Anal. Chem. 2001, 73, 379-383
Technical Notes
A Comparison of Liquid Hydrocarbon Calibration Standards in Piston Cylinders and Standard Cylinders with Eductor Tubes Robert Benesch and Tracey Jacksier*
Air Liquide, Chicago Research Center, 5230 South East Avenue, Countryside, Illinois 60525
Natural gas liquid standards consist of various components primarily in the C1-C6 range; they are available in three types of cylinder packages: piston cylinders and two other types of standard cylinders with eductor tubes. The three cylinder packages have different operation conditions and thus a sample may behave differently in each of the packages. The piston cylinder maintains the components in a single liquid phase at a constant pressure, while the standard cylinders maintain the components as a two-phase mixture. Typically, the components may distribute between the two phases in different concentrations depending on the various thermodynamic variables, such as temperature, pressure, and volume. This study will examine the integrity of the sample in the three cylinder packages during a controlled cylinder depletion. The results for the various cylinders will be compared with a theoretical model of the experiments. Gas chromatographs (GC) are found in many industrial processes. These chemical analyzers monitor various conditions throughout the processes and are essential for the successful operation of these processes. Calibration standards are used to calibrate these instruments, and difficulties with the calibration standards may lead to problems with the analysis and thus the processes themselves. As a result, it is important to deliver accurate calibration standards and also to guarantee the stability of the calibration standard. One group of calibration standards used in the petrochemical industry consists of natural gas liquids (NGLs) which are made up of small-chain hydrocarbon components (C1-C6). Depending on the conditions, some of the components in NGLs may exist in either the gas phase, liquid phase, or both. For instance, at STP, ethane and propane are gases and n-hexane and n-pentane are liquids. The introduction of these four components into a closed vessel, such as a cylinder, would create two phases (gas and liquid) with all four components distributed differently between both phases. With the possibility of two phases being present, the type of package becomes important. If two phases are present, any change in a thermodynamic variable, such as temperature,1 pressure, or volume, may lead to a change in the composition of the liquid standard. The (1) Boulanger, R. L. Adv. Inst. Control 1990, 90, 439-449. 10.1021/ac000767+ CCC: $20.00 Published on Web 12/06/2000
© 2001 American Chemical Society
mere withdrawal of product changes the size of both the gaseous and liquid phases within the cylinder, thus altering the equilibrium properties of the mixture. In North America, NGL standards are available in three different cylinder packages: a standard gas cylinder, a standard gas cylinder with dual port valve, and a piston cylinder. In a standard gas cylinder, the NGL mixture is in contact with a pressurizing fluid (usually an inert gas such as helium) and therefore two phases are present. In the standard cylinder, there is a dip tube that extends from the valve to the bottom of the cylinder. The pressurizing fluid (generally helium) is added to supply enough pressure to withdraw the liquid through the dip tube. The cylinder is initially charged with the pressurizing fluid only when the cylinder is filled to a pressure above the vapor pressure of the mixture. Thus, as the cylinder is depleted, the pressure in the cylinder decreases. The second type of cylinder is similar to the standard gas cylinder except that it has a dual port valve. This valve allows the cylinder to be continually pressurized with a pressurizing fluid source at one port while the liquid is withdrawn through the other port. As the cylinder is depleted, the pressurizing gas fills the cylinder, maintaining a constant pressure. The third type of cylinder is a constant-pressure piston cylinder. This cylinder has a movable piston that physically separates the NGL standard from the pressurizing gas. Like the cylinder with the dual port valve, the piston cylinder is continually pressurized with a pressurizing gas and thus the pressure remains constant throughout use. The piston cylinder is operated at a pressure at which the entire mixture remains in a single liquid phase. Since only one phase is present, there is no partitioning