Modeling alternative treatments systems for confined animal feedlot operations (CAFO) in Kansas

Abstract

In 2003, the EPA changed Confined Animal Feedlot Operation (CAFO) regulations to allow an alternative treatment system for feedlot runoff if a 25-year simulation can show better performance for the alternative treatment system when compared to a traditional lagoon system. A continuous computer simulation model written in C was developed to compare the release of water, nitrogen (N), phosphorous (P) and chemical oxygen demand (COD) from a Vegetated Treatment Area (VTA) system and a conventional lagoon system in Kansas. The model simulates the transport of water and constituents in a feedlot, settling basin, lagoon, and VTA. The feedlot runoff is based on NRCS unit hydrograph method for the water balance and AnnAgNPS for the constituent runoff concentrations. The settling basin is a simple water balance with sediment, organic-N, P, and COD removed with settling of the solids. The VTA is designed to calculate infiltration with Green-Ampt Equation and overland flow with Manning's Law. The VTA removes organic-N and P with the Kentucky Filtration Model. The lagoon is derived from Koelliker et al. model water balance and pumping schedule with constituent balance from Miner et al. model. VTA portion of the model was analyzed for sensitivity and calibrated with three storms. The sensitivity analysis showed that the size of the VTA and conductivity of the soil in the VTA causes the greatest variability in the release of N, P, and COD. The calibration of the release of organic-N and P for mean particle diameter showed that a diameter of 1.5 µm produced the best agreement. The model was used for a lagoon and VTA system at two sites in Kansas with 35 years of weather data. The sites were located in Ottawa and Greenwood Counties, KS. The simulated results showed that the lagoon system performed better for both sites when comparing the mean values of total-N, total-P, and COD released over the 35 years. The VTA system performance was better when median values of total-P and COD were compared at Greenwood County. This model was designed for use in Kansas, but with modifications may be useful in other areas.