Modeling of Biofiltration Process in Drinking Water Treatment

52Jourmal of Donghua University (Eng. Ed.) Vol.27, No. 1 (2010)Modeling of Biofiltration Process in Drinking Water TreatmentCHEN Wei (陈卫)' , XIA Qiong-qiong (夏琼琼)',2*，ZHENG Xing-can (郑兴灿)3College of Enrirorment Science and Engineering，Hohai University, Nanjing 210098, China2 State Key Laboratory of Hydrology Water Resource and Hydraulic Enginering, Nanjing 210098 , China3 National Engineering Research Center for Urban Water and Wastewater, Tianjin 300074, ChinaAbstract: The aim of this research is to develop awhile most research results showed that despite therepresentative model which simulates the performance of aapplication of air scour, the biomass of biofilters would notbiofilter used to treat drinking water. A steady-sate model isbe significantly decreased by backwashing. And accordingapplied. The Monod-type substrate utilization is used, andto these research data, even if the biofilter was backwashedthe extermal and the intermal mass transfer are neglected inwith air, the most significant biomass loss was not morethis model. The model describes the process of substratethan 50%一60%5.刀. This suggested that the effect ofbiodegradation, bacterial attachment onto filter media, andperiodic backwashing on biofilter performance could bedetachment of suspended bacteria. It simulates the naturalneglected. Therefore, steady-state model is moreorganic matter (NOM) and biomass profiles in a biofilter asrepresentative and practical.a function of filter depth and fitration velocity. The keyThe steady-state modei was first introduced bybiokinetic parameters k and Ks are estimated through aRittmann and McCartyC]. It described substrate diffusionspecial experiment. The results of the model testing showwithin the biofilm, substrate utilization in the biofilm andthat the model prediction agrees well with the experimentalthe biofilm growth and decay. Later, Sacz, and Rittmanndata.then developed a pseudo analytical solution for this model,Key words: biodegradable organic matter; biofiter;which helped to work out the substrate flux into a steadydrinking water; modelingstate biofilm and biofilm thickness.CLC namber: TU991.24Document code: ASteady-state biofilm model was developed by ZhangArticle ID: 1672 - 5220(2010)01 - 0052 - 06and Huck[2] to simulate the removal of assimilable organiccarbon (AOC) in deep bed filters. Four parameters of themodel were determined by non-linear regression, thus a lotIntroductionof data was needed, and the precision of the simulationresults was always lower at other sites.In drinking water treatment, biofiltration is aWang and Summer'] developed another model forpromising way to remove natural organic matter anddeep bed filters. In this model, the diffusion of substrateachieve biological stability, thus having attracted greatinto a biofilm was assumed to be neglected because of lowattention in water industry. Since 1980s, many researchesamount of biomass present. However, the concentration ofhave been performed on it. In order to further understandsessile bacteria should be determined in every situation.the chemical and microbiological processes, manyBillen et al.国published the Chabrol model. Theresearchers have made studies of the biodegradationauthors divided biodegradable dissolved organic carbonmechanisms and put forward a lot of biofilter models.