ILASS Americas, 19 th Annual Conference on Liquid Atomization and Spray Systems, Toronto, Canada, May 2006 Pneumatic Atomization of a Viscous Solution for Spray Drying: Investigation of Droplet and Particle Formation. J. P. Hecht *, and J. A. Stamper Process & Emerging Technology, Procter and Gamble 8256 Union Centre Blvd. West Chester, OH 45069 USA D. K. Giles Department of Biological & Agricultural Engineering University of California, Davis Davis, CA 95616 USA Abstract Pneumatic atomization of an aqueous sodium silicate solution (up to 0.350 Pa s) was studied to understand particle formation during spray drying. Droplet and particle measurements were made from a pilot-plant scale nozzle. Droplet-size distribution data were collected in a parametric experiment with the factors of atomizing air pressure, liquid flow rate, liquid temperature, and measurement distance from the nozzle. As expected, the droplet size decreased with decreasing liquid flow rate and increasing air pressure. The measured droplet size typically increased with distance from the nozzle due to mid-air collision and coalescence of droplets. This effect was greater at lower air pressures. The spray-dried particles were irregularly shaped, indicating they were highly agglomerated, consistent with the atomization results. In the dryer, a higher liquid flow rate led to smaller dry particles (but larger drops in the atomization study), probably because the inlet air temperature to the dryer was correspondingly higher, leading to faster drying and reduced agglomeration. * Corresponding author
Introduction and Motivation Product quality attributes such as particle size distribution, morphology, and density all contribute to the value of products from spray drying. Often, the relationships between the process and the resulting powder attributes are complicated by multiple, simultaneous phenomena making it difficult to understand the distinct mechanisms that control quality [1]. In spray drying, particles are first formed by atomization and subsequently may expand, contract, and collide with each other in a turbulent environment containing temperature and humidity gradients. A good example of this complexity is exemplified by the drying of sodium silicate solution using pneumatic nozzles. Sodium silicate is a viscous, caustic solution that forms highly expanded, low-density powder when it is spray dried [2,3]. This material also has the tendency to form agglomerates in a spray dryer, even when a single nozzle is used. Prediction of the particle morphology resulting from a spray dryer is not currently possible using computational techniques, though significant progress has occurred [4]. Furthermore, it is not enough to run a benchscale or even a pilot scale dryer to fully understand the simultaneous mechanisms involved. In this paper, the authors study both atomization alone and run a pilotscale dryer to better understand how to control product attributes. Objectives The objective of this study was to better understand the particle-formation mechanisms in the spray dryer by studying atomization of the liquid separately in addition to running pilot-plant drying studies. Specifically, the goals were to: 1. Measure the droplet size distribution as a function of flow rate, atomization air pressure, distance from the nozzle, and liquid temperature (primarily, viscosity). 2. Measure the particle size distribution of the spraydried product and compare with the atomization results to better understand the particle-formation mechanisms. Experimental Apparatus and Procedure There were two separate experiments, each conducted independently. One experiment was to generate droplet size data during the pneumatic atomization of a sodium silicate solution. The second was to compare the droplet size data to particles that were produced via spray drying with the same nozzle. Liquid Properties The two materials used in this study were municipal water and 2.0r sodium silicate (PQ Corporation, Sodium Silicate D, 44% Solids). At 20 C the water has a viscosity of 0.001 Pa s and a surface tension of 72 mn/m and the sodium silicate solution has a viscosity of 0.350 Pa s and a surface tension of 88.8 mn/m (value given to us by PQ Corp.). The viscosity of the sodium silicate solution reduces greatly with temperature, as seen in Figure 1. These data were measured with a Cannon-Fenske viscometer. Atomization Experiments Atomization experiments were carried out in a wellventilated room using a pump, an air compressor, and drop-size measurement equipment. Figure 2 shows the layout of the system. A progressive cavity pump (Moyno 0 or Moyno L series) was used to deliver the liquid to the nozzle. The flow rate through each nozzle was measured using a pre-calibrated magnetic flow-meter (Yokogawa AD- MAG AE: AE115MG)). The values were verified using a bucket and a stopwatch. The atomization air for the system was delivered with a 750 kw rotary screw air compressor (Ingersoll-Rand: IRNH-OF) and a 2500 liter receiving tank. The atomizing air flow rates were calculated in separate experiments by first pressurizing the receiving tank and then allowing it to empty through the nozzle, recording the pressure loss in the tank with time. The ideal gas law was then used to determine the mass flow rate of air at a given air pressure. The results of these tests can be found in Table 1. The nozzles used in this study were Spraying Systems SU82 two-fluid nozzles (Fluid Cap: 251376, Air Cap: 4691312). The SU82 can be seen in Figure 3. Figure 4 shows the internal flows of a two-fluid nozzle. The primary data collected in the atomization experiments were droplet size distributions. An optical imaging system (Greenfield Speedview 700) was used to measure droplet size distributions. This device is a high-speed digital camera and a backlight strobe. It captures and digitizes 30 frames per second and sizes the droplets using digital imaging software. Prior to data collection, the device was calibrated using a standard reticle (Patterson Graticule) that contains imbedded particles of known size to determine the /pixel ratio. The reported droplet-size distributions were corrected using a calibration curve that accounted for the tendency of the device to detect larger particles over a larger distance from the camera. The measurement volume of the Greenfield was approximately a 3 mm x 3 mm plane located about 20 mm from the top of
