Critical Review
Wet Electroscrubbers for State of the Art Gas Cleaning A N A T O L J A W O R E K , * ,† W A M A D E V A B A L A C H A N D R A N , ‡ A N D R Z E J K R U P A , † J A N U S Z K U L O N , ‡,§ A N D MARCIN LACKOWSKI† Institute of Fluid Flow Machinery, Polish Academy of Sciences, 80-952 Gdan ˜ sk, Fiszera 14, POLAND, Department of System Engineering, Brunel University, Uxbridge, MIDDX UB8 3PH, UK
Current trends observed in air pollution control technology are closely related to the development of new, more efficient hybrid systems, i.e., those, which simultaneously utilize two or more physical mechanisms for dust or gaseous contaminants removal. These systems can operate more economically than conventional devices, especially in the removal of PM2.5 particles. The electrostatic scrubber (electroscrubber), discussed in this paper, is one of such types of devices, which combines advantages of electrostatic precipitators and inertial wet scrubbers, and removes many shortcomings inherent to both of these systems operating independently. The electroscrubber is a device in which Coulomb attraction or repulsion forces between electrically charged scrubbing droplets (collector) and dust particles are utilized for the removal of particles from a gas. Unlike wet electrostatic precipitators in which particles are precipitated only on the collection electrode, in electroscrubbers, the collection of dust particles takes place in the entire precipitator chamber. Compared to inertial scrubbers, the electroscrubbers can operate at lower droplet velocities, but the collection efficiency for a single droplet can be larger than 1. The paper reviews the state-of-the-art of wet electrostatic scrubbing (electroscrubbing) technique used for gas cleaning from dust or smoke particles. Three groups of problems are discussed: (1) The fundamental problems concerning the charged dust particle deposition on a charged collector, usually a drop, with a focus on different models describing the process. (2) The experimental works of fundamental importance to our knowledge referring to the scrubbing process, which can be used for validating the theory. (3) The laboratory demonstrations and industrial tests of different constructions of electroscrubbers designed for effective gas cleaning. The electroscrubber is not designed to replace wet or dry electrostatic precipitators but can be used as a complementary device following the last stage of conventional electrostatic precipitator, which helps to remove submicron particles. It was shown in the paper that a higher collection efficiency of an electroscrubber could be obtained for higher values of Coulomb number (i.e., higher electric charges on the collector and particle), * Corresponding author phone: +48 586995151; fax: +48 583416144; e-mail
[email protected]. † Polish Academy of Sciences. ‡ Brunel University. § Present address: School of Electronics, University of Glamorgan, Pontypridd, Rhondda Cynon Taff, CF37 1DL, UK. 10.1021/es0605927 CCC: $33.50 Published on Web 09/19/2006
2006 American Chemical Society
and for a Stokes number lower than 5 (i.e., low particlecollector relative velocity).
1. Introduction Removal of particles smaller than a few micrometers from indoor or industrial gases presents a serious problem. Particles of this size, such as smoke, fine powders, or oil mist, which are usually hazardous to human health, are not easy to remove by conventional methods. Existing filters, cyclones, or inertial wet scrubbers, which employ inertial forces to remove particulate contaminants are ineffective in cleaning the gases from fine particles. This is because the motion of such particles is mainly governed by drag and molecular forces, and inertial force plays a diminishing role with decreasing particle size. Tighter fibrous filters can help in finer particles removal, but they operate at a high-pressure drop. Nozzle or Venturi scrubbers require liquid droplets of high velocity, but the pressure drop is also large in such devices. Water consumption in nozzle scrubbers is about 0.05 L/m3, and in Venturi scrubbers from 0.5 to 1.5 L/m3 (1). Dry electrostatic precipitators use electrostatic forces, but charging of particles smaller than 1 µm is inefficient, and the collection efficiency sharply drops with decreasing particle size. The re-entrainment of fine particles from the collection electrode can also be observed (2 (chapter 22), 3, 4). Irrigated electrostatic precipitators can only partially solve the problem of particle re-entrainment. In such precipitators, the dust particles are charged similarly to conventional electrostatic precipitators, but the collection electrodes are washed instead of rapped. Washing removes problems with back-corona discharge, but the issue of fine particle charging still remains unsolved. Therefore, an effective control of particles in the size range from 0.01 to 2 µm, known in the literature as “Greenfield gap” (4), is still a great challenge for engineers. To solve these problems, wet electrostatic scrubber which combines advantages of dry and irrigated electrostatic precipitators, and conventional inertial scrubbers, was proposed by Penney (5). In electrostatic scrubbers, dust particles and scrubbing droplets are electrically charged to the same or opposite polarities. The charged droplets capture the oppositely charged dust particles due to Coulomb attraction forces, or, when the charges are of the same polarity, the particles and droplets are repelled to the chamber walls. Hereinafter in this paper, the scrubber using electrostatic forces will be referred to as “electroscrubber” and the precipitation process as “electroscrubbing”. Since the wet electroscrubber is designed to remove fine particles, it cannot replace wet or dry electrostatic precipitators, but can be used as complementary device following the last stage of conventional electrostatic precipitators. VOL. 40, NO. 20, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6197
FIGURE 2. Coordinate frame for the droplet-particle system.
