Contribution to dynamic simulation of activated sludge wastewater treatment plants

Transcription

1 Contribution to dynamic simulation activated sludge wastewater treatment plants Sara Pinto Instituto Superior Técnico (IST-UTL), Lisbon, Portugal, Filipa Ferreira Instituto Superior Técnico (IST-UTL), Lisbon, Portugal, António Albuquerque Universidade da Beira Interior (UBI), Covilhã, Portugal, ABSTRACT In Portugal, many wastewater treatment plants are presently operated according to predetermined schemes, with few concerns to variations activated sludge process and without optimizing its performance to achieve a better effluent quality. Little attention has been given to activated sludge models as powerful tools for wastewater treatment process understanding, design, control and optimization. The main goal this study is to contribute to understanding activated sludge process and to simulation organic carbon removal based on ASM3. Monitoring campaigns were conducted in order to characterize composition flows from seven different sections wastewater treatment plant and to investigate dissolved oxygen concentrations in biological reactors. The dynamic simulation WWTP was confronted with several limitations related to treatment plant performance and desired stability for modeling was not verified. An alternative academic approach was performed as an attempt to understand consequences different operation methodologies, in terms process efficiency. Keywords: activated sludge; ASM3; modeling; wastewater treatment. NOTATION ASM3 Activated Sludge Model Nº 3 VSS Volatile suspended solids BOD 5 Biochemical oxygen demand after 5 days WWTP Wastewater treatment plant COD Chemical oxygen demand DO Dissolved oxygen GPS-X General Purpose Simulator Q inf Influent flow IWA International Water Association Q es Flow excess sludge RAS Return activated sludge Q ras Flow return activated sludge TKN Total Kjeldahl nitrogen N total Total concentration nitrogen TSS Total suspended solids P total Total concentration nitrogen 1

2 INTRODUCTION During last three decades increased public awareness about quality waters and management hydric resources has considerably increased requirements imposed on treatment plants, reflected in more stringent effluent regulations. The activated sludge process is required to meet effluent standards while minimizing investment, sludge production and energy consumption. A problem inherent in achieving this aim is that activated sludge process is highly dynamic due to variations in influent flow rate and its composition. Many wastewater treatment plants are presently operated according to predetermined schemes with very little consideration to se variations. In general, combination a better understanding dynamic behavior processes, efficient monitoring control systems, adequate mamatical models and identification model parameters, have a significant potential for solving operational problems and meet effluent quality standards at low operational costs. Within last two decades, one most significant advances in wastewater treatment has been development dynamic mamatical models capable describing physical, chemical and biological removal pathways that occur in a wastewater treatment processes. Specifically, Activated Sludge Model Nº3 (Gujer et al., 2) was proposed in order to correct some deficiencies former ASM1 (Henze et al., 1987) and become new standard model. This model relates wastewater treatment processes to oxygen consumption, sludge production, nitrification and denitrification which take place in activated sludge systems during treatment primary domestic wastewater. ASM3 uses concept storage-mediated growth heterotrophic organisms assuming that all readily biodegradable substrate is first taken up and stored in an internal polymer component (Mhlanga et al., 29). It also includes an endogenous respiration process to describe all forms biomass loss and energy requirements not associated with growth (e.g. maintenance, lysis, predation and decay). METHODOLOGY Description Valhelhas wastewater treatment plant The Valhelhas WWTP is located in district Guarda, Portugal, and receives an average 1187 m 3 /d domestic wastewater. The treatment scheme Valhelhas is illustrated in Figure 1. Influent wastewater is screened and flow rate is measured, before going through aerated grit/grease removal systems. After this pretreatment, influent is divided over two parallel oxidation ditches (each having a depth 3 m and a volume 147 m 3 ) for biological activated sludge treatment, followed by secondary clarification in two settling tanks (each having a diameter 9 m and a volume m 3 ). The final effluent is discharged into a nearby stream after a disinfection step. The underflow from clarifiers flows back to oxidation ditches. Excess sludge is thickened prior to dewatering. 2

