Đang tải... (xem toàn văn)
bài bÁO khoa học về quá trình oxy hoá bài bÁO khoa học về quá trình oxy hoábài bÁO khoa học về quá trình oxy hoábài bÁO khoa học về quá trình oxy hoábài bÁO khoa học về quá trình oxy hoábài bÁO khoa học về quá trình oxy hoábài bÁO khoa học về quá trình oxy hoá
Water Research 98 (2016) 84e97 Contents lists available at ScienceDirect Water Research journal homepage: www.elsevier.com/locate/watres An integrated mathematical model for chemical oxygen demand (COD) removal in moving bed biofilm reactors (MBBR) including predation and hydrolysis n b, Javier R Viguri b, * Marta Revilla a, Berta Gala a SNIACE, Carretera de Ganzo S/N, Torrelavega, 39300, Cantabria, Spain Green Engineering & Resources Research Group (GER), Department of Chemical and Process & Resources Engineering, ETSIIT, University of Cantabria, Avenida Los Castros s/n, Santander, 39005, Cantabria, Spain b a r t i c l e i n f o a b s t r a c t Article history: Received 13 December 2015 Received in revised form March 2016 Accepted April 2016 Available online April 2016 An integrated mathematical model is proposed for modelling a moving bed biofilm reactor (MBBR) for removal of chemical oxygen demand (COD) under aerobic conditions The composite model combines the following: (i) a one-dimensional biofilm model, (ii) a bulk liquid model, and (iii) biological processes in the bulk liquid and biofilm considering the interactions among autotrophic, heterotrophic and predator microorganisms Depending on the values for the soluble biodegradable COD loading rate (SCLR), the model takes into account a) the hydrolysis of slowly biodegradable compounds in the bulk liquid, and b) the growth of predator microorganisms in the bulk liquid and in the biofilm The integration of the model and the SCLR allows a general description of the behaviour of COD removal by the MBBR under various conditions The model is applied for two in-series MBBR wastewater plant from an integrated cellulose and viscose production and accurately describes the experimental concentrations of COD, total suspended solids (TSS), nitrogen and phosphorous obtained during 14 months working at different SCLRs and nutrient dosages The representation of the microorganism group distribution in the biofilm and in the bulk liquid allow for verification of the presence of predator microorganisms in the second reactor under some operational conditions © 2016 Elsevier Ltd All rights reserved Keywords: Mathematical model Biological treatment Moving bed biofilm reactor (MBBR) Hydrolysis Predation Pulp and viscose wastewater Introduction A moving bed biofilm reactor (MBBR) is a type of biofilm technology used for wastewater treatment (Kaindl, 2010) In such a reactor, the biomass grows as a biofilm on small carrier elements that move around in the reactor maintaining the biomass per unit volume at a high level In aerobic processes, the biofilm carrier movement is effected by blowers Therefore, the MBBR process has the advantages of attached and suspended growth systems (Qiqi et al., 2012) A key characteristic of MBBR reactors is not only the increase in the effective carrier area that thereby directly contributes to a larger biofilm but also that it allows good conditions for the transport of substrates into the biofilm (Masic et al., 2010) Because of the extremely compact high-rate process, the hydraulic retention time (HRT) in the MBBR is low (Ødegaard, 2006) * Corresponding author E-mail address: vigurij@unican.es (J.R Viguri) http://dx.doi.org/10.1016/j.watres.2016.04.003 0043-1354/© 2016 Elsevier Ltd All rights reserved Moreover, it is a continuously operating, non-cloggable biofilm reactor with no need for backwashing, low head-loss and a high specific biofilm surface area (Rusten et al., 2006) MBBR technology has been successfully applied to many types of wastewater including paper mill wastewater (Hosseini and Borghei, 2005), pharmaceutical industry wastewater (Lei et al., 2010), municipal wastewater (Rusten et al., 1998), and fish farm wastewater (Rusten et al., 2006) and has been utilized under aerobic and anoxic conditions (Barwal and Chaudhary, 2014; Borkar et al., 2013) Different applications require different configurations using one or more reactors in-series for COD removal, nitrification and nutrient removal (Ødegaard, 1999) The type of microorganisms in these reactors depends on the conditions under study such