the camera cover. The depth of field was approximately 2 mm. Single nozzle droplet size data were collected for a range of liquid feed rates, atomizing air flow rates, liquid temperatures, and measurement distances. The measurement distance is the distance from the tip of the nozzle to the centerline of the droplet size analyzer. Safety Due to the caustic nature of the sodium silicate solution, certain safety precautions were put into place: 1. Tyvek suits were worn at all time while testing was in progress. 2. When entering the spray room, Powered Air Purifying Respirators (PAPR) units were worn. 3. Water proof, steel toe, rubber boots were worn at all time while testing was in progress. 4. At least two people were present during all spray testing. Spray Drying Trials The sodium silicate solution was spray dried in a pilotscale dryer with the nozzle used for the atomization experiments. Figure 5 shows a schematic of the spray dryer. The air flow in the spray dryer is co-current. The main collection point for the dryer was at the bottom of the drying chamber. Finer particles entrained by the air exiting the dryer were mostly recovered in a cyclone. The small fraction of particles not collected by the cyclone was ultimately collected by a baghouse. The spray dryer was operated at constant air flow rate for all runs and the exhaust temperature was held nearly constant to achieve the target bulk density of the powder. To maintain the energy balance of the dryer, increases in the liquid feed rate led to increases in the inlet dryer air temperature since the overall drying rate in the dryer is proportional to the product of the air mass flow rate and the temperature difference of the air entering and exiting the dryer. The liquid flow rate and the atomizing air pressure were easily measured externally, however the liquid feed temperature needed to be measured internally due to the potential for the feed material to heat en route to the nozzle. This was accomplished by several reversible changes that were made to the equipment for this trial. A thermocouple assembly was designed to measure the liquid temperature close to the nozzle. This was done to develop a better understanding of the liquid conditions during atomization. The thermocouple was added as a tee to the streams just upstream of the nozzle as shown in Figure 6. Special thermocouples (Omega model no. TJ180-CASS-18G-3-CC-XCBX and TJ120- CASS-18G-3-CC-XCB) were needed to withstand the potentially high temperatures in the dryer. Materials collected from the bottom and cyclone of the spray dryer were later analyzed for particle size using sieves (Gradex 2000) and particle morphology was observed using light microscopy. Baghouse material was not analyzed because it represented only a few percent of the total mass of powder that was produced. Powder taken from the drying chamber and from the cyclone was analyzed independently and results are reported as a single value the volume median diameter. The particles were assumed to have a constant density in the calculation of the volume median diameter from sieve mass data. Later, the data from the cyclone and the drying chamber were combined in the proper weighted ratio to yield a representative particle size distribution; this combined result is reported as the total volume median diameter. The particle morphology was observed by digital microscope photographs. Since drums of material were created at each spray-drying condition and each photograph only shows a few particles, four photographs were taken of each small sample in an attempt to increase the likelihood of a representative sample. All of the photos in this paper were taken at the same magnification. Results Atomization Data Figure 7 shows the volume median diameter of a water spray at 55 and 110 kg/hr and ambient (ca. 22 C) temperature as a function of atomizing air pressure and measurement distance. The droplet size decreases as the air pressure increases and for most of the data, the droplet size increases with distance from the nozzle. This effect is due to coalescence of drops in the turbulent environment and has been observed in other pneumatic nozzles [5]. This trend was consistent in all of the data obtained at the higher flow rate. Similar data for ambient sodium silicate solution are presented in Figure 8. The flow rates and air pressures are different than for water, but the same qualitative trends were observed. Atomization data for sodium silicate solution sprayed at 55 C as a function of atomization air pressure, liquid flow rate, and distance from the nozzle are presented in Figure 9. The results indicate the expected trends with regards to air pressure, liquid flow rate, and measurement distance, as discussed above.
