Experimental study of electrostatic precipitatorperformance and comparison with existingtheoretical prediction modelsS.H. Kim, K.W. Lee*Kwangju Institute of Science and Technology, Department of Environmental Science and Engineering,1 Oryong-dong, Puk-gu, Kwangju 500-712, South KoreaReceived 1 February 1999; received in revised form 21 May 1999; accepted 2 June 1999AbstractA laboratory-scale single-stage electrostatic precipitator (ESP) was designed, built andoperated in a wind tunnel. As a "rst step, a series of experiments were conducted to seek theoperating conditions for increasing the particle collection e$ciency by varying basic operatingparameters including the wire-to-plate spacing, the wire radius, the air velocity, the turbulenceintensity and the applied voltage. As the diameter of the discharging wires and the wire-toplatespacing are set smaller, the higher collection e$ciency has been obtained. In thesingle-stage multiwire ESP, there exists an optimum wire-to-wire spacing which providesmaximum particle collection e$ciency. As the air velocity increases, the particle collectione$ciency decreases. The turbulent #ow is found to play an important role in the relatively lowelectric "eld region. In the high electric "eld region, however, particles can be deposited on thecollection plates readily regardless of the turbulence intensity. The experimental results werecompared with existing theories and Zhibin and Guoquan (Aerosol Sci. Technol. 20 (1994)169}176) was identi"ed to be the best model for predicting the ESP performance. As the secondstep, the in#uence of particle contamination at the discharging electrode and at the collectionplates were experimentally measured. The methods were sought for keeping the high collectione$ciency of ESP over elapsed time by varying the magnitude of rapping acceleration, the timeinterval between raps, the types of rapping system (hammer/vibrator) and the particle reentrainment.The rapping e$ciency and the particle re-entrainment were increased withincreasing magnitude of rapping acceleration and time interval between raps. However, whenthe thickness of deposited #y ash layer is su$ciently high, the concentration of re-entrainedparticles starts decreasing abruptly due to the agglomeration force which can interact among0304-3886/99/$ - see front matter ( 1999 Elsevier Science B.V. All rights reserved.PII: S 0 3 0 4 - 3 8 8 6 ( 9 9 ) 0 0 0 4 4 - 3deposited particles. The combined rapping system is found more e!ective for removingdeposited particles than the hammer rapping system only. ( 1999 Elsevier Science B.V. Allrights reserved.Keywords: Electrostatic precipitation; Turbulent #ow; Rapping; Particle re-entrainment; Collection e$-ciency; Negative corona1. IntroductionElectrostatic precipitators (ESPs) are one of the most commonly employedparticulate control devices for collecting #y ash emissions from boilers, incineratorsand from many other industrial processes. They can operate in a wide range ofgas temperatures achieving high particle collection e$ciency compared with mechanicaldevices such as cyclones and bag "lters. The electrostatic precipitation processinvolves several complicated and interrelated physical mechanisms: creationof a non-uniform electric "eld and ionic current in a corona discharge, ionicand electronic charging of particles moving in combined electro- and hydrodynamic"elds, and turbulent transport of charged particles to a collectionsurface.Generally, the collection e$ciency of ESP decreases as the discharging electrodeand collection plates are contaminated with particulates. Thus, a rapping system isneeded for removing the collected particulates periodically. While there have beennumerous theoretical and experimental studies on particle collection characteristics ofelectrostatic precipitators, a relatively small number of the studies addressed thee!ects of particle accumulation both at the discharging electrodes and at the collectionplates. Both phenomena are known to in#uence adversely the performance ofelectrostatic precipitators. Many researchers, such as Deutsch [1], Cooperman [2],Leonard et al. [3], Khim et al. [4], Zhibin and Guoquan [5], and Kallio and Stock[6], conducted particle