or fractionation of the mixture. This should guarantee a constant composition throughout the use of the cylinder. The dual port valve cylinder, however, cannot be operated in the single liquidphase region because of the pressurizing gas. Also, the cylinder has a constant internal volume and the volume of the space occupied by the liquid decreases as the samples are withdrawn. The stability of a mixture of NGLs was determined in the three cylinder packages during a controlled cylinder depletion. To determine the stability, the cylinders were depleted incrementally and subsequently the compositions were analyzed at each increment. A theoretical model of the depletion utilizing equilibrium and mass balances was developed and compared to the experimental data. Analytical Chemistry, Vol. 73, No. 2, January 15, 2001 379
Table 1. Nominal Concentration of NGL Samples name ethane propane isobutane n-butane
mol % 1.05 1.00 2.10 2.00
name isopentane n-pentane n-hexane
mol % 22.03 22.94 48.88
EXPERIMENTAL SECTION Gas Processors Association (GPA) standard 2177-95 was utilized to measure the concentrations of the various components in the mixtures. A gas chromatograph (Varian Instruments, model 3800, Walnut Creek, CA) equipped with a thermal conductivity detector (TCD) was used to measure the composition of the mixtures. A 30-ft-long, 1/8-in.-diameter packed column manufactured by Alltech (Deerfield, IL) was used. The column contained a 28 wt % loading of silicone on a 200/500 CWHP stationery phase with a 80/100 mesh size. A Valco liquid sample valve, (VICI, Houston, TX), rated to 1500 psi at 75 °C, was used to introduce the liquid sample into the column. The valve was equipped with a 1-µL sample slit inside the valve. The GC was operated isothermally at 100 °C with a carrier gas (research grade helium 99.9999%, Air Liquide Electronics, Morrisville, PA) flow rate of 30 mL/min. A Varian Star Chromatography workstation was used to handle the data acquisition and integration. The NGL standards and samples were obtained from Air Liquide America Corp. (La Porte, TX). The samples of the three cylinders that were tested had similar component concentrations (Table 1). Both the standard gas cylinder and the standard gas cylinder with the dual port valve were depleted in the same manner. The mixture was at room temperature (23 °C). The cylinders were each placed on a balance (Toledo ID 5 Multirange, Columbus, OH) and samples were withdrawn at ∼500-g increments. To ensure homogeneity after withdrawal, the cylinders were mixed on a cylinder roller for 10 min and then allowed to equilibrate for 90 min. An equilibrium time of 90 min was established by examining the stability of the components after a withdrawal as a function of time. After equilibration, the cylinder was analyzed in triplicate and the results averaged. The pressure of the standard gas cylinder was monitored with a pressure transducer (0-300 psi PX 620, Omega Engineering, Stamford, CT) to determine the actual pressure loss during the depletion. Samples from the piston cylinder were withdrawn in 15-g increments. The withdrawal size for the piston cylinder was smaller then the standard cylinders owing to the smaller volume of the piston cylinder. After withdrawal, the cylinder was mixed using the incorporated vortex mixer. Although there was no change in the temperature or pressure, the cylinder was allowed to equilibrate for 30 min before a sample was taken for analysis In gas chromatography, the concentration of a component is determined by the detector response. The change in concentration of the samples during the depletion was determined by observing the change in the area of the detector response. To determine the magnitude of this change, a calibration function was generated using standards contained in piston cylinders of varying compositions of the components (Table 2). To test for experimental reproducibility, seven injections were made utilizing one of the standards. The relative standard deviation 380
Table 2. Concentration of Calibration Standards
Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
concn, mol % component
standard 1
standard 2
standard 3
ethane propane isobutane n-butane isopentane n-pentane hexane
1.041 1.006 2.007 1.989 22.023 22.958 48.78
1.194 0.8 1.799 2.198 23.948 24.946 44.892
0.797 1.198 2.198 1.795 19.956 20.954 52.873
