On the effects of liquid viscosity on the spray characteristics of spray dry nozzles

Transcription

1 ILASS Americas 29th Annual Conference on Liquid Atomization and Spray Systems, Atlanta, GA, May 2017 On the effects of liquid viscosity on the spray characteristics of spray dry nozzles K.M. Bade and R.J. Schick Spray Analysis and Research Services Spraying System Co. Wheaton, IL USA Abstract Spray dry nozzles represent a notable portion of industrial nozzles, and are typically used to generate a dried powder from a liquid or slurry. From a spray technology perspective, these nozzles are identified as hydraulic atomization nozzles that operate at high pressures in order to generate relatively small droplets with controlled distributions. In this study, nozzles which are used for spray drying in areas such as the Dairy, Food, Pharmaceutical, and Chemical industries, are characterized and analyzed over a range of operating conditions. For this effort, water-based mixtures with viscosities of cp are investigated with various nozzle types (Swirlchamber, Slotted Core, and Whirlchamber) using a Phase Doppler Interferometer instrument to acquire drop size, velocity, and number density measurements. Traditionally difficult measurements due to high spray density, these detailed results across each spray plume provide insight on the spray characteristics over the range of operating conditions. It is found that among the tested nozzle types, i) the spray characteristics with the SV SprayDry R nozzle change the least with viscosity, ii) the SB SprayDry nozzle has much more uniform characteristics across the spray pattern with the higher viscosity fluid than with water, and iii) the WhirlJet R SprayDry nozzle produces the widest spray pattern and the largest droplets. Corresponding Author: Kyle.Bade@spray.com

2 Introduction The spray drying process has been employed in industrial applications in the Dairy, Food, Pharmaceutical, and Chemical industries for many years. In practice, spray drying involves atomization of a liquid mixture into small droplets that are then dried to generate a powder. In order to maximize the efficiency of evaporation of the water portion of the liquid, fine atomization is preferred to create a large surface area to volume ratio of the drying liquid. The Sauter Mean Diameter (SMD or D 32 ), which is representative of the total volume to surface area ratio of the sprayed liquid [1], is commonly used to evaluate the expected drying efficacy because as this ratio decreases (and D 32 decreases) the droplets will dry faster. The drying process typically takes place in a chamber where flowing gas is introduced to promote evaporation; heat may also be added to increase the drying efficiency. The end result of the spray dry process is a dried particulate product, made up of the components originally mixed in liquid form. For a comprehensive review of the spray drying process, the reader is encouraged to review the Spray Drying Handbook by K. Masters [2]. Common nozzles used for spray drying are characterized as pressure-swirl atomizers, which use the internal nozzle geometry, along with relatively high liquid pressure (energy), to atomize the fluid into very small droplets in a full or hollow cone spray pattern. Atomization in spray dry processes aims to not only to achieve small droplets, but also to produce narrow-width drop size distributions [2], and to avoid large droplets that may not fully dry within the allotted process domain. Coaxial air-blast atomizers for spray drying applications, specifically pharmaceutical tablet coating processes, have received extensive attention [3] for the atomization of viscous and non-newtonian materials [4, 5] and do allow effective atomization with narrow drop size distributions. However, pressure-swirl atomizers do not use a secondary fluid to promote liquid breakup, thus providing a simpler solution, and allowing larger material throughput. Stahle et al. [6] investigated spraying of food-industry liquids with both pressure-swirl and twin-fluid nozzles, finding improved atomization efficiency with pressure-swirl nozzles with low viscosity fluid, but requiring twinfluid nozzles to adequately atomize higher viscosity materials. To better understand the mechanisms leading to breakup, or lack of, in pressure-swirl nozzles, Lee et al. [7] used low viscosity fluid (diesel, kerosene) to examine the internal air-core and related fluid turbulence, and categorized the unstable/stable flow regimes and built upon