FIGURE 1. Vertical type of electroscrubber. Three main designs of electroscrubbers have been presented in the literature: (1) Vertical scrubbers in which dusty gas flows upward or downward, and charged droplets are sprayed at the upper part of the chamber. (2) Horizontal scrubbers in which the gas flows horizontally, and charged spray is injected cross-flowing to the gas stream. (3) Venturi scrubbers, positioned vertically or horizontally, with the charged spray injected into the throat of Venturi nozzle. An example of a vertical scrubber with gas flowing upward is shown in Figure 1. Five electroscrubber systems based on various combinations of charging the dust particles and droplets have been reported in the literature: (1) Droplets and particles are charged oppositely (5, 7-9). The particles are deposited onto the droplets due to the Coulomb force. (2) Droplets and particles are charged to the same polarity (10-12). The repulsive force of the space charge of the droplet cloud drives the particles onto the chamber walls where they are washed out. (3) Nearly equal number of positively and negatively charged droplets are sprayed (13). The positively charged droplets capture negatively charged particles due to attractive Coulomb force, and finally, they coagulate with negatively charged droplets. (4) Particles are charged to either polarity and the droplets are uncharged (13). The particles are deposited onto the droplets due to an image charge on the droplet. (5) Droplets are charged to either polarity while the particles remain uncharged (14-18). The particles are deposited onto the droplets due to an image charge induced on the particle. Although highly efficient inertial and turbulent scrubbers are now available, the problem with small particles ( 16, the vorticity becomes negative over a part of the rear side that indicates liquid circulation downstream. This critical Reynolds number differs from those presented by other authors, which assume Re ) 20 (26). The stream function proposed by Tomotika and Aoi also predicts stationary vortices behind a sphere, however, this formula is not accurate for small Reynolds numbers because it generates vortices for Re f 0. In order to determine the stream function, Keh and Yu (40) have solved the fourth order differential equation of the fluid motion around a falling sphere and determined the stream function for Re < 0.1. However, this formula is useless for electroscrubbing simulation where the Reynolds numbers are larger. For the Reynolds numbers larger than Re > 1000, the viscose force becomes negligible compared with inertial force, and the equations of motion can be linearized. This type of flow is known as potential flow. For this case, Lamb (41) has derived an approximation (cf., Table 3, Supporting Information), which describes the flow field in front of a sphere with sufficient accuracy, but it does not predict vortices downstream. The particle deposition at the rear-side of the collector cannot, therefore, be predicted using this formula. Viswanathan (42), concerning particle deposition on a single droplet concluded that the model of potential flow can be used for Reynolds numbers Re > 80. Some of the expressions from Table 3(Supporting Information) were tested by Jaworek et al. (7), with regard to scrubbing process modeling. The authors concluded that for small Reynolds numbers, Oseen, Lamb, Tomotika and Aoi, and Proudman and Pearson approximations give similar results, but with increasing Reynolds number, two formulas: Tomotika and Aoi, and Proudman and Pearson become inadequate. 6200
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 20, 2006
Other effects, like drop deformation, turbulence, and swarm effects on the flow field about the sphere were considered by many authors (43-49, 2 (Chapter 24)). However, these effects have never been incorporated into electroscrubbing models. From these considerations results that drop deforms for larger Reynolds numbers, and cannot be further considered as a rigid sphere. Beard and Pruppacher (50, 51) proposed Re ) 200 as critical Reynolds number of droplet deformation. Ingebo (44), and Petrak (45) measured the drag coefficient for a solid sphere in turbulent flow, and determined that with increasing Reynolds number, the measured drag coefficient can be an order of magnitude below the standard curve. From the curves presented by Clift et al. (ref 26, p 27), results that a droplet smaller than 1 mm (Eo¨tvo¨s number ) 0.135) can be regarded as rigid sphere for Reynolds number Re < 1000. The droplet can be distorted due to its motion in a flowing gas (52), and due to an electric field. Hsiang and Faeth (53) determined experimentally that the deformation is less than 10% when the Weber number is smaller than 1. Warnica et al. (54) have analyzed photographs of droplets, taken in a turbulent flow, and concluded that the droplet radii deviates by less than 10% from the sphere for Reynolds number 10 < Re < 100, and Weber number 0.007 < We < 0.075. It should be noticed that these critical values are higher than those met in electroscrubbers (except Venturi scrubbers), and the models based on the motion of rigid sphere are, therefore, fully justified. 