<<

Eng. Japan, 16, 324 (1983). 38, 567 (1983). ll) Taylor,G. I. andA. D. McEwan: /. FluidMech., 22, 1 (1965). 13) Yoshida, F., M. Yamaguchi and T. Katayama: /. Chem, Eng. 12) Terasawa, H., Y. H. Mori andK. Komotori: Chem. Eng. ScL, Japan, 19, 1 (1986).

HIGH-PRESSURE TERNARY LIQUID-LIQUID EQUILIBRIA FOR THE SYSTEMS OF WATER, (1-PROPANOL, 1-BUTANOL AND 1-PENTANOL) AND 1,1-DIFLUOROETHANE AT 323.2K

TAKASHI NAKAYAMAAND HIROSHI SAGARA Research & Development Div., JGC Corporation, Yokohama 232 KUNIO ARAI AND SHOZABUROSAITO Department of Chemical Engineering, Tohoku University, Sendai 980

Key Words: Liquid-Liquid Equilibrium, 1,1-Difluoroethane, 1-Propanol, 1-Butanol, 1-Pentanol, Water, , Fusel Oil, Distribution Coefficient

For the separation of fusel oil from low-concentration broths, ternary liquid-liquid equilibria at 323.2 K were measured for the systems of water and 1,1-difluoroethane with (l-propanol, 1-butanol and 1- pentanol). The liquefied gas 1,1-difluoroethane can achieve much higher distribution coefficients of l-propanol, 1- butanol and 1-pentanol than ethanol. Therefore, l-propanol, 1-butanol and 1-pentanol will be selectively removed from low-concentration ethanol aqueous solutions with liquefied 1,1-difluoroethane at low ratios. The liquid-liquid equilibrium data were correlated by the UNIQUACand NRTLequations. The UNIQUAC equation gave better correlation of the experimental data than did the NRTLequation. However, the two-phase regions estimated by the UNIQUACequation were larger than the experimental results for the systems of water + l-propanol + 1,1-difluoroethane and water + 1-pentanol + 1,1-difluoroethane.

we selected 1-propanol (PrOH), 1-butanol (BuOH) Introduction and 1-pentanol (AmOH) as representative alcohols In the refining of ethanol from low-concentration contained in fusel oil, and measured ternary liquid- fermentation broths, ethanol enrichment and the re- liquid equilibria for the systems H2O-PrOH-DFE, moval of impurities such as fusel oil from ethanol are H2O-BuOH-DFE and H2O-AmOH-DFE. The mea- usually accomplished by energy-intensive distillation. sured alcohol distribution coefficients are compared To find an alternative gas-liquid or liquid-liquid with the ethanol distribution data reported in a extraction process, manyresearchers have studied the previous work.7) In addition, the liquid-liquid dehydration of ethanol with supercritical ethane,6) equilibrium data are correlated with the UNI- supercritical or liquefied dioxide4'9'1M3) and QUAC,1} NRTL10) and LEMF (local effective mole liquefied l,l-difluoroethane.7) However, the sepa- fraction equation)5} solution models. ration of impurities from dehydrated or aqueous ethanol has not yet been reported. 1. Experimental In the present work, we selected liquefied 1,1- The static method was used for obtaining high- difluoroethane (DFEhereafter) as an extraction sol- pressure liquid-liquid equilibrium data. The experi- vent for the separation of fusel oil from ethanol mental apparatus and procedure are almost the same aqueous solutions, since it is immiscible in water and as reported in a previous paper.7) A schematic dia- is expected to show higher selectivity for the more gram is shown in Fig. 1. Liquid samples are with- lipophilic fusel oil than ethanol. DFE also has low drawn from the equilibrium cell ®to the samplers toxicity. To establish a basis for fusel oil extraction, (2) through capillary tubes and then flushed to the evacuated sampling lines. The pressure decrease in Received June 29, 1987. Correspondence concerning this article should be addressed to T. Nakayama. the equilibrium cell that occurs during sampling is

