Physicochemical interaction and its influence on deep bed filtration process

Joumal of Ensironmental Sciences Vol. 16,No.2.pp.297- :301 .2004Article ID: 1001 0742( 2004)02-0297-05CLC number: X701 Document code: APhysicochemical interaction and its influence on deep bed filtration processGUO Jin-long"，MENC Jun, LI GUI-ping, LUAN Zhao- kun, TANG Hong xiao(Stale Key Laboratory of Environmental Aquatie Chermisty, Research Center for Eco- Environnental Sciences, Chinese Acadeny of Sciences, Bijing100085，China. E- mail: jlguo15 @ hotmail. com)Abstract: The capillary model was used to analyze the hydraulic conditions in the deep bed fitration process. Thephysicochemical interaction forces between the filter media and suspended particles and their influence on deep bedfitration process, were also, studied theretically. Through the comparison of the .hydraulic and physicochemicalforces, the key infuencing factors on the fitration process were proposed and investigated. Pilot study of the micro-flocculation deep bed fitration was carried out in' the. No. 9 Potable Water Treatment Plant of Bejing, and theexperimental results of hydraulic head loss，particle distribution and entrapment were presented. The theoreticalprediction was reasonably consistent with the experimental results under different conditions , which indicated that theregulation and control of micro-flocculation and deep bed filtration could berealized by the evaluation of thephysicochemical interactions. Further theoretical and experimental research should be carried out to investigate theinteinteraction mechanism and its application in the deep bed fitration and other cases.Keywords: capillary; micro-locculation; fitration; physicochemical; interactionIntroductionstructura( or hydration) force between the particle and thefilter grain and their infuence on the filration process anFiltration process in the production of potable water isanalyzed theoretically. The theoretical results are used toprimarily aimed to remove suspended particles from waler ，explain the experimental data, and good prediction could beand it is the last unit process that will directly affect the tapined."waler quality ( Pontius, 1990). Deep bed filtration, which is1、Theoretical analysisan efficient and cost effective method for the treatment of lowturbidity and low temperature water, has been widely studied1.1. Capillary model and hydraulic slopeand used( Ives, 1980; Shandalov， 1997; Yao， 1971; Bai,When water transfers through the porous media a2000) .lamninar flow in the packed bed filration process， theDeep bed. fitration involves a variey of complexsuspended paricles in it will interact with the filter grain .mechanisms in the paricle remnoval process, and the surfaceAnd the particle will be removed if the total interaction forceinteractions belween su8pended particles and filter grain haveis strong enough to bond the particle to the grain surface. Atbeen thoroughly investigated by many researchers( Payalakes，the beginning of filtration operation, the particles will adhere1974; Huang, 1999; Jegatheesan,'2000; 0' Melia, 1997;to the high- atraction positions in the packed bed. The forcesRajagopalan,1976; Srnivasa， 1998). The particles to beon the filer pore surface will be about equal when stableremoved are generally much smaller than the pores in theoperation stage is arrived, then the pores in the fiter bed canpacked bed, so the physicochemical interaction' taken placebe considered a8 circular shaped capillaries ( Cuo，2002).in the filer will predominantly inluence the particle removal.The following equations can be formulated through comparisonThe shear force caused by_ water flow will also affect theof capillary hypothesis with actual filtration system ( Jing,particle adsorption and desorption to the filter grain2000):(Warszynski, 2000; Guo, 2002). Because of its influence(1)on the surface potential and size distribution of the suspended(2)lcles,，the chemicalpretreatment is essential to deebhedwhere δ is the filter bed porosily, f