Dealing with a Model Monster: How to Overcome Challenges to Modeling Complex Nutrient Removal Upgrades

Faced with a 1 mg/L Total Phosphorus effluent permit limit and an unusually high influent phosphorus concentration due to industrial discharges, the City of Salem, Ohio was forced to consider various advanced nutrient removal upgrade options for their existing 4 mgd secondary biological treatment plant with the following general process flow schematic (Figure 1).The City initially planned to install a coagulant chemical feed system to comply with the TP limit; however, upon further review of the option, the city became concerned with the significant increases in O&M cost and unmanageable sludge production levels that would result from its implementation. In addition, a prior demonstration project showed limited biological phosphorus removal performance in the activated sludge system. These concerns initiated a comprehensive plant evaluation to determine if process reconfiguration could achieve sufficient biological phosphorus removal (Bio-P) to minimize the amount of chemical addition, reduce sludge disposal tonnage, and improve overall plant operations. As a part of the evaluation process, computer modeling software (GPS-X, Hydromantis, 2011) was used to assess current conditions at the plant (Figure 2) and evaluate proposed process modifications for the WWTP. The modeling process was eventually helpful to designers in evaluating and refining design ideas for the plant, developing a three-phase upgrade plan to address solids handling bottlenecks, biological upgrades, and ancillary improvements including recycle stream treatment; however, the complexity of suggested design ideas made initial modeling runs too time consuming for practical use. Through the project's development, several techniques were utilized to increase the execution speed of model simulations such that model run times were decreased, without lessening the integrity of model results. This paper examines the various process configurations and modeling techniques used to reduce model run time, while still producing adequate modeling results to support evaluation of the design ideas.