Condensation of Acetol and Acetic Acid Vapor and ... - ACS Publications


Condensation of Acetol and Acetic Acid Vapor and...

0 downloads 92 Views 2MB Size

Article pubs.acs.org/IECR

Condensation of Acetol and Acetic Acid Vapor and Nitrogen Using Sprayed Aqueous Liquid Leland C. Dickey,*,§ Akwasi A. Boateng, Neil M. Goldberg, Charles A. Mullen, and David Mihalcik Eastern Regional Research Center, Agricultural Research Service, United States Department of Agriculture, 600 E. Mermaid Lane, Wyndmoor, Pennsylvania 19038, United States ABSTRACT: A cellulose-derived fraction of biomass pyrolysis vapor was simulated by evaporating acetol and acetic acid with heated nitrogen. After generating the vapor/nitrogen mixture, it was superheated in a tube oven and condensed by contact with an aqueous mist created by an ultrasonic spray nozzle. The rates of condensation and fractions of vapor condensed were determined for various fluid flow rates to estimate liquid flow rates necessary to obtain complete acetol vapor condensation. The effect of noncondensible gas on the condensation was determined by varying the ratio of nitrogen to acetol vapor. For standard conditions, no effect of nitrogen content on rate was seen for vapor contents greater than 10 mol %, which would include any practical fast pyrolysis vapor. Increasing the concentration of condensed vapor in the aqueous solution spray, up to 30 wt %, had little effect on the condensation rate.

1. INTRODUCTION Production of liquid fuel by fast pyrolysis of biomass is the subject of ongoing development efforts worldwide.1 Some of the work has included a quench of pyrolysis vapor using liquid sprays.2−4 Condensation of a mixture of hot vapor, on fine droplets of sprayed liquid, has not been much examined. A recent patent application5 sketches an outline of this process and provides a target solvent flow rate to optimize the condensation for several solvents, including water, by matching the latent heat of the vapor to that of the evaporating liquid. The evaporation/condensation scheme will lower the cooling cost by replacing the need to chill a heat transfer surface to exchange sensible heat with the vapor. Achieving the implied equilibrium conditions5 may require prolonging vapor contact with the liquid droplets delaying condensation and possibly allowing conversion of pyrolysis vapors to less desirable compounds. Another recent publication6 compared molecular dynamics simulations with experimental conditions to resolve inconsistencies in the kinetics of mass transfer at gas−water interfaces. The authors noted that a remaining question is the coupling of heat and mass transfer between the condensing molecules and a liquid surface. Older, less detailed studies have been published describing the heat transfer between a water spray used to quench hot hydrocarbon streams.7 Upgrading bio-oil by molecular distillation has been shown to remove the most troublesome acids.8 Acetol (hydroxyacetone) and acetic acid (AA) are bio-oil components present in relatively large amounts (10 and 6% of the oil from corn cobs).9 They were selected for this study to determine if they could be condensed from a mixture, similar to those reported for fast pyrolysis9 containing 70−90% nitrogen. Also conditions that would condense minimal amounts of AA while condensing nearly all of the acetol were sought. To do this, acetol would have to condense at a temperature above the AA dew point, and AA would have not to be entrained by the condensing acetol. AA and other lower dew point compounds are generally considered to be undesirable constituents of This article not subject to U.S. Copyright. Published 2012 by the American Chemical Society

pyrolysis oil and are targets of efforts to reform or eliminate them.

