Multiple Series Cyclones for Fine Dust

An ASAE Meeting Presentation Paper Number: 054014 Multiple Series Cyclones for Fine Dust D.P. Whitelock, Agricultural Engineer USDA-ARS Southwestern Cotton Ginning Research Laboratory, dwhitelo@nmsu.edu M.D. Buser, Agricultural Engineer USDA-ARS Cotton Production and Processing Research Unit, mbuser@lbk.ars.usda.gov Written for presentation at the 2005 ASAE Annual International Meeting Sponsored by ASAE Tampa Convention Center Tampa, Florida 17-20 July 2005 Abstract. Cyclones are commonly used in the processing industry as primary particulate emissions abatement devices. A study was conducted to evaluate the effectiveness of one, two, three, or four 1D3D cyclones in series on airstreams heavily loaded (273 g/m 3 ) with fine particulate (<10 µm). The overall efficiency of a single cyclone (91%) was significantly lower than that of the series configurations (98%). Also, a single cyclone was consistently less efficient (91%) and had a higher static pressure loss (1356 Pa) than the No. 1 cyclone in the series configurations (94% efficiency and 881 Pa static pressure loss). Particle size distributions showed that there was a significant shift toward smaller sized particles in dust captured by the cyclones (8.78 to 1.86 µm) and the dust emitted by the cyclones (3.23 to 1.52 µm) as the number of series cyclones increased from one to four. A secondary cyclone increased overall efficiency significantly with an insignificant rise in static pressure loss, while additional cyclones raised efficiency only about 1% more and increased pressure losses by 150%. Keywords. Cyclones, Series, Sequential, Efficiency, Particulate, PM 10. The authors are solely responsible for the content of this technical presentation. The technical presentation does not necessarily reflect the official position of the American Society of Agricultural Engineers (ASAE), and its printing and distribution does not constitute an endorsement of views which may be expressed. Technical presentations are not subject to the formal peer review process by ASAE editorial committees; therefore, they are not to be presented as refereed publications. Citation of this work should state that it is from an ASAE meeting paper. EXAMPLE: Author's Last Name, Initials. 2005. Title of Presentation. ASAE Paper No. 05xxxx. St. Joseph, Mich.: ASAE. For information about securing permission to reprint or reproduce a technical presentation, please contact ASAE at hq@asae.org or 269-429-0300 (2950 Niles Road, St. Joseph, MI 49085-9659 USA).

INTRODUCTION Cyclone inertial separators are used extensively in the processing industry to remove particulates from dust laden air streams. Much research has been performed to enhance the effectiveness of these devices in capturing particulate matter (PM) (Green et al., 2000; Baker et al., 1996; Parnell, 1990; Parnell, 1980; TCGA, 1965). To further reduce PM emissions, several studies have explored pre-separators to remove large material (>100 µm) before the dust laden air enters the cyclone (Mihalski et al., 1993; Columbus, 1994; Baker et al., 1995; Wang et al. 2004; Buser et al., 2005). Another method of reducting PM emissions is using cyclones in series, the first cyclone being the pre-separator. Gillum et al. (1982) tested both a 2D2D cyclone design and a 1D3D cyclone design as secondary collectors to a 2D2D cyclone. The collection efficiency for the primary cyclone averaged 99.6% and was not significantly different between treatments. The collection efficiencies were significantly different between 2D2D (45.7%) and 1D3D (54.0%) secondary cyclones. This research also showed that at 15.2 m/s (3000 fpm) inlet velocity the pressure drop across the preliminary cyclone (896 Pa [3.6 in. w.g.]) was not significantly different between the treatments; however, the pressure drop across the two