Date of Graduation


Document Type


Degree Name

Master of Science in Biological Engineering (MS)

Degree Level



Biological and Agricultural Engineering


Brian E. Haggard

Committee Member

G. Scott Osborn

Second Committee Member

Julian Fairey


Applied sciences, Earth sciences, Arkansas


This study investigated the effects of source water quality in Beaver Lake on the amount of chlorine (Cl) needed to develop decision support system to help guide chlorination practices in pre-treatment of source water. Chlorine demand assays were performed on water samples from Beaver Lake collected from the intake structure at Beaver Water District from March 2014 through August 2015, and using data from these assays, the two points of interest in this study were the Cl dose at which Cl residuals began to accumulate and the mean Cl demand occurring after that dose. Three methods of analysis were used to characterize relationships between source water quality, Cl dose, and mean Cl demand: (1) Pearson correlation, (2) nonparametric change point analysis (nCPA), and (3) regression tree (RT) analysis. The Cl dose and mean Cl demand generally increased with increasing concentrations of natural organic matter (NOM), iron (Fe), turbidity, TSS, and bacterial counts and generally decreased with increasing pH, conductivity, and alkalinity. Nonlinear, threshold-based relationships were also identified between Cl dose, mean Cl demand, and source water quality parameters, where Cl dose and mean Cl demand increased with NOM properties, Fe, manganese (Mn), total nitrogen (TN), temperature, turbidity, and TSS and decreased with increasing conductivity, pH, and alkalinity. Additionally, RT analysis revealed hierarchical structure to exist between Cl dose, turbidity (16.2 NTU), and pH (8.15 SU) which explained 35% of the variation in Cl dose as well as between mean Cl demand, turbidity (7.93 NTU), conductivity (138 µS/cm), sample date (May 8th), and TOC (2.65 mg/L) which collectively explained 66% of the variation in mean Cl demand. The relationships identified previously were incorporated into decision trees to potentially guide chlorination used in pre-treatment, which identified multiple ranges for Cl dose and demand based on thresholds in source water quality parameters.