Determining peak flow recurrence in combined basins with limited flow data using genetic algorithm calibration

The City of Columbus, Ohio, completed a comprehensive Wet Weather Management Plan (WWMP) to mitigate hydraulic deficiencies in the City’s main trunk sewers and to address combined sewer overflow (CSO) in the Scioto and Olentangy Rivers. The recommended solution includes a combined flow deep tunnel that will capture combined sewage overflows from the downtown CSO regulators. This combined flow tunnel will ensure that peak flow from the downtown combined sewer basins is captured up to a specific peak flow recurrence level. Due to the lack of long-term flow monitor and downtown rainfall data, it was difficult to predict peak flow recurrence. The design team proposed a procedure where we used the available two to three-years of quality-checked flow monitor and downtown 15-minute rain gauge data between the years 2000 and 2003 to calibrate a SWMM model using the PCSWMMM Genetic Algorithm Calibration (GAC). We then used the long-term hourly rain data collected by the National Weather Service at Port Columbus in conjunction with the calibrated SWMM model to generate 56 years of flow records.

We applied the GAC procedure to calibrate SWMM Runoff module parameters for each combined sewer basin downtown Columbus. Calibration was performed on the largest ten to twelve storm events recorded between years 2000 and 2003. These storm events covered a wide-range of short, medium, and long durations with high rainfall intensities. Runoff parameters were controlled within suitable uncertainty ranges. The GAC was applied until the peak flow values of the calibrated storms were within ±10% of the observed peak flow values. The hourly rain data from Port Columbus was used with the calibrated Runoff modules to predict the highest peak flows that occurred over the 56-year period. The Cunnane plotting position equation was then applied to the predicted peak flow data to generate a flow recurrence curve. The predicted peak flow recurrence was then corrected to account for the difference between using the one-hour time step rain data from Port Columbus records versus 15-minute time step rain data from the downtown rain gage that was used during the GAC process. The proposed procedure allowed for a reasonable prediction of the two-, five-, and ten-year recurrence peak combined flow while overcoming the lack of adequate long-term flow data.