You are on page 1of 15

Article

pubs.acs.org/jced

Reference Quality Vapor−Liquid Equilibrium Data for the Binary


Systems Methane + Ethane, + Propane, + Butane, and +
2‑Methylpropane, at Temperatures from (203 to 273) K and
Pressures to 9 MPa
Eric F. May,*,† Jerry Y. Guo,† Jordan H. Oakley,† Thomas J. Hughes,† Brendan F. Graham,†
Kenneth N. Marsh,† and Stanley H. Huang‡

Centre for Energy, School of Mechanical & Chemical Engineering, University of Western Australia, Crawley, WA 6009, Australia

Chevron Energy Technology Company, Houston Texas 77002, United States

ABSTRACT: A specialized cell designed for vapor liquid


equilibrium (VLE) measurements at cryogenic temperatures
and high pressures was used to measure new (p,T,x,y) data
for binary mixtures of methane + ethane, + propane, +
2-methylpropane (isobutane), and + butane from (203 to 273)
K at pressures up to 9 MPa. A literature review of VLE data for
these binary mixtures indicates that a significant number have
large uncertainties; however, because estimates of uncertainties
in measured phase compositions are often not quantified
sufficiently, it can be difficult for equation of state (EOS)
developers to identify which data sets are of poor quality.
Robust quantitative uncertainties were estimated for the VLE
data acquired in this work, which allowed the identification of
literature data sets that should not be included in future EOS development. The new data were compared with the predictions
of the Peng−Robinson (PR) EOS and the Groupe European de Recherche Gaziere (GERG-2008) multiparameter EOS. The
former describes the new data measured at low pressures within experimental uncertainty but deviates systematically from the
data as the bubble point pressure is increased; for the binary mixtures containing either of the butanes, the maximum relative
deviation of the data from the PR EOS amounted to nearly 10 % of the methane liquid mole fraction. The GERG-2008 EOS was
better able to describe the new high-pressure data for the CH4 + C3H8 system than the PR EOS. However, for both the CH4 +
C4H10 mixtures, the GERG EOS deviated from these data by an amount twice as large as the PR EOS because of ambiguity about
which VLE literature data sets should be used in model development. The new data resolve these ambiguities and should
facilitate the development of improved EOS as needed, for example, in simulations of low temperature natural gas separation
processes.

■ INTRODUCTION
Significant capital expenditure and operating costs are involved
(223 to 303) K. Furthermore, the equation of state (EOS) used
to calculate this multicomponent VLE should be computation-
in the production of liquefied natural gas (LNG). Simulations ally efficient because scrub column simulations also require the
of these facilities have the potential to reduce the total cost of iterative solution of nonlinear equations describing the material
ownership if they are sufficiently optimized. A crucial operation and energy balances on each of the trays within it.
within an LNG processing train is the cryogenic distillation Currently, the thermodynamic model used most commonly
column known as the scrub column, or demethanizer. The for simulating processes such as LNG scrub columns is the
column’s functions are (i) to control the concentration of heavier Peng−Robinson equation of state (PR EOS).1 Its use is common
hydrocarbons (propane and above) in the vapor overheads for two reasons: it is cubic in the molar volume and can thus be
product, which goes on to the main cryogenic heat exchanger solved without iteration, and its predictions of VLE for multi-
for conversion to LNG, and (ii) maximize the recovery of component mixtures are equivalent in accuracy to those of more
hydrocarbon liquids in the bottoms product, which can be an complex EOS that require an iterative solution.2 However, cubic
important source of revenue and/or the refrigerant used in the
liquefaction process. The scrub column is one of the most
Special Issue: Memorial Issue in Honor of Anthony R. H. Goodwin
difficult unit operations in an LNG plant to simulate accurately
because it requires calculation of vapor−liquid equilibrium Received: July 17, 2015
(VLE) for multicomponent mixtures at high pressures from Accepted: September 14, 2015
about (4 to 6) MPa and over a wide range of temperatures from Published: September 29, 2015

© 2015 American Chemical Society 3606 DOI: 10.1021/acs.jced.5b00610


J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

EOS have a well-known deficiency when it comes to the


prediction of liquid volumes, which more complex models3 have
been developed to address. The Groupe European de Recherche
Gaziere 2008 (GERG-2008)4 multiparameter EOS is the
reference model for the prediction of natural gas properties.
For VLE predictions within its stated range of validity, the
claimed relative uncertainties are generally between (1 and 5) %.
Kunz and Wagner4 stated that the paucity of quality VLE data
available for mixtures limited the accuracy achievable in the
development of the GERG-2008 EOS, and that VLE data at low
temperatures for mixtures of CH4 + C4H10 would be particularly
useful for improving the description of richer natural gases.
The first objective of this work was to review the available
literature VLE data for the key binary mixtures relevant to
LNG scrub column simulations, which are those involving
methane (CH4) and one of [ethane (C2H6), propane (C3H8),
isobutane (iC4H10), or butane (nC4H10)], and compare them
with the predictions of the cubic EOS most commonly used in
those simulations. As discussed below, the common feature of
these comparisons is that the cubic EOS is tuned as well as it
can be to the constituent binary systems given (a) the significant
scatter present in the literature VLE data and (b) the decreasing
ability of cubic EOS to describe VLE accurately as the mixture
critical point is approached.5,6 In contrast, the more complex
functional forms of multiparameter EOS means they are able
to better capture the curvature of the phase envelope even quite
close to the critical point. However, in some ways this additional
flexibility means that identifying less accurate data sets is even
more important for the development of multiparameter EOS.
Given the significant scatter found in the literature VLE data,
we report here the development of a specialized apparatus
to produce reference quality VLE data at low temperature,
Figure 1. (a) Measured and predicted (curves) bubble-pressures for
high-pressure conditions spanning those found in LNG scrub methane (1) + ethane (2) as a function of the measured liquid mole fraction
columns. The analytic method was used to produce new VLE of methane, x1: green +, Gupta et al.10 T = 270 K; red ○, Wei et al.11 T = 270
data including quantitative uncertainty estimates for each of K; blue □, Raabe et al.12 T = 270 K; green △, Janisch et al.13 T = 270 K; ⧫,
the four key binary mixtures. The data are compared with the Davalos et al.14 T = 250 K; red ●, Wei et al.11 T = 250 K. (b) Deviations
existing literature and clearly allow the identification of data of the measured methane liquid mole fractions, x1, in panel (a) from the
sets that should not be included in EOS development, as well as corresponding calculated methane liquid mole fraction, x1,calc, of the default
suggesting those data sets that should receive an increased HYSYS PR EOS as a function of x1.
weighting. To compare the quality of many data sets acquired
over a wide range of conditions, deviation plots are used with the better at one temperature or another. Figure 1b overcomes these
baseline being the default PR EOS implemented in the software limitations by showing the deviations of the measured (methane)
Aspen HYSYS7 because of its widespread use in the simulation liquid mole fraction, x1 from the EOS prediction, x1,calc: it is clear
of LNG scrub columns. Comparisons of the data with the that the different data sets at both temperatures are generally
GERG-2008 EOS are also made and reveal pathways for future consistent within a reasonable estimate of their uncertainties; that
improvements of this and similar models.


the EOS deviates systematically from the data with increasing
methane liquid mole fraction, x1; and that the slope of this
REVIEW OF LITERATURE VLE DATA systematic deviation is larger at 270 K than at 250 K.
The conventional and convenient way of presenting VLE data for The results of the comparisons between literature VLE data
binary mixtures and comparing them with EOS predictions is and the default HYSYS PR EOS predictions for the principal
through a pressure−composition (p, x) plot. However, while binary systems containing methane are presented in Figure 2 to
(p, x) plots are helpful for providing a global overview of a binary Figure 5: these systems contain the information about the dominant
system’s VLE and its description by a thermodynamic model, binary interactions occurring in natural gas and LNG and, thus, are
they are limited by the large scale of the pressure axis needed to the most important for scrub column simulations. The focus here is
represent the normally wide range of bubble-point conditions. on the bubble-point curve because for these systems it appears that
An example of this is shown for the methane (1) + ethane (2) the dominant deficiency in EOS predictions occurs for the liquid
system in Figure 1a, which presents a (p, x) plot that contains phase in terms of absolute mole fraction deviations. The deviations
several data sets measured at 250 and 270 K together with the of EOS predictions for the vapor and liquid phases are similar
corresponding predictions of the default PR EOS implemented on a relative basis but because the uncertainties of experimental
in AspenTech HYSYS.7 While this (p, x) plot indicates a composition measurements are often constrained to have a
reasonable agreement between the measurements and the model minimum absolute value (e.g., 0.001 mole fraction), the deviations
predictions it is difficult to assess the consistency of the different of EOS predictions from the vapor phase data are usually closer to
data sets or whether the agreement of the model with the data is this minimum experimental uncertainty.
3607 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

Figure 2. Deviations of measured CH4 liquid mole fractions (x1) from


the values calculated with the default HYSYS PR EOS (x1,calc) for
CH4 (1) + C2H6 (2) as a function x1: red ◊, Wichterle15 190 < T/K <
200; blue ×, Wichterle and Kobayashi,8,9 130 < T/K < 213; green ○,
Wilson16 T = 111 K; +, Davalos et al.14 T = 250 K; purple □, Miller
et al.17 160 < T/K < 180; brown △, Gupta et al.10 260 < T/K < 280;
orange ∗, Wei et al.11 210 < T/K < 270; red block cross, Raabe et al.12
240 < T/K < 270; ▽, Janisch et al.13 140 < T/K < 270.

