CHAPTER 3 METHODS
3.2 Numerical method
3.2.4 Turbulent flow field in a pilot scale wet ESP
In order to examine the effect of variation of design parameters for ESPs on particle collection efficiency, a pilot scale wet ESP was sumulated which had 750-1100 mm in length, 150-300 mm in width, and 6 m in height. There are 3-6 discharge wires each of which had the diameter of 1.3-2.5 mm. The wire to wire spacing and wire to plate spacing was 50-75 and 75-150 mm, respectively. A total of 21,744 (302 in x-direction, 72 in y-direction) non-uniform rectangular girds were used in this calculation domain.
The turbulent flow field model was used in this case. The time-averaged equations for mass, momentum, and energy are
) 0
where μt is turbulent viscosity (kg/m-s), ε is turbulent dissipation rate (J/s-kg),and k is turbulent kinetic energy (J/s-kg). The parameters used in k-ε model are as follows:
σk σε Cμ CD C1 C2
1.0 1.3 0.09 1.0 1.44 1.92
CHAPTER 4
RESULTS AND DISCUSSION
4.1 Experimental results for the particle collection efficiency of the wet ESP 4.1.1 Flow rate, uniformity and thickness of the water film
Surface texture and liquid flow rate are two important factors that affect the uniformity of the water film (Tsai et al. 2008). As mentioned previously, the contact angle was greater than 100° for the smooth collection electrodes or uncoated sand-blasted electrodes. It was found that water film on these electrodes was not uniform. In comparison, for the TiO2 coated, sand-blasted electrodes, scrubbing water was found to be uniform over the collection surfaces at the liquid flow rate per collection surface area (0.075×0.17 m2) of 2.31 L/min/m2. Similar findings were reported by Tsai et al. (2008), Yu et al. (2001) and Liu et al. (2002). It was also found that channeling occurred when the liquid flow rate per collection surface area was lower than 2.31 L/min/m2. These findings demonstrate that both surface texture and liquid flow rate are important for creating a uniform water film. Comparing with the liquid flow rate of 4.44-6.22 L/min/m2 used by Saiyasitpanich et al. (2006; 2007), the present design saved 48-63 % of the scrubbing water. The thickness (w) of the water film was calculated to be 0.1±0.18 mm by the method presented in Tsai et al. (2008). With this small thickness, the water film was still found to decrease the corona current compared to a dry ESP with the same design except without scrubbing water or purge air flows. Figure 4.1 shows the comparison of the corona current between the dry and wet ESPs, at various supplied voltages. It is seen that at the same voltage, the corona current was decreased after supplying the scrubbing water on the collection plate surfaces, due to the resistivity of the water film. Hence, the wet ESP collection efficiency is expected to be lower than the dry ESP operating at the same voltage.
Therefore, for the initially clean condition the operation voltage has to be increased for the
present wet ESP to have the same collection efficiency as the dry ESP. As will be shown later, the advantage of using scrubbing water flow and purge air flow becomes obvious when the collection surfaces are loaded heavily with TiO2 nanopowder.
4.1.2 Particle collection efficiency of the wet ESP at different aerosol flow rates and applied voltages
Figure 4.2 shows the collection efficiency of the wet ESP as a function of corn oil diameter when the aerosol flow rate is 5 and 10 L/min and the applied voltage is 4.3 kV under the initially clean condition. The residence time of the aerosol in the present wet ESP was calculated to be 0.39 and 0.19 s for aerosol flow rates of 5 and 10 L/min, respectively. As seen in Figure 4.2, decreasing the flow rate from 10 to 5 L/min (or increasing the residence time from 0.19 to 0.39 sec) had a substantial effect on the collection efficiency as the Deutsch-Anderson equation dictates. At the flow rate of 5 L/min, the collection efficiency was 96.9-99.7 % for particles from 16.8 to 615 nm in electrical mobility diameter. The minimum collection of 96.9 % corresponds to the particle diameter of 126.3 nm. As the flow rate is increased to 10 L/min, the minimum collection efficiency is reduced to 84.1 % at the corresponding diameter of 70 nm.
According to particle charging theory, particle charge decreases with decreasing particle diameter, while the mechanical mobility increases rapidly with decreasing particle diameter (Hinds 1999). Therefore, the collection efficiency of ESPs for particles in the size range of 0.1 to 1 μm has a U-shape efficiency curve, as shown in Figure 4.2. Saiyasitpanich et al. (2006) also found a U-shaped collection efficiency curve for the wet ESP with a minimum of about 95 % for particles between 80 and 250 nm in diameter when the supplied voltage was 70 kV, gas residence time was 0.4 s, and the particle concentration was 15.9 mg/Nm3.