as the origin of the wastewater, the treatment process, and the nutrient dosage, among others Modelling is an important step for the synthesis, design and decision making related to wastewater treatment processes For biological wastewater treatment, a mathematical model can be M Revilla et al / Water Research 98 (2016) 84e97 used to predict the performance of a biological treatment plant, to determine important variables and critical parameters and/or to help with troubleshooting A model that describes the MBBR process must include the biological processes in the biofilm and the bulk liquid because the biomass exists in two forms, suspended and attached to a carrier For general purposes, the biofilm model by Wanner and Gujer is a great tool for understanding biofilm processes in a quantitative manner (Wanner, 1996) Moreover, this type of model is generally adequate to describe a macroscopic conversion (Wanner et al., 2006) in a biofilm system and gives a reasonable description of the layered biofilm structure (van Loosdrecht et al., 2002; Masic, 2013) Biological processes describing the interaction between autotrophic and heterotrophic microorganisms are commonly considered by activated sludge models (ASM) The ASM models consider bacteria as the sole active biomass The activities of all other microbial community members (protozoa, metazoa, phages, etc.) are hidden in a simple decay process responsible for the reduction of active biomass This decay process is the sum of several independent processes such as maintenance, lysis due to phage infection and predation (van Loosdrecht and Henze, 1999) The inclusion of predation is not necessary for the successful use of current activated sludge models (Moussa et al., 2005) However, the role of predators clearly affects the performance of a treatment plant and can be especially critical for obtaining a good quality effluent with low suspended solids (Tamis et al., 2011) In the moving bed process, the type of biofilm that develops depends on the organic loading rate applied (van Haandel and van der Lubbe, 2012) Kinner and Curds, 1987, examined the predators communities inhabiting RBC biofilms exposed to various organic loading rates; predators were observed mainly in compartments with low loadings Despite many studies of the microbial ecology of activated sludge systems and mathematical modelling, little work has been reported on the interaction between bacteria and other microorganisms in the microbial community of activated sludge, especially the role of protozoa (van Loosdrecht and Henze, 1999) The role of protozoa in activated sludge has been investigated by authors such as Moussa et al., 2005; Ni et al., 2009, 2011; Hao et al., 2011, who developed a simple procedure for the determination of the activity of these predators in suspended mixed cultures These authors proposed a model to describe a mixed culture in which bacteria and predators (protozoa and metazoa) coexist In this paper, the predation process is based on the studies of Moussa et al., 2005 and Hao et al., 2011, that simplify the description of the complex reality of the predator-prey relationship, including all types of predators in a single type and assuming that the predation process is a function of the bacterial concentration However, no work has included the predation phenomena in a mathematical model for an MBBR Taking into account the different origins and characteristics of wastewater that can be treated in an MBBR plant and the different possible plant configurations, a general model of an MBBR process requires the inclusion of the predation mechanism This work presents a model that considers the interaction between bacteria and predator microorganisms in the MBBR process The integrated mathematical model for MBBR proposed in this work combines the following: (i) biological processes describing the interaction between autotrophic, heterotrophic and predator microorganisms via the model of Moussa et al., 2005; (ii) a biofilm model by Wanner and Gujer, 1986; and (iii) a bulk liquid model (Masic et al., 2010) Because the proposed model can be useful for wastewaters of different origins, plant configurations and operational conditions, the SCLR values (soluble COD loading rate) 85 proposed by Ødegaard (1999) are taken into account to consider the predation growth mechanism in an MBBR reactor Similarly, the reference values proposed by