Modeling BOD removal in free watersurface constructed wetlands with variable residence time-based methods

Abstract: Constructed wetlands (CWs) have been applied for several years in the removal process pollutants from secondary treatment plants and stormwater. In this paper, the Variable Residence Time (VART) model for solute transport in streams was extended to simulate the BOD removal process in free water surface flow constructed wetlands (FWS CWs). The double layer concept of the basic VART model was extended to FWS CWs, contemplating the wetland as a similar layer system with water column (WC) and root zone (RZ). The RZ was further divided into advection dominated upper root layer, and diffusion dominated the lower root zone. Two distinctive models were developed based on this, namely VART-BOD-F model that incorporates the first-order decay of BOD and VART-BOD-M model that combines multiple Monod kinetics in the wetland system. Unlike conventional wetland methods, both models involve dynamic root zone concept with changing reactive depth. The application of mass exchange between the water column and dynamic diffusion root zone was the distinct feature of the models. From time series data based simulation test results, coefficient of determination (r2) and root mean square error (RMSE) values for the VART-BOD-F model ranges from 0.79 to 0.91 and 0.69 to 1.84, respectively, while for VART-BOD-M model varies from 0.83 to 0.96 and 0.63 to 1.77 respectively. Furthermore, the performance of the second model (VART-BOD-M model) was tested later with another data obtained from Manzala Wetland, Egypt (Deng et al., 2016). When the two models compared, including with the exiting DND model, the VART-BOD-M model, which involved various mechanisms showed better values of r2 and RMSE, and capable of following reasonably best fit concerning observed effluent concentration. Although the VART-BOD-M model has better efficiency than the VART-BOD-F model, it involved several parameters and certain assumptions that might have compromised the results and analysis. Hence, the VART-BOD-M model can be a better tool for simulating the BOD removal process in FWS CWs when there is data available for defined parameters