is the specific surfacefiltration. The infuence of flocculent on particle aggregationarea of the filter bed, m"，n is the number of capillariesand its attachment to the filter was also investigatedextensively to improve the filtration efficiency( Berre，1998 ;per unit surface area of the filter bed， m~, d. is theHabibian, 1975; Leprince, 1984; 0' Melia， 1967; 1985;capillary diameter, m. For non-spherical filter grain， theTang, 1987). Filtration is a promising unit process deservespeeific surface area can be expressed as(Jing, 2000):further study， and there are still plenty of room fors = 6a(1 - 8)1d,(3)improvement in filtration and its relative areas .where a is the surface shape factor and d, is the equivalentOn the basis of uniform medial filration,the turbiditydiameter of a filter grain, nremoval and head loss in filtrationprocess was analyzed bySubstituting Equation (1) and (2) into (3)，thecapillary model thereically ( Jing, 2000). And in thisfollowing results can be obtained:paper, the capilary model is combined with thephysicochemical forces between the suspended particles and3a(1 - 8)d(4)filter. grain to. analyze the micro-flocculation deep be9a'(1 - 8)2filration. The shear force distibution and hydraulic slope in=πδ●d(5)the capillary is studied fistly. Then, the physicochemical中国煤化I aillary must be equal toforces including London- van der Waals altraction force ,Q1Ag，where 区is theelectric double- layer force， Bom repulsion force andMYHCNMH Gn/s. Q is the apparentFoundation iem: The National Natural Scienres Foundation of China(No. 50178009; 2977027) and the Special Funds for the Scentifie and Social Practice of CAS (forInnovative Research); * Corresponding author98GU0 Jinlong et al.Vol. 16filtration velocity, m/s， Ag is the pore area per unit area ofand suspended particles (m), z is the distance between afilter medium A; = π(d.12)'n = δ. As mentioned above.particle and the capillary wall (m),A. is the characteristicthe flow in the capillary is considered as laminar flow. andwavelength of the interaction (m), defined as 2πc/w,,c isthe velocity distribution of laminar flow in a circular shapedthe light velocity, 2.9979 x 10'm/s, w。is the dispersioncapillary can be expressed as u= rJ(r- r)14μ, where rofrequency, for most materials λ。takes a value of 100 nmis the capillary radius. The average velocity is ( Wen,(Jegatheesan, 200),k is the inverse Debye length (m ')，1992) :which can be approximated as ( Stumm，1981): KVI nm，I is the ionic strength， mole/L. kp is thewhere r is the distance from an arbitrary point to the capillaryBolzmann's constanl， 1.3807 x 10-23 J/K， T is thecenter, m，γ is the fluid density, kN/m', and μ is thesuspension temperature, K，Z is the charge number of thedynamic viscosity of the fluid. J is the hydraulic slope: J =eletrolyte used，e60hf/l，where h, is the hydraulic head loss， m，l is the10~"C, ψ and ψ2 are the surface potential of particles andepilary lengh, m.fiter grain respectively, 01 is the collision diamneter (m), K;Through the analysis conducted above, the theoreticaland h are empirical constants， E is the suspensionhydraulic slope during unifom media filtration could beachieved( Wen, 1992)permitivity (E = ε,"∈o), Eo is the pemittivity of the vacwumJ = 32. up/rd.(7)and e, is the dielectric conslant of the suspension. A. is theFor the uniform flow in pressurized circular capillany,Hamaker constant when a particle (p), and a filter grain (c)the relation belween shear force τ and hydraulic loss can beare separaled by water ( w) ( Nm'1s2 ). Assuming thegiven as: τ= γ'r. J12, and the average hydraulice head losssuspended particles are mainly composed of silica, then theof laminar flow in the capillary can be written as: hj= J xHamaker constant between the particle and the capillary wallI. Thus the shear force distribution in the circular shapedcapillary can be obtained from Equation (7):Hamaker constant of the water and quartz are5x 10~20 and 1，=16.μ.i●r1d'(8)x 10~”J, respectively ( Stumm， 1981; Zhou， 1996)， soIt can be concluded from Equation (8) that the shealAm=(A?