Helness and Ødegaard (2005), are taken into account to consider the hydrolysis in the bulk liquid Finally, the regeneration of nutrients due to predators is also considered in the model (Lindblom, 2003) Wastewater from the pulp and paper industry is characterized by a high COD content that can range from approximately 1000 to 4200 mg/l (Swamy et al., 2011) In general, this type of wastewater contains lignin (40%), carbohydrates (40%) and extractives (20%) The activated sludge process is one of the most common systems for the biological treatment of pulp and paper industry effluent; however, the main disadvantage of an AS process is the bulking of the sludge (Rankin et al., 2007) The pre-treatment of wastewater that has a high organic load with biofilm formation systems such as MBBR is used to control the phenomenon of bulking In the pulp and paper industry, modelling of a biological treatment plant can be used to develop more efficient operational conditions and can help determine a more efficient nutrient dosage (Boltz et al., 2011; Lindblom, 2003) In this work, the proposed model is applied to a full-scale MBBR plant that treats wastewater from a cellulose and viscose industrial plant with large amounts of organic matter Integrated mathematical model for MBBR The integrated mathematical model presented in this paper is a multi-species and multi-substrate biofilm and bulk liquid model for an MBBR reactor The state variables of the integrated model proposed are composed of the concentrations of soluble compounds (Si) and particulate compounds (Xi) (Henze et al., 2000) The nomenclature for the model state variables is given in Table The integrated mathematical model takes into account biological conversion processes observed in Fig 1, which describes the transformation process and the interactions between three groups of microorganisms (i.e., autotrophs, heterotrophs and predators) The stoichiometric matrix and process rate equations for all of the processes in the integrated mathematical model can be found in Table and Table 3, respectively, and the kinetic, stoichiometric and other parameters used in the integrated model are described in Table All particulate compounds in the model have been expressed as COD fractions, except for solids Xcellulose The conversion between COD and total suspension solids (TSS) has been evaluated assuming stoichiometric conversion parameters of 0.75 and 0.90 gTSS/g COD (Boltz et al., 2011) TSS, filtered COD (CODf) and total nitrogen (TN) have not been introduced as variables but were computed from the state variables by Equations (1, and 3), respectively TSS ẳ 0:75 XI ỵ 0:75 XS ỵ 0:90 XH ỵ 0:90 XAut ỵ 0:90 Xpredators ỵ Xcellulose (1) CODf ẳ SF ỵ SA ỵ SI (2) TN ẳ SNO3 ỵ SNH4 ỵ SND (3) 2.1 Biological processes 2.1.1 Predator growth The impact of predator microorganisms has been investigated in MBBR microbial communities, and it has been found that even 86 M Revilla et al / Water Research 98 (2016) 84e97 Table State variables of the mathematical model State variable Soluble compound i SF SA SNO3 SNH4 SPO4 SI SND SO2 Particulate compounds i XH XAut XI XPredators XS Xcellulose Unit Definition Reference model g g g g g g g g COD/m3 COD/m3 N/m3 N/m3 P/m3 COD/m3 N/m3 O2/m3 Readily biodegradable organic compound Fermentation products Dissolved nitrate ion Dissolved ammonium ion Inorganic soluble Phosphorus Inert soluble organic compounds Soluble biodegradable organic nitrogen Dissolved oxygen ASM2 ASM2 ASM2 ASM2 ASM2 ASM2 ASM1 ASM2 g g g g g g COD/m3 COD/m3 COD/m3 COD/m3 COD/m3 TSS/m3 Heterotrophic microorganisms Autotrophic microorganisms Inert organic compounds Predator microorganisms Slowly biodegradable organic compounds Slowly biodegradable compounds present in the original wastewater ASM2 (Henze et al., 2000) ASM2 (Henze et al., 2000) ASM2 (Henze et al., 2000) Moussa et al., 2005 ASM1 (Henze et al., 2000) Morgenroth et al., 2002 minor operating condition changes could cause a dramatic shift in the composition of these predators (Goode, 2010; Fried et al., 2000) Authors such as Villareal et al., 1975 and Kinner and Curds, 1987 have conducted studies in which organic material is either low or the limiting substrate These authors showed that the number of bacteria increased until a maximum value was reached due to the depletion of organic material, and later, the number of bacteria (Henze (Henze (Henze (Henze (Henze (Henze (Henze (Henze et et et et et et et et al., al., al., al., al., al., al., al., 2000) 2000) 2000) 2000) 2000) 2000) 2000) 2000) decreased and that of the predators increased Consequently, in this study, the different SCLR values proposed by Ødegaard, 1999 have been considered to evaluate the presence of predators in the biofilm and the bulk liquid of an MBBR reactor, as shown in Fig Other authors such as van Haandel and van der Lubbe, 2012, used the same classification Predator growth is included in the proposed model according to Fig Flow diagram of the external and the internal soluble components, conversion processes and interactions between the three microbial group in the MBBR reactor Solid lines represent growth processes and dashed lines represent inactivation process M Revilla et al / Water Research 98 (2016) 84e97 87 Table Stoichiometric matrix for the mathematical model Compounds i/ Soluble compounds Si SO2 Particulate compounds Xi SF SA SND SNH4 SPO4 SNO3 XH XA XPredators XS XI Process Pj Y 1.Aerobic iN,BM iP,BM Y1H Y1H growth on SF iN,BM iP,BM 2.Aerobic Y1H Y1H growth on SA 1YH iN,BM iP,BM 3.Anoxic Y1H 2:86 YH growth on SF 1Y iN,BM iP,BM 4.Anoxic YH 2:86 YHH growth on SA 5.Fermentation 1 ỵ1 6.Inactivation iP;BM iP;XI f XI iN;BM iN;XI f XI A XAut 7.Aerobic iP,BM iN;BM Y1A 4:57Y YA YA growth 8.Inactivation iP;BM iP;XI f XI iN;BM iN;XI f XI Xpredators 9.Aerobic 1 ỵ YP f XI ị þ f XI iN;BM iN;XI f XI iN;BM ½YP ð1 f XI Þ iP;BM iP;XI f XI iP;BM ẵYP f XI ị growth on XH 10.Aerobic 1 ỵ YP f XI ị ỵ f XI iN;BM iN;XI f XI iN;BM ẵYP f XI ị iP;BM iP;XI f XI iP;BM ẵYP f XI ị growth on XAut 11.Inactivation iP;BM iP;XI f XI iN;BM iN;XI f XI 12 Hydrolysis ỵ1 of Xs 13 1 ỵ1 Ammonication P Pj yij P (1/day) Conversion rates ri ¼ P j yij (g/m3day) SI and Xcellulose remain constant without conversion processes Observed growth Uoi ¼ f r XH ỵ1 ỵ1 ỵ1 ỵ1 ỵ1 1 f XI f XI ỵ1 1 1 YP f XI Þ f XI f XI f XI 1 YP ð1 f XI Þ f XI 1 f XI f XI 1 i Table Process rate equations for the mathematical model Process Pj Heterotrophic microorganisms XH 1.Aerobic growth on SF Process rate equation SO2 uH SO2 ỵK O2;H SF SF SNH4 SPO4 SF ỵKP SF ỵSA SNH4 ỵKNH4;H SPO4 ỵKPO4;H XH SA SA SNH4 SPO4 SA ỵKA SF ỵSA SNH4 ỵKNH4;H SPO4 ỵKPO4;H XH 2.Aerobic growth on SA SO2 uH SO2 ỵK O2:H 3.Anoxic growth on SF O2;H hNO3 uH SO2KỵK O2;H 4.Anoxic growth on SA 5.Fermentation SF SA SNO3 SNH4 SPO4 SF ỵKF SF ỵSA SNO3 ỵKNO3 SNH4 ỵKNH4;H SPO4 ỵKPO4;H KO2;H SA SA SNO3 SNH4 SPO4 hNO3 uH SO2 ỵKO2;H SA ỵKA SF ỵSA SNO3 ỵKNO3 SNH4 ỵSNH4;H SPO4 ỵKPO4;H KO2;H KNO3 SF qfe SO2 ỵK XH O2;H SNO3 ỵKNO3 SF ỵKfe 6.Inactivation Autotrophic microorganisms XAut 7.Aerobic growth b H XH 8.Inactivation Preadator microorganisms XPredators 9.Aerobic growth on XH b A XA 10.Aerobic growth on XAut SO2 uP SO2 ỵK O2;P 11.Inactivation Hydrolysis 12.Hydrolysis of XS bP XP Ammonification 13.Ammonification of SND SO2 uA SO2 ỵK O2;A SO2 uP SO2 ỵK O2;P SNH4 SPO4 SNH4 ỵKNH4;A SPO4 ỵKPO4;A XH XH ỵXA XA XH ỵXA XH XH XA XH XA XS =XH ị Kh ẵXX XH XS =XH ịỵKX Ka SND XH In the biofilm Xi is replaced by the multiplication of fir Moussa et al., 2005, who proposed that i) the predators grow aerobically on the degradable (1-fXI) fraction of the heterotrophic and autotrophic bacteria, and ii) the predation rate is a function of the bacterial concentration 2.1.2 Hydrolysis process The hydrolysis of slowly biodegradable compounds increases the readily biodegradable soluble compounds (SF) available to bacteria (Morgenroth et al., 2002) Direct contact between slowly biodegradable compounds and microorganisms is necessary Because the model proposed in this work will be used for wastewater from the pulp and paper industry, two types of slowly biodegradable compounds have been defined: i) Xcellulose and ii) XS (Morgenroth et al., 2002) Hydrolysis of Xcellulose strongly depends on the sludge retention time (Ruiken et al., 2013) Because in MBBR reactors the sludge retention time is short and the cellulose fibres are large, it is assumed that Xcellulose is not hydrolysed and passes through the MBBR reactors unconverted Slowly biodegradable organic compounds (XS) not diffuse into the biofilm, and it is assumed that the hydrolysis takes place in €s, 1993; Larsen and the bulk liquid (Rohold and Harremoe €s, 1994) Harremoe Hydrolysis in the bulk liquid is simulated depending on the SCLR value (Helness and Ødegaard, 2005) as shown in Fig 88 M Revilla et al / Water Research 98 (2016) 84e97 Table Parameters used at the mathematical model Parameters Heterotrophic (H) Stoichiometric coefficients Yield coefficient Kinetic parameters Maximum growth rate Inactivation rate constant Saturation coefficient for growth on SF Saturation coefficient for growth on SA Saturation coefficient for fermentation of SF Saturation coefficient for phosphate Saturation coefficient for ammonium Saturation coefficient for oxygen Saturation coefficient for nitrate Maximum rate for fermentation Reduction factor for denitrification Autotrophic (A) Stoichiometric coefficients Yield coefficient Kinetic parameters Maximum growth rate Inactivation rate constant Saturation coefficient for phosphate Saturation coefficient for ammonium Saturation coefficient for oxygen Predator microorganisms (P) Stoichiometric coefficients Yield coefficients Kinetic parameters Maximum growth rate Inactivation rate constant Saturation coefficient for oxygen Hydrolysis Hydrolysis rate constant Saturation coefficient for XXS Others Stoichiometric coefficients Nitrogen content of biomass Phosphorus content of biomass Nitrogen content of inert matter Phosphorus content of inert matter Fraction of inert biomass Biofilm parameters and diffusion coefficients Diffusion coefficient of SO2within biofilm Diffusion coefficient of SFwithin biofilm Diffusion coefficient of SAwithin biofilm Diffusion coefficient of SNH4within biofilm Diffusion coefficient of SNO3within biofilm Diffusion coefficient of SPO4within biofilm Diffusion coefficient of SNDwithin biofilm Detachment coefficient Mass transfer boundary layer Biofilm density Biofilm surface area Symbol Unit Used Value YH g COD/g COD 1 0.63 uH bH KF KA Kfe KPO4,H KNH4,H KO2,H KNO3 qfe ƞNO3 Day Day1 g COD/m3 g COD/m3 g COD/m3 g P/m3 g N/m3 g O2/m3 g N/m3 gCOD/gCOD day - YA g COD/g N -1 0.4 20 0.01 0.05 0.2 0.5 0.8 Reference Henze et al., 2000 ASM2 Henze Henze Henze Henze Henze Henze Henze Henze Henze Henze Henze et et et et et et et et et et et al., al., al., al., al., al., al., al., al., al., al., 2000 ASM2 2000 ASM2 2000 ASM2 2000 ASM2 2000 ASM2 2000 ASM2 2000 ASM2 2000 ASM2 2000ASM2 2000 ASM2 2000 ASM2 0.24 Henze et al., 2000 ASM2 uA bA KPO4,A KNH4,A KO2,A Day Day-1 g P/m3 g N/m3 g O2/m3 0.15 0.01 0.5 Henze Henze Henze Henze Henze YP g COD/g COD 0.335 Lindblom, 2003 uP bP KO2,P Day-1 Day-1 g O2/m3 2.2 0.1488 0.2 Ni et al., 2011 Lindblom, 2003 Lindblom, 2003 Kh Kx Day-1 g COD/g COD 0.10 Henze et al., 2000 ASM1 Henze et al., 2000 ASM1 iN,BM iP,BM iN,XI iP,XI fXI g g g g - DO2 DF DA DNH4 DNO3 DPO4 DND m2/day m2/day m2/day m2/day m2/day m2/day m2/day 1/m day mm g COD/m3 m2 l Ll r AF N/g COD P/g COD N/g COD P/g COD 2.2 Biofilm model The biofilm model in this study is based on Wanner and Gujer (1986) (Goode, 2010; Masic, 2013), and it i) describes the dynamics and spatial distribution of the microbial species and substrates in the biofilm, ii) predicts the evolution of the biofilm thickness and iii) describes detachment of the biomass due to sloughing and shear stress The following assumptions have been made regarding the biofilm: i The biofilm density is constant with depth (Horn and Lackner, 2014) ii The introduction of a slowly biodegradable compound (Xs) is considered as a particulate compound in the biofilm (Vanhooren, 2001) iii The biofilm grows perpendicular to the substratum et et et et et al., al., al., al., al., 2000 2000 2000 2000 2000 ASM2 ASM2 ASM2 ASM2 ASM2 0.06I, IIe0.03 III 0.01 I, II e0.006 III 0.017 I,II e0.008 III 0.005 I, II e0.003 III 0.1 Calibrated parameter Calibrated parameter Calibrated parameter Calibrated parameter Henze et al., 2000 ASM2 1.75 10-4 8.3 10-5 8.3 10-5 1.36 10-4 1.28 10-4 5.44 10-5 10-5 1,000 100 20,000 479,790 Wanner and Gujer, 1986 Wanner and Gujer, 1986 Wanner and Gujer, 1986 Boltz et al., 2011 Boltz et al., 2011 Geesey, 1994 Jeppsson, 1996 Masic et al., 2010 Boltz et al., 2011 Horn and Lackner, 2014 AF ¼ 900m2/m3* % fill ca* VMBBR iv Monod kinetics are used to describe the conversion rate of a soluble compound and the growth and inactivation of the microorganism groups v The biofilm and the suspended biomass in the bulk liquid are governed by similar kinetic parameters vi The attachment rate of the suspended solids in the bulk liquid to the biofilm surface has not been considered because the net balance of solids indicates that detachment is a more significant process (Goode, 2010) 2.2.1 Mass balance