Institute: District of Columbia
Year Established: 2013 Start Date: 2013-03-01 End Date: 2014-02-28
Total Federal Funds: $12,056 Total Non-Federal Funds: $38,624
Principal Investigators: Arash Massoudieh
Abstract: Activated Sludge Models (ASMs) are widely used for the design and optimization of various unit processes in wastewater treatment plants. As mechanistic models, their main goal is to predict the water treatment plant behavior under different conditions, and they are employed as tools for design and operation optimization of activated sludge systems. Optimization of Activated Sludge processes can substantially reduce the energy use and the amount of BOD and nutrients being discharge into receiving waters. This issue becomes even more important for plants that treat combine sewage and therefore experience a high level of fluctuations in both the volume and composition of wastewater (e.g. in District of Columbia). Optimization of the operation of Activated Sludge system in Blue Plains wastewater treatment plant is even more important in face of the efforts to restore the Chesapeake Bay. The outcomes of ASM models are directly influenced by various sources of uncertainty including but not limited to the uncertainty associated with model parameters. One of the most important challenges in making ASM models applicable to design problems is identifying the values of its many stoichiometric and kinetic parameters. This task is essential in designing optimal processes while minimizing the risks of exceeding effluent criteria. The goal of this project to develop an automatic Bayesian parameter estimation method for probabilistic estimation of activated sludge process parameters, utilizing the extensive data collected at the Blue Plains Wastewater treatment plant located in Washington DC. The tool that will be developed will include a flexible reactor modeling system combined with a Bayesian parameter estimation method that is able to consider the different sources of uncertainties in the system. As opposed to the deterministic parameter estimation models used in the past that provide point estimates of the ASM model parameters, the proposed Bayesian technique provides the Probability Density Functions (PDFs) of the reaction expression parameters. When the PDFs of the process parameters are identified the outcome can be used for risk-based design and optimization of biological treatment systems. The developed Bayesian parameter estimation framework will be applied to a wide range of data collected at Blue Plains WWTP in the past to determine the most representative model structures and the confidence intervals around the parameter estimates. The application of this tool will improve the experimental setups in order to achieve higher confidences about the parameters and the forms of rate expressions. Also the quality of the ASM predictions will improve and the results can be used in risk-based decision making. Optimizing the processes in the Blue Plains waste water treatment plan will result in minimizing the constituents (particularly nutrients) going into Potomac and reducing the energy consumption and cost of operation at Blue Plains and therefore helping the economy of DC citizens. The outcome of this study will be used as preliminary data for a larger proposal to EPA, NSF or the District Department of Environment.