- AH,,)=8.58x 10~"(J).force in the capillary center equals zero, and it is in directUnder the condition of 20° centigrade， the interactionproportion to the distance belween the capillary center andforces between a paricle and the capillary wall in thecalculation position. The maximum value 8puild, is reachedunivalent electrolyte solution can be simplified as follows :on the inner wall of capillary . Supposing that the separationF =1.43 x 10~ap(1 + 0.28z)●distance between the capillary wall and particle is z，then. r(14)= d,12- z，and the fluid shear force in position Z can beF. =-0.334 I'2 a, x 0.74 rP.sxx tanh(9.756ψ )calculated by the following equation:(9)tanh(9.75642),(15)1.2 Physicochemical forces between the particle and(16)capillary wallF. =- 6.28 ap Kh x exp(- z/h).(17)Duringfiltrationprocess ,influencing2 Experimental studyphysicochemical forces on the particle attachment onto thewall of filter bed pores can be divided into two groups2.1 Experimental methodsBoupsaccording to their function scope ( Ives,.1980; Jegatheesan，Pilot study of micro- focculation deep bed filration was2000). The Bor repulsion force( Fand structural (orcarried oul in the No. 9 Potable W aler Trealment Plant ofhydration) force ( F) are lermed as short- range forces due toBeiing, the raw water of which was from Miyun Reservoir.their infuence on particles being dominant only if theThe schematic flow chart for the micro-locculation deep bedparticles are less than 5 nm away from the interactionfiltration is shown in Fig. Isurface. However, the influence of London-van der W aalsaltraction force( F.) and electric double-layer force( F. )( either atractive or repulsive) exist even when the distanceTurbidimeterbelween the particle and interaction surface is about 100 nm，so they are called long- range forces. During filtrationprocess, these forces will affect the particle adsorption anddesorption on the filter pore suface。The equations govemingRecorderthese forces are given below :Relarded London-van der Waals force( Gregory， 1981;FoculantJegatheesan,200):F. =(Appa,/6)[1 + 28(z/2,)]11z*[1 + 14(z/2.)]"1. (10)Elctical double-layer force ( Gregory, 1975; Jegatheesan,2000):F. =- 64rap εK[k。T/Ze]'tanh[ Zeψ,14hgT]tanh[ Zeψ2/4kp T]exp(- kz) .I EffluentBorm force ( Ruckenstien，1976; Jegatheesan, 2000):Raw watrBackwsh waterFg =- A.papσ{(180z").(12)中国煤化工oculaton dep bed ftratonHydration force ( Israelachvili, 1992; Jegatheesan, 2000)(13)JYHC N M H Gap( Model 7523-37, Cole-where Fg,F.,F, F are the Born force . electrical double-Parmer Instrument Cor.)，the coagulant was mixed with rawlayer force, hydration force and London-van der Waals forcewaler in the static in-line mixer( Model I -40C-4-12-2, Koflorespectively (Nm/s* ),a.,ap are the radi of the fiter grainCor.)，and the coagulated water flowed into the packed filterNo.2Physicochemical interaction and its infuence on deep bed filtration process299column hrough the pipelines. The efluent turbidity wasmonitored by onlinelurbidimeter( Model 8220, Greatat. Lal| ake8(CoLdl)Co. Ld),. and the results were sent to the recorder ( Model056- 3002, Hitachi Cor. ) for analysis and printing.2.2 MaterialsThe filter column; which was made of plexiglass, was5000 mm high with the_ inner diameter of 200 mm. The←0-70cminterval between the sample ports was 300 mm, and 200 mm急20十0-140cmheight stone was used as support for the packed filter bed米0-250cmThe equivalent diameter of the filter media, which was madeforn anthracite, was 2.7 mm. The height of the filer bed2-3 3-4 4-55-6 6-10was 2500 mm and the experiment was carried out under theParticle diameter, μmfilration velocity of 6.7 mm/s. The particle size distributionFig. 3 Particle removel eficiencey along the filer bed with 0.5 mg/L PACin the raw water, coagulated water and efluent was measureddosage(Multisizer II ，Couler Elctronics) and analyzed. The sizedistribution of the suspended particles in the source andfiltered water is given in Table80-二李=Table L Partice size distribution in the source and flered walerFiltered water50上Pariclr diumeler ,Source water.10。