3 Figure 1 Flow diagram liquid phase Valhelhas WWTP Constraints Several events and operational problems occu urred in WWTP before campaign. Examples see problems are listed as follows: and duri ing moni itoring a technical failure in one aera ator (D-E in Figure 1) occurred and during prev vious month campaign reactor operated with only one aerator, compromising aeration and stirring conditions; a discharge dry sludge (about 7-8 m 3 ) upstream Valhelhas treatment plan nt was verified, as a result deactivation old Manteigas WWTP. Consequently, sand filters clog gged and disinfection step had to be interrupted; ree is settling slud dge in oxid dation ditches when aerators are turn ned f, due to an inefficient stirri ing system. Foa aming sludge can be also obse erved at surface mixed liquor, as shown in Figure 2; sensorss responsible for control DO in oxidation ditches float at surface mixed liquor. This influences DO measurements and consequently aeration cycles; higher values are me easured and do not represent environmental conditions at bottom ditch; sludge in one clarifiers tends to rise (see also Figu ure 2); control excesss sludge removal from bottom clarifiers depends on experience operator and refore high solids retention times may occur; finally, during campaign due to unknown reasons, one oxidation ditch hes was ope erating under continuous aeration, having both aerators working simultaneously. Figure 2 Foaming sludge in oxidation ditch (left) and rising sludge in clar rifierr (right) 3

4 Monitoring campaign The monitoring campaign took place in 14/15 December 29 and included collection (during 1 day) and analyses seven discrete samples at several locations WWTP, as indicated in Figure 1. The influent and final effluent samples were analyzed for following parameters: TSS, VSS, COD, BOD 5, N total, P total and fecal coliforms. Samples collected from intermediate sections WWTP were analyzed for TSS, VSS and COD. Influent and recirculation flow data were obtained from treatment plant operation logbooks. The aeration system in each ditch (consisting two mechanical surface aerators) was designed to ensure good aeration and stirring conditions mixed liquor; yet it has been incapable doing so. As soon as air-f period begins, sludge starts to settle. Therefore, it was reasonable to suppose that sludge settling could enhance establishment anoxic and even anaerobic conditions, especially in bottom ditch, which would impact overall process efficiency. Assessing performance aeration system in biological reactors seemed to be great relevance. Measurements dissolved oxygen (DO) were carried out, using an Oxi 33 sensor (WTW, Germany). DO was measured at different sections each oxidation ditch (as illustrated in Figure 1) and at three different depths (corresponding to surface, medium depth and bottom), for different temperatures and/or aeration conditions. Modeling tools Model simulations were performed using stware package GPS-X (Hydromantis, Inc., 26) with ASM3 as biological model (Gujer et al., 2). The IWA one-dimensional model was used to simulate clarifier. This model considers a multi-layer approach associated to a double exponential settling function (Takács et al., 1991) to specify solids flux due to sedimentation. The influent flow was described by a BODbased model, to which input data consists on values BOD 5, TSS and TKN. Design and operational data were obtained from plant records Valhelhas WWTP, from April 28 to December 29. Limitations dynamic simulation The dynamic simulation Valhelhas WWTP was confronted with several limitations: a) scarcity good analytical and flows information, and lack flow measurement infrastructures, which were necessary for accuracy mass balances to process units; b) over dimensioning system regarding to influent characteristics (flows and loads); and c) deficient operation. The desired stability for modeling was not verified and consequently, simulation this WWTP as it was in operation was not possible. Alternatively, an academic approach was carried out as an attempt to understand consequences, in terms process efficiency, considering following aspects: only one line in operation, in order to simulate process efficiency in case maintenance one oxidation ditch; characteristics wastewater influent (depicted in Figure 2) as typical and representative; performance biological reactor as a complete mixed tank with a DO setpoint control 2 g O 2 /m 3 ; 4

