The Use of Wastewater Models to Manage Risk Thursday, January 23, - PDF document

1/23/2020 1 The Use of Wastewater Models to Manage Risk Thursday, January 23, 2020 1:00 –3:00 PM ET 2 1

1/23/2020 How to Participate Today • Audio Modes • Listen using Mic & S peakers • Or, select “ Use Telephone” and dial the conference (please remember long distance phone charges apply). • Submit your questions using the Questions pane. • A recording will be available for replay shortly after this webcast. 3 Today’s Moderator John B. Copp Ph.D. Primodal Inc. Hamilton, Ontario 4 2

1/23/2020 Uncertainty / Risk – Jan. 23, 2020 An MRRDC Short Course: Use of Wastewater Models to Manage Risk • Topics: • Principles of Uncertainty Evaluation • DOUT Uncertainty Analysis Framework • Case Studies • Steady State • Dynamic 5 Uncertainty / Risk – Jan. 23, 2020 An MRRDC Short Course: Use of Wastewater Models to Manage Risk • Speakers: Lorenzo Lina Bruce Peter Benedetti Belia Johnson Vanrolleghem Waterways Primodal Inc. Jacobs Université Laval 6 3

1/23/2020 Lorenzo Benedetti, Evangelina Belia, Ph.D., P .Eng. Ph.D. Waterways d.o.o. Primodal US Inc. Lekenik, Croatia Kalamazoo, Michigan 7 Introducing the principles of uncertainty evaluation and the DOUT uncertainty analysis framework Evangelina Belia, Primodal Inc. Lorenzo Benedetti, Waterways 8 4

1/23/2020 IWA/WEF DOUT Group Core Group Working Group Y. Amerlinck JB Neethling D. Bixio M. O’Shaughnessy C. Bott A. Pena-Tijerina M. Burbano B. Plosz B. Chachuat L. Rieger J. Copp O. Schraa X. Flores-Alsina A. Shaw Lina Belia Lorenzo Benedetti Bruce Johnson Sudhir Murthy S. Gillot G. Sin T. Hug S. Snowling J. Jimenez G. Sprouse B. Karmasin K. Villez D. Kinnear J. Weiss J. McCormick N. Weissenbacher. Marc Neumann Peter Vanrolleghem Stefan Weijers H. Melcer 9 Motivation • Conventional steady state design • How is risk currently handled? Steady State Design Influent constituents Process-based equations Effluent standards Empirical WWTP’s Required parameters equations dimensions Operation parameters Experience- based rules Safety factors Talebizadeh M. (2015) Probabilistic design of wastewater treatment plants. PhD. Thesis. modelEAU-Université Laval, Québec, QC, Canada 10 5

1/23/2020 Paradigm shift Steady State/ WWTP’s dimensions Dynamic Influent constituents Mathematical models + Compare to Effluent Required parameters standards Statistical methods Operation parameters Safety factors Talebizadeh M. (2015) Probabilistic design of wastewater treatment plants. PhD. Thesis. modelEAU-Université Laval, Québec, QC, Canada 11 Risk and Uncertainty • Risk = expectation of losses associated with a harmful event Example: = Risk of failure (exceeding effluent permit) Risk = [Probability of failure] * [Cost of failure] • Probability : is it "likely" or "unlikely“ that the event will happen? Example: Probability of a design to meet effluent standards Probability is the expected likelihood of occurrence of an event • Uncertainty assessment and propagation are: Quantification of probabilities Quantify risk = assess uncertainty = quantify probability 12 6

1/23/2020 Levels of uncertainty Walker, W.E.; Harremoes, P.; Rotmans, J.; van der Sluij s, J.P.; van Asselt, M.B.A.; Janssen, P.; Krayer von Krauss, M.P. (2003). Defining uncertainty: a conceptual basis for uncertainty management in model-based decision support. Integrated Assessment vol. 4, issue 1, 5-18. 13 Statistical Uncertainty • Parameter uncertainty Hauduc et al. (2010): Database of AS M1 & AS M2 calibrations b ANO d -1 dDesk 14 7

1/23/2020 Scenario Uncertainty  What is going to happen at my plant in the next 30 years? • New industry • New treatment technologies • New legal requirements • … .. 15 Key Definitions • Variability • Uncertainty • Propagation in models 16 8

1/23/2020 Variability • “Lack of consistency or fixed pattern” • A measurable quantity that varies in time – timeseries • Variability is intrinsic, cannot be reduced MODEL 17 (Statistical) Uncertainty • “Refers to epistemic situations involving imperfect or unknown information” • “A state of limited knowledge where it is impossible to exactly describe the existing state or a future outcome” • Probability Density Function (PDF) • Uncertainty can be reduced by more research 18 9