collection measurements of ESP. However, they concentratedmostly on the e!ects of both turbulent mixing and secondary wind in multiwiresingle-stage electrostatic precipitators. Speci"cally, Cooperman [2] considered reentrainmentand longitudinal turbulent mixing e!ects, Leonard et al. [3] the "nitedi!usivity, and Zhibin and Guoquan [7] the non-uniform air velocity pro"le. Amongthem, only Zhibin and Guoquan [7] measured the collection e$ciency of a singlestageESP covering a wide particle size range. Even though their experimental dataare considered to be practical and useful, their experimental conditions were notidenti"ed clearly.In the present study, well-de"ned collection e$ciency data for an ESP are presentedcovering the particle size range of 0.1}100 lm. The particles used in the present studycame from the Bo-Ryung power plant in Korea. In addition, the ESP performancewas evaluated in terms of optimum operating conditions. Finally, the optimumrapping conditions were sought under which the rapping e$ciency increases and theparticle re-entrainment decreases.4 S.H. Kim, K.W. Lee / Journal of Electrostatics 48 (1999) 3}25Fig. 1. Schematic diagram of the wind tunnel for the eight wired single-stage ESP performance test.2. Review of theoretical models2.1. Particle chargingFig. 1 shows the laboratory-scale electrostatic precipitator. The particle chargingsystem consists of discharge wires with diameter (D8) and two grounded parallelplates of length (¸). A high negative voltage (<8) is applied to the corona dischargewires, and suspended particles of diameter (d1) #ow with air between the plates ata velocity (;) in the y-direction. In the whole range of particle sizes, both "eldcharging and di!usion charging mechanisms contribute to signi"cant charges [8,9].In these theoretical analyses, it is nearly correct to sum the rates of charging from thetwo mechanisms and then solve for the particle charging as follows:dq1dt"q4q A1!qq4B2#d21eN04 S8k¹pmexpA! 2qed1k¹B (1)where q1 is the particle charge, q4 is the saturation charge,N0 is the average number ofmolecules per unit volume, e is the electronic charge ("1.6]10~19 C), b is the ionmobility ("1.4]10~4 m2/V s), e0 is the permittivity of free space ("8.85]10~12 F/m), d1 is the diameter of particle, k is the Boltzmann constant ("1.38]10~23 J/K), ¹ is the absolute temperature ("293 K), m is the mass of a particle("(p/6)d31o1), and o1 is the particle density ("2.25]103 kg/m3).2.2. Theoretical models of particle collection ezciencyTheoretical models of ESPs were provided by Deutsch [1], Cooperman [2],Leonard et al. [3], Zhibin and Guoquan [7] and others. The Deutsch model forS.H. Kim, K.W. Lee / Journal of Electrostatics 48 (1999) 3}25 5calculating the particle collection in an ESP assumes complete mixing by turbulent#ow and thereby uniform concentration pro"les. In order to improve the drasticassumption of in"nite di!usivity in the Deutsch model, many researchers tried todevelop "nite di!usivity models by dealing with the convective-di!usion equationwith various boundary conditions.Cooperman [2] developed a theory which modi"es the Deutsch model to accountfor the e!ects of turbulence and particle turbulent di!usion. The major limitations ofthe Cooperman model lie absence of a general method to estimate the re-entrainmentfactor and the particle di!usivity. Leonard et al. [3] developed a more complicatedtwo-dimensional model using the method of the separation of variables from theconvective-di!usion equation. He assumed uniformity of velocity components ofcharged particles and particle di!usivity. This assumption fails to adequately describethe particle di!usivity near the collection plates, where it is governed mainly by themolecular transport and, therefore, the di!usivity near the wall is signi"cantly lowerthan the di!usivity in the turbulent core. Zhibin and Guoquan [7] suggested a newmodel for the single-stage ESP which takes into account the e!ect of turbulencemixing by electric wind. Predicted collection e$ciencies of the above theoreticalmodels are summarized as follows:gDe"1!exp(!De), (2)gCoo"1!expC;¸2D!SG A;¸2DB2#(1!R)PeA¸=B2HD, (3)gLeo"1!P10PA m!DeJ2De/PeBdm, (4)gZhi"1!S Pe4pDeP10expC!Pe4De(m!De)2Ddm, (5)where