2. METHODS 2.1. Vapor/Nitrogen Feed Stream. A vapor/nitrogen mixture was created from a heated stream of nitrogen flowing into 250 mL flasks containing acetol, supplied by Sigma-Aldrich (St. Louis, MO) or AA, supplied by Fisher Scientific (Pittsburgh, PA) as shown in Figure 1. The total nitrogen flow was heated to 30−40 °C by passing it through a 15 M coil of 0.635 cm diameter copper tube wrapped with a heating tape. The tape temperature was controlled using a thermocouple mounted after the coil outlet and a temperature controller (SOLO SL4848 series). The heated nitrogen could be divided into two streams that flowed through rotameters to flasks on a hot plate/stirrer. The flasks were weighed at the beginning and end of a run. Thermocouples were mounted through rubber stoppers, extending past the bottom of the stopper into the vapor space above the liquid. The pressure of the vapor and gas stream flowing from the flasks was measured using a differential manometer (Omega HHP886W) connected to tees in the lines leading downstream. Nitrogen/vapor mixtures flowed from the flasks through a Stratos tube mixer (15 elements, Koflo Corp), inside a 90 W oven (Supelco 2-3800) to the middle 2.54 cm diameter arm of a 7.62 cm diameter glass tee, the spray/vapor contacting vessel (SVCV). The average temperature inside the SVCV was estimated from the temperature of the spray liquid and condensed vapor leaving through a tube rising through a port in a flange on the bottom of the SVCV. Later runs were condensed in a 10.2 cm diameter tee. The larger tee enabled installation of a gas/vapor outlet and a thermocouple to Received: Revised: Accepted: Published: 5067

August 10, 2011 February 29, 2012 March 13, 2012 March 13, 2012 dx.doi.org/10.1021/ie2017465 | Ind. Eng. Chem. Res. 2012, 51, 5067−5072

Industrial & Engineering Chemistry Research

Article

Figure 1. Diagram of condensation testing equipment.

values than might be effective because of the equipment size, even with the spray at minimum accurate rates. Two runs were begun by sending only heated nitrogen into the SVCV for 2 h. The water loss was 0.21 ± 0.01 g/min. This loss was water vapor or, mostly, droplets flowing out of the SVCV with the nitrogen and was taken as an upper limit for water loss when the temperature in the SVCV was low, 20−40 °C. 2.2. Condensate Measurement. The condensation rate and fraction condensed were calculated using a measurement of the condensed vapor in the spray liquid. For a single vapor in nitrogen, the concentration of either acetic acid or acetol could be calculated from refractive index measurements and refractive index correlations determined with the same instrument (eqs 1 and 2). For acetol

measure the exiting flow, on the flange covering the top of SVCV. The nitrogen flow rate was 3.7 L/min and the vapor rate varied from 0.1 to 3.4 L/min. Liquid that collected at the bottom of the SVCV drained through a glass laboratory condenser to a collection vessel. The weight of the collection vessel was measured with a maximum 6 kg/0.1 g balance (Sartorius 6001). Liquid was pumped with a tubing pump from the collection vessel through a pulse damper to the spray nozzle. The ultrasonic spray nozzle (Sono-Tek 60 kHz) was mounted on the upper end of the SVCV, with the liquid feed flowing vertically downward to the atomizing tip. Nitrogen and uncondensed vapor flowed from the SVCV through another glass condenser to a flask in a cooling bath. The bath provided chilling water for the two glass condensers. An important advantage of the ultrasonic nozzle compared to a pressure driven nozzle is that the spray forms a mist in the nitrogen/ vapor stream and thus the liquid resides in the condenser mostly as droplets. The droplets from a pressure-driven nozzle go directly to the chamber walls so that the vapor/liquid contact is more of a wetted wall than vapor/droplet. The vapor mass sent to the SVCV was taken as the difference of the flask(s) contents’ weights before and after a run. The mass of vapor condensed in the SVCV was determined by weighing the collection vessel and contents before (usually containing 1 L of water to begin) and along with measures of condensate concentration: refractive index (r.i.), at 15 min intervals during a run and Karl Fischer water content measurement at the run’s end. The runs usually were 350 min or until the acetol or AA in the flasks ran out. When a mixture of acetol and acetic acid vapor was fed to the SVCV, their mass ratio was controlled by adjusting the nitrogen flow rates to the flasks containing the liquids. The mass ratio of condensable vapor (both acetol and AA) to nitrogen was increased by increasing the temperature of the hot plate under the flasks (three settings were used). The vapor mass flow rate with this system limited the vapor/spray mass ratio to lower

wt% = 0.0147x 2 + 1.0775x

(1)

For acetic acid wt% = 0.0443x 2 + 1.500x

(2)

where x is refractive index in brix. The spray liquid composition was determined from readings of liquid weight and refractive index taken every 15 min. When both acetol and acetic acid vapors and nitrogen flowed to the SVCV, the concentrations of acetol and acetic acid were determined by the simultaneous solution of the three equations that follow: the total vapor wt % from the Karl Fischer measurement (eq 3), and inverted versions of eqs 1 and 2.

