Managing Risk: how can process models help - Evangelia Belia, Primodal US Inc. Michael Lunn, Primodal US Inc.
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Managing Risk: how can process models help Evangelia Belia, Primodal US Inc. Michael Lunn, Primodal US Inc. WWAdCon 2021
Outline • What is risk? • How does our industry currently manage risk? • How do other industries manage risk? • Can our industry implement the same methods? WWAdCon 2021
What is risk? Risk = expectation of losses associated with a harmful event Example: = Risk of failure (exceeding effluent permit) Risk = [Probability of event happening] * [Cost] Example: = [likelihood of exceeding effluent permit] * [cost] Cost Examples (monetary + non-monetary): WWAdCon 2021 Non-compliance; Loss of reputation; Financial loss; Not winning a contract or contract annulment; Loss of operating license; ….
Sources of risk Originating from the environment: • Weather/climate effects • Collection system characteristics • Wastewater characteristics (flows, loads, temp, pH, fractions, …) • Growth or loss within collection area (growth rate, changes in I/I, changes in industry) • Discharge permits/Regulations Originating from the WRRF: WWAdCon 2021 • Biological: bacterial kinetics, floc structure • Physical: non-ideal process behavior (e.g. non-ideal mixing) • Chemical-physical: e.g. precipitation stoichiometry and kinetics • Unexpected control system behaviour • Mechanical failures • Operational problems
How does our industry currently manage risk? • Design guidelines • Design criteria • Safety factors • Redundancy standards • Scenario analysis WWAdCon 2021 • Modelling • Contract documents / contract delivery method
Example of safety factors and redundancy requirements SRT safety factor 1.4 to 1.8 takes into account (ATV-DVWK-A 131 guidelines): a) potential variations of the maximum growth rate caused by certain substances in the wastewater, short-term variations and/or pH shifts, and b) the variations of ammonium load Unit Process EPA, 1974 10 State Standards, 2014 WWAdCon 2021 Primary Must be able to process 50% of Requires multiple units capable of sedimentation plant flow with 1 (largest) unit independent operation for plants basins out of service with flows higher than 0.1 MGD, Redundancy but no redundant unit required requirements Secondary Must have 75% of rated capacity Requires multiple units capable of sedimentation with 1(largest) unit of service independent operation for plants basins with flows higher than 0.1 MGD, but no redundant unit required
How does our industry currently manage risk? Steady State Design Influent constituents Process-based equations Conventional Effluent standards steady state design Design parameters Empirical equations WWTP’s WWAdCon 2021 dimensions Operational targets Experience- based rules Safety factors Mansour Talebizadeh (2015) Probabilistic design of wastewater treatment plants. PhD. Thesis. modelEAU-Université Laval, Québec, QC, Canada
Compounding of safety factors 300 90 80 250 70 Daily average BOD (mg/L) Daily average flow (MGD) 200 60 50 150 40 100 30 20 50 Design load Target effluent 10 0 Mar-07 Jul-07 Oct-07 Jan-08 May-08 Aug-08 Nov-08 Feb-09 Jun-09 Sep-09 Dec-09 0 Safety factors Redundancy Date WWAdCon 2021 4 4 3 3 Safety factor Safety factor 2 2 1 1 0 0 Influent load Unit process Influent load Redundancy Unit process Risk registers Redundancy Effluent criteria Risk registers Effluent criteria selection safety factors selection safety factors and compliance and compliance testing testing
Drawbacks of current design approach • Steady-state: • ignores plant dynamics and plant evolution • a single value does not provide quantifiable information about the reliability of a process design. • Lumping risk in a few parameters: • missing relative importance of individual sources of risk • inflexible designs WWAdCon 2021 • Often assumes a combination of worst case conditions that may never happen • No information on likelihood or frequency of any particular load reaching the plant within the selected design horizon • No information on the likelihood or frequency of non-compliance • Does not address newly developed processes
How do other industries handle risk? WWAdCon 2021
Can our industry implement the same methods? Why now? • Safety factors have not changed for decades • Increased use of simulators for design and optimization • Strict effluent guidelines & increased energy efficiency • Awareness among practitioners • Scientific approach to balance risk/benefits WWAdCon 2021 However: • Simulators lack reliability evaluation • Need to couple with statistical methods
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 WWAdCon 2021 • Uncertainty assessment and propagation are: Quantification of probabilities Quantify risk = assess uncertainty = quantify probability
Levels of uncertainty Donald Rumsfeld WWAdCon 2021 As we know, there are known knowns. There are things we know we know. We also know there are known unknowns. That is to say : We know there are some things we do not know. But there are also unknown unknowns, the ones we don't know we don't know.