secondary cyclones was different (1010 Pa [4.06 in. w.g.] for the 2D2D and 1115 Pa [4.48 in. w.g.] for the 1D3D). Gillum and Hughs (1983) showed that the combined collection efficiency of two cyclones in series, 2D2D primary and 2D2D or 1D3D secondary, did not vary when the inlet velocity ranged from 11.8 to 18.3 m/s (2323 to 3602 fpm), but the total system pressure drop for the lower inlet velocity (1207 Pa [4.85 in. w.g.]) was half that of the higher inlet velocity (2852 Pa [11.46 in. w.g.]). Both tests mentioned above were run with gin trash consisting of all the trash collected in a seed-cotton system cyclone hopper. Also, primary cyclone collection efficiency was based on trash fed to the system and trash collected by the primary cyclone and secondary cyclone collection efficiency was based on trash collected by the secondary cyclone and secondary cyclone emissions measured using Environmental Protection Agency Method 5 procedures (U.S. EPA, 2004). In two studies, Columbus (1993) evaluated a 2D2D primary cyclone in series with a 1D3D secondary cyclone. The cyclones were arranged to capture PM emitted from a seed cotton separator. Modified high-volume samplers were used to isokinetically sample the dust-laden air before and after the primary cyclone and after the secondary cyclone. Filter weights were used to calculate collection efficiency and particle size distribution (PSD) analyses were performed on the PM captured on the filters. The mass median diameter of the dust entering the cyclones in both studies was about 10 µm and the inlet concentrations ranged from 39 to 154 mg/m 3. Results from the two studies showed that the overall collection efficiency of the 2D2D primary cyclone averaged 90.5% and 94.2% and the collection efficiency of the 1D3D secondary cyclone averaged 69.0% and 38.5%. The PM 10 (particulate 10 µm) collection efficiency of the 2D2D primary cyclone averaged 84.5% and 91.5% in each study and that of the 1D3D secondary cyclone averaged 70.6% and 28.8% for each study. The studies to date exploring series cyclones have used 2D2D designs for the primary cyclone and no studies have explored more than two in series. This study was prompted by ongoing research at a feed supplement processing facility that handles large quantities of very fine material. The objective of this study was to evaluate the performance of a single 1D3D cyclone and compare its performance to multiple (up to four) 1D3D arranged in series. Materials and Methods The test was set-up as a randomized complete block design with five replications. The test was blocked by replication. The four treatments: one, two, three, or four cyclones in series were randomly assigned within each block for a total of 20 runs. Statistical analyses were performed using SAS General Linear Models (SAS, 1999). 2

A schematic of the testing system is shown in figure 1 and included: volumetric dust feeder; one, two, three, or four (shown in fig. 1) 30.5-cm (12-in.) diameter cyclones in series; centrifugal fan; Y-valve to filter bank or bypass; centrifugal fan to atmosphere. The cyclones were 1D3D design (fig. 2) with 2D2D design inlets. All cyclones were identical, except the No. 4 cyclone had a D/4 trash outlet. Before each run, conveying air temperature, relative humidity, and barometric pressure were recorded. Airflow through the cyclones was measured with a hot-wire anemometer and adjusted to approximately 11.3 m 3 /min (400 cfm) by means of a fan butterfly valve; 16.3 m/s (3200 fpm) cyclone inlet velocity resulting. Static pressure was measured with a magnahelic gauge at the entrance and exit of each cyclone to determine the individual cyclone pressure drops and near the fan entrance to determine total system pressure drop. Eight, 20.3 25.4 cm (8 x 10 in.), pre-weighed glass-fiber filters were loaded into the filter bank. Approximately 