Reliability-Based Evaluation of the Robustness of Constructed Wetlands for Sustainable Sanitation: Implications for Design and Operation.

Constructed wetlands (CWs) are low-cost, low-energy nature-based solutions for wastewater treatment and are commonly regarded as robust; however, this perception is often based on removal efficiencies or mean effluent concentrations, without accounting for effluent variability or the probability of maintaining concentrations below regulatory limits. This study applied the reliability coefficient (RC) to 34 vertical-flow constructed wetland (VFCW), horizontal-flow constructed wetland (HFCW), and hybrid constructed wetland (HCW) systems treating domestic wastewater identified through a systematic literature review. Removal efficiencies and final effluent concentrations were evaluated against Brazilian discharge and non-potable reuse benchmarks. Reliability was calculated separately for each system-parameter combination at a 95% target non-exceedance probability. Grouped comparisons examined only descriptive patterns, dispersion, and overlap among configurations. Sensitivity to alternative target probabilities and analytical applications to design and monitoring were also examined. Mean removal efficiencies ranged from 79.8%-94.5% for biochemical oxygen demand (BOD), 76.9%-87.8% for chemical oxygen demand (COD), 73.6%-90.8% for total suspended solids (TSS), 59.6%-65.9% for total nitrogen (TN), and 68.0%-75.4% for total phosphorus (TP). Under the stringent scenario, mean effluent concentrations were below the selected limits in 21/22 systems for BOD, 33/34 for COD, 22/22 for TSS, 4/25 for TN, and 6/18 for TP. Under the assumed log-normal model, incorporating effluent variability at 95% reduced probabilistic compatibility to 20/22, 27/34, 21/22, 2/25, and 4/18 systems, respectively. Grouped mean RC values ranged from 0.38 to 0.69, but substantial within-configuration dispersion and overlap showed that no single RC can represent a flow configuration or treatment train. At target probabilities of 95% or higher, increasing variability consistently reduced the RC; at lower probabilities, the non-monotonic response of the equation altered system rankings. Incorporating the RC into first-order HFCW and VFCW models converted regulatory limits into system- and parameter-specific allowable mean effluent concentrations, providing an analytical basis for reliability-informed design and monitoring, although the framework was not experimentally validated as an independent sizing method. The results support comparatively wide operating margins for organic matter and suspended solids, whereas sustained nutrient compatibility remained less frequent. Removal efficiency, residual concentration, and reliability should therefore be interpreted jointly using RC values specific to each system and parameter.