3.2 3.4 3.6 3.8 4 4.2 4.4 4.
Applied Voltage (kV)
6 0
0.04 0.08 0.12 0.16
Corona Current (mA)
Initial clean condition Dry ESP Wet ESP After 2 hours particle loading test
Dry ESP
Figure 4.1 Corona current as a function of applied voltage in the dry and wet ESPs.
100 200 300 500 50
30 20
Particle diameter (nm) 70
80 90 100
Collection efficiency (%)
Aerosol flow rate 5 L/min 10 L/min
Figure 4.2 Collection efficiency of the present wet ESP for corn oil particles at the aerosol flow rate 5 and 10 L/min and the applied voltage of 4.3 kV. Each test was repeated 6 times.
Figure 4.3 shows the collection efficiency of the present wet ESP for corn oil particles in the size range of 16.8 to 615 nm under different applied voltages at the aerosol flow rate of 5 L/min. As shown in the figure, the corn oil particle collection efficiency also shows a U-shape efficiency curve for particles from 16.8 to 615 nm in electrical mobility diameter and the efficiency decreases with decreasing applied voltage because of the reduction of the electric field strength. When the applied voltage was decreased from 4.3 to 4.1 kV, the collection efficiency decreased slightly from 96.9-99.7 % to 93.9-99.4 %. The collection efficiency was further decreased to 74.2-82.4 % and 11.4-35.5 % for the same particle diameter range when the applied voltage was reduced to 3.9 and 3.8 kV, respectively. Based on these experimental results, it is certain that the present wet ESP can be operated to efficiently control fine and nanosized particles at an aerosol flow rate of 5 L/min, applied voltage of 4.3 kV, and scrubbing water flow rate per collection surface area of 2.31 L/min/m2.
Figure 4.4 shows the collection efficiency of the present wet ESP for corn oil particles in the size range of 16.8 to 615 nm under different applied voltages at the aerosol flow rate of 10 L/min. The collection efficiency can be increased from 86.9-99.4 % to 98.2-100 % when the applied voltage was increased from 4.3 to 4.9 kV. This result suggested that the increase of aerosol penetration due to the decrease of residence time of particles in ESPs can be diminished by increasing the applied voltage. The spark over occurs when the applied voltage is increased to 5.5 kV, which is the operation limitation for the present wet ESP.
In the present wet ESP, high particle collection efficiency (99 %) was found for small nanoparticles (16.8 to 29.4 nm) with an applied voltage of 4.3 kV. This is attributed to the high electrostatic precipitation efficiency ηelec (dp) and diffusive deposition ηdiff (dp) in this size range, which can be calculated by the following equations:
%
% 100 )(%)
( ,
in OFF out in p
diff C
C d C
(4.2)
where Cout,OFF is the outlet particle number concentration of the wet ESP without supplying high voltages. As can be seen in Figure 4.5, the electrostatic precipitation efficiency decreased from 99.4 % to 97.2 % when the particle diameter increased from 16.8 to 615 nm.
The diffusive deposition was found to be negligible for particles greater than 29.4 nm, and it increased from 4.0 to 17.4 % as particles decreased from 29.4 to 16.8 nm. These results demonstrate that diffusive deposition plays an important role for very small nanoparticles in the wet ESP.
100 200 300 500 50
30 20
Particle diameter (nm) 0
20 40 60 80 100
Collection efficiency (%) 3.8 kV
3.9 kV 4.1 kV 4.3 kV
Figure 4.3 Collection efficiency of the present wet ESP for corn oil particles under different applied voltages. The aerosol flow rate is fixed at 5 L/min. Each test was repeated 6 times.
100 200 400 600 80
60 40 20
Particle diameter (nm) 70
80 90 100
Collection efficiency (%)
Applied voltage (kV) 4.3
4.5 4.7 4.9
Figure 4.4 Collection efficiency of the present wet ESP for corn oil particles under different applied voltages. The aerosol flow rate is fixed at 10 L/min. Each test was repeated 6 times.
