Abstract
|
:
|
Activated sludge (AS) processes are one of the most common methods used to treat municipal wastewater (WW). Mathematical models have been used to predict the complex interactions of AS processes. Because of the complexity of interactions and the diversity of microorganisms involved in WW treatment, an accurate description of the processes can result in extremely complex models. From practical and operational viewpoints, these complex models may not be very useful. For example, wastewater characterization in a manner suitable for use in such complex models is extremely involved and is not done routinely at wastewater treatment plants. This research hypothesizes that superposition of simplified models in wastewater treatment along with generalized wastewater characterization techniques are capable of accurately predicting characteristics of wastewater effluent. This research evaluates this hypothesis by proposing and evaluating a simplified model based on a combination of the International Association on Water Quality (IAWQ) Activated Sludge Model 1 (ASM 1), simplified WW characterization methods, and a secondary clarifier model to predict the chemical oxygen demand (COD) and ammonia concentrations in wastewater effluent. Activated Sludge Model 1 includes the fundamental processes in a single-sludge system that carries out carbon oxidation, nitrification and denitrification. It involves eight processes incorporating thirteen soluble and particulate components. Two treatment plants in central Mississippi (Trahon and Savanna) were used to evaluate the model. Trahon treatment plant utilizes an oxidation ditch (OD) process, and Savanna treatment plant utilizes a contact-stabilization (CS) process combined with oxidation lagoons. The lagoons are used to store WW that exceed the treatment capacity of the plant. Some treatment is expected to occur in the lagoons, therefore, the lagoons were considered in the model. The number of reactors used to simulate the conditions in the OD process was considered and adjusted during the calibration. The components that contributed to the suspended solids (SS) concentrations in the clarifier's effluent were calibrated based on the SS historic data. Monthly averaged data (February 1996 to May 1998), based on daily analysis and records, was collected from the treatment plants. The data was divided into two sets: one set (February 1996 to September 1996) was used to estimate the kinetic and stoichiometric parameters, and the other set (October 1996 to May 1998) was used to evaluate the model. Sensitivity analyses of the parameters revealed that the maximum specific growth rate of autotrophs and decay rate of heterotrophs were the most sensitive parameters in ASM 1. The two characterization methods, fixed fractions and dependent fractions, were based on fractionating the total COD and TKN into biodegradable, non-biodegradable, soluble, and particulate components that can be used in ASM 1. With few exceptions, the model predicted COD and ammonia effluents similar to the reported data in trend. The largest difference between the predicted and reported COD removal efficiency using the fixed fractions wastewater characterization method was 3.53% for the OD and 7.89% for the CS; and the largest difference for ammonia removal, using the fixed fractions method was 11.25% for the OD and 26.58% for the CS. The most likely reason for the differences (exceptions) is inhibitions encountered by the microorganisms responsible for organic matter and ammonia removal which were not accounted for in the model. The fixed fractions method predicted higher concentrations of biodegradable COD, organic nitrogen, and inert soluble COD in the influent than the dependent fractions method. However, the difference did not affect the results significantly. The organic substrate needed for denitrification originates from readily biodegradable substrate in the influent WW, from hydrolysis of slowly biodegradable particulate substrate throughout the AS, and from endogenous respiration of the biomass. Using dai y data from the treatment plants, the model developed in this research was compared to a model developed by Dupont and Henze (1992). Based on the COD and ammonia effluent concentrations and using the average relative deviation (ARD), the simplified model predicted concentrations closer to the reported than Dupont and Henze model. The COD ARD for Dupont and Henze model was 35.61% for the July 1990 period and 53.29% for the January 1991 period. For ammonia, the ARD was 242.26% and 21.00% for the same respective periods. For the Trahon plant, the simplified model COD ARD using the fixed fractions method was 36.72% while the ammonia ARD was 94.44%. For the Savanna plant, the simplified model COD ARD using the fixed fractions method was 34.31% while the ammonia ARD was 69.64%.
|