Evaluation of the accuracy of the EPA model for BOD5 prediction in various climatic regions of Texas

Abstract

This project focused on evaluating the effectiveness of the EPA's first-order BOD₅ removal model for predicting BOD₅ reductions in residential constructed wetlands. Monthly grab sample data from nine constructed wetlands designed using the EPA BOD5 model demonstrated removal rates for BOD₅, TSS, fecal coliform, ammonium, and phosphorus of 80.0%, 70.9%, 87.1%, 34.7%, and 22.7%, respectively. These data provided the basis for evaluating the EPA model. A sensitivity analysis of the EPA model indicated that the model is sensitive to changes in both hydraulic design and removal kinetic parameters, particularly average flow rate, porosity, depth of water, average winter temperature, temperature correction factor, and influent BOD₅ concentration. Therefore, selecting appropriate parameters for the site where the wetland will be constructed is crucial to the performance of the wetland system. An evaluation of the EPA model's ability to predict effluent BOD₅ concentrations given the design flow rate and influent concentration indicated that the model is more effective for making long-term predictions of BOD₅ reduction. For further evaluation of the EPA model's ability to predict effluent BOD₅ concentrations, model predictions were run using both the system design flow and the actual observed flow. Again, predictions over a longer time period demonstrated better agreement with measured effluent concentrations. Because the model assumes no variation in flow and the wetland treatment volume remains constant, the predictions for BOD₅ reduction associated with the much lower flows of residential wetlands resulted in an overestimation of BOD₅ reduction. However, within all wetland systems, there exists some minimum threshold at which further reduction of contaminants cannot be achieved. The EPA model does not account for this threshold, and therefore, removal predictions of BOD₅ can be unrealistic for low flow conditions. Because wastewater temperature is a critical design parameter, soil temperatures at 20-cm and 90-cm depths were used to predict effluent concentrations at the two of the nine sites. Comparison of BOD₅ predictions using soil temperature at a depth of 90 cm and air temperature compared to monthly effluent concentrations at one of the two sites resulted in a Root Mean Square Error values of 10.45 and 18.86, respectively.