Figure 10 shows the volume median diameter of sprays of sodium silicate solution at 55 C as a function of measurement distance and atomizing air pressure at a flow rate of 350 kg/hr. These data are some of the data from Figure 9, but plotted in a different way. Again, higher air pressure led to a finer spray and the drop size increased with distance from the atomizer due to coalescence. However, this plot clearly shows that the extent of agglomeration within the spray is reduced at higher air pressures, i.e. at 780 and 970 kpa, the droplet size increase is a lower magnitude than for sprays at lower pressures. This effect probably results from the formation of a more dilute spray due to the higher air flow rates, as shown in Table 1. Droplets in a more dilute spray are less likely to collide and coalesce. Particle Formation Data Analysis of the sodium silicate particles formed in the spray dryer using the same nozzle and liquid described above and comparison with the atomization data requires some interpretation. First, the particles are highly expanded. This expansion occurs when the drops reach the boiling temperature and a steam bubble forms within each drop. When the drops are cooled, the steam in the bubble will condense, but the particle may not collapse if the particle has dried to a solid form [6]. The degree of expansion depends primarily on the exhaust temperature of the spray dryer [3], which was held nearly constant in this study. In the case of sodium silicate particles in this study, the shells are only a few thick. Secondly, the drop coalescence that is reported in the results given by Figures 7-10 above will only occur as coalescence if the drops are still liquid when they collide and combine. If the drops are already partially dry, then the drops will combine to form agglomerates. This concept is illustrated in Figure 11. Of course, these are two limiting cases and a wide range of intermediate morphologies are possible. The analysis is further complicated by the fact that the inlet air temperature of the dryer is higher for a higher liquid feed (as explained above), which will cause drying and expansion to occur more rapidly. Figure 12 shows a series of microscope photos for sodium silicate particles made by spray drying with the same nozzle and feedstock as described above at a liquid temperature of 55 C, but the atomization air pressure was varied. In these data, the feed rate of the liquid was held constant at 290 kg/hr and so was the inlet air temperature to the dryer. The photos clearly show agglomeration has occurred and that the particles are larger at the lower atomization pressure. The particle size data are consistent with the photos. The particles shown in Figure 13 were formed by atomization at the same pressure, but at different liquid feed rates. The temperature of the feedstock was 55 C, but the dryer air temperature was substantially higher at the higher feed rates. The results show that the particles appear more agglomerated in the high-feed rate case. Another observation is that the parent spheres, which comprise the agglomerates, appear to be larger in the lower feed-rate case. A comparison of particles made by atomization at 270 kpa and two feed rates is shown in Figure 14. Again, the lower feed rate gives larger particles, which is not intuitive for a spray. The difference is probably due to another phenomenon in addition to atomization, such as a reduction of agglomeration due to faster drying in the high-feed-rate case, since the inlet temperature of the spray dryer is higher for higher feed rates. The particle photos do show a larger parent droplet size and less agglomeration in the low-flow rate case, indicating that the extent of coalescence may have been higher due to the slower drying rate. Figure 15 shows a comparison between the droplet size from the atomization study, measured at a distance of 2.0 m and the droplet size particle size from the spray dryer with the same nozzle and feed temperature. The data are the same as shared in earlier figures. The particles are clearly much larger than the droplets. Most of this is due to the expansion of the individual drops, which is apparent from the photos. However, there are other effects that influence the comparison. For example, agglomeration can occur beyond the 2.0 m distance in the atomization experiments or even outside of the spray. Furthermore, the drop size distribution was measured