(BDOC) into three components and considered the majorThese models can be divided into steady-state[I] andmicrobiological processcs, including hydrolysis andnon-steady-statel5] types. Steady-state models consideredmetabolism of organics, interactions between bacteria andthat the biomass of biofilm were unchanged, while non-solid support, and mortality and grazing of bacteria. Thesteady-state took the sudden loss of attached biomass due tomodel required a large number of parameters which werefilter backwash into account.difficult to determine.The non-steady-state model proposed by Hozalski andThe overall objective of this paper is to develop a moreBouwer()] showed that the average biodegradable organicrepresentative steady-state biofilter model for designingmatter (BOM) removals should not be impacted until thebiofiltration processes in water treatment. The modelamount of biomass lost during backwashing exceeds 60% ,should be able to describe the fundamental bioprocesses in中国煤化工Recived date: 2008-03- 19Foundation items; National Natural Science Foundation of China (No.MYH. CNM H Guio Pa tor GnadureStudents of Jiangsu; Province of China (No. 1061B06013)* Correspondence should be addressed to XIA Qiong-qiong, E-mail: xqqwater@126. comJourmal of Donghua University (Eng. Ed.) Vol.27, No.1 (2010) 5biofilters and simulate the natural organic matter (NOM)to be homogenous and have the same properties. Thconcentration as well as bacterial concentration correctly.external mass transportation and diffusion inside thebiofilm are neglected.1 Experiment and Methods2.1 Biodegradation of substrateAs the concentration of sessile bacteria is much higher1.1 Biofiltration apparatusthan that of suspended bacteria, substrate utilization isNOM in drinking water treatment is composed ofconsidered mainly by the bacteria attached on the fitervarious organic substances, most of which are unknown andmedia. According to Michalis-Menten kinetic， that relatively low concentration. Therefore other method issubstrate utilization rate is:more suitable to determine the concentration of NOM. In-μmx .S.X. _ -k.S. Xm(1this research, dissolved organic carbon (DOC) was used torepresent the NOM.The DOC removal in a biofiltration process waswhere Pam is the maximum specific growth rate of bacteria(h"), Ks is the Monod half velocity constant (mgDOC/L),investigated in a pilot plant, which took the effluent waterk is the maximum specific substrate utilization rate (mgDOC/of a sedimentation tank in a water treatment plant as theCFU/h), Yw/s is the growth yield (CFU/mgDOC).raw water. The experiment was performed in a granularactivated carbon (GAC)/sand filter of 1.8 m height and2.2 Attachment of suspended bacteria onto filter100 mm inner diameter (GAC layer thickness: 60 cm, sandmedialayer thickness: 40 cm) (Fig. 1). The filter had sevenThe attachment rate of suspended bacteria onto filtersampling ports which were located along the GAC bed atmedia is" described by the theory of Langmuir-typethe depth of 10, 20, 30, 40, 50, and 60 cm respectively.adsorption.The biofilter was naturally set up and operated about eight(2)months before sampling.where ka is the adsorption cofficient (L/CFU/h),Yangzi Floculation Sedimentation Fast sand DisinfectionX.由is the maximal capacity for bacterial adsorptionRiver polymer alumfilration(CFU/L).2.3 Detachment of the bacteria by fluid shearFor the detachment process, the bacteria detachment回也is considered mainly due to fluid shear, and the shearingClearwellrate is derived from the detachment model of Speitel andDiGianofe].Sampling portsru=(b.+b; usS )xu，(3)Ks +Swhere b. is the biofilm shearing coefficient (d-1), b'is aFig.1 Drinking water treatment scheme ofdimensionless parameter between 0 and 1. The twofull scale plant and pilot-plant biofilterparameters can be estimated from experimental data.1.2 Analytical methodsAnother method to estimate b. is using the followingDOC was determined by measuring the total organicequation proposed by RittmannI0 :carbon (TOC) of samples filtrated through a 0.45 μmb,=2.29X10-[ L1-27g*,4)membrane. Heterotrophic plate count bacteria weredfes ameasured using a spread plate procedure with R2A agar,whereμ is the fluid viscosity (g/(cm ●d))， v is theincubated at room temperature for seven days"0.superficial fluid velocity (cm/d), E is the bed porosity, d,is the diameter of media (cm) and a is the specific surface2 Modelingarea of the bed (cm~').2.4 Growth and decay of bacteriaKinetic modeling of the utilization of the dissolvedThe growth rate is described according to Monod-organic matter was applied, and the fundamentalbioprocesses considered were as follows: (1) biodegradation中国煤化工of substrate by the ssile bacteria in the biofilm;5)(2) attachment of suspended bacteria onto fiter media;TYHCNMHG(3) detachment of the attached bacteria by fluid shear;A first- order constant of decay rate is considered in the(4) bacteria growth and decay. The bacteria are assumedmodel54Joumal of Donghua University (Eng. Ed.) Vol. 27, No.1 (2010)rx. dcey=一ksXm.(6)aX.at=erm-rau+rx.g +rx.deay.(9)2.5 Mass balance of a packed biofilm columnDrinking water biofilters are typical plug flowAs the mass profile in biofilter is assumed to be inreactors. Combining the transport of subtrate andsteady-state, the time derivatives in Eqs. (9)-(11) are set(suspended) bacteria, mass balance can be established.0, and these equations can be simplified asUnder the hydrodynamic conditions of biofiltration, thedS=上(10)effect of dispersion is neglected.dx vEThe substrate balance in a small segment of biofilm canbe set asdX==+(-rm),(11)asva(7)erm-rau+rx.g +rx.awy=0.2.6 Model parametersThe balance of suspended bacteria is described asModel parameters are important for biofiter simulation.ax=_ ra(8)The parameters used in the model are listed in Table 1. Of allthe parameters in Table 1, some biofilter specifications wereIn the same way, the balance of attached bacteria isknown directly, some parameters were determined fromgiven belowexperiment, and others were taken from literatures.Table 1 Paramcter valuc used for model simulationParameterValueUnitSourceBiofilter specificationsBed depth of GAC0.6mColumn diameter0.1Porosity(e)GAC diameter (ac)0.78Water viscosity(4)1.0828X10-1Biokinetic constants6. 30X10-10(17.5 C)Maximum specic substrate utilization rate (k)mgDOC/CFU/hExperiment5.36X10-10(11.5C)Growth yield (YN/s)3. 09X108CFU/mgDOCRef. [11]4.05(17.5 C)Half velocity constant (Ks)mgDOC/L4.7011.5 C)Decay cofficient (ks)0.065Ref. [4]Adsorption and detachment constantsAdsorption cofficient (ku)1.4X10-10L/CFU/hRef. [12]Maximal capacity for bacterial adsorption (X,由) 2.7X1012CFU/LBiofilm shearing cofficient (b)4.6X10-8h-1CalculatedDimensionless shearing coefficient (的.5Bacteria propertiesDensity (Pp)1.1X10-3mg/LRef. [13]Diameter (dp)1.5X10-8Infuent concentration of suspeoded bateria (Xm,x=o) 1. 2X107Other paranetersHamaker constant (H)1X10-20Bolzmann constant (kb)1. 38X10-23J/KGravity acceleration (g)9.8中国煤化工-As the model is Monod-type dependent, the keyestimCNM H G,parameters (Ks and k) therefore play an important role inAwruungw TvIuiw-quatiui, the specific substratethe biofilter simulation. The values of Ks and k wereutilization rate isJournal of Donghua University (Eng. Ed.) Vol.27, No.1 (2010) 55R=_ds_ 二.kS(13)Xdt~ Ks+S0 Exp。ooo0。dSModelo,InEq. 13, as is approximated as So-，Sisapproximated asSotSe. Where s。and s。are the itial2。and end subtrate concentration respectivcly (mgDOC/L),。689680and l represents the time period for biodegradation.InEq.13, s is regarded as an independent variable,R=5.36E-10S/4.70+5)R isa dependent variable, and then Ks and k can beestimated by non-linear regression ( using Gauss-Newtonmethod) , which is based on the minimization of the sum of10squares of the difference between predicted and mcasuredS/(mgL)specific substrate utilization rates:Fig.2 k and Ks estimations at 11.5Cmin2(14)<10-10An appropriate experiment was performed for Ks/k|0 Exptest, according to the above technique. A cocktail of。。