100 200 300 500 50
30 20
Particle diameter (nm) 0
20 40 60 80 100
Collection efficiency (%)
Diffusive deposition (%)
Electrostatic precipitation (%) Total collection efficiency (%)
Figure 4.5 Electrostatic precipitation and diffusive deposition efficiencies of the polydisperse corn oil particles in the present wet ESP when the aerosol flow rate and the applied voltage are 5 L/min and 4.3 kV, respectively. Each test was repeated 6 times.
4.1.3 Particle loading test
The particle collection efficiency for corn oil particles was tested by the dry and wet ESP at an aerosol flow rate of 5 L/min and an applied voltage of 4.3 kV for a long operation time of 10 hours. The collection efficiency in the dry ESP was found to decrease only slightly from 99.2-100 % to 98.7-100 % for 10.4-149 nm particles after the 10-hour test (loaded corn oil:
5.04×10-5 mg/plate). Particle accumulation on the collected plates and wires was found to be negligible due to the small loaded particle mass. Similar tests for the wet ESP also showed that the collection efficiency was as high as the initial condition (99.1-98.4 % for 10.4-149 nm particles) after 10-hours of loading corn oil particles. Therefore, heavy loading conditions on the collection plates had to be generated by dispersing powders, such as TiO2 nanopowder used in this study.
The collection efficiency was tested for the dry and wet ESP with an aerosol flow rate of 5 L/min, an applied voltage of 4.3 kV, and with a TiO2 loading quantity of 0.6±0.06 g/hr/plate (4±0.06 g/m3). As shown in Figure 4.6, the collection efficiency decreased slightly from 96.9-99.7 % to 92.1-99.3 % for electrical mobility diameters of 16.8-615 nm in the dry ESP after 30 minutes of particle loading. However, after one hour of loading, the aerosol penetration increased sharply from 0.9-1.6 % to 12.4-55.4 % for the same range of electrical mobility diameters. The particle collection efficiency finally decreased below 35.0 % for all particles (Figure 4.6). This resulted from the reduction of the electric field strength due to the accumulation of TiO2 particles on the collected plates and the discharge wires. As shown in Figure 4.1, it is evident that the corona current, and hence the electric field strength, decreases dramatically under the same applied voltage after two hours of heavy particle loading because of the additional resistivity of the dust layer.
Experimental results showed that after two hours of TiO2 nanoparticle loading, the total number concentration of reentrained particles was 612.2 #/cm3 (16.8 to 615 nm) at the first
sampling interval of 135 seconds and then reduced to almost zero after that (Figure 4.7).
Many particles were seen to be dislodged and redispersed back into the gas stream because the electrostatic voltage difference across the TiO2 dust layer was very low. These reentrained particles were too small to settle by gravity and would leave dry ESPs (Richards, 1995).
Under the same loading conditions as the dry ESP, the collection efficiency of the present wet ESP for corn oil particles was measured to be as high as 97.6-99.5 % and 94.9-99.9 % for particles from 16.8 to 615 nm in electrical mobility diameter after 30 and 120 minutes particle loadings, respectively, as shown in Figure 4.8. The particle loaded mass used in this study, 4±0.06 g/m3, is much higher than 2.8-15.9 mg/Nm3 used in Saiyasitpanich et al.
(2006; 2007). By direct observation, the continuous scrubbing water was seen to wash the collection plates clean and maintain a uniform water film. The pulse air jet was also capable of cleaning the discharge wires regularly. These were key factors in maintaining a high collection efficiency of the present wet ESP under heavy particle loading conditions.
100 200 300 500 50
30 20
Particle diameter (nm) 0
20 40 60 80 100
Collection efficiency (%)
Initial
After 30 min After 60 min After 120 min
Figure 4.6 Collection efficiency for corn oil particles in the dry ESP at different TiO2
nanopowder loadings. The applied voltage and aerosol flow rate are 4.3 kV and 5 L/min, respectively.
0 200 400 600 Particle diameter (nm)
0 40 80 120
Number concentration (#/cm3 )
Background Reentrainment
Figure 4.7 Particle reentrainment test in dry ESPs. The aerosol flow rate and applied voltage were 5 L/min and 4.3 kV.
100 200 300 500 50
30 20
Particle diameter (nm) 40
60 80 100
Collection efficiency (%)
Initial
After 30 min loading After 120 min loading
Figure 4.8 Collection efficiency for corn oil particles in the wet ESP at different TiO2
nanopowder loadings. The applied voltage and aerosol flow rate are 4.3 kV and 5 L/min, respectively.