earlier drop size measurements by Lefebvre and Wang [8]. Unfortunately, these studies focused on liquids of very low viscosity ( 1 cp) for fuel/engine applications, and therefore offer limited use for highly viscous material sprays. Spraying Systems Co. began designing and manufacturing spray dry nozzles at the beginning of the spray dry process adoption, and offers a wide array of SprayDry R nozzle types and capacities to accommodate nearly any application. The drop size distribution produced by each pressure-swirl spray drying nozzle will typically depend on the pressure, capacity, nozzle geometry, and liquid rheology (density, viscosity, and surface tension). For the purposes of this investigation, the results will focus on nozzle geometry and liquid viscosity variations, which expand upon a previous water-only investigation by Bade et al. [9]. The results of the present study demonstrate the great importance of material viscosity when evaluating the optimal nozzle for a spray dry process. Experimental Setup Nozzles The investigations carried out in this study include pressure-swirl nozzles commonly used in spray drying processes and manufactured by Spraying Systems Co. R. All nozzles were selected based on similar atomization technologies and targeted to provide similar flow rates at the same operating pressure for simple comparison, approximately 725 lpm at 70 bar, the details of which are noted in Table 1. Three types of Spraying Systems Co. SprayDry R nozzles were used, the primary distinction between these nozzles is the internal flow path geometry, defined by using a Swirlchamber, Slotted-Core, or Whirlchamber; identified as the SV, SB, and WhirlJet R (WJ) nozzles, respectively. Figures 1-3 show an example of each nozzle body and the associated internal components of each type of nozzle; more information on these nozzles is available in the Spraying Systems Co. SprayDry R nozzle catalog, Bulletin No. 695A [10]. The SV SprayDry nozzle uses a Swirlchamber, creating a flow path that forces the liquid to enter a shallow (axially) chamber which imparts a radial flow velocity in the fluid before exiting through a round orifice. Figure 1 provides an image of the assembled SV nozzle body, Swirlchamber, and exit orifice insert. The SB SprayDry nozzle uses a Slotted-Core which generates a rotating fluid by forcing the fluid through angled slots around the outer surface of the Slotted-Core, before exiting through a round orifice. 2

3 Nozzle (abbr.) SV-6 SB-40 WJ-2 Part Number SVI-106YM+SVS6 SIBY40+SBBY32-MFP AA104-1/2-AA-TCX2-8 Orifice Diameter mm (in.) 2.70 (0.106) 2.50 (0.098) 4.00 (0.156) Internal Swirlchamber Slotted-Core Whirlchamber P bar (psi) 70 (1000) 70 (1000) 70 (1000) Q lph (gph) 722 (189) 736 (193) 744 (195) Table 1: List of nozzles used for testing with abbreviated names, full part numbers, exit orifice diameters, internal geometry type, and rated pressure/flow performance. Note, while the abbreviations used here for each nozzle use a single numeral identifier, each nozzle incorporates 2 parts (the orifice and the insert), which are fully identified in the Part Number column. Figure 2 provides an image of the assembled SB nozzle body, as well as the Slotted-Core and exit orifice insert. The WhirlJet SprayDry nozzle directs the liquid through one, or two, off-center entry ports on the Whirlchamber into a tall (axially) chamber which imparts a radial flow velocity in the fluid before exiting through a round orifice. Figure 3 provides an image of the assembled WhirlJet SprayDry nozzle body, as well as the Whirlchamber and exit orifice insert. Each of the SprayDry nozzle types carries its own benefits and limitations over the rated flow rate range, resulting in differences in spray angle, volume distribution, droplet velocity (residence time), and drop size distribution; which are examined in this investigation. The SV SprayDry nozzle allows the lowest minimum rated flow rate (1-60 lpm at 70 bar), the SB SprayDry nozzle has the smallest low-to-high rated flow rate range ( lpm at 70 bar), and the WhirlJet SprayDry nozzle provides the widest rated flow rate range ( lpm at 70 bar). The purpose of this investigation is to investigate the details of the spray characteristics with each nozzle style, and to provide an explanation for the benefits, given the relative similarly of each design. Figure 1: SV SprayDry nozzle body, with a Swirlchamber and orifice insert Figure 2: SB SprayDry nozzle bodies (left), with a Slotted-Core and orifice insert (right) Spray Materials In order to investigate the effects of material viscosity on the resulting spray characteristics using these SprayDry nozzles, two material compositions have been used. First, standard city tap water was used, which was confirmed to have a viscosity of approximately 1.00 cp at a test temperature of 20 C. Second, a mixture of water and Methylcellulose (water+mc) material was used to create a 1000 cp material. The primary advantage of using dissolved methylcellulose to generate the increased viscosity material is that the resulting material density and surface tension remain essentially the same as pure water, and the two testing materials i ) pure water and ii ) water+mc can be used to isolate the effects of viscosity differences. Figure 3: WhirlJet SprayDry nozzle body (left), with a Whirlchamber and orifice (right) 3

4 Figure 5: PDI and nozzle setup with coordinate system definition Figure 4: Pressure (P ) vs. flow rate (Q) for each nozzle and material. Figure 4 provides the rated performance curve, pressure vs. flow rate, for the three nozzle styles and selected chamber/insert sizes. The curves are provided for water based performance, and the symbols demonstrate the measured flow rate at 70 bar for each nozzle. In terms of the actual reductions in flow for each nozzle, the SV, SB, and WJ flows were reduced by 8.5%, 9.9%, and 15.6%, respectively. Phase Doppler Interferometer (PDI) Setup Phase Doppler Interferometry (PDI) measurements were acquired, using an Artium Technologies PDI-MD-300 instrument to provide measurements of droplet size, velocity, and local volume flux. Similar acquisition methods were used as those outlined as optimal by Bade and Schick [11], with the transmitter/receiver lens combination of 1000/1000 mm, and the receiver located at a 40 off-axis position, which allowed for a sufficient drop size measurement range and resolution. 5,000-10,000 samples were acquired at each point to provide converged statistical values (±1%). Bade and Schick [11] also provide an extensive investigation of the accuracy of volume flux measurements using the PDI instrument which were found to be accurate to within approximately 2% when proper droplet trajectory considerations are used in the data collection methods. The ability of the PDI instrument to simultaneously measure local drop size, velocity, and flux provides an attractive source of point measurements within a spray plume; and the methods outlined by Bade and Schick [12, 13] allow these point measurements to be used to generate planar spray characteristics for more general comparisons. As a matter of practicality, all PDI measurement points were acquired at a single distance from the nozzle (z=-600 mm), and at constant intervals from the nozzle center point along a single radial path (+x). Figure 5 demonstrates the testing setup for these investigations as well as the coordinate systems with (x,y,z)=(0,0,0) defined at the center of the nozzle exit orifice. Post-Processing Methods Weighted averaging is a reasonable and robust method for combining many individual pointmeasurements into meaningful combined statistics; and in this case, planar-statistics. The methods used to generate weighted average values which are representative of the planar spray plume characteristics follow typical mathematical processes, and are identified in detail by Bade and Schick [13]. For the purposes of this paper, two weighting parameters are used in the post-processing of the pointwise data and each represents a relevant physical characteristics of the investigated spray plumes: i) the discrete area that each measurement point is expected to represent, and ii) the local volume flux at the location of each measurement. The use of the two parameters together weight the averaged results toward the spray characteristics where the majority of the sprayed volume is present. The phase Doppler method allows accurate simultaneous measurement of drop size, velocity, and volume flux [11] and therefore allows these weighting methods to be implemented efficiently without additional measurements or assumptions. Furthermore, drop size distributions are acquired at each point using the PDI, and these point-wise-distributions are combined using the methods outlined by Bade and Schick [12, 13] to generate planar-representative drop size distributions. Common drop size statistics and distribution terminology is used throughout these analyses; the unfamiliar reader is directed to the ubiquitous text by Lefebvre [1]. Results and Discussion Each nozzle was sprayed with water as well as the 1000 cp material and an image was acquired during the detailed spray measurements; these images are provided in Figures 6-8. It is clear that with water, the SV-6 and SB-40 sprays are of a similar shape and size, while the WJ-2 spray is wider and 4