VOL. 21 NO. 2 1988 129 Fig. 1. Schematic diagram of experimental apparatus automatically compensated for by the action of the diaphragm (3) connected to both equilibrium cell and buffer tank ®. Heavier polar componentstended to condense in the samplers and therefore we found it necessary to wrap the samplers with heat tapes. The present experimental procedure was checked by measuring binary liquid-liquid equilibria for the H2O-BuOH and H2O-AmOHsystems at 323.2 K. As shown in Fig. 2, the binary mutual solubility for both systems (solid lines) increases slightly with increasing pressure. The present measurements are in good agreement with the literature values.2'3'8'12* Materials used DFE having a quoted purity of 99.7+mol% was supplied by Daikin Ind. Co., Ltd. Special reagent-grade alcohols were supplied by Mole traction ot BuOH or AmOH WakoPure Chemical Ind. Ltd. and were used without further purification. Gas chromatograph analysis Fig. 2. Liquid-liquid equilibria for the H2O-BuOH and H2O-AmOHsystems at 323.2K. Experimental data for the with a thermal conductivity detector indicated the H2O-BuOH system: V, Fuehner;2) A, Hill et ai;3) D, purities to be 99.9mol%, 99.9mol% and 98.5mol% Othmer et al.;8) O, this work. Experimental data for the for 1-propanol, 1-butanol and 1-pentanol, respec- H2O-AmOHsystem: A, data smoothed by S^rensen et al.;12) tively. Water content of reagent 1-propanol, 1-bu- #, this work tanol and 1-pentanol was 0.1, 0.1 and 1.5mol%, respectively. Water was ion-exchanged and purified has two immiscible binary pairs and no plait point. As by distillation. DFEwas passed through a molecular the alcohol carbon numberincreases, the slopes of tie sieve 5A bed for calibration of the gas chromato- lines become greater. graph, but was otherwise used without purification 2.1 Extraction of fusel oil components from ethanol in the experimental measurements. aqueous solutions As shown in Fig. 6, the measured alcohol distri- 2. Results and Discussion bution coefficients are compared with those for the Tie line data are listed in Tables 1, 2 and 3. water-ethanol-DFE system.7) The plait points esti- Representative liquid-liquid equilibria at 323.2 K are mated by Treybal's method14) for the H2O-PrOH- shown in Figs. 3, 4 and 5, respectively for the H2O- DFE and H2O-EtOH-DFE systems are given in PrOH-DFE, H2O-BuOH-DFE and H2O-AmOH- Table 4. The alcohol distribution coefficient increases DFE systems at 1.32MPa. The H2O-PrOH-DFE as the alcohol carbon number is raised. This is to be system is classified as liquid-liquid equilibrium Type expected since alcohols with higher carbon numbers 1,15) which has one immiscible binary pair and a plait are more lipophilic in nature. On the other hand, with point, while the H2O-BuOH-DFE and H2O- increasing pressure the solute distribution coefficients AmOH-DFEsystems are classified as Type 2, which increase, except those for ethanol. An increase in

130 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN Table 1. Liquid-liquid equilibria for the H2O-PrOH-DFE system at 323.2K

Mole fraction [-] Pressure TT . , tm-t- . , H2O-nch phase DFE-nch phase

H2O PrOH DFE H2O PrOH DFE

1.32 0.9960 0. 0.0040* 0.0117 0. 0.9883* 0.9776 0.0174 0.0050 0.0289 0.0323 0.9388 0.9582 0.0363 0.0055 0.0790 0.1295 0.7915 0.9490 0.0450 0.0060 0.2385 0.2577 0.5038 0.9452 0.0483 0.0065 0.3863 0.3116 0.3021 0.9387 0.0543 0.0070 0.5036 0.3136 0.1828

6.08 0.9958 0. 0.0042* 0.0145 0. 0.9855* 0.9783 0.0166 0.0051 0.0303 0.0334 0.9363 Fig. 3. Liquid-liquid equilibrium for the H2O-PrOH- 0.9604 0.0341 0.0055 0.0808 0.1284 0.7908 DFE system at 323.2K and 1.32MPa 0.9484 0.0453 0.0063 0.2436 0.2572 0.4992 0.9444 0.0491 0.0065 0.3928 0.3121 0.2951 0.9239 0.0665 0.0096 0.6211 0.2700 0.1089

* H2O-DFE binary mutual solubility data at 323.2K were reported in a previous work.7)

Table 2. Liquid-liquid equilibria for the H2O-BuOH-DFE system at 323.2K Mole fraction [-] Pressure . ___ . TMP1 H2O-nch phase DFE-nch phase H2O BuOH DFE H2O BuOH DFE

1.32 0.9917 0.0040 0.0043 0.0215 0.0317 0.9468 0.9877 0.0080 0.0043 0.0734 0.1323 0.7943 0.9865 0.0091 0.0044 0.2021 0.2734 0.5245 Fig. 4. Liquid-liquid equilibrium for the H2O-BuOH- 0.9850 0.0112 0.0038 0.3638 0.4131 0.2231 DFE system at 323.2K and 1.32MPa 0.9838 0.0161 0. 0.5497 0.4503 0.

