Separation of Fischer-Tropsch Wax from Catalyst c Using Supercritical Fluid Extraction Quarterly Technical Progress Report 1 pril 1996-3 June 1996 Prepared by PC Joyce and MC Thies Clemson University Department of Chemical Engineering Clemson, SC 29634-99 Prepared for the Pittsburgh Energy Technology Center United States Department of Energy Contract No DE-FG22-94PC942 19 Richard E Tischler, Project Manager *---* US DEPRTMENT OF ENERGY PTENT CLERNCE NOT REQURED PROR TO PUBLCTON OF THS DOCUMENT
DSCLMER Portions of this document may be illegible in electronic image products mages are produced from the best available original document
DSCLMER This report was prepared as an account of work sponsored by an agency of the United States Government Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof
L This report was prepared as an account of work sponsored by an agency of the United States Government Neither the United States nor any agency thereof, nor any of their / employees, makes any warranty, express or implied, or assumes any legal liability or responsibility of any third party s results of such use of any information, apparatus, product, or process disclosed in this report, or represents that its use by such a third party would not infr-inge privately owned rights PTENT STTUS US/DOE Patent Clearance is not required prior to publication of this document 2
# Executive Summary Vapor and liquid equilibrium compositions have been measured for the nhexane/squalane system) at 1966 'C, 2511 'C, 313 'C, and 35 O C at pressures ranging from 55 to 548 bar Mixture critical pressures at 356,482, and 548 bar were obtained by the visual observation of critical opalescence within the view cell Because of the large difference in molecular weight between squalane (MW = 42283) and hexane (M6= 8618), the reproducibility of the reported mole fractions are highly dependent on the squalane content in the phase of interest For liquid phases containing less than 5 mol % squalane, samples are reproducible to within an error f7% in squalane composition; for those containing more than 5 mol % squalane, the error was always less than e% For the vapor phase, deviation between samples never exceeded +7% of the squalane composition, and for most samples reproducibility was better than +3% The reported temperatures are accurate to within k3 "C,and pressure variations during a run were always less than +O 14 bar The experimental data were modeled using two equations of state, Peng-Robinson QPR) and the Statistical ssociating Fluid Theory (SFT) Using the P-R equation with the optimized binary interaction parameters, the n-hexane concentration in the liquid phase is predicted to an average error of 6% n the vapor phase, the squalane concentration is predicted to an average error of 23% lthough the interaction parameters are relatively small, they do not follow a well-behaved trend that lends itself to extrapolation to other, unmeasured temperatures The interaction parameter for SFT is better behaved, as it is essentially constant over the temperature range explored The fit to the liquid-phase concentration is excellent at all temperatures (ie, no more than 2% devation in hexane concentration), but the predicted concentration of squalane in the vapor phase is off by an unacceptably large 73% f, as we had originally intended, SFT is to be used to model ow proposed SCF extraction process, it must be capable of accurately predicting the partioning of the Fischer-Tropsch waxes between the liquid and vapor phases (ie, K-values) Whether the inability of SFT to model hexane-squalane simply means that we need to quantify the effect of branching, or that the basic model itself is flawed, will be the subject of future work 3
Technical Objectives / The objective of this research project is to evaluate the potential of SCF extraction for separating the catalyst slurry of a Fischer-Tropsch (F-T) slurry bubble column (SBC) reactor into two fractions: (1) a catalyst-free wax containing less than 1 ppm particulate matter and (2) a concentrated catalyst slurry that is ready for recycle or regeneration The wax will be extracted with a hydrocarbon solvent that has a critical temperature near the operating temperature of the SBC reactor, Le, 2-3 'C nitial work is being performed using n-hexane as the solvent The success of the project depends on two major factors First, the supercritical solvent must be able to dissolve the F-T wax; furthermore, this must be accomplished without entraining the solid catalyst Second, the extraction must be controlled so as not to favor the removal of the low molecular weight wax compounds, ie, a constant carbon-number distribution of the alkanes in the wax slurry must be maintained at steady-state column operation To implement our objectives, the following task structure is being implemented: Task 1: Equilibrium Solubility Measurements a apparatus modification and construction b experimental measurement of selected model systems c catalyst/wax separation studies Task 2: Thermodynamic Modeling a programming and testing of SFT equation for nonassociating systems b programming and testing of SFT equation for associating systems c modeling measured results with the SFT equation d pure component and mixture SFT parameter determination for selected model systems Task 3: Process Design Studies a integration of our SFT program into a process simulation package b process configuration studies using above simulation package 4
Detailed Description of Technical Progress ' Task la pparatus Modification and Construction utomation of the small-scale SCF apparatus is still under consideration Some details of the control system still need to be examined c Vapor-liquid equilibrium experiments for the n-hexane/squalane system were completed Measured compositions and corresponding pressures for the n- hexanelsqualane binary at 1966 'Cy 2511 'C, 313 'Cy and 35 OC are given in Table and are depicted on a pressure-composition diagram in Figure 1 For clarity, the vaporphase compositions are plotted in Figures 2 and 3 Both the liquid and vapor phases have been checked for consistency by the methods discussed in the previous quarterly report and are observed to be internally consistent (Figures 4 and 5) Because of the large difference in molecular weight between squalane (MW = 42283) and hexane (MW = 8618), the reproducibility of the reported mole fractions are highly dependent on the squalane content in the phase of interest This occurs because our analytical technique for determining phase compositions is based on wt % For liquid phases containing less than 5 mol % squalane, samples are reproducible to within an error k7% in squalane composition; for those containing more than 5 mol % squalane, the error was always less than 32% The reproducibility of the vapor-phase compositions was in general excellent when one considers the small amounts of squalane that were present in the vapor phase, especially at the lower temperatures Deviation between samples for a given temperature and presssure never exceeded f7 % of the squalane composition, and for most samples reproducibility was better than +3 % The reported temperatures and pressures are believed to be accurate to -13 OC and -114 bar, respectively Temperatures were measured with a scheme consisting of top- and bottom-phase thermocouples and a secondary standard RTD that was recently calibrated against a primary standard Rosemount 162CE RTD The Heise pressure gauge used for our measurements was calibrated with a Budenburg 38H dead weight tester 5
Table Vapor-liquid equilibrium for the n-hexane/squalane system Exp Data Calculated by P-R Calculated by SFT mole fraction mole fraction mole fraction Press n-hexane squalane bar liquid vapor T = 1966 OC 552 689 823 945 134 169 127 1324 1344-1538 1655 1731 1741a 436 526 68 679 714 729 787 841 85 921 956 978 1 17 15 13 12 11 11 97 97 96 62 4 7 454 547 631 697 743 76 823 871 879 949 987 22 OOOf 9 17 16 15 14 13 11 11 62 19 41 41 37 27 41 43 46 36 35 31 32 24 241 32 291 311 299 319 138 111 2-523 429 511 586 647 692 79 776 833 843 942 998 36 29 24 21 19 18 16 13 12 7 4-17 -28-37 -47-31 -27-13 -9-7 23 44 862 1138 1613 213 2517 2889 3413 3516 3565b 1 151 1999 2517 334 3482 3896 424 4626 4695 4823 827 1517 2172 2861 3551 424 493 5137 5343 5481 428 514 63 1 734 813 878 954 971 985 323 466 55 613 71 756 82 845 89 9 1 946 239 357 473 571 66 728 81 826 856 189 161 155 169 186 232 457 629 147 92 78 73 75 86 1 13 162 262 32 543 362 256 238 242 26 3 19 455 547 733 896 14 a Vapor pressure of pure n-hexane Vapor-liquid critical point 445 55 1 699 81 1 881 93 1 317 45 563 666 757 827 886 193 337 459 575 682 787 T=2511 "C 179-39 58 166 72-38 167 18-113 183 14-76 26 84-84 255 6-19 T = 313 OC 84-18 -87 79-33 13 85 24 164 1 87 333 125 8 453 164 94 64 233 14 792 T = 35 OC 296-193 -182 237-55 -74 241-3 -13 277 7-145 352 34 354 512 81 65 41 1 52 634 745 827 896 994 324 444 54 627 73 761 811 85 892 899 225 369 48 578 66 732 796 814 832 5 42 35 33 34 36 24 36 29 27 27 3 34 41 49 68 73 175 121 17 16 114 132 165 18 198-41 -23 5 15 17 2 42 4-46 -18-23 29 7 11 6 2-2 6 35 14 12 5 6 14 28-791 -89-817 -828-832 -838-835 -866-875 -89-99 -737-738 -767-86 -821-843 -949-69 -628-63 -635-651 -66-685 -698-74 -758-517 -527-55 -562-562 -586-637 -67-73 O 6
Previous work at Clemson has shown that equilibrium conditions are obtained with our continuous-flow apparatus design Confirmation that equilibrium conditions do indeed / exist for our current system of interest was obtained by collecting samples at varying flow rates at 313 "C and 348 bar s shown in Table 11, no effect of flow rate on composition was observed Table 11 Measured equilibrium concentrations and flow rates for the n-hexanekqualane system at 313 "C and 3482 bar total flow rate ( m k ) 1 2 3 liquid mole fraction n-hexane 757 756 757 vapor 477 474 478 f l No effort planned for this quarter Systems No effort planned for this quarter ~ SFT Parameters Two equations of state were tested for their ability to correlate our experimental data, Peng-Robinson (PR) and Statistical ssociating Fluid Theory, or SFT PengRobinson is a popular cubic equation of state that has been in both academic and industrial use for almost 2 years, t works well for ideal or nearly ideal solutions that are well-defined SFT, developed in the late SOs, is a noncubic equation of state based on perturbation theory However, unlike earlier perturbation models, the reference fluid incorporates both chain length and molecular association The advantages of SFT are reported to be its ability to model undefined mixtures and highly nonideal systems, and 7
the fact that both its pure component and interaction parameters are well-behaved However, compared to cubic equations of state it is, to say the least, complex 8 n order to use PR in its most conmionly applied form, a knowledge of both pure component critical properties and acentric factors is required lthough these data are readily available for n-hexane, they are not for squalane Furthermore, several popular methods for estimating T, and P, could not be used because no boiling-point data are rr available for squalane Thus, the method of Fedors', which requires only structural information, was employed and gave a T, of 848 K nother structural contribution method by mbrose' was used to calculate a P, of 6 bar To determine the acentric factor, Hutchenson's* program for the PR equation, which requires pure component vapor pressures as the input, was used Using the calculated critical properties and two vapor pressure values from the literature, we obtained an acentric factor of 9695 With this information, flash calculations were then performed for the n-hexanehqualane system using PR Binary interaction parameters (tjij) were determined for each temperature by optimizing the fit of the equation to the experimentally measured liquid and vapor compositions s seen in Table 111, the 6,s are relatively small; however, they do not follow a well-behaved (eg, linear) trend Based on these results, the parameter values that one would use for extrapolating to unmeasured temperatures are uncertain Using the P-R equation with the optimized interaction parameters, the n-hexane concentration in the liquid phase is predicted to an average error of 6% n the vapor phase, the squalane concentration is predicted to an average error of 23 % Results at 35 "C are shown in Figure 6 and are representative of the fits obtained at the other temperatures T ("C) PR SFT 1 2 1966 "C -3 2 2511 "C -5 1 313 "C 2 Reid, Prausnitz, and Poling, Properties of Liquids and Gases, 4th ed, Chapter 2 Hutchenson, PhD Dissertation, Clemson University, Dec 199 8 35 "C 5 2
