Multiobjective Optimization Models for Distributing Biosolids to Reuse Fields

The District of Columbia Water and Sewer Authority (DCWASA) operates the Blue Plains Wastewater Treatment Plant located in Washington, DC. It serves more than two million Washington Metro Area customers, and treats more than 330 million gallons a day of raw sewage from area jurisdictions, including Montgomery and Prince George's Counties in Maryland, and Fairfax and Loudoun Counties in Virginia. Each day, DCWASA produces approximately 1,200 tons of biosolids or byproducts of wastewater that have been treated to reduce pathogens and can be used as fertilizer for agricultural purposes. These generated biosolids require removal from the treatment facility and distribution to reuse fields located in Maryland and Virginia. In spite of the benefits of reuse, biosolids are generally considered by many as potentially malodorous. Recently, DCWASA has received complaints from the surrounding communities and needed to minimize biosolids odors. However, trying to minimize biosolids odors could result in costly treatment processes. Therefore, one needs to determine how to minimize the odors while at the same time minimizing the treatment costs. This compromise of balancing the competing objectives of odors and costs results in a two-objective or more generally, multiobjective optimization problem. In this dissertation, we develop multiobjective optimization models to simultaneously minimize biosolids odors as well as wastewater treatment process and biosolids distribution costs. A weighting method and constraint method were employed to find tradeoff, so called Pareto optimal, points between costs and odors. Schur's decomposition and special order set type two variables were used to approximate the product of two decision variables. A Dantzig-Wolfe decomposition technique was successfully applied to break apart and solve a large optimization model encountered in this dissertation. Using the Blue Plains advanced wastewater treatment plant as a case study, we find several Pareto optimal points between costs and odors where different treatments (e.g., lime addition) and biosolids distribution (e.g., to what reuse fields biosolids should be applied) strategies should be employed. In addition, to hedge the risk of equipment failures as well as for historical reasons, an on-site dewatering contractor has also been incorporated into the model. The optimal solutions indicate different uses of the contractor (e.g., percent flow assigned) when dewatering cost employed by DCWASA varies. This model can be used proactively by any typical advanced wastewater treatment plants to produce the least malodorous biosolids at minimal costs and to our knowledge, this is the first instance of such a model