1、2009 ASHRAE 831ABSTRACT Particulate matter (PM) emissions from the poultry industry are a major pollution concern. Development of an effective PM mitigation technology is urgently needed. In this study, the potential of an electrostatic precipitator (ESP) for collecting PM emissions from poultry fac
2、ilities was evaluated using a commercial ESP unit under simulated laboratory conditions. The effects of operating parameters such as char-ger voltage, superficial air velocity and PM concentration on the performance of a two-stage plate ESP were evaluated. An empirical model for predicting ESP perfo
3、rmance was devel-oped. Preliminary factor screening analyses have shown that charger voltage (7 kV to 10 kV) and superficial air velocity 1 m/s (200 ft/min) to 5 m/s (1000 ft/min) affect ESP perfor-mance significantly while PM concentration (2.5 mg/m3to 5 mg/m3) had negligible effect (=0.05). The ES
4、P collection efficiency increased with voltage and decreased with velocity. The relationship between charger voltage and superficial air velocity and PM collection efficiency for particles within size ranges of 0.3-0.5, 0.5-1, 1-5, 5-10, 10-25, 25 , PM1, PM5, and PM10were apparently nonlinear. The E
5、SP performance approached a peak point at 9 kV charger voltage and 1 m/s air velocity. The optimized operating condition to be used at an actual farm would depend on the desired degree of PM collec-tion. To collect at least 90% of all particles, the operating conditions were 9 kV and 2.5 m s-1. The
6、power consumption of the ESP to collect 90% of the total particles is 90 watts/m2(29 BTU/h/ft2) of duct cross-sectional area. The ESP unit produced very negligible amount of ozone (5kV), industrial ESP power requirement is very minimal (16 watt m-2 5.1 BTU h-1ft-2 of treated duct cross-section, Vata
7、vuk, 1984) since the electric current consumed to charge and collect PM is very low (25 m with a Table 1. Screening Factorial Design for the ESP Operating ConditionsTreatmentVoltage (kV)Velocity ms-1(ftmin-1)PM Loadmgm-3(gr ft-3)A 7 1.25 (246) 2.5 (0.0011)B 7 1.25 (246) 5 (0.0022)C 7 2.5 (490) 2.5 (
8、0.0011)D 7 2.5 (490) 5 (.00022)E 10 1.25 (246) 2.5 (0.0011)F 10 1.25 (246) 5 (0.0022)G 10 2.5 (490) 2.5 (0.0011)H 10 2.5 (490) 5 (0.0022)Table 2. Full Factorial Design for the E-v Response CurveTreatmentVoltage (kV)Velocity ms-1(ft min-1)A 7 1 (200)B 8.5 1 (200)C 10 1 (200)D 7 3 (600)E 8.5 3 (600)F
9、10 3 (600)G 7 5 (1000)H 8.5 5 (1000)I 10 5 (1000)ASHRAE Transactions 835counting limit of 3.3 x 107particles m-3(9.3 x 10-5 particles ft-3). Sampling was performed using iso-kinetic sampling heads for accurate PM measurement. TSP PM mass concentration measurements were conducted at 1.25 and 2.5 m s-
10、1 (246 and 492 ft min-1) to calibrate the rotational speed of the dust feeder to generate 2.5, and 5 mg m-3 (0.0011 and 0.0022 gr ft-3). The temperature and relative humidity readings of the instrument were calibrated and used for all the measurements. Data AnalysisEfficiency Calculations. The perfo
11、rmance of the ESP is determined by particle collection efficiency measured at six particle size ranges (25 m) using the formula:(4)where= the particle collection efficiency at the ith range,COi= the outlet particle concentration at the ith range number m-3(number ft-3),CIi= the inlet particle concen
12、tration at the ith range number of particles m-3(number ft-3), andPi= particle penetration at the ith range.The average collection efficiency of classified ranges which includes PM1, PM5, PM10and the total particles measured were calculated (PMis mean particles whose aero-dynamic diameters are 25 m,
13、 PM1, PM5, PM10, and total particle size ranges are shown in table 4. Results indi-cated that voltage was highly correlated with the measured Table 3. Specifications of the Module Two-Stage Electrostatic PrecipitatorDimensions (W x H) in mm (in.) 400 x 640 (16 x 25)Weight in kg (lb) 2.26 (4.98)Numbe
14、r of Plates 71Number of Ionizers 9Design Ionizer Voltage in Vdc 8150Design Collector Voltage in Vdc 4075Pressure Drop at 680 m3h-1in Pa (in H20) 12.5 (0.05)Plate Width in mm (in.) 76 (3)Collector plate Spacing in mm (in.) 3.6 (0.14)Charger plate to wire spacing in mm (in.) 15 (0.59)i1COiCIi- 1 Pi=ii
15、 AVE,iCiiCii- 1 Pi AVE,=i AVE,PiAVE,i1 b0expEn1vn2- b1En3vn4=836 ASHRAE TransactionsFigure 3 Schematic diagram of the ESP laboratory evaluation set-up.ASHRAE Transactions 837collection efficiencies, with correlation coefficients ranging from 0.46 to 0.89. Air velocity showed higher correlation only
16、for collection efficiencies of particles greater than 5 m, with correlation coefficients ranging from -0.62 to -0.57. PM concentration, temperature and relative humidity all showed very weak correlations with collection efficiency with corre-lation coefficients ranging from -0.15 to 0.15 for PM conc
17、en-tration, -0.09 to 0.13 for temperature, and -0.27 to -0.08 for humidity, respectively.Results suggest that collection efficiency tend to increase with voltage and decrease with velocity as indicated by the positive and negative correlation coefficients computed for each respective factor. The tre