5068

xacetol = 0.77wt%acetol − 0.0024(wt%acetol )2

(1′)

xAA = 0.54wt%AA − 0.0024(wt%AA )2

(2′)

x mix = xacetol + xAA

(3)

dx.doi.org/10.1021/ie2017465 | Ind. Eng. Chem. Res. 2012, 51, 5067−5072

Industrial & Engineering Chemistry Research

Article

where xmix is the refractive index of both components in the solution, xacetol is the refractive index of acetol, and xAA the refractive index of AA. The method was compared with concentrations calculated using composition of the two condensates in the liquid determined using only the refractive index for data from a series of runs where it was expected that all of the acetol had condensed. A comparison, Figure 2, of the two methods

Figure 3. Vaporization heat/nitrogen heating rate, nitrogen flow rate, 4.6 g/min.

from the condensation was balanced mainly by conduction through the wall of the condenser and water evaporation from the sprayed liquid. With the use of heating and cooling rates calculated from flow rates and temperatures in the condenser, it was determined that condensing AA vapor while evaporating some water from the spray liquid required a spray rate below 20 mL/min. At higher spray rates the sensible cooling was sufficient to condense the vapor (with the bath cooling on). There were two points of temperature measurement within the SVCV: typically 78−79 °C at the inlet and 43 °C at the outlet, and they did not vary when the system reached steady state; average values were used to prepare the plot shown in Figure 5. In Figure 5, the times sign (×) and delta symbol (Δ) ordinate values (heating−cooling) are net SVCV cooling provided by conduction through the (uninsulated) SVCV glass wall. An estimate of the rate for this process was based on the thermal conductivity of borosilicate glass =1 W/(M·K). For our tee, the wall is about 0.59 cm thick and the internal surface is 434 cm2 so conduction through the walls is about 440(ΔT), J/min, where ΔT is the temperature difference between the ambient temperature, which was always within a few degrees of 24 °C, and an average temperature inside the condenser, near the wall. Temperature differences of only a few degrees are sufficient to account for the difference between the calculated heating and cooling.

Figure 2. Compensation of refractive index and Karl Fischer rate calculation methods.

supports the accuracy of the first method for mixtures of acetol and AA, for these runs, and the range of condensation rates that can be obtained, with our system. The mass ratio acetol/AA varied from 0.1 to 5.3. The fraction of acetic acid condensed was calculated by dividing the acetic acid in the spray liquid by the acetic acid fed; that is, from the change in weight of the flask containing the liquid evaporated into the nitrogen stream. Condensation rate was determined by plotting the condensed vapor weights calculated from the refractive index measurements versus the elapsed time of the sample and fitting the points. Future testing would allow unambiguous determination of AA content using a continuous reading pH meter or other instrument suitable for distinguishing acetol and AA. 2.3. Calculation of Energy Flows from Temperature Measurements. The hot plate under the flask containing the liquid acetol or AA supplied the heat to raise the temperature of the liquid and nitrogen bubbling through it. For a series of runs with a nitrogen flow rate of 4.23 g/min, the ratio of energy to evaporate the liquid in the flask/the energy to heat the nitrogen was 3 when the flask contained acetol, 8 for acetol and 20 for water, after the temperature in the flask became steady and before the amount of liquid began to reduce its ability to absorb heat. As can be seen in Figure 3, it took 2 h or more, for the AA runs to reach steady state. Initially it was likely that some liquid drops were aspirated by the nitrogen and blown into the superheater where they evaporated. The steady state of the entire system was obtained when the temperature of the nitrogen/vapor flowing into the superheater stopped rising. Figure 4 shows data for four runs, with nitrogen flows of 4.63 g/min, various mole fractions of vapor, and amounts of heat delivered to the flask contents. Reducing the hot plate heat delivery rate reduced the AA vapor temperature, but had little effect on the pressure. The measured pressures were well above the pressure for saturated vapor (calculated from published values for the Antoine equations for acetol and AA). The pressure in the condenser stayed just above the gas discharge pressure. The heat released