5 (a) experiences a notable collapse upstream of the measurement region. This collapse and any air entrainment effects are not considered detrimental to the measurements because typical spray dry processes take place in tall drying towers. The 1000 cp spray images make clear that all sprays are less dense with the increased material viscosity, but maintain a similar angle/coverage; characteristics that are evaluated further with the detailed measurements from the PDI. The results of the point-wise measurements from the center-to-edge of each spray, along the +x axis, provide a detailed view of character of each spray plume. All spray plumes are assumed to possess axisymmetric characteristics. In Figure 9, the local number density distributions are provided across each spray plume. The solid lines with circle markers represent the water (1 cp) results, while the dashed lines with triangle markers represent water+mc (1000 cp) results; this convention is maintained for Figures All three nozzle types demonstrate a peak spray density between 90<x<100 mm with water, indicative of hollow cone spray patterns, and the WJ spray is nearly 2 times wider than the SV and SB sprays. The spray widths are similar for all nozzles with 1 and 1000 cp material, however, the concentration magnitudes are significantly reduced due to the reduced flow rates and increased drop sizes. Examining the axial velocity results in Figure 10, with water, the SV and SB distributions have a much larger velocity at the spray plume center than the WJ nozzle, which produces a much wider and uniform velocity distribution with about half the peak velocity. The SV and WJ sprays have a slightly off-center (x 20 to 40 mm) peak velocity typical of hollow cone spray patterns. The SB nozzle demonstrates a peak velocity at the plume center. These center vs. off-centered peak velocity results were verified with repeated measurements. The higher viscosity velocity distributions show a range of similarities and differences compared to the water-only results. The SV spray has a significant shift changing from a near-center peaked velocity with a fast decrease in x with water, to a relatively flat and uniform velocity with the water+mc. Interestingly, the average velocity is nearly identical for the SV nozzle with both 1 and 1000 cp spray material, but clearly the distribution and max/min characteristics change significantly. The SB nozzle shows a very different effect with increased viscosity, the 1 cp and 1000 cp velocity distributions are nearly identical with an increase in velocity for the 1000 cp material only near the spray edge. Fi- (b) Figure 6: Images of the SV-6 nozzle spray with a) water and b) 1000 cp material (a) (b) Figure 7: Images of the SB-40 nozzle spray with a) water and b) 1000 cp material (a) (b) Figure 8: Images of the WJ-2 nozzle spray with a) water and b) 1000 cp material 5

6 Figure 9: Radial location (x) vs. number density (ND) for each nozzle and material ( 1 cp, cp) Figure 10: Radial location (x) vs. axial velocity (V z ) for each nozzle and material. nally, the WJ nozzle spray shows a similar radial trend in axial velocity with 1 and 1000 cp materials, with a reduced magnitude with the 1000 cp material. An important note, the WJ spray pattern with 1000 cp material generated a similar width/angle to that generated with water, however, testing facility side-wall interference and subsequent spray recirculation prevented reliable measurements for x>80 mm; therefore, all measurements for x>80 mm are an estimation based on the 1 cp data (shown as a dotted line). Figure 11 demonstrates the change in axial velocity with viscosity by generating a discrete area and volume flux weighted average of the point measurements within each spray plume. As noted previously, the SB nozzle demonstrates a similar axial velocity regardless of viscosity. The SB nozzle has a slightly increased velocity with higher viscosity, due to the increased velocity of the spray-edge droplets. The average WJ spray velocity is decreased, without a notable change in the radial distribution, at least from 0<x<80 mm where measurements were possible. The Volume Median Diameter is selected