Table 3. Liquid-liquid equilibria for the H2O-AmOH-DFE system at 323.2K Mole fraction [-] H2O-rich phase DFE-rich phase

H2O AmOH DFE H2O AmOH DFE

1.32 0.9953 0.0007 0.0040 0.0158 0.0120 0.9722 0.9942 0.0019 0.0039 0.0776 0.1292 0.7932 0.9940 0.0022 0.0038 0.2074 0.2723 0.5203 0.9935 0.0028 0.0037 0.2280 0.3175 0.4545 0.9959 0.0041 0. 0.3923 0.6077 0.

6.08 0.9952 0.0008 0.0040 0.0180 0.0130 0.9690 0.9940 0.0019 0.0041 0.0790 0.1338 0.7872 0.9939 0.0022 0.0039 0.2020 0.2818 0.5162 Fig. 5. Liquid-liquid equilibrium for the H2O-AmOH- 0.9938 0.0026 0.0036 0.2351 0.3154 0.4495 DFE system at 323.2K and 1.32MPa 0.9958 0.0042 0. 0.3989 0.6011 0.

ecules in the liquefied gas phase. pressure causes a larger increase in the liquid density Selectivity curves for the water-alcohol-DFE sys- of the liquefied gas phase than in that of the water tems at 323.2K and at pressures of 1.32, 6.08 and phase. This maybe partly attributed to an increase in 10.13MPa are shown in Fig. 7. The selectivity /? of the attractive force between DFEand solute mol- solvent for alcohol is defined as

V0L 21 NO. 2 1988 131 Fig. 6. Alcohol distribution coefficient curves for the water-alcohol-1,1-difluoroethane systems at 323.2 K. Ex- perimental data: O, #, Nakayama et al.;7) A, A, å¡, V, Fig. 7. Selectivity curves for the water-alcohol-1,1- T, this work. Estimated plait points by Treybal's method: C, difluoroethane systems at 323.2 K. Experimental data: O, à", ©,A,A Nakayamaet al:^ A, A, D, V, å¼,this work. Estimated plait points by Treybal's method: C, ©, A, A Table 4. Plait points estimated by Treybal's method for the systems H2O-PrOH-DFE and H2O-EtOH-DFE at 323.2K concentration ethanol aqueous solutions with DFE results in low solvent ratios and will lead to a more c Pressure Mole fraction S ystem Component r _ energy-efficient and economical separation process, in [M Pa] [-] which fusel oil extraction is perfomed prior to ethanol H2O-PrOH-DFE 1.32 H2O 0.924 separation. PrOH 0.067 2.2 Correlation of data DFE 0.009 The ternary liquid-liquid equilibria containing a 6.08 H2O 0.909 liquefied gas were correlated at each pressure by the PrOH 0.080 UNIQUAC, NRTL and LEMF equations. Each DFE 0.011 binary interaction parameter of immiscible pairs H2O-EtOH-DFE 1. 32 H2O 0. 368 (H2O-DFE, H2O-BuOH and H2O-AmOH) was de- EtOH 0.323 DFE 0.309 termined from the binary data in Table 1, 2 and 3, 10.13 H2O 0.385 respectively. The other interaction parameters of mis- EtOH 0.313 cible pairs were determined by reproducing the ter- DFE 0.302 nary mutual solubility as well as the solute distri- bution coefficients.7) Binary parameters of solution models are tabulated P=D2IDX (1) in Table 5. For all models there appears to be a slight pressure dependence in the fitted coefficients. The where D2 and Dx are distribution coefficients of mutual solubility curves calculated by the solution alcohol and water, respectively. Higher selectivities models and the tie lines calculated by the UNIQUAC for 1-propanol, 1-butanol and 1-pentanol with respect model are shown in Figs. 3, 4 and 5. The UNIQUAC to ethanol can be obtained in the low-alcohol con- equation gives an excellent representation of the centration range in the water phase. Each selectivity H2O-BuOH-DFEsystem, while the NRTLequation also increases and approaches a constant according gives better results than LEMF. The UNIQUACand with decreasing alcohol concentration in the water NRTLequations also give better correlation for the phase. In addition, the selectivities are not affected by H2O-AmOH-DFEsystem than LEMF. However, pressure, except that for ethanol in a range of solute the solubility curves in the liquefied gas phase are not concentration less than 25 wt%in the water phase. reproduced precisely. On the other hand, the The extraction of fusel oil components from low- UNIQUAC and LEMF equations give better cor-

