Utilising adaptive real-time control in removing organic chemicals in stormwater biofilters

Real-time control (RTC) can improve pollutant removal in stormwater biofilters, but existing strategies are largely static and rule-based, limiting their ability to respond to naturally variable rainfall conditions. This study developed and evaluated an adaptive RTC strategy (Adaptive-RTC) for stormwater biofilters, combining rolling-window rainfall forecasting with redox-based operational thresholds to improve water quality. Performance of Adaptive-RTC was compared with that of conventional biofilters (Non-RTC) across 11 historical rainfall events with diverse hydrologic characteristics, monitoring ten trace-level organic chemicals (TrOCs) and four common stormwater pollutants (e.g., nutrients). The Adaptive-RTC increased the proportion of time under favourable aerobic and anoxic conditions by up to 51.6% under short antecedent dry weather periods (< 4 days). These redox improvements translated into improved treatment performance, with removal of aerobically biodegradable TrOCs increasing from 75.8 ± 22.5% in Non-RTC biofilters to 89.9 ± 13.7% under Adaptive-RTC (p < 0.01). Significant improvements were also observed for total nitrogen (TN) and total organic carbon (TOC) (p < 0.01). Forecast uncertainty influenced outcomes: under false-positive forecasts, removal of aerobically biodegradable TrOCs declined to as low as 75.4% because of unnecessary early releases and shortened retention times. Nevertheless, Adaptive-RTC still achieved higher overall pollutant removal than Non-RTC across pollutant groups, indicating resilience under the forecast inaccuracies observed in this study. Practical considerations for implementation include ensuring long-term sensor reliability, optimising sedimentation pond and outlet design, and accounting for forecast uncertainty. Overall, this study highlights the potential of Adaptive-RTC as a practical stormwater management strategy and supports its future implementation.