4 - Presentation of Water Forward 2024 plan update methodology — original pdf

Backup

Water Forward WAVE evaluation Water Forward Task Force Meeting March 5, 2024 Water Forward: Planning for Uncertainty  Develop a range of plausible future scenarios  Find common near-term water management strategies (WMSs) that perform well over many scenarios  For long-term (WMSs), develop an adaptive management plan with key decision points  Continue to update the plan, re- evaluate, and adapt s o i r a n e c s g n i l e d o M Range of demands Possible climate futures Droughts worse than the drought of record Regional supply trends Water Forward 2024 Decision points including regular updates to the WF Plan Scenarios of plausible future water needs A C D B E 2120 WF24 Methodology Overview Goal Define the needs Evaluate potential solutions Identify potential solutions Select the best solution & implementation strategy We are here Ongoing community engagement and equity work Tasks Develop range of future scenarios WF18 implementation evaluation and update Preliminary needs assessment Identify, screen, and characterize additional WMS Use optimization model to identify top-performing WMSs over all scenarios Construct 50-yr portfolios for further evaluation 50-Year portfolio trade-off analysis Conduct vulnerability assessment to identify system stressors Preferred 50-yr portfolio & equity and affordability analysis Adaptive Management Plan (AMP) Outcome Updated WF18 strategies and range of baseline needs WMSs variables and rules for testing Evaluated 50-year portfolios and identified system stressors Water Forward 2024 Plan Timeline 2022 Fall 2024 WF24 Methodology Overview Goal Define the needs Evaluate potential solutions Identify potential solutions Select the best solution & implementation strategy Ongoing community engagement and equity work Tasks Develop range of future scenarios WF18 implementation evaluation and update Preliminary needs assessment Identify, screen, and characterize additional WMS Use optimization model to identify top-performing WMSs over all scenarios Construct 50-yr portfolios for further evaluation 50-Year portfolio trade-off analysis Conduct vulnerability assessment to identify system stressors Preferred 50-yr portfolio & equity and affordability analysis Adaptive Management Plan (AMP) Outcome Updated WF18 strategies and range of baseline needs WMSs variables and rules for testing Evaluated 50-year portfolios and identified system stressors Water Forward 2024 Plan Timeline 2022 Fall 2024 Water management strategy Assessment and Vulnerability Evaluation = WAVE WAVE Team Members Rob Lempert Michelle Miro Swaptik Chowdhury RAND Will Support the WAVE with Multi-Objective Robust Decision Making (MoRDM)  We live in a fast-changing, hard-to-predict world  We can shape the future, even when we can’t predict it with confidence • But many analyses don’t reflect this wisdom  MoRDM is a multi-objective, multi-scenario analysis that improves decisions under deep uncertainty Deep uncertainty occurs when the parties to a decision do not know or do not agree on the likelihood of alternative futures or how actions are related to consequences MoRDM Identifies Robust Portfolios Through Iterative Optimization and Stress Testing Process aims to inform better decisions, not better predictions Specify goals, WMS, and scenarios Stress test portfolios to find those robust over many objectives and scenarios Find WMS portfolios that meet objectives in individual scenarios Deep Uncertainty May Seem Daunting, But Considering It Is Empowering  Inappropriate use of ”predict-then-act” methods risks • Over-confidence, • Missed opportunities • Portfolios brittle against surprise Deep Uncertainty May Seem Daunting, But Considering It Is Empowering  Inappropriate use of ”predict-then-act” methods risks • Over-confidence, • Missed opportunities • Portfolios brittle against surprise Deep Uncertainty May Seem Daunting, But Considering It Is Empowering  Inappropriate use of ”predict-then-act” methods risks • Over-confidence, • Missed opportunities • Portfolios brittle against surprise  MoRDM helps identify • Low regrets actions • Adaptive and flexible plans • Steps to keep options open Deep Uncertainty May Seem Daunting, But Considering It Is Empowering  Inappropriate use of ”predict-then-act” Plan over multiple futures methods risks • Over-confidence, • Missed opportunities • Portfolios brittle against surprise Find robust strategies  MoRDM helps identify • Low regrets actions • Adaptive and flexible plans • Steps to keep options open Using MoRDM, we will build robust portfolios of water management strategies  Inputs: • Characterized WMS • Preliminary needs WAM outputs • Mini-WAM model • Rhodium multi-objective optimization library • Optimized portfolios of WMSs • Metrics describing reliability, resiliency, vulnerability and cost of portfolios  Tools  Outputs Goals Many scenarios One scenario Portfolios of WMS balance