Detention-based stormwater drainage management using a dynamic programming approach in aselected area in Dhaka

Abstract

Unplanned urbanization and encroachment of storm runoff detention areas pose serious challenges to stormwater drainage management in Dhaka City. The arbitrary-rule based stormwater management in the Hatirjheel detention system, the largest gravity-drainage detention basin in the city, is often inadequate to drain internal storm runoff in extreme flood years when the downstream regulator gates are closed due to high external river water level, and temporary pumping is required. A Dynamic Programming model is developed using a forward recursive algorithm and following an Incremental Dynamic Programming approach to optimize the weekly pumping requirements. System constraints are defined by the gate-closing water level, and the allowable maximum and minimum water levels in the system. Runoff from 12 sub-catchments of the system is calculated with runoff ratios estimated from satellite photo interpretation and field observation. Model application results indicate that pumping required at the optimum condition is significantly lower than that required by the current practice. In extreme flood year 1998, the pumping requirements under optimum condition and current practice would be 26.118 Mm3 and 35.925 Mm3, respectively. The feasibility of using the DP model for real-time stormwater management decision making is tested based on 14 optimal Runoff Scenarios. The pumping decision at the current stage can be taken by matching the observed runoff trend in the current year with one of these scenarios. Low-lying areas around the Hatirjheel detention basin are vulnerable to local inundation due to high water levels in the basin that prevents gravity drainage from these areas. Inundation maps prepared from a DEM indicate that 1.15 km2 area is inundated for a water level of +7.0 m PWD. A household-survey based flood damage assessment indicates that the potential damage in an extreme condition is higher for the high income group. The potential damage for the industrial area could not be assessed due to the lack of adequate information on the variations with depths and durations.