along the centerline of the spray, whereas the particles represent the result of drying the entire spray. This is significant because the drop size at the edges of the spray are typically larger than the drop sizes at the center of the spray for pneumatic nozzles [5] One can estimate the relationship between the drop diameter and the resulting particle diameter, if the bulk density of the powder, the density of the dry sodium silicate, and the density of the feed solution are known. However, a value of the packing fraction is needed for this and that value difficult to estimate given the unusual particle morphology. This comparison, which is not presented here, gives a reasonable agreement between the two sets of results. Conclusions The experimental atomization data collected on a large pneumatic nozzle follow the expected trends for pneumatic nozzles; the droplet size increases with higher
feed rate, higher liquid viscosity, and lower atomization air pressure. The droplet size also increased with distance from the nozzle due to the coalescence of droplets. This effect was less profound at high air pressures due to dilution of the spray by large air flows. The apparent coalescence behavior of the drops in the atomization experiment is strongly supported by the appearance of agglomerates in the spray-dried powder. The photos clearly show a high degree of agglomeration. The dry particle size was similarly affected by air pressure. Higher air pressure led to smaller particles. However, a higher liquid feed rate led to larger drops in the atomization study, but did not lead to larger particles in the spray-drying study. In some cases, e.g. the results presented in Figure 13, the particle size was noticeably smaller. A likely explanation for this result is that the higher drying air temperature dried the particles before (liquid) coalescence could occur. So when the droplets or partially-dried particles collided with each other, more agglomeration occurred, rather than coalescence and in the case of very rapid drying, the colliding particles may have been too dry to stick to each other. Understanding complex particle-formation mechanisms and how to manipulate them to achieve desirable product properties is not always possible using spray-drying trials alone. The combination of atomization studies in conjunction with spray drying trials has been helpful towards developing a deeper understanding of the process. References 1. Masters, K., Spray Drying in Practice, SprayDryConsult International ApS (www.spraydryconsult.com), Denmark (2002) 2. Masters, K., Spray Drying Handbook, 3 rd Ed. Halsted Press, 563-564 (1979) 3. Balfanz, et. al. Production of Spray-Dried, High Bulk Density Hydrous Sodium Silicate Mixtures, U.S. Patent 4,022,074 (1977) 4. Verdurman, et. al., Drying Technology, 22, 6, 1403-1461 (2000) 5. Hecht, J.P., et al., Controlled Agglomeration using Two Impinging Sprays, ILASS Americas, 18th Annual Conference on Liquid Atomization and Spray Systems, Irvine, CA, May 2000 6. Hecht, J.P. and King, C.J., Influence of Developing Drop Morphology on Drying Rates and Retention of Volatile Substances. 1. Single-Drop Experiments and 2. Modeling, Ind. End. Chem. Res., 39, 1756-1775 (2000)
Tables and Figures 0.50 Viscosity, Pa s 0.40 0.30 0.20 0.10 0.00 10 20 30 40 50 60 70 80 90 Temperature ( C) Figure 1. Sodium silicate viscosity as a function of temperature Atomizing Air Pressure, kpa Flow Rate, kg/hr 970 230 830 200 690 170 550 140 410 280 72 140 36 Table 1. Air flow data for the experimental system
Figure 2. The atomization test stand.
7.9mm Figure 3. SU82 Nozzle
Air Liquid Figure 4. Nozzle diagram (Source: Spraying Systems Co., Inc) Atomizing Air Feed Hot Air Exhaust Blower Baghouse Spray Dryer Cyclone Rotary Valve Flap Valve Rotary Valve Dryer Bottoms Collection Dryer Cyclone Collection Dryer Baghouse Collection Figure 5. Schematic of pilot-scale spray dryer
Liquid Feed Thermocouple Figure 6. Hardware used to measure temperatures at the nozzle
180 160 140 120 80 55 kg/hr / 1.2 m 55 kg/hr / 1.5 m 55 kg/hr / 1.8 m 55 kg/hr / 2.0 m 60 40 0 50 150 200 250 Air Pressure, kpa 180 160 140 120 80 110 kg/hr / 1.2 m 110 kg/hr / 1.5 m 110 kg/hr / 1.8 m 110 kg/hr / 2.0 m 60 40 0 50 150 200 250 Air Pressure, kpa Figure 7. Volume median diameter of droplets in a water spray at ambient temperature (ca. 22 o C) is shown as a function of atomizing air pressure, liquid flow rate, and measurement distance.