Figure 2 shows a liquid phase deviation plot for the CH4 (1) +
C2H6 (2) binary for 330 literature VLE data measured between
110 K and 280 K. The root mean square (rms) deviation for x1 in
Figure 2 is 0.007, and the distribution of the deviations does not
exhibit any clear trend with composition. These results suggest
that the default PR EOS is adequate over this range of conditions.
Potentially some improvement could be achieved by filtering
out the 1972 data of Wichterle and Kobayashi,8,9 which exhibit Figure 3. (a) Select bubble pressures for the methane (1) + propane (3)
significantly more scatter than the other data sets. system as a function of the measured liquid mole fraction of methane x1:
Figure 3 shows a comparison of measured and predicted ◊, Wichterle and Kobayashi24 T = 192 K; ◊, Wichterle and Kobayashi24
methane liquid mole fractions x1 for CH4 (1) + C3H8 (3): the T = 214 K; ∗, Kalra et al.25 T = 214 K; ○, Webster et al.26 T = 230 K; ×,
deviation plot, Figure 3b, contains 429 literature data measured Benham et al.27 T = 255 K; □, Price et al.28 T = 255 K; +, Reamer et al.29
between 130 K and 344 K. The rms deviation for x1 in Figure 3b T = 278 K; △, Wiese et al.30 T = 278 K; △, Wiese et al.30 T = 311 K; △,
is 0.015, and the distribution of the deviations exhibits an increas- Wiese et al.30 T = 344 K. (b) Deviations of measured CH4 liquid mole
ing positive trend with increasing liquid methane mole fraction. fractions (x1) from values calculated with the default HYSYS PR EOS
Figure 4 shows a comparison of measured and predicted liquid (x1,calc) for CH4 (1) + C3H8 (3) as a function of x1: red ◊, Reamer et al.29
compositions for CH4 (1) + iC4H10 (4): the deviation plot in 278 < T/K < 294; blue ×, Akers et al.31 174 < T/K < 273; green ○,
Benham and Katz27 200 < T/K < 255; +, Price et al.28 144 < T/K < 283;
Figure 4b contains 145 literature data measured between 110 K
purple □, Wichterle15 130 < T/K < 213; brown △, Wiese and
and 378 K. The majority (94 points) of the literature VLE Kobayashi30 278 < T/K < 344; orange ∗, Wichterle and Kobayashi24
data for this binary system at conditions relevant to LNG scrub 130 < T/K < 213; red ▷, Kalra and Robinson25 T = 214 K; ▽, Joffe32
columns comes from a single source,18 whose measurements 277 < T/K < 344; green block x, Raimondi33 T = 244 K; black block
range from (198 to 378) K. The only other two literature VLE cross, Webster and Kidnay26 230 < T/K < 270.
data sets available for this system are of less relevance having been
measured at high temperatures19 from (311 to 377) K or very
low temperatures20 from (110 to 140) K. The rms deviation for Figure 3 to Figure 5 highlight the commonality of both the
x1 in Figure 4b is 0.025 and the distribution of the deviations also limitations in the literature VLE data quality and the inability of
exhibits an increasing positive trend with increasing x1. the cubic PR EOS to represent these binary mixtures at high mole
Figure 5 shows a comparison of measured and predicted fractions of CH4, which, for a given temperature, correspond to
higher pressures on the bubble point curve.


methane liquid mole fractions for CH4 (1) + nC4H10 (5): the
deviation plot in Figure 5b contains 416 literature data measured
between 138 K and 411 K. The rms deviation for x1 in Figure 5b EXPERIMENTAL SECTION
is 0.031. The isothermal data shown in Figure 5a are a small Specialized Cryogenic VLE Apparatus. An experimental
subset of those in Figure 5b; they were selected because they schematic of the specialized cryogenic VLE apparatus is shown
illustrate key problems with data quality, with cubic EOS, and schematically in Figure 6. While similar in concept to the VLE
importantly, occur at temperatures near those used in LNG scrub apparatus used by Kandil et al.39 and by Hughes et al.,40,41
columns. At 244 K there is a very large discrepancy between the substantial improvements in the ability to control and manipulate
data of Roberts et al.21 and Wang and McKetta22 and those of both system temperature and the acquired samples were
Elliot et al.23 The HYSYS default PR EOS appears to be tuned to implemented to significantly reduce experimental uncertainty.
the latter data set at lower pressures but transitions to the former An equilibrium cell (EC) machined from a single billet of
data sets as the critical point is approached. stainless steel grade 316 served as the pressure vessel with a
3608 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

Figure 4. (a) Bubble pressures for the methane (1) + isobutane (4) Figure 5. (a) Select bubble pressures for the methane (1) + butane (5)
system as a function of the measured liquid mole fraction of methane, system as a function of the measured liquid mole fraction of methane, x1:
x1: × (all colors), Barsuk et al.18 (temperatures indicated on figure); ◊, Roberts et al.21 T = 211 K; +, Wang et al.22 T = 211 K; ∗, Kahre34 T =
―, bubble curves calculated with the HYSYS PR EOS. (b) Deviations 211 K; ○, Elliot et al.23 T = 211 K; ◊, Roberts et al.21 T = 244 K; +, Wang
of measured CH4 mole fractions (x1) from the values calculated with the et al.22 T = 244 K; ○, Elliot et al.23 T = 244 K; ◊, Roberts et al.21 T =
default HYSYS PR EOS (x1,calc) for the CH4 (1) + iC4H10 (4) binary as a 278 K; +, Wang et al.22 T = 278 K; ○, Elliot et al.23 T = 278 K; ―,
function of x1: red ▲, Olds et al.19 311 < T/K < 378; blue ×, Barsuk bubble curves calculated with the HYSYS PR EOS. (b) Deviations of
et al.18 198 < T/K < 378; green □, Haynes20 110 < T/K < 140. measured CH4 mole fractions (x1) from the values calculated with the
default HYSYS PR EOS (x1,calc) for the CH4 (1) + nC4H10 (5) binary as a
maximum operating pressure of 30 MPa. The cell had an internal function of x1: red ◊, Nederbragt35 252 < T/K < 316; blue ×, Sage
diameter of 3 cm and a volume of approximately 60 cm3. The et al.36,37 294 < T/K < 394; green ○, Rigas et al.38 T = 311 K; +, Roberts
et al.21 211 < T/K < 411; purple □, Wang et al.22 178 < T/K < 378;
outer surface of the cell was plated with 1 mm thick copper to brown △, Elliot et al.23 144 < T/K < 278; orange ∗, Kahre34 166 < T/K
improve heat transfer and temperature uniformity. A foil-type < 283; ▽, Raimondi33 138 < T/K < 310.
heating element was wrapped and glued to the outer surface
of the cell using high thermal conductivity epoxy suitable for
cryogenic operation. A 100 Ω platinum resistance thermometer on the bottom surface. In this way the stirrer motor was used to
(PRT) was glued to the cell’s external surface directly underneath mix the sample fluid. The lid of the cell was machined carefully
the heating foil, and was used as the sensor for temperature to fit a custom, cryogenically compatible fill valve (V1), with a
control (TC). Wells were bored in the top and the bottom of nonrotating stem that was flush with the inner surface of the cell
the equilibrium cell to house two 100 Ω PRTs (T1) and (T2). lid when closed. This minimized any dead volume associated
All three PRTs were calibrated over the temperature range from with the fill valve. This fill valve was operated by a stepper motor
(77 to 300) K with an uncertainty of u(T) = 0.02 K prior to (M2) (Phytron UHVC-80). A pressure transducer (P1) (Kulite
mounting in the EC. However, when mounted in the cell the model CT-190) was also housed in the lid to minimize associated
uncertainty of the temperature readings increased as discussed dead volume. This transducer utilized a strain-gauge on a silicon
below. The normal operating temperature profile of the diaphragm and was suitable for operation at temperatures from
apparatus established during the acquisition of samples for a (77 to 393) K. It was calibrated in situ by comparison with a
VLE measurement, as recorded by each thermometer shown in reference quartz-crystal pressure transducer (P2) (Paroscientific
Figure 6 relative to the set point of the TC, is listed in Table 1. Digiquartz model 1002K-01), located outside the Dewar at
An attachment to the bottom of the cell was used to mount a ambient conditions, with a full scale of 14 MPa and a relative
cryogenically compatible, variable speed motor (stirrer motor). uncertainty of 0.01 % of full scale as stated by the manufacturer.
The motor was used to generate a rotating magnetic field, which The manufacturer stated the Kulite transducer could be used
in turn drove a Teflon-coated magnetic bar sitting inside the cell to determine pressure with a relative uncertainty of 0.5 % for
3609 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

Figure 6. Schematic of the specialized cryogenic VLE apparatus, which had improved temperature control and sampling systems relative to the setup
described by Kandil et al.39 Symbols are explained in the text and Table 1.

