyn. 1974, 6. (13) Rúzǐ čka, K.; Fulem, M.; Rúzǐ čka, V. Recommended Vapor Pressure of Solid Naphthalene. J. Chem. Eng. Data 2005, 50, 1956− 1970. (14) Castellanos-Díaz, O.; Schoeggl, F.; Yarranton, H. W.; Satyro, M. A.; Lovestead, T. M.; Bruno, T. J. Modeling the Vapour Pressure of Biodiesel Fuels. Proceedings of the International Conference of Chemical Engineering and Technology, Tokyo, Japan, May 29−31, 2012. (15) Castellanos-Díaz, O.; Schoeggl, F.; Yarranton, H. W.; Satyro, M. A. Heavy Oil Deep Vacuum Fractionation. Proceedings of the XIII Petrophase conference, Saint Petersburg, FL, June 10−14, 2012. 3035
dx.doi.org/10.1021/ie303397y | Ind. Eng. Chem. Res. 2013, 52, 3027−3035