4.2 Numerical results based on Lagrangian method for the nanoparticle collection efficiency of ESPs
4.2.1 Particle charging and particle trajectory in the ESP
In order to examine the applicability of the present model based on the Lagrangian method for predicting particle charging and particle trajectory, the numerical results were compared with the values in two benchmark problems. In the first case, number of charges acquired by particles by the diffusion, field and combined charging mechanisms were calculated, and then compared with the analytical solutions and numerical solutions in Hinds (1999). When the electric field strength is 5×105 V/m, ion number concentration is 1013 (m-3), and charging time is 1 s, the number of particle charges is calculated as shown in table 4.1. As shown in the table, the number of charges acquired by particles with various particle sizes by diffusion charging mechanism calculated by Lawless (1999) and Fuch’s (1943) equations in the present model match with numerical solutions in Hinds (1999) very well. The combined charges calculated by Lawless equation (1999) and Pauthenier and Moreau-Hanot equation (1932) in the present model match with the numerical solutions in Hinds (1999) as well.
In the second benchmark problem, the particle charging and trajectory in an ESP were calculated and compared with the numerical results in Goo and Lee (1997). In this ESP, the wire to wire spacing was 5.08×10-2 m, the wire to plate spacing was 2.5×10-2 m, the radius of corona wire was 0.05×10-3 m, the air velocity was 2.0 m/s, and the applied voltage was 12.78 kV. Figure 4.9 shows the particle trajectory in the ESP in Goo and Lee (1997). As can be seen, the present numerical results match with the simulated results in Goo and Lee (1997). Figure 4.10 shows the number of charges acquired by particles along the particle trajectory as shown in Figure 4.9. As shown in figure 4.10, the present numerical results also match with the simulated results in Goo and Lee (1997).
Table 4.1 Comparison of different models for particle charging by field, diffusion, and combined charging.
Number of elementary units of charged acquired
Diffusion charging Field charging Combined charging
Particle 40 3180 1630 1630 1630 259000 260439 259785 264000 261164 262674
0d 0.2d 0.4d 0.6d 0.8d 1d X direction
0d 2d 4d 6d 8d
Y direction
Figure 4.9 Comparison of the simulated particle trajectory between the present numerical results and those in Goo and Lee (1997) (Line: the present numerical results; open symbol:
the numerical results in Goo and Lee (1997)).
0 2 4 6 8 Y direction (d)
0 500 1000 1500 2000 2500
Particle charge (#)
Starting position (X) 0.44 d 0.33 d 0.25 d 0.15 d 0.04 d
Starting position (x) Goo and Lee (1997)
0.04 d 0.15 d 0.25 d 0.33 d 0.44 d
Figure 4.10 Comparison of the simulated particle charging between the present numerical results and those in Goo and Lee (1997) (Line: the present numerical results; open symbol:
the numerical results in Goo and Lee (1997)).
4.2.2 Comparing the particle collection efficiency in the dry ESP of Huang and Chen (2002) and Chang and Bai (1999)
Figure 4.11 shows comparison of the aerosol penetration between the present numerical values and the experimental data of Huang and Chen (2002). The charging model of Lawless (1996) was used to predict particle charges for particles with 0.006≦dp≦10 μm. As shown in Figure 4.11, when the applied voltage is -15.5 kV, the simulated results are in reasonable agreement with the experimental data with deviation of 2.75-16.07 % for particles with diameter of 40-100 nm. However, for particles with diameter below 40 nm, the present model based on Lagrangian method using combined charging model to predict particle charges underestimates the aerosol penetration with deviation larger than 20 % as compared with the experimental data. When the applied voltage is -18.0 kV, the deviation between simulated results and the experimental data are also larger than 20 % for particles with diameter below 70 nm. Besides, the partial charging effect on the aerosol penetration for the particles with dp
≦20 nm at both applied voltages can’t be predicted correctly. These results coincide with the theoretical results in Lawless (1996), who concluded that the limit of applicability of the combined charging model is 100 nm (Lawless 1996). Thus, the appropriate charging model, Fuchs model (1963) and the model of Marlow and Brock (1975), should be adopted to calculate charges of particles with dp≦100 nm. The simulated results based on these two models are shown in section 4.3.