5 recirculation RAS and remove excess sludge at constant rates 6% and 2% wastewater influent, respectively; dry wear conditions and liquid temperature 1 ºC; period simulation 1 day; default values ASM3 for kinetic and stoichiometric coefficients. RESULTS AND DISCUSSIONS Influent wastewater characterization Figure 3 shows analytical results (measured and interpolated values) wastewater obtained during campaign and influent flow. As it can be seen from figure, period with higher inflow is from 13: to 18:. This may be due to fact that treatment plant serves mainly a rural population and has no industrial contribution. Concerning wastewater composition, it was noticed that proportion between BOD 5, COD and TSS was rar atypical in comparison with usual values raw wastewater (Henze, 1997). This is consistent with historical analytical results provided from records WWTP, where concentrations BOD 5 and TSS are considerably low, while concentration COD is usually high (data not shown). During period from 22: to 1:, wastewater was observed to be hard biodegradable, i.e. with BOD 5 /COD lower than.2. It is believed that readily biodegradable organic matter is consumed during conveyance wastewater in sewers (between 3 to 4.5 h), as a result microbial activity. Concentration [g/m 3 ] T [h:min] Average influent flow [m 3 /h] Average influent flow BOD5 COD TSS VSS N total P total Figure 3 Concentrations influent wastewater components and average influent flow during campaign 14/15 December, 29 (fulfilled points: measured values; unfulfilled points: estimated values) Measurements dissolved oxygen in oxidation ditches The results measurements DO presented in Table 1 relative to 17 December were measured during air-on period while those relative to 11 December were measured during air-f period; values section G correspond to measurements just after 2 min. aeration, which explains higher DO concentrations. The DO concentration at 13 ºC was very low and in most cases near to none. Although results suggest that anaerobic conditions might occur in ditch, this is not completely clear; measurements reduction-oxidation (or redox) potential should be carried out to identify which redox reactions occur within aquatic environment. 5

6 It was also observed that during 1-day campaign that Oxidation Ditch 2 was operating under continuous aeration conditions having both aerators working simultaneously, as can be depicted from Table 1 (for 15 December). The aeration control is automatic but aerators did not respond as programmed to DO control, concentration which (approx. 8.5 g O 2 /m 3 ) was substantially above setpoint (2 g O 2 /m 3 ). Consequently, environmental anoxic conditions could not be established and denitrification might not have occurred; this may have caused an increase nitrate in final effluent. Measurements relative to 17 December correspond to air-on period under normal aeration conditions. Measurements redox potential under normal operation conditions would be necessary in order for a full characterization aerobic/anoxic/anaerobic dynamics in biological reactors to be possible. Table 1 Results measured dissolved oxygen in oxidation ditches during campaign and in accordance with sections measurement as indicated in Figure 1 (n.a.: not assessed) Depth (from surface water) Section measurement OXIDATION DITCH 1 OXIDATION DITCH 2 Dissolved Oxygen Section Dissolved Oxygen measurement (T=13 ºC) (T=11.1ºC) (T=11.1 ºC) (T=1.9 ºC) [m] [g O2/m 3 ] [g O2/m 3 ] [g O2/m 3 ] [g O2/m 3 ] A I B J n.a C K D L E M F N G O H P n.a Analysis treatment process efficiency Table 2 summarizes monitoring results wastewater influent and final effluent relative to campaign. In general, average concentration values COD, BOD 5, TSS, N total, P total and fecal coliforms in effluent significantly exceed limit values for emission. Also, maximum percentage component removal is considerably lower than legal requirements, except in case BOD 5 which percentage removal fits range 7-9%. The concentration fecal coliforms in effluent was high above legal requirements. However, this can be explained by fact that filtration/uv disinfection step was deactivated during campaign due to clogging sand filters. 6