Table 1. Thermal Profile and Control of the Specialized (LC) was used to sample the liquid phase; it extended nearly to
Cryogenic VLE Apparatus Established During the Acquisition the bottom of the cell and had a total length of 20 cm (3.5 μL
of Samples for a VLE Measurement, as Recorded by each volume). Although it was not measured, the volume of the liquid
Thermometer Relative to the Set Point of the TC phase inside the VLE cell was estimated from an EOS and ranged
normally from a minimum of 1.5 cm3 up to about 20 cm3. The
control set point
PRT relative to TC functionality other end of each capillary sampling tube was located inside a
TC controls cell temperature specialized “Rapid On-Line Sampler Injector” (ROLSI) electro-
T1 not controlled monitors cell bottom temperature magnetic solenoid valve supplied by TransValor.42 The two
T2 not controlled monitors cell top temperature ROLSI sampling valves (VV, VL) were mounted on the top side
TL −5 K controls liquid capillary temperature of a steel plate approximately 5 cm above the top of the VLE cell
TV +5 K controls vapor capillary temperature lid. Cartridge heaters were used to control the temperature of the
TRC 0K controls temperature of ROLSI valves ROLSI valves, and 2 PRTs (TRC and TRS) were used as control
TRS not controlled monitors temperature of ROLSI valves and reference sensors. The temperatures of the two capillaries
TCu −2 K controls copper can temperature were controlled independently; this was found to be essential for
TCuB not controlled monitors copper can temperature obtaining representative samples of the equilibrium phases in the
TS1 0K controls temperature of GC transfer VLE cell. The capillaries were mounted on custom built copper
line: section 1 plates with grooves machined to seat the capillaries. A Peltier
TS2 0K controls temperature of GC transfer element was glued using epoxy to the side of each copper plate
line: section 2
and could be used to heat or cool the capillaries as required.
TS3 0K controls temperature of GC transfer
line: section 3 A 100 Ω PRT was attached to the surface of each capillary
LN2 sensor −2 K controls delivery of liquid nitrogen into (TL, TV) and together with the Peltier elements they were used
the Dewar in a proportional-integral (PI) loop implemented in software
to control the capillary temperatures. The vapor capillary was
the pressure range from (1 to 14) MPa. The transducer was, heated to a temperature of 5 K higher than the VLE cell, and the
however, sensitive to its local temperature and its excitation liquid capillary was cooled to a temperature of 5 K lower than the
voltage; accordingly the latter was held constant throughout the VLE cell.
experiments and the temperature-dependent calibration function The cell and ROLSI sampling valves were enclosed inside a
was checked regularly as described below. sealed copper can, which could be evacuated or filled with helium;
Two capillary tubes made from Monel 400 with internal the latter was most commonly employed because it assisted with
diameters of less than 0.015 cm were also mounted in the cell lid. achieving temperature uniformity across the EC. A foil-type
One of the capillaries (VC) was used to sample the vapor phase heating element that could be driven with up to 150 W, was
in the cell; it extended 1.5 cm below the bottom of the cell lid and used to control the temperature of the copper can, which was
had a total length of 13 cm (2.3 μL volume). The other capillary monitored by a control PRT (TCu) as well as a sensing PRT
3610 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

(TCuB). The copper can was enclosed inside a reflective radiation opening time. The specific opening times used when acquiring
shield, which separated it from a sealed stainless steel (SS) can. samples varied slightly depending on the cell pressure and the
The inside of the SS can was evacuated to maximize the thermal phase being sampled but generally an opening time of 0.05 s was
isolation between the copper can and the inner wall of the SS can. used to acquire a sufficient amount of sample that did not
The SS can was placed inside a cryogenic Dewar equipped with an saturate the GC column or its detectors. The temperatures of the
automatic liquid nitrogen dosing pump that maintained a sampling valves and of the transfer lines were set using the
constant level of liquid N2 in the bottom of the Dewar; usually ROLSI control box. To avoid inadvertent condensation of heavy
the outside of the SS can was only in contact with the N2 boil-off components in the transfer lines, each of the three sections of the
vapor. transfer lines were thermally controlled using 100 Ω PRTs (TS1,
A helium carrier gas line (shown in red) was also connected to TS2, and TS3) and wire heaters.
each of the ROLSI sampling valves, and helium (BOC, mole The GC used was a Varian model CP-3800 equipped with two
fraction purity 0.99999) flowed continuously through each of the capillary columns and two flame ionization detectors (FID): one
valves and into their respective gas chromatograph (GC) (FID V) for samples from the vapor phase inside the EC and the
columns. When the valves were actuated, the carrier gas would other (FID L) for samples from the liquid phase inside the EC.
ensure that the samples from the vapor and liquid phases within Further details of the instrument and method used to analyze
the cell were swept along heated transfer lines into the GC samples taken from the EC are listed in Table 2.
columns. The ROLSI sampling valves were actuated using a Uncertainty Estimates. The VLE apparatus produces
control box (VSC), which allowed specification of the valve measurements of four quantities: temperature, pressure, and
opening time with a resolution of 0.01 s. Thus, for a given the liquid and vapor phase compositions. Thus, it is essential that
pressure in the VLE cell, the amount of sample withdrawn by the transducers used to determine these four quantities are
opening the valves could be adjusted by varying the specified calibrated reliably and that robust estimates of the uncertainties
in these measured quantities are made. Each of the mixture VLE
experiments lasted several weeks and involved multiple cycles of
Table 2. Details of the Varian CP-3800 Gas Chromatograph
temperature and pressure, which can often cause inadvertent
and Method Used for Sample Analysis
drifts in the calibrations of the thermometers and pressure gauge.
column length and diameter 25 m, 0.53 mm Accordingly, before and after a set of mixture experiments made
column head pressure (constant) 83 kPa with the VLE cell, in situ calibrations (or validations) of the
column packing PoroPlot Q Kulite pressure transducer and the PRTs TC, T1 and T2 were
injection temperature 473 K conducted using pure fluids.
injection split ratio 10:1 When excited with a 12 V DC power source, the Kulite
initial and final oven temperatures 323 K, 473 K pressure transducer, P1, produced a voltage output that varied
oven temperature ramp rate 20 K· min−1 linearly with the system pressure. However, both the slope and
FID temperature 473 K offset of this linear function varied with temperature (as well as

Table 3. Details of the Mixtures Prepared Gravimetrically in This Work, Including the Purity of the Component Fluids and
Standard Uncertainties in the Mixture Component Mole Fractionsa
CH4 C2H6 C3H8 iC4H10 nC4H10
source BOC Coregas Air Liquide Coregas Coregas
grade UHP 4.0 N35 N35 N35
mole fraction purity 0.99999 0.9999 0.9995 0.9995 0.9995

Gravimetric Mixture 1: CH4 + C2H6 (Estimated Cricondentherm Temperature 265.7 K)


masses added to evacuated cylinder/g 21.309 37.113
component mole fraction 0.5183 0.4817
component mole fraction uncertainty 0.0008 0.0008

Gravimetric Mixture 2: CH4 + C3H8 (Estimated Cricondentherm Temperature 287.6 K)


masses added to evacuated cylinder/g 29.661 23.731
component mole fraction 0.7745 0.2255
component mole fraction uncertainty 0.0007 0.0007

Gravimetric Mixture 3: CH4 + iC4H10 (Estimated Cricondentherm Temperature 284.4 K)


masses added to evacuated cylinder/g 46.744 15.851
component mole fraction 0.9144 0.0856
component mole fraction uncertainty 0.0007 0.0007

Gravimetric Mixture 4: CH4 +nC4H10 (Estimated Cricondentherm Temperature 287.5 K)


masses added to evacuated cylinder/g 24.993 6.259
component mole fraction 0.9354 0.0646
component mole fraction uncertainty 0.0012 0.0012
a
Mixtures were prepared in high-pressure cylinders that had masses of about 1030 g when evacuated. A stainless steel ball from a ball bearing was
placed inside the cylinder and used to ensure the sample was mixed.

3611 DOI: 10.1021/acs.jced.5b00610


J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

Table 4. Relative Response Factors of the Two Flame the precise excitation voltage used). During mixture experiments
Ionization Detectors (FID) Used to Convert the Ratios of with V1 closed, the cell pressure was determined from the Kulite
Integrated Detector Response Areas into Mole Fraction transducers voltage output, V, and the measured temperature T,
Compositions Using eq 3 for Components CH4 (1), C2H6 (2), using
C3H8 (3), iC4H10 (4), and nC4H10 (5)
p = (w0 + w1·T ) ·(V + w2 + w3·T ) (1)
FID-L FID-V
κ2/κ1 2.031 ± 0.035 2.003 ± 0.033 The parameters w0, w1, w2, and w3 were determined by calibration
κ3/κ1 2.957 ± 0.008 2.975 ± 0.035 against the reference Paroscientific Digiquartz pressure trans-
κ4/κ1 3.906 ± 0.078 3.843 ± 0.023 ducer, P2, at the temperatures 203 K, 243 K and 303 K using
κ5/κ1 3.879 ± 0.071 3.870 ± 0.098 methane or argon from vacuum to 13 MPa. The standard
uncertainty of P2 over this range was u(P) = 2 kPa based on the
manufacturer’s specifications. The rms deviation of the Kulite
pressures calculated using eq 1 from the pressures recorded by
P2 during the calibration was 12 kPa, and in subsequent checks
of P1 against P2, the variation remained within this amount.
The standard uncertainty of the pressures reported here is thus
estimated to be 0.02 MPa. (Over the course of these measure-
ments, which lasted about 9 months, the stability of the reference
transducer P2 was checked by monitoring its reading when under
vacuum. The vacuum reading obtained was always consistent
within the specified uncertainty of P2.)
The PRTs used in the VLE apparatus were all Class A and
measured using the four-wire method. Prior to their installation
in the apparatus, they were all compared against a reference PRT
in a liquid bath from (245 to 323) K, and at 77 K using liquid
nitrogen. The measured resistances of each PRT were converted
to a temperature using a quadratic function, the parameters
of which were determined for each PRT using this ex situ
Figure 7. Isochoric and isothermal measurement pathways used for the calibration. Once installed in the apparatus, the PRT temperature
acquisition of VLE data reported in this work, illustrated by the example readings were checked for drift by filling the cell with pure ethane
of the experiment with the binary system CH4 + iC4H10: □, isochoric
and varying the cell temperature between 203 K and 283 K. Pure
pathway (including the initial single-phase condition at which the cell
was loaded); ■, isothermal pathway. The solid curve represents the ethane is two-phase at these temperatures and from measure-
predicted phase envelope for the gravimetric composition. The dashed ments of its vapor pressure, the corresponding saturation temp-
curve represents the predicted phase envelope for the leanest eratures could be calculated and compared with the temperatures
composition. The bubble point curves are shown in blue and the dew readings from the PRTs. An initial in situ calibration with the
point curves are shown in red. cell heater switched off and only the copper can temperature

