Study of control strategy and simulation in anoxic-oxic nitrogen removal process

426PENG Yong-zhen et al .Vol.17dosage . There existed critical scope for Sso. When SNo wasfor the performance of the controller since it will determinemore than the critical scope, the effluent TN concentrationthe consumption of the extermal carbon. According to thedecreased largely with the increasing carbon dosage and theanalysis above, this study determined that the SNo set pointextemal carbon using eficiency was high. When Swo was lesswas 2.0 mg/L in the following simulation.than the critical scope, the effluent TN concentration1.3 Determination of the nitrate recycle flow and thedecreased lttle or did not decrease with the increasing carbonsludge recycle flowdosage and the extemal carbon using fficiency was low. IfIn the A/O process, NOx -.N produced by nitrificationthe extermal carbon dosage was continuously increased, thewould be recycled to tlanoxic zone through the nitrateCOD load in the oxic zone would be increased, and therecycle flow and finally be reduced by denitrification. Innitrification ability in the oxic zone would be impaired .order to simplify the control stralegy, it was supposed that allthe incoming NH3-N was either transformed into NOx'-N ordirectly consumed by the biomass growth. Therefore thenitrate recycle flow could be expressed by the followingequation:! 60Qm = (1- r)Q. SwH,n/(Sw.sE- SNo)- Qm- Q..在4020-In which， SNo, AE represents the efluent NOx'-Nconcentration in the oxic zone ( mg/L); SNo represents the400Totalre 4600NOx -N concentration in the anoxic zone ( mg/L); rrepresents the NH3-N utilized by the biomass growth; SNH,inrepresents the influent NH3-N concentation ( mg/L); Qm，Qm，Q, represent the influent flow, the nitrate recycle flow,Fig. 1 Three-dimensional simulation results of BP neural networkand the sludge recycle flow ( m /d) respectively .Fig.2 shows that the efect of carbon dosage on theThe nitrate recycle flow can be controlled according toefluent COD and NOx -N concentration and TN removalthe Equation (1) . The sludge recycle flow was controlled theefficiency in the anoxic zone under total recirculating rate ofsame as the influent flow.2. .97. When carbon dosage varied from 0 mg/L to 221 .582 Controller design and simulation analysismg/L，Swo decreased from 30.51 mg/L to 1.92 mg/L, theThe IAWQ ASM No. 1 is a widely used model foCOD concentration increased from 66.22 mg/L to 82.78 mg/simulation of biological nitrogen removal in wastewaterL，the TN removal effciency improved from 45.2% totreatment plants( Aspegren, 1992; Lukasse, 1998). Due to73.1 %. Continuously increasing carbon dosage to 697 mgL, SNo decreased to 0 m/L, but the TN removal eficiencyits complexity it is difcult to use the ASM No.1 directly foronly improved 3.27% with double dosage of the extermalcontroller design. In order to design the controller of the flowrate of extermal carbon addition， a simplifed model wascarbon source. With the carbon dosage increasing， CODderived with the following assumptions according to theconcentration in the anoxic zone increased from 82.78 mg/Lunderstanding and analysisof the A/O process. Theto 185. 12 mg/L which added burden to the oxic zone.assumptions were :1 ) the anoxie zone was completely85mixed; (2) there was no denitrification in the sttling tank;1830(3) the NOx -N concentration in the nitrate recycle and the省160sludge recycle was equal to the effluent NOx"-N14070宣concentration; (4) the oxygen in the anoxic zone was always昌10052zero; (5) the oxygen that enters the anoxic zone via the50influent, the nitrate recycle and the sludge recycle was not;0 F-0- NOx-Ns5 意considered. Based on these bhypotheses a model of extermal50 Fcarbon addition was developed.蕾40-。-1- TN removal fficiency50 ?According to themass balance theory ( input-output +reaction = accumulation) of denitrification， the variations of0100200300400500600700the NOx -N and the biodegradable COD could be expressedCarbon dosagr, mgCOD/L influentby the following equations:Fig.2 Efeet of carbon dosage on the ffuent C0D, NOx -N and TN removaldS. _ iumshs..y. PH.udi =VSsin-efficieney in anoxie zone with constant total recirculating rateUnder the constant total recirculating rate， SNo.)(Swo .)n。Xp.u + U,(2)s + Ss/\KNo+ SNOdecreased with the carbon dosage increasing .When SNo中国煤化工_ 1- Yreached to about 2 mg/L，even through the carbon dosageCNMHG2. 86Y_MH.m起increased twice as much，SNo decreased lttle and the TNremoval efficiency improved lttle. And dosing too muchx(K:+ s,八Ko+ Swo) ngXB.H，external carbon would increasing the effluent COD of theQ =Qm+ Q.