7 From Table 2 it is evident ent that efficiency TSS removal is very low, which is consistent with problem resuspension slud dge in clarifiers (see Figure 2). The effluent samples collected at 23: and 2: revealed ealed very high concentrations ons total nitrogen. It is believed that this is due to a measurement error, sinc ce no significant nt alterations atio were observed on or components. Table 2 Summary mary measurements men wastewater ater influent nt and final effluent carried out during campaign 14/15 December at Valhelhas WWTP PARAMETER UNITS Average St.Dev. Maximumm Minimum Average St.Dev. Maximumm Minimum ph BOD5 g/ /m COD g/ /m 3 TSS g/ /m 3 VSS g/m 3 Ntotal N g/m 3 WASTEWATER INFLUENT (1) EFFLUENT (7) Ptotal g/m Fecal coliforms MPN/1 ml 4.9E+7 3.7E+7 9.2E+7 2.1E+6 1.1E+7 6.9E+6 1.6E+7 1.3E+6 In general, analytical al results from monitoring ng campaign aign indicate that process efficiency WWTP is stro ongly related to events ent and operational problems previously described. Model application The layout Valhelhas WWTP (shown in Figu ure 4) corresponds onds to a simplified system, where only one biological reactor and one clarifier were mod deled. Figure 4 Simplified layout Valhelhas lhas WWTP used for modeling The results s obtained from dynamic model ASM3 are presented in Figure 5, respectively to operation flows, concentrations centra ns TSS S and DO in biological reactor, r, solids retention time and global concentrations in final effluent. These e results are intended ndedd to represent esent an example model application, considering however real inf nfluent data and design parameters ame rs (e.g. recirculation ion and removal excess sludge rates, concentration total suspended d solids mixed liquor). Moreover, y consist an attempt to understand consequences, in terms processs efficiency, operation with only one line. In general, discharge final effluent would be in compliance with legal requirements relatively to component concentrations cent tions and/or component one removal. However, concentrations 7

8 total nitrogen in effluent seemed to be higher than what was expected, which may be due to atypical characteristics wastewater influent used as input data. Q [(m 3 /d] Flows TSS [g/m 3 ] TSS and DO in Mixed Liquor DO [g O 2 /m 3 ] SRT [d] Solids Retention Time T [h.min] Qinf Qras Qes Qeff T [h:min] Concentration [g/m 3 ] TSS DO Concentrations in final effluent T [h:min] T [h:min] COD TSS BOD5 N total Figure 5 Results dynamic simulation with ASM3 (T=1 ºC), considering: 1 line; Qras/Qinf=.6; Qes/Qinf.2 CONCLUSION The dynamic simulation Valhelhas WWTP was confronted with several limitations (e.g. scarcity good analytical and flows information; and deficient operation). Consequently, desired stability for modeling was not verified and simulation this WWTP as it was in operation was not possible. Alternatively, an academic approach was carried out as an attempt to understand consequences, in terms process efficiency, considering different operation methodologies and a simplified layout. As a result, global quality final effluent could oretically be improved and operation costs minimized if only one treatment line was used. AKNOWLEDGEMENTS The financial support Fundação para a Ciência e Tecnologia (FCT) as part MOGIS project (reference number PPCDT/AMB/56349/24) is gratefully acknowledged. 8

9 REFERENCES Gujer, W., Henze, M., Mino, T., & van Loosdrecht, M. (2). Activated Sludge Model Nº 3. Scientific and Technical Report Nº9. IWA, London. Henze, M., Grady, C. J., Gujer, W., Marais, G., & Matsuo, T. (1987). Activated Sludge Model Nº 1. Scientific and Technical Report Nº1. IAWPRC, London. Hydromantis, Inc. (26). GPS-X technical reference: Version 5.. Hydromantis, Inc. Mhlanga, F., Brouckaert, C., Foxon, K., Fennemore, C., Mzulwini, D., & Buckley, C. (29). Simulation a wastewater treatment plant receiving industrial effluents. Water SA, 35 (4), Spanjers, H., Vanrolleghem, P., Olsson, G., & Dold, P. (1998). Respirometry in control activated sludge process: Principles. IAWQ Scientific and Technical Report Nº 7. IAWQ, London. Takács, I., Patry, G., & Nolasco, D. (1991). A dynamic model clarification thickening process. Water Research, 25 (1),