Chemical Dosing of Small Water Utilities Using Regression Analysis

Abstract

A method for determining chemical dosing based on multiple regression analysis is presented. Due to smaller operating budgets, small plants often lack the full‐time laboratory technicians and testing equipment necessary to assess rapidly changing influent conditions. Therefore, this method is particularly applicable to small treatment utilities. It was found that alum and prelime dosage levels can be predicted from correlations with dosage levels of other chemicals in the treatment operation, physical parameters describing the influent conditions, and previous values of the dosage. From the confidence limits of the predicted dosage, it is also possible to assign dosage levels based upon the risk of underdosing. A comparison with the effectiveness of streaming current detectors indicates several influent conditions under which the regression approach may be more appropriate. By basing treatment decisions on statistical models of previous successful treatment operation, plant efficiency will be maximized—resulting in fewer plant upsets and more consistent water quality.