3. RESULTS AND DISCUSSION 3.1. General Observations. Twenty-four runs are included in Figure 2. A mass balance was made on each run, and the mean mass lost was 4.4%, with a standard deviation of 2.7. There was no clear trend in mass loss despite the modifications of the setup and variations of vapor composition and condensation conditions. The source of loss was estimated to have been water vapor not condensed by the second condenser and acetol vapor that may have pyrolyzed in and remained in the superheater. 3.2. Approach to Equilibrium. Equilibrium models used to predict product compositions and flow rates from condensers assume that an (average) temperature can be used to represent liquid, gas, and vapor exiting the condenser. For example the Henry’s law constant, K, for acetol has been reported,10 and assuming the log of the constant varies linearly with the reciprocal of the temperature,11 a dimensionless K 5069

dx.doi.org/10.1021/ie2017465 | Ind. Eng. Chem. Res. 2012, 51, 5067−5072

Industrial & Engineering Chemistry Research

Article

Figure 4. Pressure and temperature of nitrogen/saturated vapor mixtures.

Figure 5. Dependence of conductive heat loss on spray rate.

Figure 6. Water evaporation rate, without cooling, using 7.6 cm diameter condenser.

(=Ca/Cw) for acetol at 24 °C is 74400. Ca is the vapor concentration of acetol, and Cw is the concentration of acetol in the solution. At 24 °C, Ca is about 0.04 kg/M3 and Cw, 5 × 10−7. This is far too low to indicate mass transfer limitation to acetol concentration in the solution coming from the vapor. Consequently the analysis proceeded on the basis of heat transfer-driven condensation. Liquid phase diffusion is slow compared to condensation, and the droplet outer layer temperature will rise as vapor condenses on or near it, while the droplet interior remains near the freshly sprayed droplet temperature. Thus the temperature of the liquid exiting the condenser will be lower than that at the droplet surface, especially when a significant fraction of the cooling in the condenser is due to conduction through the wall. Visual observation of the spray confirmed that the low spray inertia kept the droplets from streaming directly to the wall. The residence time in the SVCV was 10−15 s. Most of the condensation occurred on the drop surfaces with a bulk temperature between the dew point and the incoming liquid temperature. No condensation occurred during experiments unless liquid droplets were sprayed into the SVCV. This was demonstrated during a control run where water flowed through the nozzle, without power to the nozzle, and no droplets formed. 3.3. Evaporation of Water. With the bath off, water evaporation rates were calculated from the weight of the circulating liquid and the condensed vapor in it as determined from r.i. measurements (Figure 6). A spray rate of 2.8 g/min, 40 times the condensation rate (0.07 g/min), condensed 0.8 weight fraction of acetol vapor fed

to the SVCV; for the acetic acid, which condensed completely, the ratio was 7.5 (spray rate 1.5 g/min, condensation 0.20 g/ min). These were single vapor condensations carried out under the same conditions, except spray rate. On the basis of a (linear) fit of the evaporation rate versus the temperature of the nitrogen vapor stream flowing into the condenser, the nitrogen/vapor temperature had to be at least 133 °C, to evaporate water. Increasing the nitrogen/vapor inlet temperature increased the evaporation rate about 0.5% of the vapor flow per °C. The heat −cool = 500 J/min point in Figure 5 (the bath-off point) corresponds to the AA line in Figure 6. The 0.11 g/min of water evaporation at 2260 J/g provided 240 J/ min of cooling and the AA condensation took 80 J/min of cooling. These rates were low relative to conduction through the wall: 11 500 J/min and nitrogen cooling: 11 400 J/min. 3.4. Condensation Rate and Fraction Condensed. One goal of this condensation study was to determine the effect of condensation temperature and spray rate on the condensation of low dew point vapor, represented by AA vapor. When both acetol vapor and AA vapor were fed to the SVCV, and the condensed mass exceeded the acetol mass fed, we assumed that acetol vapor condensed completely and AA condensation made up the difference. The plot of fraction condensed (with an error of about 10%) versus condensation rate (Figure 7) showed no decrease of condensed fraction with condensation rate increase, up to about 2.5 g/min, when the spray rate was 10 mL/min or higher. In the plot legend, vmf is mole fraction of vapor in the nitrogen/vapor mixture flowing to the SVCV. For the mixed runs at low vmf (dark diamonds), the fraction condensed 5070

