ESET ALEMU WEST Consultants, Inc. Bellevue , Washington Presentation Outline Purpose for DSS Project Area Description Construction Steps Experimental Design Results and Implications Purpose of DSS Construct a system that represents the physical configuration and operational aspects of a of the Jackson Hydropower Project System wide management tool used for identifying and evaluating operation alternatives in a multipurpose system Provide a basis for effecting a collective understanding between the different departments involved with managing the project Demonstrate the value of forecasts o Hydrologic Forecasts (Ensemble Streamflow) o Energy Price Forecasts Demonstrate the time of year (season) when forecasts are the most valuable Project Description Project Description ▶ The project is located on the Sultan River which drains into the Skykomish River in the Snohomish River Drainage ▶ The system is fed by snowmelt and is characterized by a double humped peak in the fall and spring ▶ It is operated for water supply, flow regulation and hydroelectric power generation ▶ It provides the city of Everett with water supply and about 8% of its electricity ▶ Spada Lake has a storage of 153,000 acre feet while Lake Chaplin has about 18,000 acre feet of storage Project Description Project Description Instream flow requirements are enforced at different segments of the Sultan River and take precedence over hydropower generation. Project is operated based on a July-June water year in four different zones. Operations target to maintain the pool within State 3 1460 State 1 State 2 1440 1430 State 3 1420 1410 State 4 1400 State 1-2 Boundary State 2-3 Boundary State 3-4 Boundary 26-Jun 17-May 7-Apr 26-Feb 17-Jan 8-Dec 29-Oct 19-Sep 10-Aug 1390 1-Jul Water Surface Elecation (ft) 1450 Construction Steps Components of a Decision Support System Operational Models Simulation Model Optimization Model Forecast Generation and Integration Evaluation of Values of Forecasts Real-Time Operation Support Operational Models Simulation Model Built with GoldSim simulation software Captures system operations at the project with hydraulic formula and conditional statements Inputs streamflow and energy prices time-series , instream flow requirements and starting pool elevations Runs on a daily timestep to represent real-time operations of reservoir releases, environmental flow requirements, routing priorities Used to set operational guidelines for optimization model Simulation Model Operational Models Optimization Model Built with Lingo linear programming language Captures the hydraulic and operational elements of the system in a mathematical framework Input variables are forecasts of streamflow and energy prices Decision variables used are power tunnel releases, timing of releases Calculates the quantity and timing of reservoir releases that maximizes energy production Optimizes system operations within 60 days Uses simulation model output for monthly storage targets and hydraulic capacities as constraints Forecast Generation Retrospective Streamflow Forecasts Produced using Ensemble Streamflow Prediction (ESP) method Generated by running a hydrological model (DHSVM) with historical meteorological data of the Sultan River Basin Model is run with using observed precip and temp from the first 7 days and climatology for the following days Produces streamflow traces that have equal probability of occurring Captures the high daily variation in actual streamflow Produced for a period of two months and updated weekly Forecast Generation Retrospective Energy Price Forecasts Generated using historic daily spot prices from the mid-Columbia energy market Produce weekly averages of recent (2008-09) 60-day forecasts of daily spot prices Calculate the average weekly error between the weekly averages and actual spot prices Apply average weekly error by forecast lead time to historic spot prices for the years evaluated (2001-04) Disaggregate synthesized weekly forecasts to a daily time-step Integration of Forecasts Impacts of each forecast on reservoir operations Streamflow forecasts Energy Price forecasts quantity of water released for energy production timing of releases to capture energy price peaks Evaluation of the value of individual and combined forecasts 1. Forecasted Energy Prices + Actual Streamflow 2. Forecasted Streamflow + Actual Energy Prices 3. Forecasted Energy Prices + Forecasted Streamflow 4. Actual Energy Prices + Actual Streamflow Forecasts Integration of Forecasts Framework for operating the models Retrospective 60 day Forecast Input Streamflow Forecasts Energy Price Forecasts Simulation model Updated weekly reservoir storages Activated by Change in Week Monthly Reservoir storages Update system weekly with observed streamflow Optimization model Derive weekly operations Evaluation of Values of Forecasts Investigations are conducted for three years representing a range of hydrological conditions Simulations based on Rule Curves used as baselines for measuring the improvement of reservoir operations Reservoir operations are improved though refining Quantity of releases Timing of releases Metrics used for quantifying the improvement in skill in operation Cumulative revenue generated Quantity of energy produced Values of Forecasts Revenue Generated from Integration of Forecasts $20,000,000 $18,000,000 $16,000,000 $14,000,000 $12,000,000 $10,000,000 $8,000,000 $6,000,000 $4,000,000 $2,000,000 $0 2001–2002 2002–2003 Simulation 2003–2004 Mean Forecast Energy + Forecast Streamflow Forecast energy + Perfect streamflow Forecast streamflow + Perfect energy Perfect Energy + Perfect Streamflow Actual avoided cost Evaluation of Values of Forecasts Assess the value of forecasts in periods of pre-specified operational procedures Review the range of optimal operation policies based on an ensemble streamflow traces Calculate probabilities of elevation targets in drawdown/ refill periods Ensemble Forecasts at Summer Drawdown Ensemble Forecasts at Spring Refill Real-Time Operation Support Monday Morning-Decision Making 1. Obtain the streamflows forecasts for next 60 days from a data center (RFC/other government agencies) 2. Select ensemble streamflow forecasts to be used 3. Run a combination of simulation and optimization model 4. Record the ensemble optimal releases for individual forecasts 5. Evaluate the mean, median and distribution of releases 6. Consider other factors such as state of the snowpack and energy market and other operational factors to make an informed decision There are significant economic benefits to be gained from effectively incorporating forecasts. Reservoir operations can be improved with use of forecast information. Simulation model provides insight into the probabilistic range of historical reservoir storages based on current rule curves. DSS supplements the overall management process of making operating decisions for a multipurpose project. Questions