Steady-state probabilistic design Sources if uncertainty Steady State “Safety Factor” Influent constituents redistributed Effluent standards Design guidelines + daylighting the Design parameters Models WWTP’s Effluent + WWAdCon 2021 dimensions performance specific risks Operational targets Statistical and risk methods Safety factors Mansour Talebizadeh (2015) Probabilistic design of wastewater treatment plants. PhD. Thesis. modelEAU-Université Laval, Québec, QC, Canada
Dynamic probabilistic design with Monte Carlo PDF: Probability density function CDF: Cumulative distribution function WWAdCon 2021 Unit sizes
Variability and Uncertainty – model output 0.8 0.7 in blue: temporal variability 95%ile - MC due to influent variability 0.6 0.5 in red: output uncertainty NH4 [mg/L] single WWAdCon 2021 0.4 simulation band due to parameter 0.3 uncertainty 0.2 0.1 5%ile - MC 0 1 2 3 4 5 time [d]
Uncertainty analysis of results 100 90 Frequency + confidence (variability + uncertainty) • 5% sure NH4 < 3 mg/l 96% of the time 80 70 • 95% sure NH4 < 3 mg/l 78% of the time 25 60 Alt1 Mixed Alt1 80 Nominal Duration [%] 50 Worst case 20 70 40 60 WWAdCon 2021 PONC for NH4 (%) PONC for TN (%) Alt2 15 50 30 Alt2 Alt3 40 20 10 Alt3 30 10 20 5 0 0 1 2 3 4 5 6 7 8 Concentration [mg/l] Alt4 Alt5 10 Alt4 Cumulative curves for NH4 concentration at the effluent: 0 6 6.2 6.4 6.6 0 6 6.2 Total Cost (Million Dollar) Total Cost (Mi daily composite samples 50th, 5th and 95th %iles Mansour Talebizadeh (2015) Probabilistic design of wastewater treatment plants. PhD. Thesis. modelEAU- Université Laval, Québec, QC, Canada
Proposed design procedure Define project Preliminary objectives evaluation of selected Compile list of sources Identify alternatives of uncertainty for the Prioritise Effluent standards and Reduce frequency of compliance Use design guidelines for design objective preliminary sizing and Design horizon approximate costing Characterize sources Run dynamic of uncertainty with simulations using Model Monte Carlo PDFs WWAdCon 2021 Propagate Compile effluent CDFs Define scenarios Calculate output Analyse metrics Multi-criteria decision Synthesize Quantify the probability of and selection of best Communicate (non)compliance alternative Estimate detailed total cost Talebizadeh, M. (2015). Probabilistic design of wastewater treatment plants. PhD. Thesis. Département de génie civil et de génie des eaux, Université Laval, Québec, QC, Canada. https://corpus.ulaval.ca/jspui/handle/20.500.11794/26196
Lag in implementation of explicit risk analysis • Considered high costs and time-consuming - computers reduce considerably costs and time • Lack of data - the more uncertain the input data, the more helpful accounting for uncertainty and evaluating the associated risk • Decision makers reluctant to accept these techniques (not confident they will WWAdCon 2021 help them make better decisions) - may stem in part from lack of understanding of the techniques, requires multidisciplinary teams 1988 NIST: “A Comprehensive examination of the different approaches to treating uncertainty and risk in project evaluation would show how the application of risk analysis techniques to uncertain data can improve management decision making.”
Opportunities of explicit risk analysis • Risk is made explicit • Daylights conservatisms buried in the various design assumptions • Allows plants to be right-sized • Allows to design a system with a desired level of risk • Transparency • Decision making process • Quantitative evaluation of statistical likelihood of achieving a particular effluent/performance criteria WWAdCon 2021 • Approach can be steady-state (applying Monte Carlo to basic design equations) or dynamic (probabilistic design with dynamic Monte Carlo simulator runs) • Benefit-cost-risk analysis • Showing how benefits, costs and risks are spread among stakeholders • Integration of socio-economic & statistical sciences • Fore-sighting methods, life cycle analysis, environmental economics, ...
Scientific and Technical Report (STR) WWAdCon 2021 (Publication in 2021)
Presenter contact information Evangelina Belia Ph.D., P. Eng. Primodal Inc. WWAdCon 2021 US & Canada belia@primodal.com www.primodal.com
You can also read