dx.doi.org/10.1021/ie2017465 | Ind. Eng. Chem. Res. 2012, 51, 5067−5072

Industrial & Engineering Chemistry Research

Article

This may be because the solubility of the vapors in water made the condensation insensitive to temperature and flow rates in the range where they were varied. We expected that generally the pyrolysis vapors will not interfere with each other during condensation. If this is not true then the analysis of condensation experiments (with a view to fractionation) could be very complicated. Although many compounds comprise pyrolysis vapor, they may be represented as a relatively small number of groups with similar dew point. A possible difference between the two-component results we report and the results of complete vapor condensation might be caused by droplet coating with insoluble higher dew point compounds. The coating might prevent dissolution of the water-soluble compounds and therefore change any effect of solubility on condensation. Further work including adding a vapor of an insoluble compound (in aqueous solutions) could show whether the features noted here are changed by an insoluble coating of the droplets. Nearly all the acetol/low AA feeds with vapor content greater than 0.2, vapor feed temperature up to 230 °C, and a ratio of spray mass to vapor mass ratio of 1 or more, condensed. The lowest fraction condensed was obtained at the highest spray rate for runs shown in the Figure 8, so spray rate was not a limiting parameter. The larger SVCV did as well as the smaller one, even at higher vapor content feeds. In general, spray cooling with water may not be able to condense less than 60% of AA, while completely condensing acetol vapor, possibly due to high AA solubility in water. For the runs in which both acetol and AA vapor were fed, at a spray rate of 5 mL/min, increasing the cooling bath temperature from 1.8 to 5 °C, (which controlled the temperature of the spray liquid), reduced the AA vapor fraction condensed from 0.91 to 0.74. These values bound the range where acetol can be completely condensed without condensing all the AA since we established that acetol will completely condense with a spray rate of 3 mL/min and a bath temperature of 24 °C. Runs were made in which the aqueous condensate solution from the preceding run was used, instead of water, to begin the run. In this way concentrations of AA and acetol up to 30% were obtained with slight, if any, change in water evaporation or condensation rate. This was unexpected as both acetol and AA are less volatile than water and might have reduced evaporation from the liquid droplets.

Figure 7. Vapor fraction condensed from nitrogen/vapor, using 7.6 cm diameter condenser.

decreased as condensation rate increased, suggesting a rate limit on the capacity of the SVCV. The limit is likely due to the residence time rather than cooling capacity. The limit is consistent with the runs at a condensation rate of 1 g/min. The fraction condensed decreases as the vmf increased, indicating that a vmf = 0.05 may have been the maximum for complete condensation at 1 g/min. Complete AA condensation was obtained at a lower spray rate if the vapor feed rate was lower. At sprayed liquid flow rates of 10−20 mL/ min, the liquid (sprayed) temperature increased 30−40 °C during a run. Only when the flask temperature was 60−65 °C (the black square in Figure 7) did acetol fail to completely condense, and this run started with unevaporated acetol from the preceding run, so there may have been some degradation of the acetol in the flask. The vapor feed rate for complete acetol condensation using “standard conditions”, (∼190 °C vapor temperature into the SVCV, nitrogen flow rate of 4.2 g/min, liquid spray rate 20 mL/min) was between 1 and 0.45 g/min. The temperature of the liquid stream leaving the SVCV increased 1−2 degrees during these runs. The vapor temperature and nitrogen flow rate were similar to condenser feed conditions reported for a small scale pyrolysis unit.9 The fraction of the vapors that condense should correlate with condenser temperature which is dependent on liquid and vapor flow rates, compositions, and temperatures. Defining a condensation temperature is not simple because temperature varies within the SVCV. However we show that the molar fraction condensed increased for both acetol and AA with increased vapor mole fraction fed, as shown in Figure 8. Three points shown were for runs with the larger SVCV. Using dimensionless factors to correlate the mass ratio of spray with vapor content did not change the plot (Figure 8) significantly.