18、nd also agreed with the analysis of the parameters made in equation 3. In addition, the logical explanation for the observations mentioned is that increasing charger voltage increases the rate of ion generation which facilitates particle charging while increasing air velocity decreases retention tim
19、e and increases particle re-entrainment.The relatively high correlation coefficients between volt-age and collection efficiency also suggests that operating volt-age had strong influence on collection efficiency. However, the correlation coefficient between air velocity and collection efficiency inc
20、reased between 0.5 m to 10 m only while equation 3 predicts that increasing particle diameter should lead to an increase in effect. Oglesby and Nichols (1978) explains that deviations such as these are caused by PM re-entrainment which has stronger influence as particle size is increased. This obser
21、vation was not in agreement with the predictions of equation 3 since it did not account for the effect of reentrainment.The effects of environmental factors like PM load, temperature, and humidity (not directly implied in equation 3) were also evaluated. Results showed that the initial perfor-mance
22、of the ESP was not affected by PM loads ranging 2.5 to 5 mg m-3(0.0011 to 0.0022 gr ft-3), inside the air conditioned room whose temperature varied from 24oC to 27oC (75 to 81 oF), and relative humidity varied 18% to 62%. However, the wider range of temperature and humidity variations typically enco
23、untered in poultry houses may have an effect on ESP performance as suggested by Salam (1992). In the study, vari-ations in temperature and humidity were minimized to a degree which did not affect ESP performance.Effects of Voltage and Superficial Air VelocityFigure 4 shows the contour plots of the m
24、easured ESP collection efficiencies for particles sized 0.3 m-0.5 m, 0.5 m-1 m, 1 m-5 m, 5 m-10 m, 10 m-25 m, 25 m. In general, the plots confirmed observations obtained during the screening tests that collection efficiency increased with volt-age while it decreased with velocity. However, these act
25、ual contour plots revealed greater detail on the effect of charger voltage and superficial air velocity on ESP collection effi-ciency. By observation, the curves were apparently nonlinear whose contours approached a peak point near the region of the 9 kV charger voltage and 1 m s-1(200 ft min-1) air
26、 velocity. The same trend was observed for the contour plots of the measured number averaged collection efficiency (PM1, PM5, PM10, and total particles) as affected by charger voltage and superficial air velocity (Figure 5). The shape of the number averaged efficiency curves shown in Figure 5 were,
27、however, more sigmoidal compared to those presented in Figure 4, which can be described as hyperbolic.Statistical analyses also showed that the effect of temper-ature 24 to 27oC (75 to 80oF) and humidity (18% - 62%) vari-ations encountered during the test did not significantly affect the outcome of
28、these results ( =0.05). Effect of Operating Time on ESP PerformancePM concentration was proven to have negligible effect on the initial performance of the ESP when tested at 10 kV and 2.5 m s-1(490 ft min-1) based on analysis of variance at =0.05. However, further investigation reveals that ESP perf
29、ormance declines with time due to the accumulation of particles on the plates that tend to coagulate and get re-entrained with the airflow. The plot in figure 6 shows that the Table 4. Results of the Factor Correlation Analysis of the Screening ExperimentEfficiency at Size Range Voltage Velocity Loa
30、d Humidity Temperature0.3m-0.5m 0.89 -0.08 -0.11 -0.08 0.130.5m -1m 0.86 -0.29 0.15 -0.17 0.021m -5m 0.8 -0.44 -0.02 -0.19 -0.065m -10m 0.59 -0.62 -0.09 -0.27 -0.210m -25m 0.46 -0.59 -0.16 -0.28 -0.225m 0.47 -0.57 -0.15 -0.26 -0.09Total 0.87 -0.16 -0.04 -0.26 -0.09PM10.88 -0.13 -0.06 -0.26 -0.09PM50
31、.87 -0.16 -0.04 -0.26 -0.09PM100.87 -0.16 -0.04 0.13 -0.08838 ASHRAE TransactionsFigure 4a Contour plots of the actual ESP collection efficiencies (in SI units) as affected by charger voltage and superficial air velocity for particles: (a) 0.3 -0.5 , (b) 0.5 -1 , (c) 1 -5 , (d) 5 -10 , (e) 10 -25 ,
32、(f) 25 .m m m m m m mm m m mASHRAE Transactions 839Figure 4b Graphs a through f in I-P units. 840 ASHRAE TransactionsFigure 5 Contour plots of the actual ESP collection efficiencies (in SI and I-P units) as affected by charger voltage and superficial air velocity for: (a) PM1, (b) PM5, (c) PM10, and
33、 (d) total particles.ASHRAE Transactions 841collection efficiency of particles greater than 10 m decreased from 90% to 80% after 30 minutes of operation. On the other hand, the collection efficiency of particles between 5 m and 10 m began to decline from 90% to 85% after 45 minutes. Finally, the col
