Multiobjective Optimization of Seasonal Operating Rules for Water Grids Using Streamflow Forecast Information

Abstract

Multiobjective simulation-optimization is a useful tool for determining operating rules for water supply that are optimal for multiple management objectives and expected conditions over the planning period. Previous studies have shown the benefits of streamflow forecasts in improving optimized objective performance of reservoir operating rules. This study demonstrates a simple method for incorporating publicly available streamflow forecast information in operational planning for a water grid. Multiobjective optimization is used to find operating rules for a case study that are optimal for three management objectives—maximizing minimum system storage, minimizing operational cost, and minimizing spills from reservoirs—and for forecast inflow scenarios. These forecast-optimized rules are compared to those optimized using inflow scenarios from the historical distribution, representing operation in the absence of forecast information. The results across four seasonal (3-month) planning periods indicate that, on average, operating rules optimized using forecast streamflow information perform slightly better in terms of the management objectives than those optimized using historical inflow information. Using multiple scenarios of inflow that span the forecast distribution increases the robustness of the operating rules and reduces the risk of underperformance because of forecast inaccuracy. The results suggest that incorporating streamflow forecast information into multiobjective simulation-optimization has the potential to provide improvements in seasonal operating rules for a water grid.