Simulation and assessment of sludge concentration and rheology in the process of waste activated slu

Available online at www.sciencedirect.comJOURNAL OFENVIRONMENTAL。ScienceDirectISSN L001 0747ISN 10-0743JES iJoural of Eavironmental Sciences 21(2009) 1639-1645www.jesc.ac.cnLELESimulation and assessment of sludge concentration and rheology in the processof waste activated sludge treatmentXIA Mingfang', WANG Zhiwei2*, WU Zhichao2, WANG Xinhua2,ZHOU Zhen2, LU Jilail1. State Key Laboratory of Pollution Control and Resource Reuse, Nanjing Universir, Nanjing 210093, China. E-mail: mingfang. xia@ sina.com2. State Key Laboratory of Pollution Control and Resource Reuse, School of Emvironmental Science and Engineering.Tongji University, Shanghai 200092, ChinaReceived 10 February 2009; revised 20 April 2009; accepted 27 Apri 209AbstractThe process of using fat sheet membrane for simultaneous sludge thickening and digestion MSTD) was employed. The variationsof sludge concentration and rheology were characterized and simulated. Based on mass balance analysis, mathematical models weredeveloped and successfully used to predict and evaluate the variations of sludge concentration and the digestion eficiency in the MSTDprocess. The apparent viscosity of sludge could be modeled as functions of mixed liquor suspended solids and shear rates. The sludgein the MSTD process showed both shear thinning and viscoplatic behaviour, and under various shear rates diferent rheological modelscould be chosen to predict their flow behaviour. It was also found that sludge concentration and viscosity had signifcant correlationswith membrane fouling in the MSTD process.Key words: membrane fitration; membrane foulig; sludge thickening; sludge digestion; waste activated sludgeDOI: 10.1016/S1001-0742(08)62467-5Introductiontime (SRT) with gravity thickening process, lower quan-tity of sludge storage and higher energy cost with DAFIt is widely accepted that conventional activated sludgethickening compared with gravity thickening, and muchprocess is a cost- effective and efficient method for wastew-higher energy cost and advanced maintenance require-ater treatment; however, it produces excess biomass asments with centrifugal thickening technology (Zhang etwaste activated sludge (WAS) that is dificult and expen-al, 2000; Metcalf & Eddy Inc., 2003; Hua et al, 2005).sive to handle and dispose of, particularly when wastewaterIn order to solve some problems dealing with conventionaltreatment plants (WWTPs) were adapted for biologicalthickening technologies and to incorporate thickening andnutrient removal. Rai et al. (2004) reported that the cost ofdigestion into a single reactor, an innovative process ofWAS treatment and disposal accounted up to 60% of theemploying flat-sheet membrane for simultaneous sludgetotal operating cost in WWTPs. WAS treatment became athickening and digestion (MSTD) was proposed in ourmore dificult and complex problem in WWTPs atributedprevious research (Wang et al, 2008). The preliminaryto the stringent efluent criteria and restrictions to landillstudy demonstrated its feasibility for WAS treatment, inWAS.which good sludge thickening and digestion efficiency andThickening and dewatering are usually performed fcsuperior effluent quality were achieved. Although micro-volume reduction of WAS in WWTPs. After thickening,filtration, ultra-filtration, nanoflitration and other processesa stabilization step, such as sludge digestion, is practicedcoupled membrane solid-liquid separation such as mem-to achieve stabilization, detoxification and minimizationbrane bioreactor (MBR) have been intensively studied andof WAS, especially for medium and large-scale WWTPswidely used for the treatment of municipal wastewater,(Zhang et al, 2000). Sludge thickening is generallyindustrial wastewater and surface water/drinking water inconducted by physical means, includinggravity thicken-past decades (Mehta and Zydney, 2005; Melin et al, 2006;ing, dissolved air flotation (DAF) thickening, centrifugalKim中国煤化工，the apliecion ofthickening, etc. Among the typical sludge thickening tech-membrdigestion is a newnologies, several problems and disadvantages are existing,attermpl|YHCNMH Ge.g. the large footprint and low thickening eficiency andIn this suuy, specuIc aluenuon was paid to simulate andthe release of phosphorus under long sludge retentionassess the variations of sludge concentration and rheolo-gy in the membrane for simultaneous sludge thickening* Corresponding autbor. E-mail: zwwang@tongji.edu.cn1640XIA Mingfang et al.Vol. 21and digestion (MSTD) process by employing mathemat-It was about fteen days of one cycle from the initialical models. The variations of sludge concentration andfresh sludge to the final thickened and digested sludge.the digestion efficiency were predicted and evaluated by The sludge retention time (SRT) of the MSTD reactormathematical models. Sludge rheology behavior and thewas 15 d (one cycle), and the hydraulic retention timecharacteristics of sludge flow behavior under different(HRT) about 1 d. The filtration operation of the pilot-shear rates were studied. The results were expected to scale MSTD reactor was conducted by pump suction withprovide a sound understanding of sludge concentration and the constant flow rate (15 L/(m2-.h)) of membrane fuxrheology variations in MSTD process and to facilitate its(membranes were chemically cleaned with NaClIO solutiondesign and operation for WAS treatment.(0.5% (VV), 2 h duration) every two cycles, i.e, every 30d to recover their permeability). Intermittent filtration, 101 Materials and methodsmin filtration and 2 min pause, was carried out during thewhole experimental period. The CFV along the membrane1.1 Operation of MSTD processsurface was maintained at 0.2- 0.4 m/s by coarse bubbleA pilot-scale MSTD reactor located at Quyang Munic-aeration. When the transmembrane pressure was higherthan 30 kPa, chemical cleaning in-place procedure wouldipal Wastewater Treatment Plant (WWTP) in Shanghai,be performed. The temperature in the MSTD reactor wasChina, was studied. The reactor configuration (Fig. 1)about 20- -23°C during the operation.nd membrane materials were the same as those reportedpreviously (Wu et al, 2009). Activated sludge was firstly1.2 Rheological behavior modelpumped into a sludge storage tank from an aerobic basinSludge is often considered as a non-Newtonian fuid,of Quyang Municipal WWTP and then fed into the MSTDand its rheological behaviour can be described by thereactor. The properties of the activated sludge could beBingham model (Eq. ()), the Ostwald model (Eq. (2)),found elsewhere (Wu et al.. 2009). In order to supplythe Herschel-Bulkley model (Eq. (3)), and the Sisko modeloxygen demanded by the microorganisms and to induce a(Eq. (4)) (Guibaud et al, 2004; Hasar et al, 2004; Mori etcrosflow velocity (CFV) along the membrane surface, airal, 2006; Laera et al, 2007).was provided by a compressor through an air diffuser. Theefluent filtered through membrane modules was obtained, dhby suction pumps connected to the modules. The effluentτ=To+k(1)flow rate and transmembrane pressure were monitored. Awater level sensor was used for maintaining a constantτ=x()(2)water level of the MSTD reactor over the experimentalperiod.(出)(3)When the mixed liquor suspended solids (MLSS) con-centration in the MSTD reactor reached approximately 35Ivg/L, the thickened and digested sludge was drained outτ=4B>(出)(4)while the influent activated sludge and the permeate fowwere stopped. After that, the reactor was flled with thewhere, T (Pa) is the shear stress, dv/dx (s-I) is the shear .fresh activated sludge, and then the next cycle was started.rate. The consistency index k represents the cobesivenessof the fluid, and flow behaviour index n far from one meansControl systemhigh deviation from Newtonian behaviour (n = 1 for New-tonian fuids) and the yield stress TO indicates the resistanceof the sludge to the deformation until sufficient stress isWASPressure gaugeapplied to exceed the yield strength of the solid phase. The二自自parameter uB is the high shear limiting viscosity when theInfluent pumpEffluent pumpshear rate imposed on the fluid tends to an infinite value.The ratio between shear stress and shear rate is definedas apparent viscosity U4a):募|Ha=(5)曾厦/Flow meterThe parameter values of n, k, τo and μg of the four名| /models could be obtained by ftting the experimental dataof various shear rates under a series of ML ss concentra-tion中国煤化工-lels) or by adoptingAir diffusernon-lr Herschel-BulkleyMSTD reactorBlowerand sHC N M H Guld be crrelated toMLSS concenurauon unrougn unear or exponential-powerFig. 1 Diagram of the pilot-scale membrane for simutancous sludgelaws. Then, the four models mentioned above could bethickening and digestion (MSTD) reactor. WAS: waste activated sludge.described as a function of the shear rate and variable MLSSNo. 12Simulation and assessment of sludge concentration and rheology in the process of waste activated sludge treatment1641and used to assess their simulations by ftting experimentalinfuent flowrate, Xo,a (g/L) and Xea (g/L) are infuentand efluent active biomass concentration, respectively;S。(g/L), and Se (g/L) are infuent and efluent soluble13 Analytical methodssubstrate concentration measured as soluble COD (SCOD)Chemical oxygen demand (COD), MLSS and mixedof sludge supematant, respectively; Ko (d-) is endogenousliquor volatile suspended solids (MLVSS) were mea-decay cofficient; Y (g VSS/g COD) is biomass yieldsured according to standard methods of Chinese NEPAcoefficient, and Xn (g/L) is active biomass concentration(1997). CFV was determined using Cup-type Currentin the reactor.Meter (LS45A, Chongqing Hydrological Instrument Inc.,The mass balance equation for non- biodegradableChina). Sludge apparent viscosity was measured by avolatile suspended solids (nbVSS) in the system could berevolving viscosity meter (NDJ, TJ Environmental Facilitysimilarly developed and expressed by Eq, (7):Inc., Shanghai, China) under various shear rates. All abovementioned analyses were conducted in duplicates, andv= QXs- QX&s + foKgVX。.(7average values were reported.dThe relations of sludge concentration and viscosity withmembrane fouling in the MSTD process were studiedwhere, dX/dt (g(L-d)) is the rate of nbVSS change in reac-according to the method described in our previous studiestor; Xoi (g/L) and Xes (g/L) are infuent and eluent nbVSS(Wang et al, 2006; Wu et al, 2007). Statistical analysis,concentration, respectively; fa is fraction of biomass thatincluding Pearson and Spearman's. rank correlations, wasremains as cell debris.carried out using the software SPSS 11.0 produced byAnother mass balance equation of inert inorganic sus-SPSS Incorporation (USA) to characterize the effects ofpended solids in the system can also be developed asMLSS concentation, viscosity on membrane fouling. Cor-shown in Eq. (8):relations are considered statistically significant at a 95%confidence interval (P < 0.05)."v = Qxoi - QXei(82 Results and discussionwhere, Xi (g/L) is the inert inorganic suspended solids;2.1 Mathematical simulation of sludge concentrationXo.i (g/L) and Xx.i (g/L) are influent and efuent inertinorganic suspended solids, respectively.and destruction processThe models for evaluating the variations of MLVSS and2.1.1 Mass balance analysisMLSS in the reactor could be expressed as Eqs. (9) andThe flow diagram together with thenomenclature used(10), respectively.in the following mass-balance equations are shown inFig. 2. In the MSTD process, the sludge concentration inVXv.v,= VXo,v +t(v些+ v些)(9)”dt)reactor is mainly dependent on three factors: the thickeningeffects of membrane separation which enables the effluentfree of ss; the biomass growth consuming soluble sub-vx=vx+(v些+v些+v)(10)strate of infuent mixed liquor; and the digestion effectscaused by endogenous decay. According to the principle,where, Xxv, (g/L) is the MLVSS concentration in theaccumulation = inflow - outflow + generation, a massreactor at time 1 (d), Xo.v (g/L) is the influent MLVSSbalance equation for active biomass within the systemconcentration, X; (g/L) is the MLSS concentration in theboundary can be established as Eq. (6):reactor at time t, and., (g/L) is the infuent MLSSconcentration.Ov= QXa- QXea + YQ(S。-S。)- KgVXa(6)According to the fact that total MLVSS in the reactorduequals the active biomass concentration Xg plus the nbVSswhere, dX/dr (g(L.d)) is the rate of active biomass change concentration x and the terms of Xea and X&i both equalin reactor; V (L) is MSTD reactor volume; Q (Ld) is 0 (membrane efluent is free of Ss), substituting Eq. (6)and Eq. (7) into Eq. (9) produces the fllowing Eq. (11)that can be used to evaluate the variations of the MLVSSconcentration in the process.Efluent waterInfluent studgeQ,So Xx, Xea=0,Xon Xol XaXv;Xa.v(1+藏)+最Y(So -S)中国煤化工~(11)7=THCNMHGSystem boundarySimilar equation for modeling MLSS variations in sys-FE2 Diagram of MSTD process with model nomenclature.tem could be obtained by substuting Eqs. (6), (7) and (8)1642XIA Mingfang et al.Vol. 21into Eq. (10).design of MSTD process, if two control parameters suchas HRT and MLVSS or MLSS are chosen, the operationX;=x(1+藏)+献Y(So-So)time for one cycle, i.e., SRT could be determined through1 +utKa(1-fa)(12) Fig. 4. It will, to a great extent, faciltate the design andaroperation of the MSTD process. It is worth pointing outthat another important variable for aerobic digestion designand operation is volatile solids reduction (destruction)2.1.2 Modeling of sludge concentration variationswhich will be discussed in the following sections.The experimental data of MLSS, MLVSS concentrationsand modeling values in MSTD during three operational2.1.3 Sludge destructioncycles (15 d of each) are ilustrated in Fig.3. It can beA major objective of aerobic digestion is to reduce theobserved that model predictions ft very well with exper- mass of the solids for disposal. This reduction is assumedimental data. The three cycles in this study had similarto take place mainly with the biodegradable content of thevariation trends of MI ss and MLVSS concentration in theWAS, while there maybe some destruction of the nonor-MSTD process. During one cycle of operation with HRT 1ganics as well Metcalf & Eddy Inc., 2003). The MLVSSd, MI SS concentration continuously increased from about and MLSS reduction rate in the MSTD reactor could be4 to 34 g/L and MLVSS increased from about 3 to over 22calculated through Eqs. (13) and (14), respectively.In order to further elucidate the relationship amongsludge concentrations, HRT and operation time, Fig. 4&exav- VXv,was plotted based on Eqs. (11) and (12). It can be seenEDva =(13)from Fig. 4a that MLVSS concentration increased morerapidly under lower HRT as operation time increased. Iti Q;Xovis attributed to the fact that the thickening effects tendto dominate with the decrease of HRT. Similar trend ofEQxo-VX;MLSS variation can also be observed in Fig. 4b. During the(14)名Q.x●Measuired MLSS。Measured MLVSS写50下MLSS model value - - - MLVSS model valuewhere, EDw. and ED: are the MLVSS and MLSS reductionrates at time 1, respectively, and Q (L/d) is the influent40卜 Cycle 1Cycle 2Cycle 3sludge flow at time I.Figure 5 shows the experimental data of MLVSS, MLSSdestruction rate and the model simulation values based on百20一Eqs. (13) and (14), demonstrating that the experimentaldata are roughly in agreement with the model predictedvalues. It could also be found that after 15 d operationabout 42% of MLVSS and 39% of MLSS reduction could1020304050be achieved, and this could meet vector attraction require-Operation time (d)ments of 40 CFR Part 503 for sewage sludge treatment,Fig3 Comparison between theoretical model predictions and exper-ie., a minimum of 38% reduction in volatile solids duringimental data of mixed liquor suspended solids (MLSS), mixed liquorbiosolids treatment (USEPA, 1992).volatile suspended solids (MLVSS) variatioas in MSTD process duringSolids destruction is primarily a direct function of boththree operational cycles.liquid temperatures in the reactor and SRT. Figure 6 shows6(30220 -3兰20- -0k <中国煤化工.0一3.2.525152.0*1HCNMHmC(o010HRTCA“可01.0Fig4 Correlations of MLVSS (a) and MLSS (b) concentation variation with operation time and hydraulic retention time (HRT) in MSTD process.No.12Simulation and asessmet of sludge concentration and rheology in the process of waste activated shudge treatment164370same degree-days. In order to meet the vector attractionrequirements of a minimum of 38% reduction in volatile60Model value of ML.VSS destruction ratesolids, the MSTD process needs about 240 degree-days- Model value of MLSS destuction rate8 50while conventional aerobic digester demands over 400Cycle 1Cycle 2Cycle 3degree-days. The higher digestion eficiency achieved in色40the MSTD process could be attributed to the fact thatin this process the influent undigested sludge under low费20concentration was continuously fed into the reactor andthe infuent undigested sludge would be blended with1the previously digested sludge existing in the reactor tocontinue the digestion process. In fact, it has been proven10203050that the destruction eficiency of MLSS and MLVSS inOperation time (d)aerobic digestion can be enhanced by adding digestedFig, 5 Comparison between theoretical model predictions and experi-sludge into the digester filled with undigested WAS andmental data of MLSS, MLVSS destruction rate in MSTD process duringthe digested sludge could serve as the source of viable cellthree cycles.mass needed for the degradation of organic solids (Khaliliet al.. 2000; Wang et al, 2008). The MSTD process couldnaturally utilize the mechanisms and thus the digestionefficiency was higher compared to the conventional aerobic80digestion process.The correlations of MLVSS or MI SS destruction ratewith obtained MLVSS or MLSS and HRT in MSTD40process are plotted in Fig. 7. The solids destruction ratetended to increase with increasing HRT. The increaseof operation time (i.e, SRT) could also result in theenhancement of solids reduction rate and increase the00 600 900 1200 1500 1800thickened sludge concentration at the end of each cycle. IfTemperaureX SRT(CXd)MLVSS destruction rate is chosen, the