Downloaded from orbit.dtu.dk on: Dec 2, 217 A systematic methodology for controller tuning in wastewater treatment plants Mauricio Iglesias, Miguel; Jørgensen, Sten Bay; Sin, Gürkan Publication date: 212 Document Version Publisher's PDF, also known as Version of record Link back to DTU Orbit Citation (APA): Mauricio Iglesias, M., Jørgensen, S. B., & Sin, G. (212). A systematic methodology for controller tuning in wastewater treatment plants. Poster session presented at 8th IFAC Symposium on Advanced Control of Chemical Processes, Singapore, Singapore. General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. Users may download and print one copy of any publication from the public portal for the purpose of private study or research. You may not further distribute the material or use it for any profit-making activity or commercial gain You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Controller tuning in wastewater treatment plants Miguel Mauricio-Iglesias Sten Bay Jørgensen Gürkan Sin
Introduction and motivation Tuning a control loop is the adjustment of its control parameters to the optimum values for the desired control response There is a large number of methods that can be used for tuning. O Dwyer, A. Handbook of Controller Tuning Rules Åström, K. J. and Hägglund, T. PID Controllers: Theory, Design and Tuning Flow (m3/d) Wastewater treatment plants (WWTP) can be tuned according to any of those methods. They use mostly PI controllers 4 2 2 4 6 8 1 12 14 Time (days) 2 Continuous disturbances in the inflow The purpose of this contribution is to use process engineering knowledge of WWTP and its influent dynamics to improve the tuning of the controllers!
Overview of the presentation 1) Presentation of the WWTP 2) Methodology. Example 3) Process review 4) Open loop analysis 5) Closed loop design 6) Evaluation 7) Conclusion 3
Model WWTP Benchmark simulation model 1 4 Provide basic information about the plant capacity. Typically flow rate Typical effluent requirement. Etc. This is Copp 22
Methodology Methods/steps Screening of disturbances Influent characterization. Time series analysis, spectral decomposition Variable scaling and linearization Control objectives Process review Analysis of disturbances Analysis of loop interations Open loop assessment Loop shaping Tuning of parameters (e.g. PI) Closed loop design Model-based simulation Experimental evaluation Evaluation 5 Implementation
Methodology Methods/steps Screening of disturbances Influent characterization. Time series analysis, spectral decomposition Variable scaling and linearization Control objectives Process review Analysis of disturbances Analysis of loop interations Open loop assessment Loop shaping Tuning of parameters (e.g. PI) Closed loop design Model-based simulation Experimental evaluation Evaluation 6 Implementation
Process review. Screening of disturbances 11 variables Wastewater -soluble inert organic matter -readily biodegradable substrate -particulate inert organic matter -slowly biodegradable substrate -active heterotrophic biomass -total ammonium nitrogen -soluble biodegradable organic nitrogen -part. biodegradable organic nitrogen -alkalinity -total suspended solids -total flowrate S I S S X I X S X BH S NH S ND X ND S ALK T SS Q 7
Process review. Screening of disturbances 11 variables Wastewater -soluble inert organic matter -readily biodegradable substrate -particulate inert organic matter -slowly biodegradable substrate -active heterotrophic biomass -total ammonium nitrogen -soluble biodegradable organic nitrogen -part. biodegradable organic nitrogen -alkalinity -total suspended solids -total flowrate S I S S X I X S X BH S NH S ND X ND S ALK T SS Q 8 Sin et al. 211
Process review. Influent characterization & dynamics Dry weather 5 4 Daily pattern 5 4 Monday-Friday Weekly pattern Weekends Flow (m3/d) 3 2 Flow (m3/d) 3 2 1 1 1 2 3 4 Time (h) 1 2 3 4 5 6 7 Time (days) 9 Gernaey et al. 211