14.5 kg (32 lb) of test dust was weighed to the nearest 4.5 g (0.01 lb) and then placed in the volumetric feeder that rested on a digital balance. The volumetric feeder was set to meter approximately 2.7 kg (6 lb) of test dust per minute. This loading rate is high, but typical for the processing facility currently being studied. Tests 1 and 2 were run for about two minutes through the filter bypass and two minutes through the filters. After these two runs, the remaining tests were run for three and one-half minutes through the filter bypass and 30 s through the filters because of the high PM loading rate. After four minutes, pressure drops across each cyclone and the total system static pressure were measured. After each run, the PM captured by each cyclone was weighed to the nearest 4.5 g (0.01 lb). The mass of PM metered, indicated by the volumetric feeder balance, was recorded. PM remaining in the feeder was carefully removed and weighed, and this value was subtracted from the starting test PM weight; used as a check for the volumetric feeder balance value. Samples of the PM fed into the system and captured by the cyclones were collected for PSD analyses using a Coulter Counter Multisizer III (Beckman Coulter, Fullerton, CA). The eight filters were removed from the filter bank and placed in anti-static envelopes for weighing (nearest 10-6 g) and PSD analyses. PSD analyses followed procedures described by Buser (2004) Results Table 1 summarizes the controlled variables and measured parameters. There were significant differences in inlet velocity among the treatments, number of cyclones in series. These differences were, for practical purposes, very small ( 0.2 m/s [40 fpm]). The PM loading averaged 273 g/m 3 and was not significantly different among treatments. Although the airflow was consistent, the static pressure drop across the cyclones tended to decrease as the number of cyclones in series increased. The average pressure drop across the No. 1 cyclone was significantly higher when in a single cyclone configuration (1356 Pa [5.45 in. w.g.]) than when in a two or three series cyclone configuration (946 Pa [3.80 in. w.g.]) and was significantly less in a four series cyclone configuration (756 Pa [3.04 in. w.g.]) than the other configurations. Similarly, the pressure drop across the No. 2 cyclone was significantly less for the four series cyclone configuration (667 Pa [2.68 in. w.g.]) than the two or three cyclone configurations. Typically, a change in cyclone pressure drop from 747 to 1369 Pa (3 to 5.5 in. w.g.) would be the result of a 5.1-m/s (1000-fpm) change in inlet velocity (Parnell et al., 1994), but that was not the case in this study. The overall system static pressure (fan inlet static pressure, table 1) was numerically only slightly higher for two cyclones in series (3496 Pa [14.05 in. w.g.]) than for a single cyclone (3123 Pa [12.55 in. w.g.]). Throughout the test, ambient temperature averaged 34.7ºC (94.4ºF), barometric pressure averaged 896.6 kpa (265.5 in. Hg), and relative humidity averaged 32%. There were no significant differences in these ambient conditions among treatments. Cyclone efficiency, based the weight of PM fed into the system and the weight of the PM captured by each cyclone (table 2), followed similar trends to those discussed above for cyclone pressure drop. The efficiency of the No. 1 cyclone for a single cyclone configuration was lowest (90.9%) and that for the four 3