2.2. Collection Efficiency. Collection efficiency is a parameter used for assessment the ability of a cleaning device, regardless of its type, to remove particulate matter from the gas. In the case of electroscrubbers, the term “collection efficiency” is used in a double sense. In the first sense, it means an “overall collection efficiency”, which is the ratio of mass of particles min - mout removed by the scrubber to the mass of particles at the scrubber inlet min:
K)1-
mout min
(10)
The second concept of the collection efficiency characterizes the particle deposition on a single collector. In this sense, the collection efficiency is the ratio of total number Nd of monodisperse particles deposited on the collector to the number of particles Ns flowing through the projected area of the collector:
K)
Nd Ns
(11)
This parameter is sometimes called a “collision efficiency”, however, in this paper, following the tradition established by Kraemer and Johnstone (6) regarding electroscrubbers, the term “collection efficiency” will be used in the second sense. For a fixed spherical collector, the collection efficiency (11) is usually determined from the equations of motion of the particle as the ratio of the surface Si perpendicular to the gas velocity vector, crossed by the particles which are deposited on the collector, to the cross-section Sc of the collector.
K)
Si Sc
(12)
This definition is, however, useless when the collector is in motion, and the surface Si changes with time and place. In this case, the number of particles deposited on the collector cannot be determined unambiguously. Another method of
calculation of the collection efficiency, which can be applied to a moving as well as fixed collector, was proposed by Jaworek and co-workers (55, 56). The collection efficiency was determined as the ratio of total volume Vd from which the particles were deposited on a moving collector to the geometrical volume Vs swept by the collector:
K)
Vd Vs
(13)
This definition differs from that proposed by Kraemer and Johnstone, and is insensitive to the starting point of the trajectory, and relative particle-collector velocity. For monodisperse particles, eq 13 can be interpreted as a number Nd of particles present in the volume Vd, to the number Ns of particles present in the volume swept by the collector. Numerical calculations of the collection efficiency were based on particle trajectories determined from eq 1. Theoretical models of particle trajectories deposited on a collector used for numerical calculations of the collection efficiency are reviewed in Table 4 in the Supporting Information. The results of modeling are usually shown as a function of Stokes, Coulomb, and Reynolds numbers based on the collector diameter. The Stokes number represents an inertial deposition of a particle on the collector. In the considerations of inertial or electroscrubbers, the Stokes number is usually used in the form resulting from differential eq 1:
St )
K ) 4Kc
2Cc R p2Fpu 9ηg Rc
Because this equation was valid only for small Stokes numbers, Nielsen and Hill (57, 58) proposed a formula to which the Stokes number was incorporated:
K ) (2(-Kc)1/2 - 0.8St)2
CcQpQc 2
(16)
(14)
The Coulomb number represents the electric forces between the particle and collector, related to the Stokes drag force:
Kc )
tices produced downstream, are almost identical to those recorded experimentally for charged and uncharged objects (66). For larger Reynolds numbers, the solution of the equations became unstable, and the potential flow was assumed. The method used by Viswanathan (67) allowed determining the flow field for Reynolds number up to 400, but these results were obtained for uncharged droplets and particles. Viswanathan (67) also showed that up to 50% of particles can be deposited from the rear side of the collector. For very small Stokes numbers, the numerical solution of the equations of motion also encounters difficulties. A small particle flowing along the stagnation line can never collide with the collector because the convergence of the equations is very