0- 140 cmm0.5 mg /L0.5 mg/L-米- 0-250 cm0t88.2687.1492.382-37.578.305.142. 162.371.2Particle diameter, pm1.550.86-100.440.550.2dosage .Particle removel eficieicg along the fiter bed with 0.5 mg/L PFC10-200.060.090.03The coagulants used in the experiment were liquidcomparison of Fig. 3 and Fig. 4. For PFC flocculent, thepolyalumium chloride ( Qingdao, China， hereinafter referredparticle removal occurred mostly in the upper part of the filterto as PAC) and polyferrie chlonide( made from liquid wastescolumn， while the entrapped particles' rather evenlyof pickling industry, hereinafter referred lo as PFC). Thedistributed along the whole filter bed for PAC flocculent .Fe'* concentration in the PFC was 2.23 mol/L while theHowever, the head loss of PFC was lower than that of theAl2O; content in the PAC was 15.6 %. The alkalinities, i.PAC during the filtration operalion, which may be resultede. _the concentration ratio of OH~ to metal ion, were 1 4 andfrom the dehydration of the PFC flocs. The relationship0.31 for the PAC and PFC, respectively.between the head loss and filtration time is ilustrated asrig. 5.Since the raw. water turbidity was low， the lowerestablished limit for the head loss was set as 2000 mm, while00厂that for the efluent turbidity was 0.3 NTU.30 -女PFC.smg/- + PAC.0.5 mg/L3 Results and discussion200 F3.1 Experimental resultsIt can be scen from Fig. 2 that the particle removal00 Fefficiency was in negative correlation with its diameter whenno flocculent was dosed， i. e. the removal effciency wasdecrease with the increase of particle size along the wholefilter bed. However ，different phenomena could be observedi!5253545ss6when locculent was added. The infuences of PAC and PFCOperntion time, hon particle removal are shown in Fig. 3 and Fig. 4, fromwhich it can be concluded that the particle removal efficiencyFig. 5 Influence of operation time on the head loss of fiter bedincreases with the increase of is size.3.2 Theretical results￥The ion strength in the theoretical calculation is 0.01and the values for constant k and h are 0.00001 and 0.8”*respectively. The temperature used in the caleulation is20C. The shape factor of the filter grain is 2.08, while theother conditions are the same as those of the filter bed in the去0-140mexperimental part, i. e.2500 mm packed anthracite grain十0-210 cm .whose equivalent diamneter is 2.7 mm，and the filterbed20↑米 0-250Cmporosity is 55%. The frictiocofficient between thesuspended particles and filter is assumed as 0.2. Other02-3 3-44-5re given in the. sectionsParticle diander, m中国煤化工was presented in oher .Fig. 2 Eifet of particle diamneter on is removal without foceulent dosageTYHC N M H Gorosity on head lossAccording to the above-mentioned theoretical analysis,Different inluence of PFC. and. PAC on particlethe inluence of filter bed porosity on the hydraulic slopeentrapment along the filter bed could be obtained through theduring filtration was calculated by Equation ( 7)，and thetheoretical resuls are shown in Fig. 6. It can be seen from00GUO Jin-long et al .Vol.16 .Fig.6 that the hydraulic slope increases rapidly with thecharacteristic of the particle and filter grain has slight efectdecrease of filter porosity， especially when the porosity isn the shear force. According to the analysis of capillarylow. which means that great head loss change will occur ifnodel， the shear foree has linear relationship with thethe porosity have small changes during the filtration process .particle diameter, which could be got from Fig9.16A09.129.10 t.089.065 5453525s15049484746Filter bed porosity, %Fig. 7 Theoretical relation between the particle diameter and shear foreFig. 