series cyclone configuration was highest (94.6%), and both were significantly different than the other configurations. A comparison of the data in tables 1 and 2 revealed that cyclone configurations with less pressure drop across the No. 1 cyclone had higher No. 1 cyclone collection efficiencies. The No. 2 cyclone collection efficiency was significantly less for the four series cyclone configuration than for the three cyclone configuration. No. 2 cyclone efficiency did not follow the same trend with pressure drop as No. 1 cyclone efficiency and, more likely, lower No. 2 cyclone efficiency was a consequence of higher No. 1 efficiencies for a particular configuration. In general, overall efficiency increased significantly as the number of cyclones in series increased and jumped from 90.0% for a single cyclone to 97.2% for two cyclones in series. However, there was no significant difference in collection efficiency between the two and three series cyclone configurations or between the three and four cyclone configurations. As expected, the individual collection efficiency of sequential cyclones decreased from one cyclone to the next; starting near 94% for the first cyclone connected in series and decreasing to about 56, 22, and 13% for the 2nd, 3rd, and 4th cyclones, respectively. PSD analyses of the PM fed from each run showed that there were no significant differences in mass median diameter (MMD), geometric standard deviation (GSD), percentage of particles less than or equal to 10 µm (PM 10 ), and percentage of particles less than or equal to 2.5 µm (PM 2.5 ) among the treatments (fig.3). Results in table 2, showing that the No. 1 cyclone was significantly more efficient when in a series configuration than when alone, at first prompted the conclusion that the PM captured in those different configurations must have been different to explain the difference in efficiency. But this was not the case. There were no significant differences in the MMD, GSD, PM 10, and PM 2.5 of the PM captured by the No. 1 cyclone among treatments (fig. 4). The same was true for the No. 2 cyclone. There were significant differences detected in those variables among treatments for the No. 3 cyclone (fig. 4), but the differences were small and not of practical importance. The PM captured by the No. 1 cyclone was essentially no different from the feed dust (table 3) having MMD = 8.78 µm, PM 10 = 59.3%, and PM 2.5 = 1.2%. There was a significant difference in the GSD of the PM captured by the No. 1 cyclone (1.73) and that of the feed PM (1.81). For the second, third, and fourth cyclones, nearly all (>94%) of the material captured was less than 10 µm. Also, more than 50% of the material captured by the third and fourth cyclones was PM 2.5. This may become important as more attention is focused on PM 2.5. As dust laden air was processed through one to four cyclones, the PM collected by succeeding cyclones was smaller. Figure 5a illustrates this shift in the PSDs. It is generally accepted that succeeding cyclones in a series configuration have lower efficiencies than preceding cyclones, because preceding cyclones remove most of the larger particles resulting in lighter loading and smaller particles for the succeeding cyclones (Cooper and Alley, 1994). These PSD results from table 3 and figure 5, along with the efficiency values from table 2, support that conclusion. Similar to the trend observed for the PM captured by the cyclones, as the number of cyclones in series increased from one to four, the size of the PM that passed through the cyclones to the filters decreased significantly from 3.23 to 1.52 µm (table 4). This shift is illustrated in figure 5b. Almost no PM that passed through the cyclones with one, two, three, or four in series was larger than 10 µm. There was a similarity between the feed PM and the PM captured by cyclone No. 1. A comparison of tables 3 and 4 revealed that the PM emitted by a single cyclone, which would essentially be the feed PM for cyclone No. 2, was likewise similar to the PM captured by cyclone No. 2, but to a lesser extent. These similarities diminished as the number of cyclones in series increase and subsequently the size of particles decreased. Conclusions Tests were performed to evaluate the performance of multiple 1D3D cyclones in series. A high dust loading rate and PM characterized by a MMD of 8.7 µm and a GSD of 1.8 were used for the evaluation. At recommended airflow rates, the pressure drop across a single cyclone alone was significantly higher 4