1O«J JOURNAL OF CHEMICAL ENGINEERING OF JAPAN Table 5. Binary interaction parameters of solution models for the systems H20-Pr0H-DFE, H20-Bu0H-DFE and

Interaction parameter [K] Pressure UNIQUAC NRTL [MPa]

A,, *» An

H2O(l)-PrOH(2)-DFE(3) 1.32 1 2 203.26 -14.105 479.90 40.743 0.20 105.93 306.55 1 3 468.72 722.16 1336.2 992.54 0.24 361.27 410.21 2 3 -67.185 358.13 -217.02 381.49 0.20 283.52 153.86

6.08 1 2 206.79 - 14.364 481.21 48.404 0.20 75.930 306.55 1 3 466.37 676.22 1329.8 929.09 0.24 350.02 409.74 2 3 -48.519 325.10 -217.83 375.53 0.20 283.52 153.86 H2O(l )-BuOH(2)-DFE(3) 1.32 1 2 355.25 -59.790 2450.6 - 894.69 0.12 367.81 2 3 54.188 202.40 -26.567 609.80 0.20 267.89 H2O(l )-AmOH(2)-DFE(3) 1.32 1 2 312.10 35.291 4106.4 - 1779.6 0.06 576.01 2 3 - 34.347 293.01 -46.674 802.21 0.20 302.76

6.08 1 2 313.87 30.505 4119.5 -1793.1 0.06 100.56 2 3 - 34.497 292.49 -47.246 803.69 0.20 305.27

Aji =(up-uu)/R for the UNIQUACequation. Aj^igji-g^/R for the NRTL (o.-ay) and LEMF (a^=aj7= - 1) equations. relation for the H2O-PrOH-DFEsystem than does NRTL and LEMF models. The UNIQUACequation NRTL.However, all the models fail to represent the gave better data correlation than the NRTL and solubility curves in the liquefied gas phase. This liquid LEMF equations, but was unable to reproduce the mixture represents a solutropic system, whose solute two-phase regions for the systems H2O-PrOH-DFE distribution coefficient curve crosses unity except for and H2O-AmOH-DFE. the plait point as shown in Fig. 6. The alcohol distribution coefficient curves and selectivity curves Nomenclature predicted by the UNIQUACequation, as shown in An binary interaction parameter defined by Figs. 6 and 7, are in good agreement with the (Uji-u^/R for the UNIQUACeq. and experimental results except for the selectivity for 1- by {gji-gid/R for the NRTLand [K] propanol. A LEMFeqs. distribution coefficient of component /, In data reduction of the high-pressure liquid-liquid weight fraction of component / in the equilibria for the systems of water, alcohols and DFE, 1 , 1 -difluoroethane phase/weight fraction the UNIQUAC equation gave the best data cor- of component / in the water phase [-] relation. However, an improvement in the energy of interaction between ay'- / pair of molecules [J/mol] UNIQUACequation may be necessary for practical gas constant [J/mol - K] use, such as in the design of extractors. UNIQUACbinary interaction C onclusion parameter [J/mol] High-pressure ternary liquid-liquid equilibria at = nonrandomnessparameter H 323.2 K were measured for the systems H2O-PrOH- =selectivity DFE, H2O-BuOH-DFE and H2O-AmOH-DFE. (Subscripts) The selectivities of DFE for 1-propanol and 1-pen- / = component / tanol are not affected by pressure, but the alcohol j = componentj distribution coefficients rise with increasing pressure. Liquefied DFE can be an effective solvent for fusel oil Literature Cited extraction, since it gives higher distribution coef- 1) Abrams, D. S. and J. M. Prausnitz: AIChE /., 21, 1, 116 (1975). ficients of 1-propanol, 1-butanol and 1-pentanol than 2) Fuehner, H.: Ber. Dtsch. Chem. Ges., 57, 510 (1924). ethanol and shows higher selectivities for these al- 3) Hill, A. E. and W. M. Malisoff: J. Am. Chem. Soc, 48, 918 cohols than ethanol. (1926). The data were correlated with the UNIQUAC, 4) Kuk, M. S. and J. C. Montagna: "Chemical Engineering at