SFT was also used to model the binary system n-hexanehqualane The pure component parameters for hexane were calculated fiom the correlations for n-alkanes #, given in the publication by Huang and Radosz3 For this initial work, the same correlation was also used for squalane, because no correlation for branched alkanes exists at this time Thus, squalane was assumed to act no differently than triacontane (n-c,,) s with P-R, the fit of SFT to the experimentally measured data was also optimized c using an interaction parameter (k,) For SFT, kij was essentially optimized against the liquid composition only because varying k, had little effect on the vapor-phase concentration The k,s obtained are shown above in Table 111 Compared to PRYwe see that the interaction parameter for SFT is better behaved; for extrapolating to temperatures above or below those measured, k, can be estimated with confidence to be about 2 Unfortunately, for the correlations used, SFT does not do a good job of correlating the measured data lthough the fit to the liquid-phase concentration is excellent at all temperatures (ie,, no more than 2% devation in hexane concentration), the predicted concentration of squalane in the vapor phase is off by an unacceptably large 73 % Results for SFT at 35 "C are seen in Figure 6 complete listing of the phase compositions calculated using both equations of state, as well as the percent deviation between experimental and calculated values, is shown in Table f, as we had originally intended, SFT is to be used to model our proposed SCF extraction process, it must be capable of accurately predicting the partioning of the Fischer-Tropsch waxes between the liquid and vapor phases (Le, K-values) Whether'the inability of SFT to model hexane-squalane simply means that we need to quantify the effect of branching, or that the basic model itself is flawed, will be the subject of future work u $ No effort planned for this quarter Huang and Radosz, 'Equation of state for small, large, polydisperse, and associating molecules', & EC Res 199,29, 2284-2294 3 9
Plans for Next Quarter # Equilibrium phase compositions and mixture critical points will be measured and correlated for binary mixtures of n-hexane with an n-alkane such as hexadecane or eicosane n addition, currently available VLE data will be used to evaluate the ability of SFT to accurately predict both liquid and vapor compositions for mixtures of light aad c heavy alkanes 1
6 c W e 3 -- 1966 "C o 2511 "C 313"C O e * n rj ' n @l o 35"C O -- $ o!,+ 2 -- 4 -- 3 5 -- n L + 1 d *e 1 2 4 6 x,y (mole fraction n-hexane) 8 Figure 1 Pressure-vs-composition diagram for the nhexanekqualane system at 1966 "C, 2511 "C, 313 "C, and 35 "C The + are the mixture critical temperatures
18 16 14 12 n L tu s v n 1 8 6 + 4 2 9997 9998 9999 y (mole fraction n-hexane) Figure 2 Vapor-phase compositions at 1966 "C 1
" 6 5 4 n L d Y e 2511 "C o 313"C 35 "C 3 2 O 92 94 96 98 1 8 y (mole fraction n-hexane) Figure 3 Vapor-phase compositions at 2511 "C, 313 "C, and 35 "C 1
n 4 1 9 8 7 + 1966 "C + 2511 "C 313 "C x 35 "C 6 5 / 4 3 2 1 2 4 6 8 1 x (mole fraction n-hexane) Figure 4 Modified Henry's Law constant vs liquid composition at 1966 "C, 2511 "C, 313 "C, and 35 "C
) 4 35 LJ 2511 "C u 313"C 35"C 3 25 w S 2 3 1, 5 O a 1 a * 5 O 2 3 4 5 P (bar) Figure 5 Natural logarii i m of the enhancement factor vs pressure at 2511 "C, 313 "C, ant 35 "C 6
/ 5 --- Expt Data SFT P-R / / / 4 / 4 * n L 2 Y e 3 1 4 ) 2 "i 1 yl 2 4 6 x,y (mole fraction n-hexane) 8 Figure 6 Comparison of Peng-Robinsonand SFT equations with experimental data at 35 "C 1