for the particulate compounds by the volume fraction in the biofilm Equations (4e10) describe the mass balance for the particulate compounds (i) by volume fraction fi (t, z) in the biofilm and the boundary conditions: M Revilla et al / Water Research 98 (2016) 84e97 89 Fig The influence of SCRL values in predation and hydrolysis process in the mathematical model for MBBR reactors df i ðt; zÞ df t; zị ẳ Uoi t; zị Uot; zị f i ðt; zÞ Uðt; zÞ i dt dz (4) i ¼ S, H, Aut, I and predators X Uot; zị ẳ Uoi t; zịf i t; zị (5) to be 80% of the diffusion coefficient in water ðDW i Þ (Wanner and Gujer, 1986) The model describes the flux of soluble compounds in the biofilm according to Fick's rst law Ji t; zị ẳ Dfi dSfi t; zị dz (14) Zz Ut; zị ẳ Uot; zịdz (6) 2.3 Bulk liquid model Ut; 0ị ẳ X P fi ẳ (7) Xi r ẳ1 dLtị ẳ Ut; Lịestị dt stị ẳ l Ltị2 (8) The MBBR reactor is modelled as a perfectly mixed reactor according to Equations (15 and 16) (Masic et al., 2010) VMBBR (9) (10) dSbi tị b ẳ Q in Sin i Si Ji t; zị AF ỵ ri tị VMBBR dt (15) i ¼ F, A, NH4, PO4, NO3 and ND VMBBR dXbi tị b ẳ Q in Xin ỵ l Ltị2 AF r ỵ ri tị VMBBR i X dt (16) 2.2.2 Mass balance for the soluble compounds in the biofilm Equations (11e13) describe the mass balance for the soluble components (i) in the biofilm ðSif Þ and the boundary conditions: i ¼ S, H, Aut, I and predators 2.4 Methodology for the numerical solution of the model dSfi t; zị d2 Sfi t; zị ẳ Dfi ỵ ri t; zị dt dz2 (11) i ẳ F, A, NH4, PO4, NO3, O2, ND dSfi t; 0ị ẳ0 dz (12) i DW h b dSfi t; Lị ẳ fi Si ðtÞ Sfi ðt; LÞ dz D Ll (13) i The diffusion coefficients within the biofilm ðDif Þ are supposed The model was built using the commercial software Aspen Custom Modeler® (ACM), which allows models to be customized for specific processes The technique used to solve the system of equations is the method of lines (MOL), and the BFD1 method is the discretization method The evolution of the biofilm thickness leads to a “moving boundary” problem that requires that the biofilm thickness be normalized to as described by Wanner and Gujer (1986) The system of equations was iterated at time steps of Dt ¼ 0.1 days until 30 days to ensure that the biofilm thickness had reached a steady-state The maximum number of iterations was 100 90 M Revilla et al / Water Research 98 (2016) 84e97 2.5 Model calibration Biological wastewater treatment plants in the pulp and paper industry are designed for COD removal (Rankin et al., 2007) This enables a rather simple strategy for model calibration because only one predominant biological process exists: the degradation of organic matter (Keskitalo et al., 2010), and it is necessary to change only a few model parameters (Henze et al., 2000) In this study, the parameters iN,BM, iP,BM, iN,XI and iP,XI were adjusted at steady state with average experimental data for each scenario These four parameters are designated in Table as “calibrated parameters”, and the other parameters were obtained from the references The corresponding parameters were estimated using the Aspen Custom Modeler software, which allows rigorous models to be solved and parameters to be estimated The adjustment of the model parameters was carried out using an NL2SOL algorithm for least-square minimization of the deviation between the experimental and theoretical data Experimental section: pulp and paper full-scale MBBR plant The pulp and paper industry produces a considerable amount of wastewater of variable characteristics depending on the production process and the quality of the final product (Buyukkamaci and Koken, 2010) 3.1 Description of the full-scale MBBR treatment plant The MBBR treatment plant of the integrated cellulose and viscose manufacturing mill is shown in Fig The influent wastewater is coarsely screened to eliminate the larger solids (>6 mm) An equalization tank with a volume of 1600 m3 is used to adjust the flow rate and introduce nutrients Later, two aerobic MBBR reactors of a unit volume of 5331 m3 are employed in the treatment line Normally, the pulp and paper mill effluent contains low concentrations of nitrogen and phosphorus, especially in the readily available forms of ammonium and orthophosphate These nutrients must be added externally for efficient biological treatment (Kenny, 2010) In this study, nitrogen was added as urea with a nitrogen content of 18.4% and phosphorus as phosphoric acid with a phosphorus content of 23.7% Both