VOL 21 NO. 2 1988 133 Supercritical Fluid Conditions," p. 101, Ann Arbor Science Chemical Engineers, Japan, S-C6 (1983). Publishers, Michigan (1983). 12) S^rensen, J. M. and W. Arlt: "Liquid-Liquid Equilibrium 5) Marina, J. M. and D. P. Tassios: Ind. Eng. Chem. Proc. Des. Data Collection," p. 236, DECHEMA(1980). Develop., 12, 271 (1973). 13) Takishima, S., K. Saiki, K. Arai and S. Saito: J. Chem. Eng. 6) McHugh, M. A., M. W. Mallet and J. P. Kohn: "Chemical Japan, 19, 48 (1986). Engineering at Supercritical Fluid Conditions," p. 1 13, Ann 14) Treybal, R. E., L. D. WeberandJ. F. Daley: Ind. Eng. Chem., Arbor Science Publishers, Michigan (1983). 38, 817 (1946). 7) Nakayama, T., H. Sagara, K. Arai and S. Saito: Fluid Phase 15) Treybal, R. E.: "Liquid Extraction," p. 30, McGraw-Hill Equilibria, 38, 109 (1987). Book Company, Inc. (1963). 8) Othmer, D. F., W. S. Bergen, N. Shlechter and P. F. Bruins: Ind. Eng. Chem., 37, 9, 890 (1945). (A part of this work was presented at the 19th Autumn Meeting 9) Paulaitis, M. E., R. G. Kander and J. R. DiAndreth: Ber. of the Society of Chemical Engineers, Japan, at Nagoya, 1985 and Bunsenges. Phys. Chem., 88, 869 (1984). at The 51st Annual Meeting of The Society of Chemical Engineers, 10) Renon, H. and J. M. Prausnitz: AIChEJ., 14, 135 (1968). Japan, at Osaka, 1986.) ll) Sagara, H.: Preprint of the Fukuoka meeting of the Society of

LASER-DOPPLER MEASUREMENTS OF TURBULENCE STRUCTURE IN A DRAG-REDUCING PIPE FLOW WITH POLYMER INJECTION

HIROMOTO USUI, MATSURUKODAMAAND YUJI SANO Department of Chemical Engineering, Yamaguchi University, Ube 755

Key Words : Drag Reduction, Polymer Injection, Turbulence, Laser Doppler Velocimeter, Polymer Additive Aqueoussolutions of polyethylene oxide were injected into a pipe flow through a small tube at the center of a pipe. The turbulent characteristics in drag-reducing flow with polymer injection were measured by means of a Laser Doppler Velocimeter (LDV). The experimental results were compared with measurements both in a Newtonian fluid (water) flow and in a premixed flow with 300 ppm homogeneous polymer solution. The experimental results suggested that the polymer injection caused a thickening of the buffer layer, enlargement of macroscale turbulent eddy and suppression of fine turbulent eddy. A difference in turbulent characteristics between premixed and polymer injected systems was observed in the distributions of turbulent macroscale, skewness factor and flatness factor. The enlargement of macroscale was moresignificant in the turbulent core region in the case of polymer injection experiments. The measurements of skewness and flatness factors showed that the low-speed fluid element caused by the ejection process passed more clearly the measuring position near the outer boundary of the buffer layer in the case of polymer injection experiments.

dosing system. Injected polymer solution does not Introduction mix soon with the main flow. Heterogeneous flow Drag reduction by use of a polymer additive is condition is maintained over a considerably long important as a power-saving technology in fluid downstream distance from the polymer dosing transportation. In 1979, the Trans-Alaska Pipe-Line station. introduced the use of polymer injection technology to Drag reduction in premixed, i.e. homogeneous, reduce pumping power consumption. Other appli- dilute polymer solutions has been well documented by cations such as flow improvement in sewer systems many investigations published during the last two and slurry transportation have also been reported decades. However, reports on heterogeneous drag recently. Dosing of polymer additives into the tube reduction through polymer injection are rather scarce. flow is usually accomplished by injecting a relatively McComband Rabie10) made precise LDVmeasure- concentrated polymer solution by means of a suitable ments in a heterogeneous drag-reducing flow with centerline polymer injection. Their experimental re- Received July 3, 1987. Correspondence concerning this article should be addressed to H. Usui. M. Kodama is now with Diafoil Co., Ltd. sults showed that the change of turbulent flow charac-

134 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN