Title:

Machine Learning application in predicting nitrogen discharge from a local wastewater treatment plant

Tracking nitrogen and other chemical levels is important for wastewater treatment plants to limit pollution discharge to surface water sources. It also allows them to discover trends and address rising nitrogen levels before they reach limits set by the EPA, protecting the public and environmental health. However, infrequent sampling of the major nitrogen species (e.g., ammonia, nitrate, nitrite, etc.) prevent an accurate depiction of the effluent nitrogen. This work seeks to predict nitrogen discharge from a local wastewater treatment plant using machine learning algorithms. Data collected over 5 years with more frequent sampling such as wastewater flow, biological oxygen demand, temperature, and precipitation data was used to establish variable correlation and predict nitrogen levels in the effluent that allow for a more accurate depiction of nitrogen levels. Training machine learning algorithms with a variety of these inputs can help municipalities better alter their practices to meet the EPA standards. This research demonstrates how machine learning can be used to predict nitrogen discharge and other effluent factors in water treatment plants more accurately. The results from this work will help guide a larger effort in understanding economical and energy requirements for decreasing point source nitrogen contributions to the Great Bay.

Name

Comment*

Comments are viewable only by submitter

Captcha*