were added to the equalization tank Oxygen is introduced in an MBBR reactor by means of blowers For all of the experimental conditions, the dissolved oxygen concentration (SO2) was constant in the bulk liquid at approximately g/m3 in MBBR1 and g/m3 in MBBR2 The blower aeration was controlled by a Programmable Logic Controller (PLC) Both MBBR reactors were filled to 10% (Zalakain and Manterola, 2011) with flat shaped AnoxKaldnes™ carrier media type BiofilmChip P for biofilm growth The carrier had an effective specific surface of 900 m2/m3, nominal dimensions of 45 mm mm, a weight of 174 kg/m3 and specific gravity of 0.96e1.02 g/cm3 3.2 Analytical method The dissolved oxygen (SO2) in the bulk liquid for each MBBR reactor was monitored online by an optical oxygen sensor Oxymax W COS61, and the influent flow-rate (Q) was monitored online by an electromagnetic Flow Measuring System ProlinePromag 10 W The analysis of CODf, TN, SNO3 and SPO4 was performed using cuvette tests from Hach The CODf and TN samples were previously prepared in an LT 200 Hach Lange heating block The concentration values were obtained from the Hach Lange DR 2800 photometer The TSS determination was performed after a sample of bulk liquid was filtered on a Whatman glass micro fibre filter (GC/F) The dry weight was determined after the filter was dried at 105 C and weighed on a microbalance A Leitz Wetzlar ORTHOLUX POL microscope was used to observe the biomass attached to the carriers and biomass in the bulk liquid 3.3 Stream characterization The MBBR plant operated under three different conditions (scenarios) distinguished by the origin of industrial wastewater (pulp and/or viscose), the flow rate of the influent, and the inlet concentrations of the CODf, TSS, TN, SI, SNO3 and SPO4 The total nitrogen of the influent was mostly organic biodegradable nitrogen from the added urea Scenario I ran continuously for eight months, scenario II for two months and scenario III for four months These periods were determined by industrial production considerations For the influent stream, daily grab samples were collected in scenario I, but in scenarios II and III, the sampling was 24-h mixed samples For the outlet MBBR1 and MBBR2 streams in all scenarios, grab samples were collected in situ during operation All of the samples collected were analysed to determinate the COD and TSS concentration, but the TN, SNO3 and SPO4 were analysed in half of the samples Table shows the average influent flow rate and concentrations for each scenario (i.e., stable operational conditions) The data are expressed using different reference values (q, s, c, n and p) to maintain the confidentiality of the information Even though the inlet stream originated from industrial production, the concentration of the compounds was quite stable during the entire run time in each scenario; however, variations in the inlet concentrations lower than 15% occurred in scenarios I and II and lower than 25% in scenario III A previous study using the same wastewater (Zalakain and Manterola, 2011) showed that in the influent, the higher the CODf, the higher is SI In this study, it is assumed that SI in the influent is 25% of the CODf in scenarios I and II and 15% in scenario III Results and discussion 4.1 Simulated and experimental results for the full-scale MBBR plant The simulation of the outlet stream concentration from the fullscale MBBR plant discussed in Section 3.1 for the influent stream detailed in Section 3.3 was carried out using the model proposed in Section The plant consisted of two in-series MBBR reactors Because the same type of reactors are used in the plant, the same model is used to simulate the two MBBR units Figs and show the experimental and simulated results for the CODf and TSS for MBBR1 and MBBR2, respectively, during the operation of the inlet stream treatment Good concordance between experimental and simulated values was observed, as seen in Figs and The standard deviations (SD) between the experimental and simulated concentrations of CODf and TSS are lower than 10% for the three scenarios (Table 6) The similar behaviour of the experimental (Cexp) and simulated (Csim) concentration values with time and the SD values lower than 15% obtained in the three scenarios confirm the validity of the model Fig indicates an average CODf removal percentage of approximately 42%e65% in MBBR1 and only 14e21% in MBBR2 In MBBR2 