Process review. Influent characterization Flow (m3/d) 4 2 2 4 6 8 1 12 14 Time (days) Energy (db) Energy (db) 4 2-2 1 2 3 4 5 6 2 1 Period(h) S NH 1 2 3 4 5 6 Period(h) Spectral decomposition Flow in Energy (db).2.15.1.5 Principal component analysis Main periods associated with influent dynamics PCA spectral decomposition influent 8, 12 and 24 h 5 1 15 2 25 3 35 4 Period (h) 1 component 2 component.8.6.4.2 1 Thornhill and Horch. 27
Table 1. Scaling factors used for modelled variables Process review. Variable scaling A number of criteria used depend critically on the scaling Disturbances (d max -d min ) for dry weather, rain and storm weather CVs y max MVs (u max -u min ) Var. S I (g COD m -3 ) S S (g COD m -3 ) X I (g COD m -3 ) Value 3 65.2 45.6 Var. X S (g COD m -3 ) X BH (g COD m -3 ) S NH (g N m -3 ) Value 193 26.5 3.1 Var. S ND (g N m -3 ) X ND (g N m -3 ) S ALK (mol m -3 ) Value 6.5 1 7 Var. T SS (g COD m -3 ) Q (m 3 d -1 ) Value 199 18446 Var. D O3 (g O m -3 ) D O4 (g O m -3 ) D O5 (g O m -3 ) Value.1.1.1 Var. kla 3 (d -1 ) kla 4 (d -1 ) kla 5 (d -1 ) Value 36 36 36 11 Skogestad and Postlethwaite 27
Process review. Plant linearisation kla 3 kla 4 kla 5 G = DO 3 47.9.92 1.97 1 DO 4 13.8s + 1 16.7 67.1 5.64 DO 5 15.8 31.7 13 Full linear model (96 states) Reduced linear model (1 state) Magnitude Magnitude Magnitude.2.15 Full Reduced.1 1-4 1-2 1 1 2.2.15 Full Reduced.1 1-4 1-2 1 1 2.3.25 Full Reduced.2 1-4 1-2 1 1 2 12
Methodology Methods/steps Screening of disturbances Influent characterization. Time series analysis, spectral decomposition Variable scaling and linearization Control objectives Process review Analysis of disturbances Analysis of loop interations Open loop assessment Loop shaping Tuning of parameters (e.g. PI) Closed loop design Model-based simulation Experimental evaluation Evaluation 13 Implementation
Open loop assessment. Disturbances Closed Loop Disturbance Gain (CLDG) CDLG = G G G d 1 ( s) ( s) ( s) Magnitude 1 1 1 DO3 DO4 DO5 Ss 14 CLDG = δ i < 1 The disturbance effect is lower than y max CLDG = δ i >1 The disturbance effect is higher than y max Need of control action! g i c i > δ i Hovd and Skogestad 1994 Magnitude Magnitude 1-1 1 1 1 1-1 1 1 1 1-1 1-2 1-1 1 1 1 DO3 DO4 DO5 SNH 1-2 1-1 1 1 1 DO3 DO4 DO5 Q 1-2 1-1 1 1 1
Open loop assessment. Interactions λ Relative Gain Array (RGA) ij yi = yi λ ij = u u j j u =, k j y k k =, k i Open loop gain Closed loop gain Magnitude Magnitude Magnitude 1 1 1 1-1 1 1 1 1-1 1 1 1 1-1 Loop 1 1-2 1-1 1 1 1 1 2 1 3 Frequency (rad/d) Loop 2 λ11 λ12 λ13 1-2 1-1 1 1 1 1 2 1 3 Frequency (rad/d) Loop 3 λ21 λ22 λ23 1-2 1-1 1 1 1 1 2 1 3 Frequency (rad/d) λ31 λ32 λ33 15
Process review. Plant linearisation kla 3 kla 4 kla 5 G = DO 3 47.9.92 1.97 1 DO 4 13.8s + 1 16.7 67.1 5.64 DO 5 15.8 31.7 13 Full linear model (96 states) Reduced linear model (1 state) Magnitude Magnitude Magnitude.2.15 Full Reduced.1 1-4 1-2 1 1 2.2.15 Full Reduced.1 1-4 1-2 1 1 2.3.25 Full Reduced.2 1-4 1-2 1 1 2 16
Open loop assessment. Interactions Performance Relative Gain Array (PRGA) PRGA =Γ= G G 1 ( s) ( s) Dependent on scaling Suitable for one-way interactions Magnitude Magnitude Magnitude 1 1 1 1-1 1 1 1 1-1 1 1 1 1-1 Loop 1 1-2 1-1 1 1 1 1 2 Loop 2 1-2 1-1 1 1 1 1 2 Loop 3 1-2 1-1 1 1 1 1 2 γ 11 γ 12 γ 13 γ 21 γ 22 γ 23 γ 31 γ 32 γ 33 17 Hovd and Skogestad 1992
Methodology Methods/steps Screening of disturbances Influent characterization. Time series analysis, spectral decomposition Variable scaling and linearization Control objectives Process review Analysis of disturbances Analysis of loop interations Open loop assessment Non parametric loop design Tuning of parameters (e.g. PI) Closed loop design Model-based simulation Experimental evaluation Evaluation 18 Implementation
Closed loop design. Analysis of tuning parameters 3 PI controllers Reported parameters Vanrolleghem and Gillot 22 Loop 1 Loop 2 Loop 3 K p.28.28.28 τ Ι (min) 14.4 14.4 14.4 19
Closed loop design. Analysis of tuning parameters CLDG for flow disturbance Magnitude 1 2 1 δ 1 g 11 c 1 Loop 1 CLDG = δ i >1 at some frequencies g i c i > δ i? Magnitude 1 2 1 δ 2 g 22 c 2 1 Loop 2 Magnitude 1-2 1-1 1 1 1 1 2 δ 3 g 1 33 c 3 1-2 1-1 1 1 1 Loop 3 2 Hovd and Skogestad 1992