than that of the No. 1 cyclone in a series configuration. Because of this, the overall system static pressure for two cyclones in series was not different from single cyclone. For the PM and very high loading rate used, the overall efficiency of a single cyclone was not as high as the series configurations. This was mainly due to the additional collection efficiency of the succeeding cyclones, but also because the No. 1 cyclone in a series was consistently more efficient than a single cyclone. This difference in efficiency among a single cyclone and the No. 1 cyclone in series configurations could only be connected with the differences in pressure drop measured. There was a significant shift in the PSD (from larger to smaller particle sizes) of the dust captured by the cyclones in sequence. This was evidence for the idea that the first cyclone removes more large material, the next cyclone removes less and smaller material, and so forth. After the first cyclone, most of the dust captured was less than 10 µm. There was also a significant shift in the PSD (from larger to smaller particle sizes) of the dust emitted by the cyclones as the number cyclones in series increased from one to two. For this fine dust, almost none of the dust emitted by a single or series of cyclones was larger than 10 µm. The study showed that a substantial increase in collection efficiency over a single cyclone could be realized with two cyclones in series with less increase in static pressure loss than expected. However, adding cyclones for three or four in series increased efficiency only slightly at a considerable increase in pressure loss. References Baker, R.V., M.N. Gillum, and S.E. Hughs. 1995. Pre-separators and cyclones for the collection of stripper cotton trash. Trans ASAE. 28(5): 1335-1342. Baker, R.V., S.E. Hughs, and J.K. Green. 1996. Overview of new emission control strategies for cotton gins. In Proc. 1996 Beltwide Cotton Conf., 566-571. Memphis, TN: National Cotton Council. Buser, M.D., D.P. Whitelock, G.A. Holt, C.B. Armijo, and L. Wang. 2005. Preliminary evaluation of the baffle-type pre-separator in terms of baffle location, critical air velocity, and loading rate.. In Proc. 2005 Beltwide Cotton Conf., 469-485. Memphis, TN: National Cotton Council. Columbus, E.P. 1993. Series cyclone arrangements to reduce gin emissions. Trans ASAE 36(2):545-550. Columbus, E.P. 1994. A pre-separator for cyclones at cotton gins. In Proc. 1994 Beltwide Cotton Production Conferences, 1741-17448. Memphis, TN: National Cotton Council. Gillum, M.N., S.E. Hughs, and B.M. Armijo. 1982. Use of secondary cyclones for reducing gin emissions. Trans ASAE 25(1):210-213. Gillum, M.N. and S.E. Hughs. 1983. Velocity effects of operating parameters of series cyclones. Trans ASAE 26(2):606-609. Green, J.K., P.A. Funk, and G.A. Holt. 2000. Current recommendations for gin emission control. In Proc. 2000 Beltwide Cotton Conf., 474-475. Memphis, TN: National Cotton Council. Mihalski, K., P. Kaspar, and C.B. Parnell, Jr. 1993. Design of pre-separators for cyclone collectors. In Proc. 1993 Beltwide Cotton Production Conferences, 1561-1568. Memphis, TN: National Cotton Council. Parnell, C.B. 1980. Design of cyclone collectors to minimize dust emissions. Oil Mill Gazetteer, 16-19. Champaign, IL: International Oil Mill Superintendents Assoc. 5