of the values ranged from 3.4% for ethane and 1.2% for propane to less than 0.8% for the other compounds. Theoretical Calculation. A theoretical model of the cylinder depletion was developed using Aspen Plus (Version 10.1-0, Aspen Technology, Cambridge, MA). The software was used to calculate the equilibrium properties of the mixtures and mass balances resulting from the depletion. For the equilibrium calculations, the mixtures were treated as nonideal fluids. They were modeled using the Peng-Robinson equation of state,2 which is recommended for modeling hydrocarbon processes.3 The equilibrium equation is a modified version of Raoult’s law:4 sat φgas ) yiP ) φliq i i xPii
where φ’s are the fugacity coefficients, x is the mole fraction in the liquid, y is the mole fraction in the gas, P is the total pressure, and Psat is the vapor pressure. Aspen Plus calculates the fugacity coefficients (φ), which are a function of pressure, temperature, and composition, by utilizing various pure component parameters and binary interaction parameters of the components along with the Peng-Robinson equation of state. All parameters for the model were taken from the Aspen Plus library, the source of which is the Gmehling databank. It should be noted that the system under consideration is a multicomponent system; however, the Peng-Robinson equation of state has been shown to give relatively good results when modeling the equilibrium of multicomponent systems.5 Cylinder depletion is inherently a batch process; however, it may be modeled as a continuous process with discrete equilibrium steps. A flow diagram for the standard gas cylinder is given in Figure 1. Referring to Figure 1, the mixture enters the first equilibrium flash unit, B1, where it separates between a liquid and gas (the temperature and pressure must be specified in the flash unit). The liquid leaves in stream 3 and enters a splitter where 500 g is removed in stream 9. The remainder of the liquid stream is returned and combines with the gas stream in the flash unit B2. The temperatures in B1 and B2 are equivalent (22 °C); however, the pressure is unknown and must be chosen so that the volume of mixture in the flash unit is equal to the volume of the standard cylinder, which is 22 L. A trial-and-error process is (2) Peng, D. Y.; Robinson, D. B. Ind. Eng. Chem. Fundam. 1976, 15, 59-64. (3) Aspen Plus, Reference Manual 2, 1996; pp 3-34. (4) Prausnitz, J. M.; Lichtenthaler, R. N.; Gomes de Azevedo, E. Molecular Thermodynamics of Fluid Phase Equilibria; Prentice-Hall: Upper Saddle River, NJ, 1986. (5) Gupta, M. D.; Gardner, G. C.; Hegartz, M. J.; Kidnay, A. J. Chem. Eng. Data 1980, 25, 313.
Figure 1. Flow diagram of theoretical model of standard gas cylinder.
Figure 2. Flow diagram of theoretical model of standard gas cylinder with a dual port valve.
used to choose the correct pressure. The compositions of the stream exiting the splitter units (B4, B5) are used as the theoretical data for the compositions of the cylinders following depletion. The flow diagram for the standard cylinder with the dual port valve is slightly different from the standard cylinder (Figure 2). With the dual port valve, the pressure is constant; consequently, the pressure in all of the flash units is equivalent. Helium, used to keep the pressure constant, is added to the flash units in streams 14 and 15. The exact amount of helium is adjusted to maintain a volume of the experimental cylinder volume, 22 L. Like the pressure in the standard cylinder, the amount of helium is determined through trial and error. RESULTS AND DISCUSSION The results of the three cylinder depletions and the theoretical model are illustrated in Figures 3-6 and presented numerically in Table 3. The percent compositions are plotted versus the percent of liquid removed from the cylinder. The results clearly show that the stability of the mixture is dependent upon the type of cylinder package. The standard gas cylinder performed the poorest of the three cylinder packages evaluated. The concentration of the lighter components decreased substantially in the liquid phase (see Figures 3-5). The decrease
Figure 3. Measured and predicted ethane concentrations during depletion: SGC, standard gas cylinder.
in composition for ethane (Figure 3), propane, and isobutane began immediately. The liquid-phase ethane composition showed a 22% decrease between the initial and final values, which was the largest change observed. The hexane concentration showed an increase in composition (Figure 6). This result is observed because as the lighter components partition into the gaseous Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
381
Figure 4. Measured and predicted butane concentrations during depletion.