Table 5. Measured (p,T,x,y) Data for the CH4 (1) + C2H6 (2) Binary Mixturea
T/K p/kPa x1 u(x1) uc(x1) y2 u(y2) uc(y2)
Isochoric Path
243.58 3949 0.3099 0.0028 0.0037 0.3336 0.0033 0.0038
203.22 2124 0.3653 0.0048 0.0064 0.1333 0.0017 0.0023
213.39 2517 0.3464 0.0039 0.0052 0.1711 0.0020 0.0025
223.50 2945 0.3315 0.0034 0.0045 0.2195 0.0022 0.0028
233.56 3417 0.3188 0.0030 0.0039 0.2755 0.0018 0.0025
Isothermal Path
243.60 3942 0.3082 0.0028 0.0037 0.3340 0.0026 0.0031
243.60 4831 0.3973 0.0031 0.0039 0.2954 0.0022 0.0032
243.60 5395 0.4553 0.0031 0.0040 0.2788 0.0036 0.0038
243.61 6080 0.5273 0.0033 0.0042 0.2778 0.0021 0.0024
243.61 6469 0.5673 0.0040 0.0049 0.2731 0.0029 0.0031
243.61 6691 0.5957 0.0044 0.0055 0.2812 0.0027 0.0029
243.60 6885 0.6218 0.0126 0.0133 0.2877 0.0024 0.0028
243.60 6487 0.5754 0.0039 0.0049 0.2801 0.0022 0.0025
243.60 5564 0.4741 0.0034 0.0042 0.2745 0.0013 0.0018
243.60 5094 0.4255 0.0032 0.0040 0.2856 0.0024 0.0028
243.61 3718 0.2865 0.0025 0.0034 0.3421 0.0021 0.0028
243.60 3675 0.2812 0.0032 0.0040 0.3490 0.0040 0.0045

a
The uncertainties, u, for the measured mole fractions refer to the Type A standard uncertainties associated with sampling and detector calibration.
The uncertainties of the temperature and pressure measurements were 0.15 K and 20 kPa, respectively. These were combined in quadrature with the
sampling and detector uncertainties to calculate a combined standard uncertainty, uc.

3612 DOI: 10.1021/acs.jced.5b00610


J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

controlled resulted in the cell temperature readings from T1 and


T2 being consistent with the calculated ethane saturation
temperatures to within 0.1 K. However, when the cell heater
was used to control TC the difference in the values of T1 and T2
was observed to increase. The standard deviations of the T1 and
T2 readings from the ethane saturation temperature when the
copper can temperature was controlled were both less than
0.12 K. The temperature of the VLE measurements reported
here corresponds to the average of the T1 and T2 readings
and the corresponding temperature uncertainty was estimated to
be 0.15 K.
The uncertainty in the flame ionization detector (FID) res-
ponses dominates the uncertainty of phase equilibrium measure-
ments and, accordingly, significant efforts were made to calibrate
and reduce the uncertainty associated with the relative response
factors of the two FIDs. Mole fraction compositions were
determined from ratio measurements of the FID’s responses to
each component. Within the linear range of the detector, the
integrated detector response, Ai, of an FID to component i is
proportional to the number of moles of that component ni at the
detector.
Ai = κini (2)
39
Kandil et al. determined that the linear range of the FID
detectors used in this work corresponded to the equivalent of
A1 < 1.5 × 108 counts, and that κ1 ≈ 3.6 × 1013 counts per mole.
These values are useful for estimating whether a given sample
was, in absolute terms, too small (e.g., only representative of fluid
trapped in the sampling capillary rather than from the mixed
region of the EC), or too large (exceeding the linearity range of
the FID). However, assuming these constraints are satisfied, the
mole fraction composition, zi, of a given N component mixture Figure 8. VLE data measured in this work, with selected literature data,
analyzed with the FID can be determined without reference to for CH4 (1) + C2H6 (2). (a) Deviations of CH4 liquid mole fractions
the ni from the measured ratios (Ai/Aj) by solving the system of (x1) from values calculated with the PR EOS (x1, calc PR) as a function of
equations: x1. (b) Deviations of the C2H6 vapor mole fractions (y2) from values
calculated with the PR EOS (y2, calc PR) as a function of the y2. The values
⎛ zi ⎞ ⎛ Ai ⎞⎛ κ1 ⎞ of x1, calc PR and y2, calc PR were calculated at the experimental (p, T).
⎜ ⎟ = ⎜ ⎟⎜ ⎟ Symbols: blue ⧫, 244 K, this work; blue ◊, isochore (203 K to 244 K),
⎝ z1 ⎠ ⎝ A1 ⎠⎝ κi ⎠ (3a) this work; ∗, 250 K, Davalos et al.;14 ○, 270 K, Gupta et al.;10 △, 250 K,
Wei et al.;11 □, 270 K, Wei et al.;11 red ◊, 270 K, Raabe et al.;12 ×, 270 K,
⎛ ⎛ Ai ⎞⎛ κ1 ⎞⎞
N −1 Janisch et al.13 Curves: − − − , GERG-2008 EOS at (p,T) measured in
z1 = ⎜⎜1 + ∑ ⎜ ⎟⎜ ⎟⎟⎟ this work on the 244 K isotherm; ― GERG-2008 EOS at (p,T)
⎝ i=2
⎝ A1 ⎠⎝ κi ⎠⎠ (3b) measured in this work along the isochore for (203 to 244) K.

where i = 1 would normally be taken as the mixture component purities of the component fluids as well as the details of the
with the largest integrated detector response (methane in this gravimetric mixtures prepared in this work. There were three
work). The relative response factors of the FIDs, (κi /κ1), and main contributions to the mole fraction uncertainties of the
their uncertainties can be determined using a standard gas gravimetric mixtures: weighing, pure component impurities,
mixture with known mole fractions, usually from gravimetric and valve dead volume. The combination of these uncertainties
preparation. The mole fraction uncertainties for the measured in quadrature was used to estimate the uncertainties of the
phase compositions can then be estimated by propagating gravimetric mixture mole fractions which are reported in Table 3.
through eq 3 the uncertainties in the (κi /κ1) arising from the FID The relative response factors, (κi /κ1), for each of the FIDs are
calibration together with the statistical uncertainty arising from listed together with their uncertainties in Table 4; as described
repeat measurements of the area ratios (Ai/A1). in the following section, these response factors and their
Gravimetric Preparation of Gas Mixtures. Single-phase uncertainties were determined first by sampling gravimetric
gas mixtures were prepared gravimetrically for two reasons: as mixtures 1 to 4 while in a single-phase condition at a (p,T) far
part of the determination of the FID relative response factors, from the mixture’s dew point, and then at a (p,T) condition
and to facilitate the initial loading of the equilibrium cell with a inside the two-phase region where the detector’s sensitivity was
mixture of known composition. To prepare a mixture gravi- a maximum.
metrically, a 300 cm3 cylinder was evacuated, and then the com- Method. For each experiment, a gravimetric mixture was pre-
ponents were added into the cylinder in the order of increasing pared and loaded into the cell and controlled at 298 K or 303 K,
vapor pressure. With each step, the total mass was determined at least 10 K above the mixture’s estimated cricondentherm. This
with an electronic balance with a resolution and standard was done to obtain (1) an estimate of the single-phase fluid den-
uncertainty of 0.001 g and a full scale of 1100 g. Table 3 lists the sity using the GERG EOS,3 and (2) composition measurements
3613 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

Table 6. Measured (p,T,x,y) Data for the CH4 (1) + C3H8 (3) Binary Mixturea
T/K p/kPa x1 u(x1) uc(x1) y3 u(y3) uc(y3)
Isochoric Path
283.38 7630 0.4586 0.0010 0.0018 0.2045 0.0006 0.0009
273.48 7102 0.4589 0.0010 0.0018 0.1614 0.0003 0.0007
263.51 6597 0.4623 0.0007 0.0018 0.1197 0.0012 0.0013
263.56 6560 0.4601 0.0013 0.0021 0.1174 0.0018 0.0019
253.59 6100 0.4708 0.0011 0.0021 0.0947 0.0004 0.0006
243.61 5611 0.4857 0.0015 0.0024 0.0715 0.0001 0.0004
243.64 5635 0.4907 0.0010 0.0022 0.0719 0.0001 0.0004
Isothermal Path
243.62 5544 0.4832 0.0014 0.0024 0.0715 0.0002 0.0004
233.67 5119 0.5077 0.0012 0.0026 0.0518 0.0003 0.0004
233.61 5077 0.5037 0.0021 0.0031 0.0520 0.0003 0.0004
223.54 4622 0.5333 0.0006 0.0027 0.0368 0.0005 0.0005
223.36 4628 0.5331 0.0012 0.0030 0.0366 0.0002 0.0003
213.38 4108 0.5614 0.0005 0.0033 0.0245 0.0003 0.0003
203.40 3576 0.5952 0.0014 0.0044 0.0162 0.0005 0.0005
243.62 891 0.0710 0.0005 0.0019 0.2076 0.0003 0.0041
243.62 3909 0.3442 0.0003 0.0019 0.0743 0.0004 0.0006
243.63 5500 0.4788 0.0014 0.0024 0.0704 0.0002 0.0004
243.61 5544 0.4764 0.0015 0.0025 0.0734 0.0002 0.0004
243.62 6402 0.5565 0.0006 0.0022 0.0726 0.0006 0.0007
243.63 6989 0.6082 0.0015 0.0026 0.0768 0.0005 0.0006
243.62 7530 0.6572 0.0010 0.0025 0.0779 0.0016 0.0016
243.62 7943 0.6961 0.0013 0.0027 0.0855 0.0005 0.0007
a
The uncertainties, u, for the measured mole fractions refer to the Type A standard uncertainties associated with sampling and detector calibration.
The uncertainties of the temperature and pressure measurements were 0.15 K and 20 kPa, respectively. These were combined in quadrature with the
sampling and detector uncertainties to calculate a combined standard uncertainty, uc.