1/23/2020 Uncertainty Propagation: Monte Carlo dDesk frequency  value mean Boeij e G. (1999) Chemical fate prediction for use in geo-referenced environmental exposure assessment. PhD. Thesis. BIOMATH-Ghent University, Belgium 19 Monte Carlo simulation Deterministic Monte Carlo ‘Shot’ ... Inputs Model Simulation Distributions 200 Concentration Discrete 175 Result 150 125 Probabilistic Deterministic 100 75 50 25 0 0 10 20 30 40 50 60 70 80 Time Boeij e G. (1999) Chemical fate prediction for use in geo-referenced environmental exposure assessment. PhD. Thesis. BIOMATH-Ghent University, Belgium 20 10

1/23/2020 Monte Carlo simulation Deterministic ‘Shot’ ... Monte Carlo Inputs Model Distributions Simulation 200 ... Statististical Discrete Result Concentration 175 Analysis Result Distributions 90 % ile 150 Probabilistic Deterministic 125 100 75 50 Average 25 0 0 10 20 30 40 50 60 70 80 Time Boeij e G. (1999) Chemical fate prediction for use in geo-referenced environmental exposure assessment. PhD. Thesis. BIOMATH-Ghent University, Belgium 21 Variability and Uncertainty – model output 0.8 0.7 95% ile - MC in blue: 0.6 temporal variability due to influent 0.5 variability NH4 [mg/L] single simulation 0.4 in red: 0.3 output uncertainty band 0.2 due to parameter uncertainty 0.1 5% ile - MC 0 1 2 3 4 5 time [d] 22 11

1/23/2020 Four different ways to combine variability (steady state or dynamic simulation) and uncertainty (single or MC simulation) 23 Steady state – no MC (1 simulation) Point estimate 1 0.8 2.2 0.6 fraction x 0.4 0.2 0 0 1 2 3 4 5 6 NH4 mg/l 24 12

1/23/2020 Steady state – MC (1000 simulation) Confidence interval (uncertainty) 1 0.8 1.5 2.3 4.1 0.6 fraction x x x 0.4 5% 95% 50% 0.2 0 0 1 2 3 4 5 6 NH4 mg/l 25 Dynamic – no MC (1 simulation) Frequency estimate (variability) 1 90% 0.8 0.6 fraction 0.4 0.2 0 2.4 0 1 2 3 4 5 6 NH4 mg/l 26 13

1/23/2020 Dynamic – MC (1000 simulation) Frequency + confidence (variability + uncertainty) 100 90% 95% ile 50% ile 80 Duration [%] 60 40 20 0 3.0 4.3 0 1 2 3 4 5 6 NH4 [mg/l] 27 In Summary • Variability is something “ sure ” : we push it throught the model and we get the frequency of compliance • Uncertainty is about possible futures : with probabilities expressed by PDFs, confidence means “ in how many possible futures something is happening” 28 14

1/23/2020 DOUT uncertainty analysis framework – what impacts risk in projects Regulatory MODEL PROJECT PHASE Planning Project definition Preliminary design Data collection Detailed design Model set‐up Construction Calibration Start‐up Simulation Operations Citizens Desing Bid Build Regulator Design Build CONTRACT TYPE STAKEHOLDERS Government Design Build Operate Utility Numerical Contractor Model structure Model parameter Measurement Aggregation SOURCE OF UNCERTAINTY 29 Sources of variability and uncertainty Location Examples Details Sources Current and future predicted flow, COD, Influent data ammonia Tank volume and geometry Physical data DO set points Measured data Operational settings Performance data Effluent data, reactor concentrations Inputs Additional info Input from connected systems e.g. sewers, catchment Number of tanks in series Hydraulic Model parameters Biokinetic Maximum growth rates S ettling S ettling coefficients Influent model, hydraulic model, aeration Models system model, process models (biological, settling, ...) Model structure Waste activated sludge pumped to an anaerobic Interfaces between models digester; digester effluent pumped to sludge treatment S olver settings S oftware Numerical approximations Numerics (model technical aspects) S oftware limitations Bugs Propagation of Model output Probability of meeting effluent criteria All model uncertainties uncertainty 30 15

1/23/2020 Engineering project phase • Prioritization of the sources of uncertainty 31 Contract delivery methods • Risk allocation 32 16

1/23/2020 Uncertainty analysis methodology Identify : Uncertainty propagation : → Decision drivers → Influent variability → Metrics → Parametric uncertainty → S ources Scenario analysis Prioritize: → Fore sighting methods → S → ensitivity analysis Life cycle assessment → Expert knowledge → Multi-attribute-utility theory → Benefit-cost-risk approach Reduce: → Benchmarking and auditing → S ampling → Experimental design Synthesize and communicate results: Model: → PONC and PS E estimates → Influent → .... → CFD → Integrated modeling Adapted from: Jakeman, A.J., Letcher, R.A. and Norton J.P. (2006) Ten iterative steps in development and evaluation of environmental models. Environmental Modelling & Software. 21, pp 602-614. 33 33 Scientific and Technical Report (STR) Publication in 2020 34 17