Figure 6. Measured and predicted hexane concentrations during depletion. Table 3. Mole Percent Change between Initial and Final Values of the Component Compositions
Figure 5. Measured and predicted pentane concentrations during depletion.
phase the percent hexane in the remaining liquid increases (see Table 3). The standard gas cylinder with the dual port valve gave only slightly better results when compared to the standard cylinder. Like the standard cylinder, ethane showed the largest percent change. The only difference in the thermodynamic variables that describe the two standard cylinders is the amount of helium in the gas phase and the pressure of the cylinders; the dual port valve cylinder has more helium in the gas phase and it is also at a higher pressure. Helium is not a good solvent at moderate pressures; therefore, the only thermodynamic difference between the two cylinders is the total pressure. The standard cylinder decreased in pressure from 203 to 70 psig while the dual port valve cylinder was continuously operated at 210 psig. This large pressure differential between the two cylinders resulted in only a small change between the uniformity of the mixture. The piston cylinder provided uniformity of composition during depletion. Less than a 0.5% change was observed for any component during the depletion (Figures 3-6). This can be attributed to the existence of only one phase in the cylinder. The package design ensures that the mixture cannot fractionate. It is 382 Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
component
standard gas cylinder
standard gas cylinder w/ dual port valve
piston cylinder
ethane propane isobutane butane isopentane pentane hexane
-22 -10.6 -5.4 -3.5 -0.08 -0.3 +1.2
-19.6 -7.2 -2.5 -1.5 -0.09 -0.2 +0.7
0.00 0.00 0.00 +0.50 -0.18 -0.09 -0.16
clear that the standard cylinders perform poorly because of the presence of two phases and the compositions of the components in these phases change as the mixture is withdrawn and the volume of the gas phase increases. The results of the theoretical modeling agreed quite well with the experiments for the lighter components: ethane, propane, and isobutane. One problem associated with the simulation is that the initial compositions do not match the experiments exactly; this is especially clear for the heavier components. Convergence between the model and the exact starting experimental values was not possible. This is most likely a result of the model not being exact. However, the discrepancies between the model and experimental data are less than 3%. The models and experiments for butane have similar trends; if the model and experiment were able to start at the same initial compositions, the correlation would be improved. The model predicts an increase in concentration of the pentane for the standard gas cylinder, which is not clearly observed experimentally (Figure 5). The increase is small (1%); however, it is slightly above the experimental error (0.7%) and thus should be seen experimentally. Both models predict increases in concentrations for hexane, which are also seen experimentally. The magnitude of the increase for the standard gas cylinder model is slightly larger than the model. There can be several reasons for the discrepancies between experiments and the models. The models use binary interaction parameters, which are typically regressed from experimental vapor liquid equilibrium (VLE) data, which has experimental error. It is believed that overall, the
models did quite well in predicting the experimental compositions for the components. CONCLUSIONS Significant decreases in the concentrations of the light components in the mixture were observed for both the standard gas cylinder and the standard gas cylinder with the dual port valve. This decrease in concentration was observed almost immediately. Conversely, the piston cylinder provided a constant composition of all of the components during the depletion. The existence of multiple phases for liquid hydrocarbons of low molecular weights allows repartitioning of these components from the liquid to the gas phase as the gas volume in the cylinder increases. A model
was developed which predicted the decrease in concentration of the lighter components. ACKNOWLEDGMENT The authors sincerely appreciate the help provided by Mr. Woodrow “Woody” Mock (Air Liquide America Corp.. La Porte, Texas) and Carol Schnepper (Air Liquide, Chicago Research Center) during this study. Received for review July 5, 2000. Accepted October 25, 2000. AC000767+
Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
383