6 Theoretical relation between the hydraulie slope and filer bed porosilyThe PFC flocculent contains hydroxylic polymers,3.2.3. Influence of surface potential on the particleremovalwhich can increase the collision o[ suspended paricles andThe infuence of surface potential on the particle removalthe collision between the particles and flter grain， thusshown in Fig. 8(a) and (b) are achieved when the surfaceaccelerate the fToc growth and enhance its entrapment by thepotential of the suspended particles are- 25 mV and - 125filter. Dehydroxylation of the hydroxylic polymers in the PFCrespectvely.locs will lead to the shrink of the flocs, thus slow down theIt can be seen from Fig. 8(a) that the physicochemicalporosity decrease ( Leprince，1984; 0' Melia，1967). Asforces between the particle and filter grain surface decreasediscussed in thetheoretical'analysis, small increase of therapidly with the increase of their separation distance. Thefilter bed porosily will lead 1o rather great decrease ofinfluence of particle diameter on the physicochemical forceshydraulic slope. Therefore， the lower head loss of PFCwill be quite small when the separation interval is larger thancompared with PAC during deep bed filtration may be5 nm. However, in the range. from ! to 4 nm, the particleresulted from the dehydroxylation of PFC flocs.diameter has great effect on the physicochemical interaction3.2.2 Influence of particle diameter on the shear forceforce，and the attraction force increased sharply with therig. 7 shows the influence of paricle diameter on theincrease ofparticle diameter. The surface potential of theshear force of suspended particles, which is calculated on thesuspended particles ( flocswill be grealy increased whenbases of Equation (9), and the separation distance z belweenflocculent is dosed,the particle and capillary wall is assumed as 2 nnThesheargreatly increased due to the floc attachment on the filterforce on the suspended particle is the function of filergrain. Thus the physicochemical force between the suspendedporosity, i.e. diameter of the capillary, which will determineparticles and filter grain will have the same variation trend asthe flow velocity in the capillary . Therefore ，surfaceshown in Fig.8(a).14? 10040-AXxxxxx*XX60卜80 t00 t、 21-120 F40 t-2-160 (1234567g'910Separstiom distace, nm-▲a=sxa=2Fig. 8 Inluence of srparation dirtance on the plhysicochemical force between the paricle and filer grainAs concluded from Fig. 7, the influence of particleon particle removal under such conditions will be inertia,diameter on the shear force is linear and quite small， whichdiffusion and other interactions .means that the obslacle to particle attachment has slightlyThe repulsion force between the particle and filter grainrelation with the particle diameter. Therefore, the changes ofwill increase with the increase of particle diameter andphysicochemical forcehat the allachment on theattachment according to the theoretical analysis, and the中国煤化工for langer partcles. Theflocculent dosage will result in higher removal efficiency for: will be quite small whenlarger particles, which can be seen from Fig. 3 and Fig. 4he IYHCNMHGn4nm.whichisoflheIt can be seen from Fig. 8( b) that the physicochemicalsame trend observed in the high surface polential case. Theforce between the particle and capillary wall is minus whensurface potential of the particles and filter grain will be quitethe surface potential is - 125 mV, 80 the infuencing factorslow without flocculent dosage, and as discussed above, theNo.2hysicochemical interaction and its infuence on deep bed filtration process30removal eficiency will decrease significantly with the increase(5): 1278- 1286.of particle size under such conditions . which can interpret theIsraclachvili J1992. Adhesion forces. between surfaces in liquids andcondensable vapours{J]. Surf Sei Rep. 14: 109- 159.experimental results of Fig. 2 qualitatively.IvesK J， 1980. Deep bed frtration theory and patie(J. Filrtion and4 Conclusions. JeeatheesanV.VimeswaranS.2000.Using capillary model， the hydraulic shear forceprticles in dep bed fihration under unfavorable onditions[J]. Wat Res. 34(7): 2119-2131.distribution in the pore of filter bed was simulated, and theJingY H, Jin T G, FanJ C. 2000. Calculation model of head los in thephysicochemical