4. CONCLUSIONS Mass spray rates equal to the vapor flow rate were sufficient to condense all of the acetol and some fraction of the AA vapor. This limit was imposed by the equipment, and it is likely that lower spray/vapor rate ratios would be effective with a condenser designed to ensure mixing of the vapor with the liquid droplets (more countercurrent). The low spray rate indicates that condensate collected with the spray liquid can comprise at least 50 mass% condensed vapor. For low condensable vapor feed concentrations (0.10 mol fraction and below), the condensed fraction decreased with increased condensation rate, for essentially constant spray rates and temperature. However for condensable vapor loadings greater than 30 wt % nearly all the vapor condensed with 10−15 °C liquid at spray rates of 10 g/min or more. At higher vapor loadings which would be expected for a practical fast pyrolysis system higher spray and condensation temperatures would allow substantial evaporation of water from the liquid so that water content would be limited by the amount needed to keep

Figure 8. Condensation dependence on vapor content. 5071

dx.doi.org/10.1021/ie2017465 | Ind. Eng. Chem. Res. 2012, 51, 5067−5072

Industrial & Engineering Chemistry Research

Article

the liquid draining from the condenser. It was not possible to show a sharp separation of AA and acetol along with complete acetol condensation, but this does not mean that conditions cannot be found that will allow it.



AUTHOR INFORMATION

Corresponding Author

*Tel.: +1 215 540 0186. E-mail: [email protected]. Notes

The authors declare no competing financial interest. § Eastern Regional Research Center, retired.



ACKNOWLEDGMENTS Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. USDA is an equal opportunity provider and employer. The authors thank Michael F. Dallmer who carried out the experiments and made significant improvements to the equipment and procedures.



REFERENCES

(1) Elliott, D. C. IEA Bioenergy Agreement Task 34 Newsletter-PyNe 28; Bioenergy Research Group: Aston University, U.K. December 2010. (2) Westerhof, R. J. M.; Kuipers, N. J. M.; Kersten, S. R. A.; Swaaij, W. P. M Controlling the water content of biomass fast pyrolysis oil. Ind. Eng. Chem. Res. 2007, 46, 9238−9247. (3) Hilten, R. N.; Bibens, B. P.; Kastner, J. R.; Das, K. C. In-line esterification of pyrolysis vapor with ethanol improves bio-oil quality. Energy Fuels 2004, 24, 673−682. (4) Ji-lu, Z. Bio-oil from fast pyrolysis of rice husk: Yields and related properties and improvement of the pyrolysis system. J. Anal. Appl. Pyrolysis 2007, 80, 30−35. (5) Lawrence, T.; Das, K. C. Pyrolysis systems, methods of use thereof, and methods of bio-oil transformation. 2009; US patent application 2009/0054711 (6) Morita, A.; Garrett, B. C. Molecular theory of mass transfer kinetics and dynamics at gas-water interface. Fluid Dynamics Res 2008, 40, 459−473. (7) Ho, S.; Fisher, J. A.; Im, U. K. Modeling and simulation of the oil/water quench system in an olefin unit. Ind. Eng. Chem. Process Des. Dev. 1978, 17, 82−87. (8) Guo, Z.; Wang, S.; Zhu, Y.; Luo, Z.; Cen, K. Separation of acid compounds for refining biomass pyrolysis oil. J. Fuel Chem. Technol. 2009, 37, 49−52. (9) Mullen, C. A.; Boateng, A. A.; Goldberg, N. M.; Lima, I. M.; Laird, D. A.; Hicks, K. B. Bio-oil and bio-char production from corn cobs and stover by fast pyrolysis. Biomass Bioenergy 2010, 34, 67−74. (10) Lee, Y.; Zhou, X. Method for the determination of some soluble atmospheric carbonyl compounds. Environ. Sci. Technol. 1993, 27, 749−756. (11) Rice, C. P.; Chernyak, S.; McConnell, L. Henry’s law constants for pesticides measured as a function of temperature and salinity. J. Agric. Food Chem. 1997, 45, 2291−2298.

5072

dx.doi.org/10.1021/ie2017465 | Ind. Eng. Chem. Res. 2012, 51, 5067−5072