slow. Michael (68), and Michael and Norey (69) determined the critical Stokes number Stcrit ) 1/12 for which the inertial deposition diminishes, and all particles are flowing along the streamlines. Only interception and electrostatic forces can cause a particle to be deposited on the collector. Numerical determination of the collection efficiency is time-consuming, and for rough estimation of this parameter, some approximate formulas were, therefore, proposed in the publications. For an uncharged collector and particle, such an approximation was determined by Langmuir and Blodgett (70), Walton and Woolcock (71), and Viswanathan (42). For highly charged particle and droplet, the collection efficiency was estimated by Kraemer and Johnstone (6) from the Coulomb number Kc:
2
24π 0 ηgu0 Rc Rp
(17)
(15)
Kraemer and Johnstone (6) determined the collection efficiency, taking into account the Coulomb, image, and Stokes forces as well as the space charge effects. Nielsen and Hill (51, 58) additionally considered an effect of external electric field and dipole moment. Beizaie and Tien (59) concluded that the gravity is dominant for the particles co-flowing with gravity vector. Wang et al. (60, 61) considered the problem of particle deposition on a collector falling in a flowing gas, but their theoretical results were restricted only to two dimensions. Additional forces due to external electric field produced by a pair of electrodes were considered by Sumiyoshitani (62), Wang (63), and Shapiro and Laufer (64). From these results can be concluded that an external electric field can increase the collection efficiency when either collector or particle is only weakly charged or uncharged. In this case, the electric field produces polarization force driving the particle to the collector. In opposite situation, the field can cause a decrease in the collection efficiency. More exact calculations of the collection efficiency in wide Reynolds number range were obtained by solving the NavierStokes equation. Dau (31), and Schmidt and Lo¨ffler (32, 65) concluded from the results obtained for a fixed collector that the rear-surface deposition takes place only for a turbulent flow. For laminar flow, even for large Reynolds numbers (≈1000), the vortices generated downstream have only negligible effect on particle deposition on the collector. Navier-Stokes equations were solved by Adamiak et al. (33), and Jaworek et al. (18, 66) who determined the particle trajectories near a spherical collector in three dimensions. The results were obtained for the Reynolds number based on the collector diameter smaller than 130. It was shown that the trajectories obtained numerically, including vor-
However, the collection efficiency determined from eq 17 becomes negative for St > 2.5 and Kc ) 1. Equation 1 used to determine the collection efficiency predicts the particle trajectories in good agreement with the observations, but only for particles larger than about 1 µm. For smaller particles, additional forces at the molecular level have to be considered. Wang (63) included diffusion and thermodiffusion to the model of charged particle motion and concluded that these effects are important for the collection of particles smaller than 0.1 µm. Thermophoresis, i.e., the motion of particles due to the temperature gradient, is another mechanism operating in submicron size range. When cold droplets are sprayed into a usually warm exhaust gas, the temperature gradient near the droplet drives the particles toward its surface (72). For very small particles, for which the Knudsen number, Kn, . 1, the force was determined by Waldman and Schmitt (73), and for Kn ≈ 1 by Epstein (cf., ref 74). Viswanathan (67) has shown that for small Reynolds numbers (Rec ) 1.54), the thermophoretic collection efficiency can be higher than 10, for the particles in the size range 0.1-10 µm, the gas temperature 95 °C, and droplet temperature 10 °C. Results of numerical calculations of the collection efficiency due to diffusiophoresis and thermophoresis were presented by Deshler (75). These forces are important only for small particles (300 µm), or large particles (>5 µm) (7, 33, 74, 75, 80). Vortices downstream of the collector, occurring for large Reynolds numbers can also participate in the particle deposition (31, 81). The electrostatic deposition starts to play its role for Stokes numbers smaller than 5, and it only weakly depends on the Reynolds number based on the collector diameter (7, 33, 66). For further refinement of numerical models, the particle swarm effects, particle shape, droplet distortion, and droplet contamination should be considered.