of the single-phase fluid to ensure consistency between the flushed with the mixture multiple times to sweep the entire
samples taken from the top and bottom of the cell, and between capillary volume. A typical procedure consisted of directing the
the GC-derived composition and the known gravimetric GC carrier gas into a vented waste stream instead of onto the GC
composition. The experiment then proceeded along an column using rotating valve RV1 in Figure 6. The vapor capillary
isochoric pathway (as shown, for example, in Figure 7) with was then flushed three times with an opening time of 0.25 s
cell temperatures controlled between 203 K and 273 K or 283 K (sweeping a total amount of material about 15 times that which
(depending on the mixture’s phase envelope) in increments would be contained in the vapor capillary). The liquid capillary
of 10 K. During the isochoric phase of the experiment, repeat was flushed three times but with an opening time of 1 s
measurements at several conditions were conducted to provide (sweeping about 60 times the amount of material that would be
a consistency check; this usually included another single-phase contained in the liquid capillary). Then, the GC carrier gas was
measurement which allowed the total change in mass and overall redirected back onto the column before two samples from each
composition caused by sampling to be estimated. Usually, after phase were taken (opening times about 0.05 s) and analyzed. The
about 10 measurements along the isochore (which lasted about carrier gas lines were returned to the vent position and each of
10 days and about the removal of about 1000 samples, including the capillaries was flushed three more times with an opening of
flushing), upon returning to the initial single-phase temperature, 0.25 s. Then two further samples from each phase were taken,
the total cell pressure was about 2 % lower, while the overall analyzed, and compared with the two samples acquired after the
mixture composition had not varied significantly from its original first set of flushes. This procedure was repeated until the samples
value. At the conclusion of the isochoric experiment, measure- taken between the flush runs gave consistent results.
ments were conducted along an isothermal pathway with the Once consistent readings had been achieved between flushing,
apparatus cooled to approximately 244 K. The system pressure all further samples (usually a minimum of six per measurement)
was first raised, in increments of about 1 MPa, through the were analyzed using the GC FIDs. The ratios of the areas from
addition of methane to the mixture. Material balance was used to the integrated FID responses to each of the sample’s component
estimate the change in the overall composition of the mixture so species were calculated and used to determine its mole fraction
that the cell pressure was not increased such that the fluid became composition. A criterion of achieving a relative standard deviation
single-phase. Once the maximum pressure along the isothermal of less than 1 % in each of the measured area ratios (Ai/A1) over
pathway had been reached, the vapor phase of the mixture in the four consecutive runs was set for the measurement to be
cell was vented in steps of about 1 MPa to complete the isotherm considered successful. Often, relative standard deviations of 0.5 %
experiment. or better were achieved, which for an equimolar binary mixture,
At each measurement condition at least 12 h were allowed to corresponds to a mole fraction variation for the four samples of
reach temperature and pressure equilibrium and then no less about 0.0025.
than 2 h of mixing was applied after the temperature stabilized. The magnitude of the (κi /κ1) uncertainties for each FID and
Prior to capturing samples for analysis, the capillaries were their effect on the measured mole fraction compositions were
3614 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

minimized through the use of a rotating valve RV2 shown in


Figure 6, which could direct samples acquired from the top and
bottom of the cell to either of the two detectors, FID-V or FID-L.
In the normal detector configuration, vapor samples were
analyzed with FID-V and liquid samples with FID-L, while in the
alternate detector configuration liquid samples were analyzed
with FID-V and vapor samples with FID-L. Once a successful
measurement had been made in one detector configuration, the
position of valve RV2 was switched and the measurement was
repeated in the other configuration. (This occurred immediately
following the first measurement so no additional flushing
procedure was required.) Any difference in the composition
measurements made using the two different detector config-
urations was then attributable to uncertainty in the values of
(κi /κ1) being used for each FID, with the average of the com-
positions derived using the two configurations having a reduced
uncertainty.
This technique of measuring the same sample using two
different detectors allowed for the values of the (κi /κ1) for
each FID to be fine-tuned and their uncertainties reduced. As
discussed by Kandil et al.,39 the sensitivity of FIDs is a maximum
for equimolar mixtures; however, calibrations of FIDs cannot
usually be performed at this most sensitive condition. This is
because to prepare mixtures with accurately known compositions
gravimetrically usually requires that the heavier components are
present only in low concentrations to ensure the mixture is single
phase at ambient temperature. The ability to measure the same
sample with two different detectors allows the operator to apply
the following constraint to improve each detector’s calibration:
the composition measured by the two FIDs must be the same
even if the exact composition of the sample is unknown. Thus,
after performing the first order calibration using a single-phase, Figure 9. VLE data measured in this work, with selected literature data,
lean binary mixture prepared gravimetrically, a second order for the CH4 (1) + C3H8 (3) system. (a) Deviations of CH4 liquid mole
tuning of the calibration was performed by cooling the cell into fractions (x1) from values calculated with the PR EOS (x1, calc PR) as a
the two phase region to a condition at which the liquid phase was function of the measured CH4 liquid mole fraction (x1). (b) Deviations
approximately equimolar. Then the precise values of (κi /κ1) for of the C3H8 vapor mole fractions (y3) from values calculated with the PR
each FID could be adjusted sensitively by forcing the two liquid EOS (y3, calc PR) as a function of the y3. The values of x1,calc PR and y3, calc PR
were calculated at the experimental (p,T). Symbols: ⧫, 244 K, this work;
phase compositions measured with the detectors to agree. The
◊, isochore (203 K to 283 K), this work; □, 241 K, Akers et al.;31 △,
fine-tuned values of the (κi /κ1) with reduced uncertainties
are listed in Table 4; these values were always consistent with 273 K, Akers et al.;31 ×, 255 K, Price and Kobayashi;28 ○, 230 K,
Webster et al..26 Curves: , GERG-2008 EOS at (p,T) measured in this
the values determined during the first order calibration using the work on the 244 K isotherm; − − − , GERG-2008 EOS at (p,T)
gravimetrically prepared mixtures, and this procedure had the measured in this work along the isochore from (203 to 283) K.
effect of reducing the uncertainty of the (κi /κ1) by at least a
factor of 2.
As a final check, following the completion of the phase com- using an EOS. The total combined standard uncertainty in
position measurements with the FIDs in the alternate con- the measured compositions may be calculated by combining
figuration, the sample in the cell was remixed with the in quadrature the sampling uncertainties with the propagated
magnetically driven stirrer for 30 min and allowed to stabilize temperature and pressure uncertainty. Averaged across all of the
for another 2 h. The flushing and sampling process was then measurements reported here made using the specialized cryogenic
repeated including the use of the two configurations of the FIDs. VLE apparatus, the estimated combined standard uncertainties in
The experimental mole fraction compositions reported here the mole fraction compositions were 0.0015 for the vapor phase
and 0.0031 for the liquid phase.


correspond to the average of these four measurements (before
and after the mixing at constant temperature and pressure in two
detector configurations), with each measurement representing RESULTS AND DISCUSSION
the average of at least four samples. The uncertainties in the mole The measured VLE phase compositions and estimated
fraction compositions listed here correspond to the standard uncertainties at each temperature and pressure are tabulated in
deviation of these four measurements. These standard uncertainty Tables 5 to 8. The tables also list the combined measurement
estimates represent a conservative assessment of the contributions uncertainties, u c , which include the GC measurement
of sampling and FID calibration to the uncertainty of the measured uncertainties, u(xi) and u(yi), together with the propagated
composition. uncertainties of the temperature and pressure measurements.
The uncertainties of the measured temperature and pressure The differences in the liquid and vapor compositions reported
of the mixture also propagate into the composition uncertainty; here from values calculated with the PR EOS implemented in
the magnitude of these propagated contributions can be estimated Aspen HYSYS are shown in Figures 8 to 11, together with the
3615 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

Table 7. Measured (p,T,x,y) Data for the CH4 (1) + iC4H10 (4) Binary Mixturea
T/K p/kPa x1 u(x1) uc(x1) y4 u(y4) uc(y4)
Isochoric Pathway
273.47 8329 0.5160 0.0027 0.0030 0.0753 0.0008 0.0009
263.56 7819 0.5237 0.0025 0.0029 0.0572 0.0006 0.0007
253.60 7306 0.5371 0.0017 0.0023 0.0420 0.0007 0.0007
243.61 6785 0.5561 0.0010 0.0020 0.0302 0.0006 0.0006
243.60 6735 0.5523 0.0020 0.0027 0.0308 0.0005 0.0005
243.58 6824 0.5607 0.0019 0.0026 0.0313 0.0006 0.0006
233.57 6251 0.5817 0.0037 0.0042 0.0204 0.0008 0.0008
223.49 5714 0.6151 0.0009 0.0026 0.0155 0.0005 0.0005
213.39 5118 0.6565 0.0027 0.0041 0.0097 0.0002 0.0002
203.25 4503 0.7133 0.0002 0.0043 0.0059 0.0004 0.0004
Isothermal Pathway
243.60 8684 0.7043 0.0015 0.0024 0.0506 0.0009 0.0011
243.60 8280 0.6749 0.0021 0.0028 0.0431 0.0007 0.0008
243.60 7854 0.6345 0.0024 0.0030 0.0394 0.0008 0.0009
243.59 7847 0.6386 0.0016 0.0024 0.0381 0.0004 0.0005
243.60 6735 0.5523 0.0020 0.0027 0.0308 0.0005 0.0005
243.60 6041 0.4971 0.0006 0.0018 0.0287 0.0005 0.0005
243.60 4418 0.3734 0.0009 0.0020 0.0263 0.0004 0.0004
243.60 2642 0.2300 0.0011 0.0021 0.0305 0.0004 0.0005
a
The uncertainties, u, for the measured mole fractions refer to the Type A standard uncertainties associated with sampling and detector calibration.
The uncertainties of the temperature and pressure measurements were 0.15 K and 20 kPa, respectively. These were combined in quadrature with the
sampling and detector uncertainties to calculate a combined standard uncertainty, uc.

