Optimisation of coagulation process in water treatment plant using statistical approach

Abstract

The typical jar test practice at water treatment plants to determine the optimum coagulation conditions is statistically inappropriate as it is based on onefactor- at-a-time (OFAT) approach. This approach, which does not cover the whole experimental space, could miss the actual optimum values results in higher cost and less effective coagulation performance. Additionally, the jar test exercise is time consuming and is therefore conducted once a day or when it is really needed. This study carried out a long-term comparison between the use of Response Surface Method (RSM) and the traditional (a modified OFAT) jar test in optimising the coagulation process. The study was carried out at Bandar Tenggara WTP (BTWTP), Sungai Gembut WTP (SGWTP) and Sri Gading WTP (SRGWTP) as these WTPs have different raw water characteristics. Characterisation of natural organic matters (NOMs) in terms of dissolved organic carbon (DOC) and UV254 and their removal via coagulation process were explored. The relationship between the raw water characteristics and coagulation optimum conditions were also developed. The results showed that the RSM and traditional approaches acquired almost identical optimum coagulation conditions at BTWTP and SGWTP. At SRGWTP, RSM technique was found to be better than the traditional method as the coagulant (alum) and flocculant (polymer) optimum concentrations produced by the RSM technique were respectively 50% and 20% of the optimum values obtained by the traditional method. The raw waters of BTWTP and SGWTP were identified as hydrophilic, non-humic, with low molecular mass of NOMs. At optimum coagulation conditions (based on turbidity removal), the average DOC and UV254 removals were about 11% and 70%, respectively. The models predicting optimum pH and coagulant dosing from raw water quality data were formulated from the RSM and traditional approach and was found to be reliable. From the models, it was found that NH3–N in the raw water at BTWTP and turbidity, Mn and NH3–N in the raw water at SGWTP did not affect the optimum coagulation conditions. The detail optimisation model derivation procedures were also developed.