3. Fundamental Experiments Two groups of experiments regarding electroscrubbing can be distinguished: those, which are aimed at learning about fundamental processes taking part in the particle-collector interaction, and those which investigate the bulk effects in gas cleaning devices, including their performance. This section summarizes the first group of experiments. The second group is discussed in Section 4. In the group of fundamental experiments, the collection efficiency has been measured for a single collector (a droplet), fixed or falling. These experiments are summarized in Table 5 of the Supporting Information. The results are usually presented in dimensionless form with Stokes, Reynolds, and Coulomb numbers as variables. For uncharged particle and collector, a comparison of various experimental and computational results on the collection efficiency was carried out by Jaworek and Smigielski (82). Figure 3 presents similar results with some additional data. Analogous comparison 6202
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 20, 2006
for charged particle and collector were not hitherto published and are shown in Figure 4. Theoretical curves obtained by Jaworek et al. (7) for Kc ) 0, and Adamiak et al. (33) for Kc ) 1 are also drawn in this plot. It should be emphasized that the experiments were carried out in a limited range of Stokes number, with the Reynolds number kept constant. This is caused by the fact that both of these numbers cannot be changed independently because of the following relation:
()
Rec Fg Rc )9 St Fp R p
2
(18)
The ratio (18) is independent of gas velocity and gas viscosity, and depends only on the particle/gas density and crosssection ratios. This equation determines the relation between the Reynolds and Stokes numbers for given experimental conditions. It is, therefore, not possible to change the Stokes number by variation of gas velocity, with the Reynolds number kept constant. The only practical way to change the Stokes number is to vary the particle or collector sizes, or the particle density, which however, cannot assume unrealistic values to cover the entire St number range. The collection efficiency is, therefore, usually determined experimentally for St < 5. Four theoretical models for uncharged droplets are shown in Figure 3: Langmuir and Blodget (70), Le Dinh (83), Schmidt and Lo¨ffler (84), and Jaworek et al. (7). The discrepancies between theoretical curves and measurements are large, even for the same Reynolds numbers. It can be noticed that the experimental points are usually above the theoretical curves, independently of the model. The collection efficiency for large Reynolds numbers (>1000) is above all theoretical curves. This suggests that the potential flow used for determination of the collection efficiency is not adequate for large Reynolds numbers, and underestimates its value. The measurements obtained by Dau and Ebert (81) did not follow this rule, and are below the theoretical curve. Only the results obtained by Star and Mason (85), and Walton and Woolcock (71), for Rec < 1000 fall within the theoretical curves. The effect of electrical forces on the collection efficiency is shown in Figure 4. The approximate equations, proposed by Kraemer and Johnstone (16), and Nielsen and Hill (17) are also drawn in Figure 4. These formulas, however, give results far from exact numerical simulations and experimental results. The results shown in Figure 4 indicate that measure-
FIGURE 4. Collection efficiency for a single, charged spherical collector and charged particles. ments of the collection efficiency are much more scattered for a charged collector than for uncharged one. Two trends can, however, be inferred from this plot. The first one is of increasing collection efficiency with decreasing Stokes number and increasing Coulomb number. The second one is a diminishing effect of electrostatic forces with increasing Stokes number. It should be noticed that the experimental results are not available for the Stokes number larger than 2. The collection efficiency measured by Schmidt (86) and Wang (60, 61) was for small Coulomb numbers (