corresponding deviations of selected literature VLE data. for CH4 (1) + C2H6 (2) are presented as deviation plots in the
Differences between the phase compositions predicted at the mole fraction of methane x1 showing the difference between
experimental conditions with the GERG-2008 EOS from those the measurements and the PR EOS predicted compositions in
predicted with the PR EOS are also shown. The GERG-2008 Figure 8. The deviations of the measurements made here along
EOS was implemented in a dynamic linked library supplied by both the isothermal and isochoric pathways are consistent within
Kunz and Wagner43 and accessed through either Microsoft Excel experimental uncertainty at the same liquid mole fractions. The
or a stand-alone executable. In all cases, the calculated values rms deviation of the measured methane liquid mole fractions
were determined by first estimating the overall composition (Figure 8a) from the PR EOS is 0.0090, while the rms devia-
of the mixture, and then performing a flash calculation at the tions of the bubble point data from the GERG-2008 EOS is
experimental pressure and temperature to produce both xi,calc 0.0097 mole fraction, although the average deviations for the two
and yi,calc. For the isochoric measurements, the mixture’s overall EOS are −0.0033 and +0.0018, respectively. Given that across all
composition was taken to be that of the corresponding single- the CH4 (1) + C2H6 (2) bubble point data, the average combined
phase mixture that was prepared gravimetrically (Table 3) standard uncertainty in x1 ranges from 0.0026 to 0.0072, it is
and loaded into the cell at ambient temperatures, prior to apparent that both EOS do a reasonable job predicting the new
commencement of the isochore. For the isothermal measure- bubble point data. The new bubble point data are also consistent
ments, the experimentally measured liquid and vapor phase with the data of Davalos et al.14 and Wei et al.11 measured at
compositions were combined using an assumed value for the
similar temperatures, and the increased negative deviation of the
vapor-phase mole fraction, β, to give an overall mixture com-
literature data from the PR EOS measured at 270 K around x1 =
position. The assumed value of β was set either to an average of
0.3, is similar in nature to the increased deviation in the new
the values estimated by the flash calculations done for the
isochoric measurements (β ≈ 0.8), or it was set to a particular 244 K isothermal data from the PR EOS as the critical point is
value (usually around 0.5) to ensure that the flash calculation did approached.
not return a single-phase condition at the experimental pressure The vapor phase ethane mole fractions y2 of the measured data
and temperature. Formally, the values of xi,calc and yi,calc calculated shown in Figure 8b have an rms and average deviation from the
at a given (T, p, zi) with zi = β yi + (1 − β) xi are independent of PR EOS of 0.0040 and −0.0004, respectively, both of which are
the value of β chosen, as long as the experimental (T, p) is within comparable with the average combined standard uncertainty in y2
the two-phase envelope predicted using the EOS with the ranging from 0.0023 to 0.0035. Some temperature dependence
resulting zi. In practice, the numerical solution of the flash in the y2 deviations is apparent: the most positive occurs at
algorithm does lead to small differences in the values of xi,calc and the lowest temperature 203 K, and the deviations become
yi,calc obtained if different values of β are assumed; however, these increasingly negative as temperature increases. The measured
differences were much smaller (typically ≤10−8 mole fraction) vapor phase ethane mole fractions have an rms and average
than the experimental uncertainty. deviation from the GERG-2008 EOS of 0.0081 and −0.0051,
The CH4 (1) + C2H6 (2) measurements are tabulated in respectively. As the mixture’s critical point at 243 K is approached
Table 5. Five data were measured along the isochoric pathway (e.g., at 6885 kPa, x1 ≈ 0.62 and y1 ≈ 0.71) the GERG EOS vapor
at temperatures between 203 K and 244 K and twelve data were predictions oscillate slightly, perhaps for reasons of numerical
measured along the isothermal pathway at 243.6 K. The data stability.
3616 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

The results of the CH4 (1) + C3H8 (3) measurements are


tabulated in Table 6. They include 14 isochoric values, measured
at temperatures between 203 K and 283 K and 10 isothermal
values measured at 243.6 K. The isochoric and isothermal
measurements are consistent with each other within exper-
imental uncertainty. The rms deviation of the measured methane
liquid mole fractions x1 (Figure 9a) from the PR EOS was 0.0085.
The new data exhibit the same trend in their deviations from
the PR EOS as the literature data, and particularly those of Price
and Kobayashi28 and Webster and Kidnay26 as the bubble point
pressure increases toward the mixture critical pressure. (In all of
the mixtures considered here, as x1 increases along the bubble
point curve so does pressure). The data of Akers et al.31 are
somewhat inconsistent with the other measurements, having
a more negative deviation from the PR EOS at low values of x1
and either more positive (273 K) or more negative (241 K)
deviations from the PR EOS at high values of x1. The GERG-
2008 predictions show the same positive divergence from the
PR EOS as the critical point is approached, indicating that
the equation’s more complex functional form is better able to
emulate the physical dependence of pressure on composition in
this region. However, the rms deviation of the new x1 data from
the GERG-2008 EOS is still 0.0083 because it underpredicts
the present data as well as those of Price and Kobayashi28 and
Webster and Kidnay26 at lower pressures.
The rms deviation of the measured propane vapor mole
fractions y3 (Figure 9b) from the PR EOS is 0.0042, whereas
the rms deviation of the y3 data from the GERG-2008 EOS is
0.0032. The data deviate negatively from the predictions of both
models, with the magnitude of the deviations from the PR EOS
being largest for data measured at pressures around 6500 kPa
(y3 ≈ 0.12). The GERG-2008 EOS was better able to describe Figure 10. VLE data measured in this work, with selected literature data,
the relationship between y3 and pressure, exhibiting a shallow for the CH4 (1) + iC4H10 (4) system. (a) Deviations of the CH4 liquid
minimum in its deviations from the PR EOS at this condition mole fractions (x1) from values calculated with the PR EOS (x1,calc PR) as
near y3 = 0.12. a function of x1. (b) Deviations of the iC4H10 vapor mole fractions (y4)
Table 7 lists the results of the CH4 (1) + iC4H10 (4) measure- from values calculated with the PR EOS (y4 calc PR) as a function of the y4.
The values of x1,calc PR and y4, calc PR were calculated at the experimental
ments, ten of which were obtained along an isochoric pathway at (p, T). Symbols: blue ◆, 244 K, this work; blue ◇, isochore (203 K to
temperatures between 203 and 273 K and eight of which were 273 K), this work; red ○, 233 K, Barsuk et al.;18 red □, 253 K, Barsuk
obtained along the 243.6 K isotherm. The rms deviation of the et al.;18 ×, Barsuk et al.18 Curves: , GERG-2008 EOS at (p,T)
measured liquid methane mole fractions x1 shown in Figure 10a measured in this work on the 244 K isotherm; − − − , GERG-2008 EOS
from the PR EOS is 0.0088. Notably, the deviations of x1 at (p,T) measured in this work along the isochore (203 K to 273 K).
measured along the isothermal and isochoric pathways from
the PR EOS are identical within experimental uncertainty at the and isochoric pathways) and better captures the pressure
same liquid methane mole fractions, which indicates that the dependence of the x1 data than the PR EOS.
temperature dependence of these data are well-described by the The deviations of the measured vapor phase isobutane mole
cubic model, and it is only at high pressures that the PR EOS does fractions y4 from the PR EOS (Figure 10b) have an rms value of
not describe the new data well, reflecting the cubic’s limited 0.0018, which is comparable with the combined experimental
functional form. In contrast the x1 deviations of Barsuk et al.18 standard uncertainty. The y4 deviations from the PR EOS
measured at 273 K are significantly offset from the deviations predictions increase with pressure along the 244 K isotherm,
of their x1 data obtained along the 233 K and 253 K isotherms; which is the first clear reflection in the vapor phase of the same
the latter two isothermal data sets are only consistent with the trend that is apparent in the liquid phase. The rms deviation of
new data at the two lowest pressures measured here. In the the y4 data from the GERG EOS is 0.0037, reflecting that the
development of the GERG EOS no other thermodynamic data deviations of the GERG EOS y4 predictions exhibit a negative
in the range 198 K to 278 K of any kind were identif ied for the slope with increasing y4.
CH4 (1) + iC4H10 (4) binary system3 (our emphasis) and so it was The CH4 (1) + nC4H10 (5) measurements are tabulated in
necessary to use the data of Barsuk et al.18 in the model’s Table 8. Ten isochoric values were measured at temperatures
regression. Consequently, the rms deviation of the new data from between 203 K, and 273 K and ten isothermal values were
the GERG EOS is 0.0151, nearly twice that of the PR EOS measured at 244.5 K. The rms deviation of the measured methane
because the GERG model has an offset from the present data. liquid mole fractions from the PR EOS shown in Figure 11a was
This is even though the GERG model correctly captures the 0.0184. The deviations of the isothermal and isochoric measure-
temperature dependence in the data (as evidenced by the ments were consistent within experimental uncertainty at the
coincidence of the GERG EOS predictions for the isothermal same methane liquid mole fraction except for one point at x1 ≈ 0.62;
3617 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