Closed loop design. New tuning parameters min Μ S = f(k p,τ Ι ) for Magnitude 1 2 1 δ 1 g 11 c 1 CLDG for flow disturbance Loop 1 s.t. g i c i >1.1 δ i for ω < ω B New parameters Loop 1 Loop 2 Loop 3 K p.56.56.56 τ Ι (min) 1.1 1.1 1.1 Magnitude Magnitude 1 2 1 1 2 1 1-2 1-1 1 1 1 δ 2 g 22 c 2 1-2 1-1 1 1 1 δ 3 g 33 c 3 Loop 2 Loop 3 1-2 1-1 1 1 1 21
Methodology Methods/steps Screening of disturbances Influent characterization. Time series analysis, spectral decomposition Variable scaling and linearization Control objectives Process review Analysis of disturbances Analysis of loop interations Open loop assessment Non parametric loop design Tuning of parameters (e.g. PI) Closed loop design Model-based simulation Experimental evaluation Evaluation 22 Implementation
Evaluation 7 days dry weather + 7 days storm weather DO g/m 3 kla t+1-kla t d -1.2.1 -.1 Vanrolleghem et al 22 This work -.2 257 8 9 1 11 12 13 14 This work 2 Vanrolleghem et al. 22 15 1 5 7 8 9 1 11 12 13 14 1 2 3 4 5 6 7 Time (days) 23 Copp 22
Evaluation 7 days dry weather + 7 days storm weather DO g/m 3 kla t+1-kla t d -1 Integral of absolute error (mgo 2 d L -1 ) Total Variance of MV (d -1 ) DO 3 DO 4 DO 5 kla 3 kla 4 kla 5 Dryweather influent Vanrolleghem &Gillot (22).19.326.315 14. 31.5 31.3 This work.2.34.32 16.1 35.1 34.7 Storm weather influent Vanrolleghem &Gillot (22).193.313.323 13.2 31.5 33.3 This 24 work Copp 22.21.32.33 15.3 31.5 33.3.2.1 -.1 Vanrolleghem et al 22 This work -.2 257 8 9 1 11 12 13 14 This work 2 Vanrolleghem et al. 22 15 1 5 7 8 9 1 11 12 13 14 1 2 3 4 5 6 7 Time (days)
Evaluation 7 days dry weather + 7 days storm weather 25 2 15 1 5 DO g/m 3 kla t+1-kla t d -1 Integral of absolute error (mgo 2 d L -1 ) Total Variance of MV (d -1 ) DO 3 DO 4 DO 5 kla 3 kla 4 kla 5 Dryweather influent Vanrolleghem &Gillot (22).19.326.315 14. 31.5 31.3 This work.2.34.32 16.1 35.1 34.7 Storm weather influent Vanrolleghem &Gillot (22).193.313.323 13.2 31.5 33.3 This 25 work Copp 22.21.32.33 15.3 31.5 33.3.2.1 -.1 This work Vanrolleghem et al 22 Vanrolleghem et al. 22 This work -.2 257 8 9 1 11 12 13 14 This work 2 Vanrolleghem et al. 22 15 1 8 8.2 8.4 8.6 8.8 9 1 1.2 1.4 1.6 1.8 2 5 Time (days) 7 8 9 1 11 12 13 14 1 2 3 4 5 6 7 Time (days)
Evaluation 7 days dry weather + 7 days storm weather 25 2 15 1 5 DO g/m 3 kla t+1-kla t d -1 Integral of absolute error (mgo 2 d L -1 ) Total Variance of MV (d -1 ) DO 3 DO 4 DO 5 kla 3 kla 4 kla 5 Dryweather influent Vanrolleghem &Gillot (22).19.326.315 14. 31.5 31.3 This work.2.34.32 16.1 35.1 34.7 Storm weather influent Vanrolleghem &Gillot (22).193.313.323 13.2 31.5 33.3 This 26 work Copp 22.21.32.33 15.3 31.5 33.3.2.1 -.1 This work Vanrolleghem et al 22 Vanrolleghem et al. 22 This work -.2 257 8 9 1 11 12 13 14 This work 2 Vanrolleghem et al. 22 15 1 8 8.2 8.4 8.6 8.8 9 1 1.2 1.4 1.6 1.8 2 5 Time (days) 7 8 9 1 11 12 13 14 1 2 3 4 5 6 7 Time (days)
Conclusions We developed a systematic methodology to formally approach the tuning of controllers in WWTP It takes into account the knowledge of disturbances in WWTP & user specified performance requirement (DCVs, deltaymax) Information about the inflow is included to improve the tuning It requires a model. Which is commonly used in WWTPs process design and operation. It targets mainly existing plants but can be also used at design and planning stage. The newly tuned controller showed superior performance Some upcoming ideas Performance evaluation of the control structure including uncertainty, i.e. the combination of sensors and actuators, and how these relate to the control objectives 27
Controller tuning in wastewater treatment plants Funded by Danish Agency for Science, Technology and Innovation through the Research Centre for Design of Microbial Communities in Membrane Bioreactors (9-6723) for funding of the project Miguel Mauricio-Iglesias Sten Bay Jørgensen Gürkan Sin mim@kt.dtu.dk