When the applied voltage is -26.4 kV, the simulated aerosol penetration increases from 0.0 to 26.3 % with increasing particle diameter from 100 to 400 nm, and then decreases from 26.3 % to 0.0 % particle diameter from 400 to 1500 nm. The numerical results agreed reasonably with the experimental data with deviation of 0.54-14.57 % for particles with 100≦
dp≦1000 nm.
Figure 4.12 shows comparison of the particle collection efficiency between the present
numerical values and the experimental data of Chang and Bai (1999). In the ESP of Chang and Bai (1999), the wire to wire spacing was 300 mm, wire to plate spacing was 60 mm, and wire diameter was 0.5 mm. The aerosol flow rate and the applied voltage were 109 L/min and 27.0 kV, respectively. The numerical results are also in reasonable agreement with the experimental data in Chang and Bai (1999) with deviation of 1.49-12.46 %. These results could be attributed to the fact that Lawless’ model predicts particle charges accurately for particles in the continuum charging regime (Kn1) (Lawless 1997). Lawless concluded that when the electric field strength is very strong, diffusion charging can be neglected for particles with dp>1 μm; whereas diffusion charging should never be neglected for submicron particles (Marquard 2007). This is a very important conclusion for predicting particle charges using combined charging model in the continuum regime.
10 100 1000 10000 Particle diameter (nm)
0 20 40 60 80
Aerosol penetration (%)
Experimental data (Huang and Chen, 2002)
(-15.5 kV) (-18.0 kV) (-26.4 kV) Present model
(-15.5 kV)
(Lagrangian method) (-18.0 kV)
(Lagrangian method) (-26.4 kV)
(Lagrangian method)
Figure 4.11 Comparison of particle collection efficiency in the single-stage wire-in-plate dry ESP between numerical results and experimental data in Huang and Chen (2002) at the aerosol flow rate of 100 L/min.
100 1000 10000 Particle diameter (nm)
40 60 80 100
Collection efficiency
Experimental data (Chang and Bai 1999) Numerical values
Figure 4.12 Comparison of particle collection efficiency in the single-stage wire-in-plate dry ESP between numerical results and experimental data in Chang and Bai (1999) at an aerosol flow rate of 109 L/min and an applied voltage of 27.0 kV.
4.3 Numerical results based on Eulerian method for the nanoparticle collection efficiency of ESPs
4.3.1 Characteristics of the V-I curve, electric field and ion concentration distribution
Figure 4.13 shows the comparison of V-I curve between the theoretical values and the experimental data in the dry ESP of Huang and Chen (2002) and the wet ESP. Applying the ion mobility in the range of 1.35×10-4 ~1.90×10-4 m2/V/s for negative ions and 1.15×10-4~1.40×10-4 m2/V/s for positive ions, good agreement was obtained. The range of ion mobility was taken from previous researches and shown in Table 4.2 (Adachi et al. 1985;
Hoppel and Frick 1986; Hussin et al. 1983; Mohnen 1977; Wiedensohler et al. 1986; Wen et al. 1984; Wiedensohler and Fissan 1991). In the mobility range, the variation of the collection efficiency for particles with 2≦dp≦100 nm was found to be insignificant, which was calculated to decrease from 3.44~0.5 % to 5.70~0.0 % when the applied voltage was increased from -15.5 to -21.5 kV in the dry ESP of Huang and Chen (2002), and decreased from 3.7 to 0.0 % when the applied voltage was increased from +3.6 to +4.3 kV for 10 nm particles in the wet ESP. Thus, a fixed value of 1.35×10-4 m2/V/s for negative ions and 1.15×10-4 m2/V/s for positive ions was used to calculate the particle collection efficiency.
The present simulated electric potential and ion concentration distribution were also compared to the analytical solutions in a wire-in-tube ESP case (Marquard ea al. 2005), and numerical results were found to match with analytical solutions as shown in Figure 4.14.
Therefore, the present model is able to predict electric field strength and ion concentration distribution in ESPs. Figures 4.15 and 4.16 show the electric potential and the ion density distribution in the wire-in-plate wet ESP at an applied voltage of +3.7 kV and an air flow rate was of 5 L/min, respectively. The electric potential is shown to be symmetric with respect to
Therefore, the present model is able to predict electric field strength and ion concentration distribution in ESPs. Figures 4.15 and 4.16 show the electric potential and the ion density distribution in the wire-in-plate wet ESP at an applied voltage of +3.7 kV and an air flow rate was of 5 L/min, respectively. The electric potential is shown to be symmetric with respect to