interaction forces between the suspendedflhration process with uniform melialJ]. China Water and Wastewater, 16particle and filter grain and their influence during filtrationprocess was investigated theoretically. The computationfleretion process with uniform media[]. China Water and Wasewater, 16results were used to interpretintepret the experimentala of pilot(6): 1-4.study for miicro-floculation deep bed filtration， which wasLeprince A.. Fiessinger F, Bottero. J y, 1984. Polymerized imn chloride: ancarried out in the No. 9 Potable Water Treatment Plant ofO'Melia C R. Summ W.1967. Theory of waler flrtio[J]. J AWWA. Nov.:Beijing. The water head loss, paricle size distribution and0' Melia C R，1985. Particles. prereeits removal efficiency along the_ packed filter bed were[J]. J Envir Engrg-ASCE, 111(6): 874- -882.measured during the experiment . Theoretical calculation forO' MeliaC R，Melinda W H, Chen C, 1997. Some efeets of parnieles size inthe physicochemical forces between the suspended particleseparntion procas involving cloldl~JI. Wat Sei Technol, 36(4): 119-and filter grain, shear force on the particle caused by the flowin filter bed and the influence of the surface polential ,PayetakesAC, Tien C, Trian R M.1974(8). Trajectony ralculation of particleparticle size and separation distance were carried out. ThePayatakes A C. Rajgpalan R. Tien C.1974(b). Aplcation of porou mediacomputational results coexplainexperimentalmodels to the sludy of deep bed flration[J]. Can j Chem Fng, 52: 722一phenomena qualitatively, and through the comparison of thePontiues F W.1990. Water quality and lreatment - - A handbook of rommunitytheoretical and experimental results, it can be concluded thatwater supplies(4th ed.)lM. New York: MeGraw-Hill, Inethe_ physicochemical interaction and flow condition analysisShere-in-cel prousmedianmodei[J1.AIChE1.2261.523-533.ih thecould be used to study. the regulation and control of theRuckenstien E，Priere D C, 1976. Adsarption and desorption of particles anduniform media deep. bed filtration. Further theoretical antheir chromalographie separntion{J]. AIChE J, 22 (2): 276- -283.experimental research should be carried out to study theShandalow s, Yakirevich A. Brener A et al., 19979. Model calibratio of deep-interaction mechanism and its application in deep bedbed. flration based on pil-srale treatnent of serondanry efluent[JI. Wat Seifilration and other cases, then the quantitative computationnd analysis will be possible and the deep bed filtrationbeds in the presence of natural organice mattetJ]. Colloids Surf A. I45:process could be operated more easily and efficiently .Stumm w. Morgan J J, 1981. Aquatie chemistry . an intodurtion emphasizingReferences:chemnical equilibria in natural waters(2nd ed.)[M]. New York: John WileyBaiR B. Tien C.2000. Transient beharvior of particle deposition in granularspecies-( I ): preformed polymers by base ddition[]. Wat Res. 21: 115-Ber mrdia underwarous surace ietrationsJ iColopi SudA, 165:95-114:aggregation of hydrated ollids[J. j Cllid interfaceSci, 199: 1-12.Warszynski P, 2000. Coupling of. hydrodynamic and etrie interactions. inandsorption of colloidal particlesJl. Adv Colloid Interface Sci. 84: 47一142.Gregory. J 1975. Internction of unequal double layens at constant charge[J]. JColloid Interface Sei, 51 (1): 44- -5Wen DX. WeiYD, LiZN eal., 199 Hydrodynamiel M]. Bejing: HigherGiregory」, 1981. Aproxinate expressions for Retarded van der W als interactionYao K M. Habibian M T, 0'MeliaC R.1971. Water and waslewater fltraion:CuoLollid hreresei830 18-145.concepls and applications| . Envimn Sxi Terhnol, 5(11): 1105-1112.Zhou z K. GuT R. MaJ M, 1996. Bases of colidial chemistry [ M]. Beijing:HabibianMJ; 0' Meli C R.1975. Prticles,. polymers. and perormance inPeking University Press.Huang CP. PanJ R. HuangS H, 199 Cllision eficincie of elgae and kaolin( Received for reriew November 25, 2002. Accepled January 14. 2003)in deph flter: the eleet of surface properties of particles[J]. Wat Res, 33中国煤化工MYHCNMHG