BOM, nitrogen and phosphorus was added in bottles for室4bioreaction, and the biomass taken from the top layer ofthe biofilter was then put into the bottle. The bioreaction0 9eCtemperature was contolled, and in this project, the Ks/ktest was carried out at a middle temperature (17.5 C) and00 (Doa lower temperature (11.5 C).ogo 8。R-=.30E- 10S/(4.5+5)he estimation results are presented in Table 1 and theobserved and predicted R valucs as a function of S areshown in Figs. 2 and 3. It should be noted that comparingS/(mg:L7)the k valucs in this study with those from other literaturesis difficult, because variant units are used. However, theFig.3 k and Ks estimations at 17.5CHw (h~1) is comparable. The value of Pmx in this paper iscalculated as 0.195 h*'(17.5 C) and 0. 166 h*'(11.5C)，3 Results and Discussionwhich agree well with those from HozalskiFS] and Zhang[2].The half velocity constant, Ks at twdiffcrentFigures 4 - 9 describe the influence of filtrationtemperature in this paper is also at the same level (1mgDOC/L) with those investigated rcently in biologicalvelocity and bed depth on the DOC and bacteriaconcentration profile at 17.5 C and 11.5 C respectively.water treatment. The effect of temperature on themaximum specific substrate utilization rate (k) is usuallyThe concentration of suspended bacteria and substrate atx=0 are measured and listed in Table 1 with otherexpressed in the following form[4;parameters. The model results are tested with the datakτ=kxθT-∞,from a pilot-scale biofilter to evaluate its applicability.where, kτ is the maximum substrate utilization rate at theFigures 4 and 5 show that DOC removal part of filterdetermined temperature, kzo is the maximum substratebed increases from 0.2 m to the whole filter depth whenutilization rate at 20C, and θ is the temperature activitythe filtration velocity increases from 5 m/h to 15 m/h.coefficient.Consequently, the biodegradable DOC moves deeper intoAccording to the k values estimated at two differentthe filter bed with filtration velocity increasing.temperatures,日can be calculated as 1.027, which is in theFigures 6-9 show that both attached and suspendedlower range for trickling filters (1. 02 - 1. 08) used inbacteria concentration present the similar profile in steadywastewater treatment[I0. This reason might be that thestatesubstrate concentration in drinking water is lower and thedepth中国煤化工al the fiter上: is increased, thbacterial populations which are accustomed to this are:YHCNMHGthebed.selected, while in wastewater most bacteria are accustomedThe model results also show that the DOC removalto a higher BOM level.ate is the same at equal empty bed contact times56Joumal of Donghua Uhiversity (Eng. Ed.) Vol.27, No. 1 (2010),x10*-17.5C. v=7.5 m/h7=11.5C -- 一v=7.5m/hv=10m/h1.9--.--=15 m/h●ExperimentA Experiment1.8-房g4 t1.60.20.4”0.6-二=---Bed depth/m0.Fg.4 Simulated and measured DOC concentrationsas function of bed depth and uperficial flowFig.7 Simulated and meaured suspended bacteriarate at17.5Cconcentrations as function of bed depthand superficial flow rate at11.5 C2.27=11.5C- v=7.5 m/hae由!0...... v_10m/h7-17.5 C-- - - v=7.5 m/h0 Experiment忌32.1角、-.---v=15m/h2上圈Experimeot2.0重申雨" 0.6田、...._.0.6Fig.5 Simulated and mcasured DOC concentrationsas function of bed depth and superficialflow rate at 11.5 Cteriaand superficial flow rate at 17.5 c爷10"-..