4th UF Water Institute Symposium Abstract

   
Submitter's Name Christine Valcarce
Session Name Poster Session: Water quality protection and treatment
Poster Number 47
 
Author(s) Christine Valcarce,  UF (Presenting Author)
  David Mazyck,  UF
   
  Regression Modeling of PAC Performance
   
  Powdered activated carbon (PAC) is the most widely accepted control technology for the removal of taste and odor causing compounds in water treatment. Increasingly, this technology is being used to remove other trace organics, such as endocrine disrupting compounds (EDCs). PAC is often used because it offers advantages over other options such as membranes and advanced oxidation process in that it is not energy intensive, does not require large losses of residual water, and does not produce oxidation by-products that may be more dangerous than the parent compounds. PAC is advantageous for T&O and pesticides in surface waters because these problems are usually seasonal. Thus, PAC is applied when necessary and at dosages necessary to achieve desired removal. Although PAC is effective for trace organics, the complex nature of adsorption phenomenon limits most treatment operators to select PACs based on conventional manufacture’s specifications without taking into account the complex water characteristics and treatment process. This may result in improper performance or increased costs to meet finished water standards. On-going research aims to create predictive metrics for an easy and economical method of selecting an effective PAC for the removal of trace organics. Motivation stems from the likelihood of scrupulous regulations surrounding EDCs in the future. Furthermore, continual decreases in water quality will likely result in T&O issues becoming more frequent. Research goals include using multi-variable regression modeling to describe PAC’s performance holistically, converging the paradigms about the relative importance of physical and chemical attributes that affect adsorption. Criteria for model selection include: 1) providing 95% statistical confidence that regression coefficients of variables are useful in predicting removal 2) predictor variables must optimize the adjusted R2 and 3) model validation of future outcomes within the mean standard error of the model. Preliminary results indicate that regression modeling may be a powerful method for optimizing the selection of PACs in water treatment.