Table 8. Measured (p,T,x,y) Data for the CH4 (1) + nC4H10 (5) Binary Mixturea
T/K p/kPa x1 u(x1) uc(x1) y5 u(y5) uc(y5)
Isochoric Path
273.42 7249 0.4246 0.0008 0.0014 0.0482 0.0017 0.0017
263.48 6842 0.4331 0.0018 0.0022 0.0357 0.0016 0.0016
253.52 6431 0.4480 0.0004 0.0014 0.0265 0.0015 0.0015
244.52 6005 0.4601 0.0003 0.0015 0.0197 0.0015 0.0015
243.52 6010 0.4646 0.0013 0.0020 0.0188 0.0015 0.0015
233.50 5593 0.4880 0.0005 0.0019 0.0126 0.0015 0.0015
233.49 5604 0.4882 0.0010 0.0021 0.0146 0.0016 0.0016
223.44 5163 0.5201 0.0006 0.0024 0.0075 0.0016 0.0016
213.36 4702 0.5615 0.0028 0.0042 0.0061 0.0015 0.0015
203.25 4224 0.6217 0.0004 0.0052 0.0036 0.0015 0.0015
Isothermal Path
244.52 1311 0.1071 0.0000 0.0016 0.0289 0.0015 0.0016
244.53 3318 0.2669 0.0004 0.0014 0.0175 0.0015 0.0015
244.48 5058 0.3950 0.0017 0.0022 0.0154 0.0023 0.0023
244.52 6005 0.4601 0.0003 0.0015 0.0197 0.0015 0.0015
244.52 6593 0.5008 0.0021 0.0026 0.0185 0.0016 0.0016
244.51 8016 0.6002 0.0009 0.0021 0.0274 0.0015 0.0015
244.52 8238 0.6180 0.0015 0.0025 0.0251 0.0062 0.0062
244.52 9161 0.6869 0.0005 0.0024 0.0399 0.0015 0.0017
244.52 9913 0.7522 0.0017 0.0034 0.0515 0.0019 0.0024
244.48 10132 0.7764 0.0018 0.0037 0.0594 0.0016 0.0024
a
The uncertainties, u, for the measured mole fractions refer to the Type A standard uncertainties associated with sampling and detector calibration.
The uncertainties of the temperature and pressure measurements were 0.15 K and 20 kPa, respectively. These were combined in quadrature with the
sampling and detector uncertainties to calculate a combined standard uncertainty, uc.

at this point, however, the temperature of the isochoric mea- improved by including the new data and those of Elliot et al.23 in
surement was 203 K, some 40 K lower than the corresponding a new fitting process.
point on the isotherm, and thus this difference potentially reflects The rms deviation of the measured butane vapor mole
an effect of temperature dependence. The isothermal data set fractions (Figure 11b) from the PR EOS is 0.0036, whereas the
shows the most significant positive divergence from the PR EOS rms deviation of the y5 data from the GERG-2008 EOS is 0.0031.
of all the binary systems considered here, as the bubble-point The measured y5 deviations from the PR EOS predictions
pressure is increased toward the mixture critical point. increase with pressures along the 244 K isotherm even more
Importantly, the new data further validate the three isothermal strongly than was the case for y4 in the CH4 (1) + iC4H10 (4)
data sets of Elliot et al.,23 the temperature dependence of which system (Figure 10b). The values of y5 predicted with the GERG
EOS at 244 K deviate slightly from those predicted using the
appears to be reasonably well-described by the PR EOS even
PR EOS, with the deviations becoming more positive at higher
though each isotherm exhibits a similar divergence from the PR
pressures, partially reflecting the trend exhibited by the data
EOS predictions as the critical point is approached. In contrast albeit with a much smaller magnitude.


the new data are clearly inconsistent with the earlier data sets of
Roberts et al.21 and Wang and McKetta,22 which appear to be
CONCLUSIONS
erroneous.
Kunz et al.3 stated that none of these literature data sets A review was conducted of the available literature VLE data for
were used in the development of the GERG-2008 EOS for the the binary mixtures most relevant to the simulation of LNG scrub
CH4 (1) + nC4H10 (5) binary system; this may have been partly columns. Analysis of those data identified that EOS used to
because of the inconsistency between the data of Elliot et al.23 model such industrial operations are limited in their ability to
and those of McKetta and co-workers.21,22 Given that the only describe those data by the presence of a large number of poor
binary VLE data used in the development of the GERG EOS for quality measurements. Accordingly, a specialized cryogenic VLE
the CH4 (1) + nC4H10 (5) mixture were those of Weise et al.30 apparatus was developed and used to produce new, reference
quality data for the methane binary systems of most importance
and Sage et al.37 measured at temperatures between (278 and
to LNG scrub columns consisting of CH4, C2H6, C3H8, and
394) K, it is interesting that the GERG predictions at 244 K
iC4H10 or nC4H10. The term reference quality is used to indicate
effectively split the literature data sets.21−23 As a consequence, that (a) unlike many similar VLE data reported in the literature,
the rms deviation of the new x1 data from the GERG EOS robust quantitative uncertainty estimates have been made for the
prediction is 0.0464, which is a factor of 2.5 times larger than the measured mole fraction compositions; and (b) the accuracy of
rms deviation of the PR EOS and more than 40 times larger the present data is equal to or better than other sources available
than the average combined experimental standard uncertainty. for these systems over this range of temperature and pressure.
This is despite the fact that the curvature of the GERG-2008 EOS As a result, the new measurements have enabled the identifica-
deviations from the PR EOS predictions of x1 with increasing tion of literature data sets that should receive an increased
pressure match those of the new VLE data, and suggests that weighting as well as those that should not be included in future
the performance of the GERG EOS for this binary could be EOS development. This work has, therefore, contributed to the
3618 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data Article

For the liquid phase, the rms deviations in x1 from the PR EOS
for the four binaries ranged from (0.008 to 0.018) mole fraction,
which corresponds to (2 to 4) % of the average value of x1 ≈ 0.5
at the conditions studied. This is about two to six times the
average combined experimental standard uncertainty of the x1
data; however the deviations in x1 from the PR EOS are
systematic in nature, increasing with bubble point pressure as the
mixture critical point is approached, and reaching up to 0.05 mol
fraction in the case of the CH4 (1) + nC4H10 (5) data set. This
systematic deficiency of the PR EOS is a consequence of its cubic
form, which prevents it from describing simultaneously the shape
of the phase envelope at both low and high pressure. The default
PR EOS in Aspen HYSYS has been tuned to describe the VLE of
these binary systems at low to medium pressures essentially
within experimental standard uncertainty, and it would seem
there is little prospect of further improving the performance of
this cubic EOS at higher pressures, without sacrificing the lower
pressure performance.
The multiparameter nature of the GERG-2008 EOS means it
has the capacity to describe accurately the VLE of these binary
mixtures over a much wider range of pressure and temperature.
However, with the exception of the CH4 (1) + C3H8 (3) mixture,
the rms deviations of the measured x1 from the GERG EOS
were larger than those for the PR EOS. This was particularly the
case for the CH4 (1) + iC4H10 (4) and CH4 (1) + nC4H10 (5)
mixtures, and is likely a consequence of the uncertainty regarding
the literature data sets that should be used in the model’s
development. Prior to these measurements, we are aware of only
one other data set for any thermodynamic property of the CH4
(1) + iC4H10 (4) binary in the temperature range 198 K to 278 K,
Figure 11. VLE data measured in this work, with selected literature data,
and accordingly the GERG EOS represents the average of the
for the CH4 (1) + nC4H10 (5) system. (a) Deviations of CH4 liquid Barsuk et al.18 data, which have an anomalous temperature
mole fractions (x1) from values calculated with the PR EOS (x1, calc PR,) dependence. For the CH4 (1) + nC4H10 (5) binary at 244 K, two
as a function of the x1. (b) Deviations of the nC4H10 vapor mole fractions groups of inconsistent VLE data were available in the literature at
(y5) from values calculated with the PR EOS (y5, calc PR) as a function of the time of the GERG EOS development and so neither was used
y5. The values of x1,calc PR and y5, calc PR were calculated at the in its regression. The GERG EOS currently splits the difference
experimental (p, T). Symbols: blue ◆, 244 K, this work; blue ◇, between the erroneous data of McKetta and co-workers,21,22 and
isochore (203 K to 273 K), this work; red △, 244 K, Roberts et al.;21 ∗,
244 K, Wang and McKetta;22 red □, 244 K, Elliot et al.;23 ×, 255 K, Elliot
those of Elliot et al.,23 which are in excellent agreement with
et al.;23 red ○, 233 K, Elliot et al.23 Curves: , GERG-2008 EOS at the new data presented here. The reference quality VLE data
(p,T) measured in this work on the 244 K isotherm; − − − , GERG- presented in this work for the binary systems of primary interest
2008 EOS at (p,T) measured in this work along the isochore (203 K to to LNG scrub columns should therefore contribute signifi-
273 K). cantly to the improvement of the advanced EOS capable of
describing rich natural gases over a wider range of temperatures
and pressures.


resolution of a problem articulated by Kunz and Wagner;4
namely that the paucity of quality VLE data available for mixtures
limits the accuracy achievable in the further development of AUTHOR INFORMATION
modern EOS, especially those aimed at describing richer natural Corresponding Author
gases at low temperatures. *E-mail: Eric.May@uwa.edu.au.
When comparing the new VLE data with the predictions of
Funding
EOS it is important to consider both the absolute mole fraction
deviations and the deviations relative to the amount of the minor The research was funded by Chevron Energy Technology
component in the phase under consideration. For the vapor Company and the Australian Research Council through
phase, the rms deviations of the heavy component, yi (i = 2, 3, 4, LP0882519 and LP120200605.
or 5), data from either the PR or GERG EOS had an average Notes
value across all systems of 0.004 mol fraction, and amounted The authors declare no competing financial interest.


to between (1 and 13) % of the average value of the yi at the
conditions studied. These deviations are, however, closer in
ACKNOWLEDGMENTS
magnitude to the combined experimental standard uncertainty,
which makes it more difficult to discern whether clear systematic The authors thank Craig Grimm for helping to construct the
trends are exhibited in the vapor phase predictions of either apparatus. The authors are also grateful to Mohammed Kandil
model. and Andrew Vieler for their contributions to the project.
3619 DOI: 10.1021/acs.jced.5b00610
J. Chem. Eng. Data 2015, 60, 3606−3620
Journal of Chemical & Engineering Data