130 120 110 90 80 70 60 50 175 kg/hr 350 kg/hr 525 kg/hr Measurement Distance 1.2 m 40 600 700 800 900 0 Air Pressure, kpa 130 120 110 90 80 70 60 50 175 kg/hr 350 kg/hr 525 kg/hr Measurement Distance 1.6 m 40 600 700 800 900 0 Air Pressure, kpa 140 130 120 110 90 80 70 60 50 175 kg/hr 350 kg/hr 525 kg/hr Measurement Distance 2.0 m 40 600 700 800 900 0 Air Pressure, kpa Figure 8. Volume median diameter of droplets in a sodium silicate spray at ambient temperature (ca. 22 o C) is shown as a function of atomizing air pressure, liquid flow rate, and measurement distance.
140 120 80 60 40 Measurement Distance 1.2 m 175 kg/hr 350 kg/hr 525 kg/hr 20 0 200 400 600 800 0 Air Pressure, kpa 140 120 80 60 40 Measurement Distance 1.6 m 175 kg/hr 350 kg/hr 525 kg/hr 20 0 200 400 600 800 0 Air Pressure, kpa 140 120 80 60 40 Measurement Distance 2.0 m 175 kg/hr 350 kg/hr 525 kg/hr 525 kg/hr 20 0 200 400 600 800 0 Air Pressure, kpa Figure 9. Volume median diameter of droplets in a sodium silicate spray at 55 o C is shown as a function of atomizing air pressure, liquid flow rate, and measurement distance. Two of the data points at the high flow rate did not follow the expected trend. Although some experimental error was probably involved, no irregularities were observed during the experiment, so they are included as hollow symbols.
120 110 90 80 70 60 50 270 kpa 440 kpa 610 kpa 780 kpa 950 kpa 40 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 Measurement Distance, meters Figure 10. Volume median diameter of droplets in a 350 kg/hr sodium silicate spray at 55 o C is shown as a function of atomizing air pressure and measurement distance, plotted to illustrate the effect of coalescence.
Collision before expansion (Coalescence) Figure 11. Coalescence vs. Agglomeration Collision after expansion (Agglomeration)
Conditions 290 kg/hr 410 kpa Volume Median Diameter 372 m chamber 324 m total 250 m Conditions 290 kg/hr 340 kpa Volume Median Diameter 387 m chamber 327 m total Conditions 290 kg/hr 270 kpa Volume Median Diameter 429 m chamber 385 m total 250 m Figure 12. Microscope photos of spray-dried sodium silicate particles produced by atomization at different air pressures. One of many agglomerates is highlighted by the red circle in the upper right photo.
Conditions 435 kg/hr 680 kpa Volume Median Diameter 333 m chamber 264 m total 250 m Conditions 350 kg/hr 680 kpa Volume Median Diameter 371 m chamber 250 m total 250 m Conditions 200 kg/hr 680 kpa Volume Median Diameter 370 m chamber 276 m total 250 m Figure 13. Microscope photos of spray-dried sodium silicate particles produced by atomization at 680 kpa air pressure and different liquid feed rates. The inlet temperature to the spray dryer is higher for higher liquid flow rates.
Conditions 145 kg/hr 270 kpa Volume Median Diameter 530 m chamber 508 m total 250 m Conditions 290 kg/hr 270 kpa Volume Median Diameter 370 m chamber 276 m total 250 m Figure 14. Comparison of particles produced at two different flow rates at 270 kpa.
600 500 400 300 200 145 kg/hr 290 kg/hr 350 kg/hr drops 175 kg/hr drops 525 kg/hr drops Particles 200 kg/hr 435 kg/hr 360 kg/hr 0 200 300 400 500 600 700 800 900 0 Air Pressure, kpa Figure 15. A comparison of dried particle size with liquid droplet size is shown. The droplet size data are the same as shown in the bottom plot of Figure 8.