required HRT andFig.6 MLVSS reduction in conventional aerobic digestion processsolids concentration of the MSTD process can be obtained(reported by WEF) and in MSTD of this study as a function of digesteraccording to Fig. 7, which could facilitate the design andliquid temperature times SRT (degree-days).operation of the process.volatile solids reduction of this MSTD and conventional2.2 Sludge rheology propertiesaerobic digesters reported by Water Environmental Feder-2.2.1 Rheological models to assess the variations ofation (WEF, 1995) under various degree-days (temperatureviscositytimes SRT). It can be observed that the two digestionprocesses have similar variation trends as the degree-daysIn this study, a set of MLSS samples with concentrationsincreased, i.e, initially, the rate of volatile solids reductionof 3.30, 6.59, 9.89, 13.18, 16.48, 19.78, 23.82, and 28.58increased rapidly as the degree days increased while theg/L obtained at various operational time were used tocurve begins to fatten as the degree-days approach to ameasure their apparent viscosity at (25土1)°C under shearcertain value for the two aerobic digestion processes. Itrates 850, 1000, and 1850 s-'. The following Eqs. (13)-can also be seen from Fig. 6 that the sludge reduction(16), were respectively deduced from the Bingham model,efficiency of MSTD process is higher than that of conven-Ostwald model, Herschel-Bulkley model and Sisko model,tional aerobic digestion process for WAS treatment underrespectively, by using the apparent viscosity values under1.0~，080.6-0.640.4-自0.2+贸0.2中国煤化工250.6 0.3.6 0.3sen,i1.51209:HCNMH Gah101.8HRT101.8 1.0Fg7 Crrelations of MLVSS (a) and MLSS (b) destruction rate with obtained MLVSS/MLSS and HRT in MSTD process.1644XIA Mingfang et al.Vol. 21various shear rates and MLSS concentrations. They couldshould be overcome to induce fow due to the presence ofbe used to predict the apparent viscosity of mixed liquorsuficient suspensions (Ducla et al, 1983; Seyssiecq et al, .under diferent MI .SS concentrations.2003), and the concentrated suspensions are called as yicldstress fuids which show viscoplastic behavior, Viscoplas-tic models such as Bingham and Herschel-Bulkley modelsμa = 3249.1 exp(0146MLSS)()厂+(15)could be employed to represent its rheological behavior(0.1010MLSS - 0.0737)(Dentel, 1997). From Fig. 8, it can be seen that under 850μa = 1433.5 exp(-0.0554MLSS)xs-' Herschel-Bulkley model demonstrated good modeling0.1557 exp(0.0428ML SS)-1)(16)effects while Bingham model obviously deviated experi-(出)mental data; however, Bingham model could well simulateHa = (159.53MLSS + 3121.7>()+the apparent viscosity of various sludge concentrations(1.1807 - 0.0418MLSS)X 10-x(17)under 1850 s~'. Under 1000 s~', Herschel-Bulkley andBingham models were both applicable for modeling theirviscoplastic behavior. These results indicated that underμa = 0.1845 exp (0.1010MLSS)+different shear rates the fluids could show different rheo-(115.57MLSS + 993.81)x(18)logical behaviors, therefore, appropriate models should be()000202 ss 08732developed to simulate and characterize their flow behavior.Figure 8 ilustrates the plots of Eqs. (13)-(16) for three2.2.2 Relationship between sludge concentration, vis-cosity and membrane foulingdifferent shear rates versus MLSS concentrations. It canThe effects of sludge concentration in terms of MLSSbe observed in Fig. 8a that at shear rate 850 s-1 Herschel-and apparent viscosity on membrane fouling were studiedBulkley and Sisko models provided better estimations ofin this MSTD process. The method reported in publicationsapparent viscosity under different MLSS concentrations(Wang et al, 2006; Wu et al, 2007) was adopted tothan other two models. Under shear rate 1000 s-1 (Fig.determine the membrane fouling rate. Table 1 lists the8b), except for the Ostwald model, the rest models showedstatistical analysis results of the correlations of MLSSgood simulations of measured viscosity value. However,concentration, apparent viscosity and membrane foulingunder shear rate 1850 s~ I (Fig. 8c), Bingham and Siskorate in the MSTD process. It was found that MLSSmodels demonstrated better estimation results than the oth-concentration had significant correlations with membraneer two models. It was also found that under shear rate 1850fouling and their Pearson's correlation cefficient rp =S~ ) the Herschel-Bulkley model showed good agreement0.909 (P < 0.01). Apparent viscosity also had positivewith the experimental data under sludge concentrationcorrelations with membrane fouling with rp = 0.949 (P