Article

REFERENCES (23) Elliot, D. G.; Chen, R. J. J.; Chappelear, P. S.; Kobayashi, R. Vapor-
Liquid Equilibrium of Methane-Butane System at Low Temperatures
(1) Peng, D.-Y.; Robinson, D. B. A New Two-Constant Equation of and High Pressures. J. Chem. Eng. Data 1974, 19, 71−77.
State. Ind. Eng. Chem. Fundam. 1976, 15, 59−64. (24) Wichterle, I.; Kobayashi, R. Vapor-Liquid Equilibrium of
(2) Assael, M. J.; Trusler, J. P. M.; Tsolakis, T. F. Thermophysical Methane-Propane System at Low Temperatures and High Pressures.
Properties of Fluids: An Introduction to Their Prediction; Imperial College J. Chem. Eng. Data 1972, 17, 4−9.
Press: London, UK, 1996. (25) Kalra, H.; Robinson, D. B. An Apparatus for the Simultaneous
(3) Kunz, O.; Klimeck, R.; Wagner, W.; Jaeschke, M. The GERG-2004 Measurement of Equilibrium Phase Composition and Refractive Index
Wide-Range Equation of State for Natural Gases and Other Mixtures, Data at Low Temperatures and High Pressures. Cryogenics 1975, 15,
GERG Technical Monograph 15; Fortschr.-Ber. VDI Verlag: Düsseldorf, 409−412.
Germany, 2007. (26) Webster, L. A.; Kidnay, A. J. Vapor-Liquid Equilibria for the
(4) Kunz, O.; Wagner, W. The GERG-2008 Wide-Range Equation of Methane-Propane-Carbon Dioxide Systems at 230 and 270 K. J. Chem.
State for Natural Gases and Other Mixtures: An Expansion of GERG- Eng. Data 2001, 46, 759−764.
2004. J. Chem. Eng. Data 2012, 57, 3032−3091. (27) Benham, A. I.; Katz, D. L. Vapor-Liquid Equilibria for Hydrogen -
(5) Kiselev, S. B.; Friend, D. G. Cubic Crossover Equation of State for Light Hydrocarbon Systems at Low Temperatures. AIChE J. 1957, 3,
Mixtures. Fluid Phase Equilib. 1999, 162, 51−82. 33−36.
(6) Lafitte, T.; Apostolakou, A.; Avendaño, C.; Galindo, A.; Adjiman, (28) Price, A. R.; Kobayashi, R. Low Temperature Vapor-Liquid
C. S.; Müller, E. A.; Jackson, G. Accurate Statistical Associating Fluid Equilibrium in Light Hydrocarbon Mixtures: Methane-Ethane-Propane
Theory for Chain Molecules Formed from Mie Segments. J. Chem. Phys. System. J. Chem. Eng. Data 1959, 4, 40−52.
2013, 139, 154504. (29) Reamer, H. H.; Sage, B. H.; Lacey, W. N. Phase Equilibria in
(7) Aspen HYSYS Process Simulator, v7.3; Aspen Technology, Inc: Hydrocarbon Systems: Volumetric and Phase Behavior of the Methane-
Cambridge, MA, 2011. Propane System. Ind. Eng. Chem. 1950, 42, 534−539.
(8) Wichterle, I.; Kobayashi, R. Vapor-Liquid Equilibrium of Methane- (30) Wiese, H. C.; Jacobs, J.; Sage, B. H. Phase Equilibria in the
Ethane System at Low Temperatures and High Pressures. J. Chem. Eng. Hydrocarbon Systems. Phase Behavior in the Methane- Propane-n-
Data 1972, 17, 9−12. Butane System. J. Chem. Eng. Data 1970, 15, 82−91.
(9) Wichterle, I.; Kobayashi, R. Vapor-Liquid Equilibrium of Methane- (31) Akers, W. W.; Burns, J. F.; Fairchild, W. R. Low-Temperature
Ethane-Propane System at Low Temperatures and High Pressures. J. Phase Equilibria. Methane-Propane System. Ind. Eng. Chem. 1954, 46,
Chem. Eng. Data 1972, 17, 13−18. 2531−2534.
(10) Gupta, M. K.; Gardner, G. C.; Hegarty, M. J.; Kidnay, A. J. Liquid- (32) Joffe, J. Vapor-Liquid Equilibria by the Pseudocritical Method.
Vapor Equilibriums for the N2 + CH4 + C2H6 System from 260 to 280 K. Ind. Eng. Chem. Fundam. 1976, 15, 298−304.
J. Chem. Eng. Data 1980, 25, 313−318. (33) Raimondi, L. A Modified Redlich-Kwong Equation of State for
(11) Wei, M. S. W.; Brown, T. S.; Kidnay, A. J.; Sloan, E. D. Vapor + Vapour-Liquid Equilibrium Calculations. Chem. Eng. Sci. 1980, 35,
Liquid Equilibria for the Ternary System Methane + Ethane + Carbon 1269−1275.
Dioxide at 230 K and Its Constituent Binaries at Temperatures from 207 (34) Kahre, L. C. Low-Temperature K Data for Methane-n-Butane. J.
to 270 K. J. Chem. Eng. Data 1995, 40, 726−731. Chem. Eng. Data 1974, 19, 67−71.
(12) Raabe, G.; Janisch, J.; Köhler, J. Experimental Studies of Phase (35) Nederbragt, G. W. Gas-Liquid Equilibria for the System Methane
Equilibria in Mixtures Relevant for the Description of Natural Gases. - Butane. Ind. Eng. Chem. 1938, 30, 587−588.
Fluid Phase Equilib. 2001, 185, 199−208. (36) Sage, B. H.; Budenholzer, R. A.; Lacey, W. N. Phase Equilibria in
(13) Janisch, J.; Raabe, G.; Köhler, J. Vapor-Liquid Equilibria and Hydrocarbon Systems: Methane - n-Butane System in the Gaseous and
Saturated Liquid Densities in Binary Mixtures of Nitrogen, Methane, Liquid Regions. Ind. Eng. Chem. 1940, 32, 1262−1277.
and Ethane and Their Correlation Using the VTPR and PSRK GCEOS. (37) Sage, B. H.; Hicks, B. L.; Lacey, W. N. Phase Equilibria in
J. Chem. Eng. Data 2007, 52, 1897−1903. Hydrocarbon Systems: The Methane-n-Butane System in the Two-
(14) Davalos, J.; Anderson, W. R.; Phelps, R. E.; Kidnay, A. J. Liquid- Phase Region. Ind. Eng. Chem. 1940, 32, 1085−1092.
Vapor Equilibria at 250.00K for Systems Containing Methane, Ethane, (38) Rigas, T. J.; Mason, D. F.; Thodos, G. Vapor-Liquid Equilibria
and Carbon Dioxide. J. Chem. Eng. Data 1976, 21, 81−84. Microsampling Technique Applied to a New Variable-Volume Cell. Ind.
(15) Wichterle, I. Low Temperature Vapor−Liquid Equilibria in the Eng. Chem. 1958, 50, 1297−1300.
Methane-Ethane-Propane Ternary and Associated Binary Methane Systems (39) Kandil, M. E.; Thoma, M. J.; Syed, T.; Guo, J.; Graham, B. F.;
with Special Consideration of the Equilibria in the Vicinity of the Critical Marsh, K. N.; Huang, S. H.; May, E. F. Vapor-Liquid Equilibria
Temperature of Methane, A Monograph. Rice University: Houston TX, Measurements of the Methane + Pentane and Methane + Hexane
1970. Systems at Temperatures from (173 to 330) K and Pressures to 14 MPa.
(16) Wilson, G. M. Vapor−Liquid Equilibria of Nitrogen, Methane, J. Chem. Eng. Data 2011, 56, 4301−4309.
Ethane, and Propane Binary Mixtures at LNG Temperatures from Total (40) Hughes, T. J.; Kandil, M. E.; Graham, B. F.; May, E. F. Simulating
Pressure Measurements. Adv. Cryog. Eng. 1975, 20, 164−171. the Capture of CO2 from Natural Gas: New Data and Improved Models
(17) Miller, R. C.; Kidnay, A. J.; Hiza, M. J. Liquid + Vapor for Methane + Carbon Dioxide + Methanol. Int. J. Greenhouse Gas
Equilibriums in Methane + Ethene and in Methane + Ethane from Control 2014, 31, 121−127.
150.00 to 190.00 K. J. Chem. Thermodyn. 1977, 9, 167−178. (41) Hughes, T. J.; Kandil, M. E.; Graham, B. F.; Marsh, K. N.; Huang,
(18) Barsuk, S. D.; Skripka, V. G.; Ben’yaminovich, O. A. Liquid-Vapor S. H.; May, E. F. Phase Equilibrium Measurements of (Methane +
Equilibriums in a Methane-Isobutane System at Low Temperatures. Benzene) and (Methane + Methylbenzene) at Temperatures from (188
Gaz. Prom-st. 1970, 15, 38−41. to 348) K and Pressures to 13 MPa. J. Chem. Thermodyn. 2015, 85, 141−
(19) Olds, R. H.; Sage, B. H.; Lacey, W. N. Methane-Isobutane System. 147.
Ind. Eng. Chem. 1942, 34, 1008−1013. (42) TransValor. The ROLSI Evolution IV. http://www.rolsi.com/
(20) Haynes, W. M. Orthobaric Liquid Densities and Dielectric English.htm (accessed February 2, 2011).
Constants of (Methane + 2-Methylpropane) and (Methane + n-Butane) (43) Kunz, O.; Wagner, W. Software for the Calculation of
at Low Temperatures. J. Chem. Thermodyn. 1983, 15, 903−911. Thermodynamic Properties from the GERG-2004 XT08 Wide Range
(21) Roberts, L. R.; Wang, R. H.; Azarnoosh, A.; McKetta, J. J. Equation of State for Natural Gases and Other Mixtures; Ruhr Universitat
Methane-n-Butane System in the Two-Phase Region. J. Chem. Eng. Data Bochum: 2009.
1962, 7, 484−485.
(22) Wang, R. H.; McKetta, J. J. Vapor-Liquid Equilibrium of the
Methane-n-Butane-Carbon Dioxide System at Low Temperatures and
Elevated Pressures. J. Chem. Eng. Data 1964, 9, 30−35.

3620 DOI: 10.1021/acs.jced.5b00610


J. Chem. Eng. Data 2015, 60, 3606−3620

You might also like