to be shown to represent the drop size trends across the spray patterns in Figure 12. All nozzles generate the smallest average drop size at the spray centers, with increasing drop size toward the spray edges. The SV nozzle demonstrates the most significant change in drop size trend with increasing viscosity, wherein it produces a much more constant D V 0.5 across the spray plume with the 1000 cp material, which could prove very beneficial in spray drying applications. In contrast, the SB nozzle generates a very similar drop size and distribution with the 1 and 1000 cp materials, indicating that this nozzle may be very effective in applications where the material viscosity Figure 11: Dynamic viscosity vs. axial velocity (V z ) for each nozzle and material. Note, dotted lines are used to connect same-nozzle data points, but do not represent an expected trend. may change regularly. The WJ nozzle generates the smallest D V 0.5 values with water, but the largest with 1000 cp material. Figure 13 captures these trends of average D V 0.5 vs viscosity. Because no data is available inside of the nozzle bodies, it is difficult to assess exactly what leads to the spray plume differences between nozzles and across the wide range of viscosity. However, the SV and WJ nozzles both operate with radial inlets to a chamber which imparts swirl before exiting; alternatively, the SB nozzle employs a somewhat different flow path which uses multiple slots to impart radial velocity without an internal chamber. These geometry differences must be what leads to the very different spray plume characteristics. The WJ nozzle s Whirlchamber design has the largest internal cavity and the resulting wide spray pattern, low spray velocity, and large drop size (with 1000 cp material) indicate that this design likely sacrifices axial velocity, for increased radial velocity at the nozzle exit, slowing the spray down (axially) and widening the pattern. Additionally, it is likely that the veloc- 6

7 Figure 12: Radial location (x) vs. Median Volume Diameter (D V 0.5 ) for each nozzle and material. Figure 14: Weighted average cumulative volume fraction drop size distributions for each nozzle and material ( 1 cp, cp) Figure 13: Dynamic viscosity vs. Volume Median Diameter (D V 0.5 ) for each nozzle and material. Note, dotted lines are used to connect same-nozzle data points, but do not represent the expected trend. ity magnitude at exit is lower than the other nozzle types and/or the spray does not pre-film along the exit orifice surface as effectively as the SV and SB nozzles, resulting in larger droplets. The similarities in the magnitude and trend of the velocity and drop size with the SB nozzle across the spray pattern indicate that this nozzle geometry creates internal flow characteristics that are similar of the wide range of viscosities. This is most readily explained by the lack of a notable internal flow chamber with this style of nozzle, and while the SB nozzle does not generate the smallest droplets, its consistency is notable. Finally, the SV nozzle demonstrates the most notable change in radial characteristic distributions with an increase in material viscosity; resulting in more uniform distributions and only slightly larger droplet size. Finally, weighted average cumulative volume fraction drop size distributions are presented in Figure 14; these combined distributions were generated using the methods outlined by Bade and Schick [12, 13]. It can be seen that the water-only distri- butions for all three nozzles are similar in shape and only slightly different in magnitude, with the drop size generated by the WJ, SV, and SB nozzles increasing, in that order. Notably, the SV nozzle experiences only a small increase the the volume of larger droplets in the µm range with the 1000 cp material and, from the details of the SV distributions in Figure 12, these larger droplets are generated near the spray-edge only. The SB nozzle spray sees a more significant increase in volume of droplets in the µm range, resulting in a greater average drop size increase, but still not significant. The WJ combined distribution cannot be reasonably determined due to the limited radial span of the measurable points (x 80), but the measured portion of the radial distribution in Figure 12 indicates that the drop size will likely be increased by 2-3 times that with water. Conclusions Experimental testing was conducted with three Spraying Systems Co. R SprayDry R nozzles commonly used in industrial