water management goals WAM WAM Needs Assessment Projected water demands and supplies WMS Portfolios Mini-WAM Model Water balance model of AW system Reliability Resiliency Vulnerability Cost Rhodium Multi- Objective Optimization- Adjust individual and combinations of WMS to MAXIMIZE and MINIMIZE Future Hydrologic Sequences Regional Supply Scenarios Future Water Demand Forecasts Goals Many scenarios One scenario We carry out a multi-objective optimization to build WMS portfolios  Multi-objective optimization finds an “optimal solution” satisfying multiple objectives simultaneously  “Optimal solution” is a set of points describing tradeoffs between different objectives  We use multi-objective evolutionary algorithms  Evolutionary algorithms use more than one solution in each iteration (population-based approach) to find multiple optimal solutions Goals Many scenarios One scenario The optimization process will generate an “optimal” set of solutions which will be evaluated as portfolios s c i r t e m e c n a m r o f r e P : 1 e v i t c e j b O Ideal solution: cheap & reliable Front of optimized solutions: no objective can be improved without sacrificing another objective Objective 2: Cost e r o c S o i l o f t r o P 50-year portfolios will be evaluated using a multi-criteria decision analysis (MCDA) incorporating additional criteria Objectives Avoid severe water shortages during drought and a variety of climate change scenarios Focus on water conservation and water use efficiency Include diverse water management strategies that make use of all water sources. Performance Measures • WAM/WAVE modeling results Potable GPCD Portfolio diversity score Minimize impacts and maximize benefits of plan outcomes for marginalized communities Cost (lifecycle, capital, O&M) Equity & Affordability Tool Develop strategies that continue to protect the natural environment, including source and downstream water quality Develop strategies that are technically, socially, and economically feasible and can be implemented and operated with a manageable level of risk Net return flows Operational energy use • • • Water quality impacts • • Implementation and operational risk score System resiliency benefits Develop strategies that make use of locally available and AW- controlled water resources Volume of local and AW-controlled water resources • • • • • We leverage Rhodium to carry out WMS portfolio optimization Rhodium Multi- Objective Optimization- Adjust individual and combinations of WMS to MAXIMIZE and MINIMIZE optimization metrics  Open-source Python library developed by researchers  Contains framework for evolutionary computing in Python supporting different MOEAs • EAs such as NSGA-II, NSGA-III, MOEA/D, IBEA, Epsilon-MOEA, SPEA2, GDE3, OMOPSO, SMPSO, and Epsilon-NSGA-II • Has built-in capabilities for high-dimensional, interactive scientific visualization linking to other models • Parallelization capabilities with in-built wrapper function for WMS portfolios are assessed under each future combination of climate, supply and demand Under all futures and for each planning horizon WAM WAM Needs Assessment Projected water demands and supplies WMS Portfolios Mini-WAM Model Water balance model of AW system Reliability Resiliency Vulnerability Cost Rhodium Multi- Objective Optimization- Adjust individual and combinations of WMS to MAXIMIZE and MINIMIZE Future Hydrologic Sequences Regional Supply Scenarios Future Water Demand Forecasts Goals Many scenarios One scenario Use Regret to Compare Portfolios Across Multiple Scenarios  A portfolio’s regret in any future is the difference between its outcome in that future and the outcome of the best portfolio in that future Let’s See How This Works Strategies Stay home Go for picnic Let’s See How This Works Strategies Futures Sun Rain Stay home Go for picnic Let’s See How This Works Strategies Futures Sun Rain Stay home Bad (u11) Good (u12) Outcomes Go for picnic Good (u21) Bad (u22) u12 > u11 u21 > u22 Let’s See How This Works Futures Strategies Stay home - Regret = u21 - u11 > 0 - Regret = 0 Let’s See How This Works Futures Strategies Stay home - Regret = u21 - u11 > 0 - Regret = 0 - Go for picnic - Regret = 0 Regret = u12 – u22 > 0 Use Regret to Compare Portfolios Across Multiple Scenarios  A portfolio’s regret in any future is the difference between its outcome in that future and the outcome of the best portfolio in that future  Use Multi-objective optimization to find WMS portfolios will small reliability, resiliency, vulnerability, and cost regret over a wide range of scenarios Next steps  Analytical steps: • Complete characterization of WMSs • Input all WMSs into optimization model • Form portfolios and analyze trade-offs and vulnerabilities  Presenting preliminary WAVE results at May WFTF  Presenting final WAVE results at July WFTF • Use WAVE results to build 50-year portfolios  Use WAVE results to perform vulnerability evaluation and develop adaptive management plan Questions?