Real-time Regulation of Detention Ponds via Feedback Control: Balancing Flood Mitigation and Water Quality

TL;DR

This work develops a physics-based RTC framework that couples a quasi-2D watershed-hydrodynamic model with reservoir routing and an adaptive Model Predictive Control (MPC) strategy to operate valve actuators and movable gates in detention ponds. The MPC switches between flood mitigation and water-quality detention-time objectives based on predicted inflows, enabling improved peak-flow reduction and longer detention times compared with passive control. Validation against HEC-RAS 2D shows good agreement, and scenario analyses demonstrate substantial flood attenuation (up to 79%) and proxy water-quality improvements during dry periods, while continuous 1-year simulations indicate robust performance under spatially distributed climate inputs. The approach highlights the practical potential of retrofitting detention ponds with real-time, predictive control to balance flood resilience and runoff treatment in urban catchments.

Abstract

Detention ponds can mitigate flooding and improve water quality by allowing the settlement of pollutants. Typically, they are operated with fully open orifices and weirs (i.e., passive control). Active controls can improve the performance of these systems: orifices can be retrofitted with controlled valves and spillways can have controllable gates. The real-time optimal operation of its hydraulic devices can be achieved with techniques such as Model Predictive Control (MPC). A distributed quasi-2D hydrologic-hydrodynamic coupled with a reservoir flood routing model is developed and integrated with an MPC algorithm to estimate the operation of valves and movable gates. The control optimization problem is adapted to switch from a flood-related algorithm focusing on mitigating floods to a heuristic objective function that aims to increase the detention time when no inflow hydrographs are predicted. The case studies show the potential of applying the methods developed in a catchment in Sao Paulo, Brazil. The performance of MPC compared to alternatives with either fully or partially open valves and gates are tested. Comparisons with HEC-RAS 2D indicate volume and peak flow errors of approximately 1.4% and 0.91% for the watershed module. Simulating two consecutive 10-year storms shows that the MPC strategy can achieve peak flow reductions of 79%. In contrast, passive control has nearly half of the performance. A 1-year continuous simulation results show that the passive scenario with 25% of the valves opened can treat 12% more runoff compared to the developed MPC approach, with an average detention time of approximately 6 hours. For the MPC approach, the average detention time is nearly 14 hours indicating that both control techniques can treat similar volumes; however, the proxy water quality for the MPC approach is enhanced due to the longer detention times.

Figures (13)

• Figure 1: Hydrologic conceptual model where $q_{\mathrm{in}}$ and $q_{\mathrm{out}}$ are inflows and outflows [$\mathrm{L \cdot T^{-1}}$], $i_{\mathrm{p}}$ is the rainfall intensity [$\mathrm{L \cdot T^{-1}}$], $e_{\mathrm{TR}}$ is the real evapotranspiration [$\mathrm{L \cdot T^{-1}}$], $f$ is the infiltration rate [$\mathrm{L \cdot T^{-1}}$], $f_{\mathrm{g}}$ is the groundwater replenishing [$\mathrm{L \cdot T^{-1}}$], $\psi$ is the suction head acting at the wetting front [$\mathrm{L}$], $\Delta \theta$ is the soil moisture deficit [-], $h_{\mathrm{ef}}$ is the effective water depth [$\mathrm{L}$] while $h$ is the total depth [$\mathrm{L}$], $h_{\mathrm{0}}$ are the losses through plant interception and initial abstractions [$\mathrm{L}$], $L(t)$ is the effective depth of the saturated zone [$\mathrm{L}$], and $f_{\mathrm{d}}$ is the cumulative infiltrated depth [$\mathrm{L}$]. Two fundamental equations are solved for the atmosphere-soil (1) interface and the wetting front-soil interface (2).
• Figure 2: Inflow and outflow hydrographs of a relatively large and relatively small stormwater reservoir with water quantity and quality control, where $\Delta t_{\mathrm{d}}$ is the required detention time. The factor $\alpha_{\mathrm{p}}$ can be tuned and be used to represent a desired peak flow reduction under minor flood events with maximum predicted inflows smaller or equal than $q_{\mathrm{max}}^{*}$. The relatively large reservoir can store all inflow hydrograph and later release after a detention time threshold is reached, while the relatively small reservoir does not have the storage capacity to do so and has to be operated focusing on flood mitigation.
• Figure 3: Inflow hydrograph and the definition of the maximum inflow $q_{\mathrm{m}}$ for each prediction horizon $p_{\mathrm{h}}$. The weights to be used in Eq. \ref{['equ:optimization_problem']} are defined by each maximum inflow $q_m$ for each prediction horizon as shown in Eq. \ref{['equ:weights']}. Therefore, the control theoretical goal is changed according to the predicted flood magnitude. Note that $q_{\mathrm{m}}^3$ violates the threshold for large flood $q_{\mathrm{max}}^{**}$, increasing the focus of the control for flood mitigation, while $q_{\mathrm{max}}^1$, for example, has a maximum predicted inflow in the prediction horizon smaller than the threshold for minor floods.
• Figure 4: Study Area Map in São Paulo City - Brazil. The reservoir has an inlet channel and receives headwater from the Aricanduva watershed. During large events, runoff is spilled by a rectangular crest spillway.
• Figure 5: Land use and land cover map (a), hypsometric map of the watershed (b). Elevation was filled and resampled to $\mathrm{50~m}$. Following this, the D-8 watershed algorithm was run, and the watershed boundary was created. All other maps were derived by clipping available rasters into this resulting polygon.