Date of Award
Doctor of Philosophy (PhD)
The United States Environmental Protection Agency (U.S.EPA) has identified nutrient pollution as the leading cause of use impairment in U.S. waters. Consequently, for improved nutrient management, the U.S.EPA recommends ensuring point sources comply with their permit limits. However, there is limited understanding of compliance with discharge limits, primarily due to great spatial and temporal variability in effluent nutrient concentrations as well as discharge permit limits. Further, the regulatory climate for nutrients is rapidly changing in several states of the country, with the adoption of more stringent discharge limits for wastewater treatment plants. This research presents a performance-based statistical modeling approach to understand the spatial and temporal variability of nutrient compliance (specifically nitrogen species) with changing regulations, in treated wastewaters of the United States.
A hierarchical model is built using Generalized Linear Models (GLMs) and Kriging, and effluent ammonia concentrations from Discharge Monthly Report (DMR) data from more than 100 wastewater treatment plants across the US. Compliance with current ammonia permit discharge limits is seen to be determined by the flow rate and its compliance history. The probability, frequency and magnitude of risk of non-compliance with ammonia discharge limits is modeled using GLMs and Extreme Value Theory (EVT). The probability, frequency and magnitude of risk of non-compliance with ammonia discharge limits is found to be determined by both the flow rate and compliance history, in addition to the fractional use of design capacity. Wastewater treatment plant compliance with decreasing ammonia discharge limits is assessed using a regression trees. Once again, the compliance history and the flow rate are seen to affect compliance with both existing and lowered discharge limits. Some states, such as Colorado, are considering broader regulations, for all nitrogen species, by regulating levels of Total Inorganic Nitrogen (TIN) in effluent wastewaters. A Hidden Markov Model (HMM) and multinomial logistic regression based modeling framework is presented to predict TIN concentrations in treated wastewaters, using data from an operating wastewater treatment plant in Colorado, US. Effluent TIN concentrations are found to be a function of climate variables (such as minimum air temperature and precipitation), seasonality, effluent ammonia concentrations and effluent TIN concentrations in the previous month.
The performance-based models presented in this research can be beneficial to several stakeholders; they can be useful for predictive purposes or reliability analysis on both a single treatment plant or multi-plant level. While they can help individual plant operators ensure compliance with changing nutrient regulations, monitoring and enforcement efforts can now be better channelized towards frequent and egregious violators. Additionally, for various proposed (lowered) discharge limits, these models can be implemented to delineate reliable sources of demand and supply for a point source-to-point source nutrient credit trading scheme.
Suchetana, Bihu, "Performance-Based Modeling of Spatial and Temporal Variability of Treated Wastewater Quality for Improved Nutrient Management" (2018). Civil Engineering Graduate Theses & Dissertations. 374.