34、lection of all the particles greater than 0.3 m showed signs of decreased performance after 90 minutes. This problem can be remedied by frequent cleaning of the plates. However, an interval of less than 1 hour may not be practical when applied in actual farm conditions. Further research is needed to
35、 solve this problem. Effect of Dust Type on ESP PerformanceFigure 7 shows the difference in performance of the ESP for collecting poultry PM and standard fine dust. Results showed that poultry PM was easier to capture by electrostatic precipitation compared to standard fine road dust. However, stand
36、ard fine dust (CMD = 9 m) had larger particles compared to poultry PM (CMD = 1 m) suggesting that comparison was based on composition (i.e. organic versus inorganic) was not entirely valid. It must also be noted that although the manufacturer provided the ESP with a higher efficiency rating ( 99%),
37、it was modified for this test by removing the pre and post filter and operated with a variable voltage power supply.At 2.5 m s-1(490 ft min-1) superficial air velocity, poultry PM collection efficiency for total particles was 50% at 8 kV and 75% at 10 kV while standard fine road dust collection effi
38、-ciency was only 10% at 8 kV and 40% at 10 kV for any particle size range. At 3.0 m s-1(600 ft min-1) superficial air velocity, poultry PM collection efficiency was 40% at 8 kV to 60% at 10 kV while standard fine dust collection efficiency was -10% at 8 kV to 25% for total particles. The average col
39、lection effi-ciency for total particles became negative since the particles less than 1 m were greater in the inlet than in the outlet while most of the number of particles belong to that range. There was also a decrease in the expected collection performance of poultry PM which dropped to only 70%
40、from the expected 90% collection efficiency at 2.5 m s-1(490 ft min-1) air veloc-ity and 9 kV charger voltage. This may be due to the effect of air humidity which went up to 73% saturation during the test. Further analysis showed that this effect was only experienced by the particles in the 0.3 m to
41、 0.5 m bin since all the parti-cles greater than 0.5 m were collected by at least 90%.Empirical Models of ESP PerformanceThe coefficients of equation 6 for predicting particle collection efficiency as a function of voltage and air velocity (E-v equations) for 0.3 m-0.5 m, 0.5 m-1 m, 1 m-5 m, 5 m-10
42、m, 10 m-25 m, and 25 m, PM1, PM5, PM10, and total particle size ranges are shown in table 5. The final coef-ficients selected obtained the maximum value of the log-like-lihood ratio for the equation and also has the least sum of square of residuals. Each coefficient estimates passed the tests of sig
43、nificance. However, for conciseness purposes, only the largest probability value (P-value) computed among all the coefficients of the equation was shown in table 5. This means Figure 6 The transient behavior of the commercial ESP set at kV operating voltage and 2.5 ms-1(500 ft min-1) operating veloc
44、ity.Figure 7 Comparison of the performance of the ESP for collecting standard fine dust and poultry PM for collection of total particles. 842 ASHRAE Transactionsthat the P-value of the of all the parameters (i.e. bo, b1, n1, n2, and n3) were at least the parameter P-value presented or better.The mod
45、el with the least log-likelihood ratio is the effi-ciency curve for the 0.3 m-0.5 m size range; the model with the maximum log-likelihood ratio is the efficiency curve for the 1 m-5 m size range. The parameter n2was consistently greater than n1for all models with an n2/n1ratio ranging from 1.3 to 2.
46、 The parameters b0and b1(if present) were set to 1 after several attempts of iteration show that their parameter estimates failed the t-test. All iterations for the parameter n4failed to converge, so it was set to zero. The t-tests on param-eter n4succeeded only for the efficiency curve for PM1and t
47、otal collection efficiency. Surface plots of the fitted models of collection efficiency for the particles which belong to 0.3 m-0.5 m, 0.5 m-1 m, 1 m-5 m, 5 m-10 m, 10 m-25 m, and 25 m bin range are shown in figure 8. The scatter plots of the actual efficiencies were also simultaneously shown for co
48、mparison. In general, the model was able to explain most of the measured data.The best fit surface based on visual inspection is that in figure 4c, which is the efficiency curve for 1 m-5 m size particles. The surface with the least desirable fit was that in figure 4a, which is the efficiency curve
49、for 0.3 m-0.5 m particles. As discussed previously, these empirical models were shown to have the maximum and minimum calculated log-likelihood ratios, respectively.Surface plots of the fitted models of collection efficiency for the PM1, PM5, PM10and the total particles are shown in figure 9. In general, the empirical models for the averaged collection efficiencies explained the data well. However, the residuals were greater at the region near the vo