the CODf removal percentage was much lower than for MBBR1 because most of the readily biodegradable components (SF) from the influent were consumed by MBBR1 M Revilla et al / Water Research 98 (2016) 84e97 91 Fig Process Flowsheet Diagram of the MBBR plant Table Influent characterization in each proposed scenario as average values Reference values q, s, c, n and p are used to maintain the confidentiality of the information Scenario Origin Q (m3/day) TSS (g/m3) CODf (g/m3) TN (g/m3) SPO4 (g/m3) SNO3 (g/m3) SI (g/m3) I II III Pulp and Viscose Pulp and Viscose Pulp 1.1q 1.0q 0.59q 1.27s 0.68s 0.28s 4.08c 3.73c 4.67c 5.28n 3.99n 2.65n 2.14p 0.84p 0.90p 0.5n 0.5n 0.5n 1.02c 0.93c 0.70c Fig Experimental concentration of CODf in the influent (-) and experimental (Cfor MBBR1; :for MBBR2) and simulated concentrations (ee for MBBR1; …… for MBBR2 outlet streams) of CODf in the bulk liquid An important increase in the TSS in MBBR1 in all three scenarios due to cell growth and the detachment of the biomass from the carriers is observed in Fig because heterotrophic growth was the predominant process studied (Schubert et al., 2013) In scenario II, a slight increase in the TSS was observed in MBBR2; however, a nontypical slight decrease was observed in scenarios I and III in MBBR2 Table shows the average experimental concentrations of total nitrogen (TN) and inorganic soluble phosphorous (SPO4) in the bulk liquid for each scenario In scenarios I and III, the average values decreased sharply in MBBR1 and increased slightly in MBBR2 because of nutrient regeneration by the predation process Such an increase has been observed in other works such as Lindblom, 2003; Rankin et al., 2007, and Tamis et al., 2011 However in scenario II, a sharp decrease in MBBR1 occurred, but no increase was seen in MBBR2 Simulated values for TN and SPO4 in the bulk liquid were also obtained from the integrated model proposed in this study The standard deviations between the experimental and simulated concentrations of TN and SPO4 are shown in Table In the three scenarios, SD values lower than 15% were obtained for TN and SPO4, but these values are higher than the standard deviations of CODf and TSS The higher SD values are probably due to the lower number of experimental nitrogen and phosphorous samples Table shows the average experimental values of SCLR and 92 M Revilla et al / Water Research 98 (2016) 84e97 Fig Experimental concentration of TSS in the influent (-) and experimental (C for MBBR1; :for MBBR2) and simulated concentrations (ee for MBBR1; …… for MBBR2 outlet streams) of TSS in the bulk liquid Table Standard deviation (SD) between experimental and simulated outlet concentrations of CODf, TSS, TN andSPO4 in the bulk liquid of the MBBR1 and MBBR2 Working conditions Standard deviation, SD (%) Scenario I MBBR1 MBBR2 Scenario II MBBR1 MBBR2 Scenario III MBBR1 MBBR2 CODf TSS TN SPO4 6.4 8.8 4.3 5.4 15.3 11.7 11.9 12.6 3.1 8.2 8.5 9.3 - 13.7 10.4 7.2 9.1 7.9 8.7 12.2 10.9 14.6 9.7 Table The average experimental values of total nitrogen (TN), phosphorous (SPO4) and nitrate (SNO3) in the influent and the average experimental values in the bulk liquid of MBBR1 and MBBR2 outlet streams Scenario I Influent MBBR1 MBBR2 Scenario II Influent MBBR1 MBBR2 Scenario III Influent MBBR1 MBBR2 * in scenario II imply that hydrolysis and predator growth are negligible (Helness and Ødegaard, 2005; Schubert et al., 2013; Ødegaard, 1999; Villareal et al., 1975; Canale, 1973) Moreover, the presence of predator microorganisms such as ciliates was observed microscopically in the MBBR2 reactor in scenarios I and III Therefore, two MBBR reactors in-series are used in this work that can be considered as a two-stage system The first stage at MBBR1 is the bacterial stage, and the second stage at MBBR2 is the bacterial-predator stage because at this second stage, the source food is composed of the bacteria that leave MBBR1 and a low COD concentration Table shows a comparison between experimental and simulated values in MBBR2 when the predation and hydrolysis were switched on and off at steady state in scenarios I and III because predation and hydrolysis occur in these scenarios The simulated values were similar to the experimental values when the predation and hydrolysis were switched on TN (g/m3) SPO4(g/m3) SNO3(g/m3) 4.2 Simulated microorganism distribution within biofilm 5.28n 0.35n 1.03n 2.14p 0.87p 0.99p 0.5n