Parnell, C.B. 1990. Cyclone design for cotton gins. ASAE Paper No. 905102. St. Joseph, MI: ASAE. Parnell, Jr., C.B., E.P. Columbus, and W.E. Mayfield. 1994. Abatement of air pollution and disposal of gin waste. In Cotton Ginners' Handbook, 172-194. W.S Anthony and W.D. Mayfield, eds. Agricultural Handbook No. 503. Washington, D.C.: USDA. SAS. 1999. SAS Online Doc. Ver. 8. Cary, N.C.: SAS Institute, Inc. Available at: v8doc.sas.com/sashtml/. TCGA. 1965. What we know about air pollution control: Special bulletin No. 1. Austin, TX: Texas Cotton Ginner's Association. U.S. EPA. 2004. Code of Federal Regulations, Title 40, Part 50, Appendix A. Standards of performance for new stationary sources - Method 5. Washington, D.C.: U.S. EPA. Wang, L., J.D. Wanjura, C.B. Parnell, Jr., B.W. Shaw, R.E. Lacey, S.C. Capareda, and M.D. Buser. 2004. Study of baffle-type pre-separator plus cyclone abatement systems for cotton gins. ASAE Paper No. 044017. St. Joseph, MI: ASAE. 6

Table 1 - Measured test variables [a] for each treatment. Static Pressure Drop, Pa (in. w.g.) System Static Number of Series Cyclones Inlet Velocity m/s (fpm) Loading (g/m 3 ) No. 1 Cyclone No. 2 Cyclone No. 3 Cyclone No. 4 Cyclone Pressure Pa (in. w.g.) 1 15.66 (3083) b 243.0 1356 (5.45) a ---- ---- ---- 3123 (12.55) b 2 15.86 (3123) a 234.2 975 (3.92) b 794 (3.19) b ---- ---- 3496 (14.05) b 3 15.72 (3095) ab 234.1 913 (3.67) b 883 (3.55) a 699 (2.81) ---- 5574 (22.40) a 4 15.86 (3123) a 236.3 756 (3.04) c 667 (2.68) c 575 (2.31) 901 (3.62) 5649 (22.70) a Pvalue 0.03 0.60 <0.0001 0.002 0.06 ---- <0.0001 [a] Means in a column followed by the same letter are not significantly different (LSD, P=0.05) Table 2 - Cyclone efficiency [a] (%) based on dust captured by each cyclone in series. Number of Series Cyclones No. 1 Cyclone No. 2 Cyclone No. 3 Cyclone No. 4 Cyclone Overall 1 90.9 c ----- ----- ----- 90.9 c 2 93.5 b 55.3 ab ----- ----- 97.2 b 3 93.2 b 59.8 a 24.7 ----- 97.9 ab 4 94.6 a 51.6 b 20 12.6 98.2 a Pvalue <0.0001 0.04 0.06 ---- <0.0001 [a] Means in a column followed by the same letter are not significantly different (LSD, P=0.05) Table 3 - Particle size distribution parameters [a] for feed dust and dust captured by each cyclone in series. Dust Type Mass Median Diameter (µm) Geometric Standard Deviation PM 10 (%) PM 2.5 (%) Feed 8.65 a 1.81 b 59.7 c 1.9 d Cyclone No. 1 8.78 a 1.73 c 59.3 c 1.2 d Cyclone No. 2 3.69 b 1.84 a 94.7 b 26.5 c Cyclone No. 3 2.30 c 1.60 d 99.9 a 57.3 b Cyclone No. 4 1.86 d 1.47 e >99.9 a 78.2 a Pvalue <0.0001 <0.0001 <0.0001 <0.0001 [a] Means in a column followed by the same letter are not significantly different (LSD, P=0.05) Table 4 - Particle size distribution parameters [a] for filters. Number of Series Cyclones Mass Median Diameter (µm) Geometric Standard Deviation PM 10 (%) PM 2.5 (%) 1 3.23 a 1.92 a 95.6 b 35.3 c 2 1.79 b 1.60 b >99.9 a 78.2 b 3 1.67 bc 1.53 b >99.9 a 81.2 b 4 1.52 c 1.47 b >99.9 a 91.2 a Pvalue <0.0001 <0.0001 <0.0001 <0.0001 [a] Means in a column followed by the same letter are not significantly different (LSD, P=0.05) 7

Figure1. Test system. Figure 2. 1D3D Cyclone. 8

MMD: 8.6 8.8 µm NS GSD: 1.80 1.82 NS PM10: 58.7 60.3% NS PM2.5: 1.7 2.0% NS Figure 3. Particle size distribution of feed dust. NS indicates no significant difference (P = 0.05) among treatments. MMD: 8.6 9.0 µm NS GSD: 1.72 1.74 NS PM 10 : 57.6 60.3% NS PM 2.5 : 1.0 1.3% NS MMD: 3.5 3.8 µm NS GSD: 1.82 1.86 NS PM 10 : 94.0 95.8% NS PM 2.5 : 25.5 29.0% NS (a) (b) MMD: 2.2 2.4 µm P=0.03 GSD: 1.57 1.63 P=0.04 PM 10 : 99.8 99.9% NS PM 2.5 : 54.3 61.1% P=0.04 MMD: 1.86 µm GSD: 1.47 PM 10 : 99.9% PM 2.5 : 78.2% (c) (d) Figure 4. Particle size distributions of dust captured by the a) No. 1 cyclone, b) No. 2 cyclone, c) No. 3 cyclone, d) No.4 cyclone. NS indicates no significant difference (P = 0.05) among treatments. 9

1.9 µm 1.5 µm 2.2 µm 3.5 µm 9.0 µm 1.7 µm 1.8 µm 3.2 µm (a) (b) Figure 5. Shift in particle size distributions of dust a) captured by cyclones in series and b) collected on filters downstream from the cyclones. Values over curves indicate the mass median diameter. 10