--.. v=10 m/h=11.5 C6fExperiment图Experiment4、号2(电0Fig.6 Simulated and measured suspended bacteria_ Fi.9 Simulated and measured attached bacteria中国煤化工t boed depth11.5C(EBCTs), which suggcsts that the performance ofHCNMHGremoving DOC is not affected by filtration velocity, asremove doC 1s6.9 min at 17.5 c and7.6 min at11.5 Clong as the EBCT keeps the same. The EBCT needed toin this pilot experiment.Jourmal of Donghua University (Eng. Ed.) Vol.27, No. 1 (2010) 57The impact of water temperature is not shownWater Treatment Processes: a Kinetic Modcling Approachdirectly. According to the calculation results, when[J]. Water Research, 196, 30(5); 1195 - 1207.temperature is decreased from 17.5 C to 11.5 C, the[3] WangJ Z, Summer R S. Modeling of Biofiltration ofNatural Organic Matter in Drinking Water Treatment[C].average removal rate of DOC decreases from 19% to 11%Proceedings of 1994 National Conference on Environmentand the effective zone of filter bed extends from 0. 59 m toEnineering, NY, 1994: 452 - 459.0. 63 m (filtration rate is 5 m/h). It seems that[4] Bllen G, Servais P, Bouill P, et al. Functioning oftemperature has not significant effect on the size of activeBiological Filters Used in Drinking Water Treatment一thezone. The prediction result is also affected by DOCChabrol Model [J]. Water Supply research Technolog4concentration of raw water. Actually, according to theAqua, 1992, 41(4); 231 - 241.results of another proccdure, the active zone will extend[ 5] Hozalski R M, Bouwer EJ. Non stcady State Simulation offrom0.53 m to 0.63 m if the DOC concentration of rawBOM Removal in Drinking Water Biofilterst Modelwater keeps the same (2. 25 mgDOC/L) at the above twoDevelopment[J]. Water Research, 2001, 35 (1);temperatures.198-210.[6] LuP, Huck P M. Evaluation of Methods for Measuring4 ConclusionsBiomass and Biofilm Thickness in Biological DrinkingWater Treatment[C]. Proccedins of 1993 Water QualityTechnology Conference (WQTC), Miami, Florida, 1993;This paper described the deveclopment of a model,1415- 1456.which simulated the steady-state substrate and biomass[7] Hozalski R M, Bouwer E J. Deposition and Retention ofprofiles in drinking water biofilters. The model was testedBacteria in Backwashed Filters[J]. Journal America Walerusing data from a pilot-scale experiment. The results showWork Association, 1998, 90(1); 71-85.that this model gives an accurate description of DOC[8] Speitel G E, DiGiano F A. Biofilm Shearing underbiodegradation and biomass distribution in biofilter.Dynamic Conditions[J]. Environmental Engineering, 1981,Therefore, it's an effective model for simulation of the113(3): 464-475.biofiltration process.[9] Standard Methods for the Examination of Water anThe key biokinetic parameters Ks and k wereWastewater[s]. 19th cd. Washington D. C: Amcricanestimated at two different temperatures (17. 5 c andPublic Hcalth Association, 1998.11.5 C) through a simple experiment in this paper. The[10] Rittmann BE. The Effect of Shear Stress on Biofilm LossRate [J]. Biotechnology and Bionginering, 1982,estimated temperature activity coefficicnt θ can be used to24(2): 501 - 506.estimate k values at other temperatures. Modeling results[11] Fry J C. Determination of biomass [C]//Methods inof DOC concentration and bacteria profiles along the depthAquatic Bacteriology. New York: Wiley, 1988: 27 - 72.of biofilters can give some insights into the biofilter[12] Uhl w. Enfluss von Schittungsmaterial undperformance. Other parameters characteristic for filterProzessparanetern auf die Leistung von Bioreaktoren beimedia and water quality parameters like bacteria and DOCder Trinkwasseraufbereitung[C]/Dissertationen aus demconcentration should be determined for the simulation.Iww. Milheim, 2000.[13] Martin R E. Quantitative Description of BacterialReferencesDeposition and Initial Biofilm Developmeat in PorousMedia[D]. The John Hopkins University, Baltimore,[1] Rittmann B E, McCarty P L. Model of Stcady-State1991.Biofim Kinetics [J]. Biotechnology and bioengineering,[14] Tchobanoglous G, Burton F L, Eddy M. Wastewater1980, 2(11); 2343-2357.Engincering; Treatment, Disposal, aod Rcuse[M]. New[2] ZhangS L, Huck P M. Modeling Biological DrinkingYork; McGraw-Hill, 1991.中国煤化工MYHCNMHG