spray drying processes. Data were acquired at 70 bar (1000 psi) using two spray materials: water (1 cp) and water+methylcellulose (1000 cp), allowing direct comparison of the spray plume characteristics with this wide range of spray material viscosities. The water data provides a direct connection to the large set of data provided by the nozzle manufacturer, while the 1000 cp results represent a viscosity similar to the slurry materials typically used with these types of nozzles in applications. It is found that all three SprayDry nozzles produce a hollow cone spray pattern with both viscosities. The spray pattern for the SV and SB SprayDry nozzles are of a similar width/angle, while the Whirl- Jet (WJ) SprayDry nozzle has a much wider spray 7

8 patten, nearly twice as wide. The SV SprayDry nozzle with the Swirlchamber insert produces surprisingly similar velocity and drop size statistics at both viscosities, while also maintaining a similar radial distribution of these characteristics. The weighted average axial velocity across the spray plume decreases by 5% and the Volume Median Diameter drop size statistic increases by only 5%. The SB SprayDry nozzle with the Slotted-Core insert provided the most significant spray plume radial change with viscosity, where the spray with the higher 1000 cp viscosity had a much more uniform radial distribution of both velocity and drop size. The SB nozzle had a 17% increase in axial velocity, but only a 0.5% change in droplet size even with the significant improvement in radial spray plume uniformity. The WhirlJet (WJ) SprayDry nozzle was the most consistent in radial trend, but had the largest change in droplet characteristic magnitudes with the change in material viscosity, experiencing a decrease in velocity by 47%, and an increase in D V 0.5 drop size by 107% (just over 2 times larger). Finally, an explanation of the internal flow path geometry differences between the various SprayDry R nozzles is provided in order to offer some insight regarding the methods of spray plume production that may lead to the differences in performance for each nozzle. Acknowledgements The authors thank Narbeh Haroutunian of Spraying Systems Co. for his work on data acquisition and processing in support of these investigations. References [1] Lefebvre, A. H., Atomization and Sprays, Combustion: An International Series, Hemisphere Publishing Corporation, [2] Masters, K., Spray Drying Handbook, 4th Ed., George Gowdin, London, ISBN , [3] Lasheras, J.C., Hopfinger, E.J., Liquid jet instability and atomization in a coaxial gas stream, Annual Review of Fluid Mechanics, 32, pp , [4] Mansour, A., Chigier, N., Air-blast atomization of non-newtonian liquids, Journal of Non- Newtonian Fluid Mechanics, 58, pp , Atomization of viscous and non-newtonian liquids by coaxial, high-speed gas jet. Experiments and drop size modeling, International Journal of Multiphase Flow, 34, pp , [6] Stahle, P., Schuchmann, H.P., Gaukel, V., Performance and efficiency of pressure-swirl and twin-fluid nozzles spraying food liquids with varying viscosity, Journal of Food Processing Engineering, 40, pp. 1-12, [7] Lee, E.J., Oh, S.Y., Kim, H.Y., James, S.C., Yoon, S.S., Measuring air core characteristics of a pressure-swirl atomizer via a transparent acrylic nozzle at various Reynolds numbers, Experimental Thermal and Fluid Science, 34(8), pp , [8] Lefebvre, A.H. and Wang, X.F., Mean drop sizes from pressure-swirl nozzles, Journal of Propulsion and Power, January, 3(1), pp , [9] Bade, K. M., Schick, R. J., T. Oberg, C. Pagcatipunan, Local and General spray characteristics of spray dry nozzles with water, ILASS Americas 28th Annual Conference on Liquid Atomization and Spray Systems, Dearborn, MI, May [10] Spraying Systems Co. R, SprayDry R Nozzles, Bulletin No. 695A, [11] Bade, K. M., Schick, R. J., Phase Doppler Interferometry Volume Flux Sensitivity to Parametric Settings and Droplet Trajectory, Atomization and Sprays, vol. 21, issue 7, pp , [12] Bade, K. M., Schick, R. J., On Generating Combined Drop Size Distributions from Point Measurements in a Spray, ICLASS - 13th Triennial International Conference on Liquid Atomization and Spray Systems, Tainan, Taiwan, August 23-27, [13] Bade, K. M., Schick, R. J., Generating Planar Statistics from Point Measurements in a Spray, Atomization and Sprays, (DOI: /AtomizSpr ) SprayDry R is a registered trademark of Spraying Systems Co. [5] Aliseda, A., Hopfinger, E.J., Lasheras, J.C., Kremer, D.M., Berchielli, A., Connonlly, E.K., 8