+ Qm,(4)anoxic zone， the COD load of the oxic zone， and theU = QeSc.operating costs. So it is important to choose the Swo set pointIn which， Ss represents the biodegradable COD27No.3Study of control strategy and simulation in anoxic-oxic nitrogen removal processconcentration in the anoxic zone ,including the extemalday(g/d); V represents the volume of the anoxic zone(m* );carbon( mg/L); Ss,ia represents the influent biodegradableμH.m represents the maximum specifie growth of theCOD concentration ( mg/L); SNo represents the NOx*-Nheterotrophic biomass(d' ); Yu represents the yield forconcentration in the anoxic zone ( mg/L); SNo. in representsheterotrophic biomass; and Ks represents the saturationthe recycle NOx -N concentration to the anoxic zone( mg/L) ;constant for biodegradable COD( mg/L).Xg.u represents heterotrophic biomass concentration in the A/2.1 Controller design0 reactor( mg/L); Qc represents the extemal carbon dosageVariations of the influent flow and composition were theflow rate (m'/d); Sc represents the concentration of thedisturbances to the control system. Fig.3 shows the controlstructure of the external carbon addition .extemal carbon( mg/L); U represents the carbon dosage pe2。Cotoller。SNo.ntQnt, erN, io Sco, inSwo_Setting tank11....Influentd| d| 0000。。o° 000。o°|00°Anoxic zoneOxic zone2m2,Fig.3 Schematie of the control structure of extemal carbon addition in the A/O processPI controller was used. Let SNo, nt be the Sso set point,two controllers both could trace the variations of the set pointK, be the proportion factor, K be the integrating factor, andwith small error, and had the good tracing ability, but PIthe control nule could be expressed by Equation (6)controller has bigger overshooting with a slight surge . Aboutthe control effects of the extermal carbon addition, when theU(t) = Kp(Sno(t) - Sso.nr(t)) +set point changed, PI controller can give a fast response withK{[(Sm(r)- su.u())dI.(6)large extent of change, so it is difficult to realize because ofthe limitation of executer. But the feed forward PI controllerIn order to quickly eliminating the efet of the inluenthas the same response speed with small extent of change, itvariations，a feed-forward part was. added to the_ PIwas easy to be realized for the executer.contoller, then we get the feed-forward PI contoller. Thecontrol nule could be expressed by Equation (7)....... SNcSNo.etU(t) = ((Q.(1) + Qm(t))Swo -Q(I)So.e(I)) - Q.(l)Ssi6(t) + Q(t)Ss(l) +K(Sm(1)- sur())+ K{[(Sw(r)- so.m())dr.(7)2.2 Simulation analysis)54The control objective was to keep Swo at the set point of,d2 mg/L. A sin wave disturbance was input with a 50% swing, x100of the reference to the infuent flow, 25% swing of thereference to the infuent biodegradable COD and 11% swing6-to the influent, ammonia nitrogen. At the same time, a cyecle-varied set point was given in order to venify the dynamicresponse and steady precision of the two controllers in thesimulation: the set point was 2.0 mg/L at the first two days,务2and was adjusted to 2.5 mg/L at the second two days, andwas adjusted to 1.5 mg/L at the third two days, and came中国煤化工back to 2.0 mg/L at the last two days.The simulation results(Fig.4 and Fig.5) show that theMHCNMHGtwo controllers both could reach the control objective and thefeed-forward PI controller had belter control effects than theFig.4 The simulation resuls of the Pl cntollerPI controller for the consideration of the disturbances of theinfluent( including the influent flow and composition). The.428PENG Yong-zhen et al.Vol.17extemal carbon dosage to the anoxic zone. Through the..... SNoanalysis of the experimental data, the demand of the effluentSNo.tquality could bemet and the carbon dosage could be optimizedby keeping Sso at the level of about 2 mg/L. Based on ASMNo.1, a simplifed mathematical model of extemal carbondosage was developed and two controllers of PI and feedforward PI were designed. Simulation results showed thafeed-forward PI controller had better dynamic response and1steady precision due to its consideration of influentdisturbances .。41,dReferences :Aspegren H，Anderson B, Nyberg U et al.. 1992. Model and sensor basedoptimization of nitrogen removal at Klangshamn wastewater trealment plant[J]. Wat Sci Tech, 26(5-6): 1315- 1323.APHA( American Pubice Halth Association)/ American Water Works Association/Water Environment Federation, 1995. Standurd methods for the examinationof water and wastewater[ M].19th ed. USA W ashingon DC.BarrosPR, Carlsson B, 1998. Iterative design of a nitrate contoller using an3-external carbon source in an activated sludge process[J]. Wat Sci Tech, 37(12): 95- -102.2-Cho J H, SungS w, Le1 B, 2002. Cascade contrl strtey for extemal cutbondosage in predeninifying process[J]. Wat Sci Tech, 45(4-5): 53- -60.IWA task group on mathematical modelling for design and operation of biologicalt,wastewater treament, 2002. Activated sludge models ASM1，ASM2 andASM3[ M]. Shanghai: Tongji University PresLindberg C F, Carlson B, 1996. Adaptive control of extermal carbon flow rate inFig.5 The simulation resulis of the fdeforwarndd PI cntolleran activated sludge process[J]. Wat Sei Tech, 34(3-4): 173- 180.Lukasse LJS, Keesman K J, Klapwijk A et al.. 1998. Optimal control of N-3 Conclusionsremoval in ASPs[J]. Wat Sei Tech, 38(3): 255- 262Samuelsson P, Carlsson B, 2001. Feedforward control of the external carbon flowIn this paper, a control stralegy to control SNo of the A/rate in an acivated sludge poess[J]. Wat Sei Tech, 43(1): 115- -122.0 process was determined by adjusting the flow rate of the( Received for review April 13, 2004. Accepted Seplember 22, 2004)中国煤化工MYHCNMHG