Is two stage GAC better than one stage GAC for removing PFAS at a DWTP? : Investigation of PFAS removal from drinking water using two stage granular activated carbon (GAC) filter

Sammanfattning: The removal of 34 per- and polyfluoroalkyl substances (PFAS) were compared in a 1 stagegranular activated carbon (GAC) filtration to a 2 stage GAC filtration in a pilot study at adrinking water treatment plant (DWTP). The PFASs that were present in the water wereperfluorooctanoic acid (PFOA), perfluorooctane sulfonic acid (PFOS), perfluoropentanoic acid(PFPeA), perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), perfluorobutanesulfonic acid (PFBS), perfluoropentane sulfonic acid (PFPeS) and perfluorohexane sulfonic acid(PFHxS). A cost comparison for the operation of a one stage GAC to a two stage GAC wascompared for PFAS4 (sum of PFOA, PFNA, PFHxS and PFOS) at treatment goals ranging from2 - 10 ng/L. The pilot was operated at three different flows and the three different bed volumes(BV)s resulting in three different empty bed contact times (EBCTs) at three different times.Therefore, the Lin & Huang adsorption model (1999) was used to model the concentrations ateach EBCT. It was found that the model worked good for PFBS, PFPeS, PFHxS, PFOS andPFOA but not for PFPeA, PFHxA and PFHpA (except for PFPeA and PFHxA during EBCT 5min) and did not work for desorbing PFASs. The removal comparison of PFASs was made,partly by comparing removal efficiencies between the first stage and the second stage GAC filterand by comparing the removal per weight of GAC per BV 1 stage and 2 stages. It was found thatthe removal efficiency decreases with decreasing chain length and increasing treated BVs forboth the first stage and the second stage. The short chain PFCAs were also desorbing after anumber of BVs. The removal per weight of GAC showed that the removal does not increasewhen comparing a one stage GAC to a two stage GAC for any the PFAS. The cost comparisonwas made using the adsorption model. It showed that it was cheaper to operate a 2 stage GAC forthe EBCT of 5 minutes and 8 minutes for the whole range of treatment goals. However, for theEBCT of 15 minutes the costs for the second stage was decreasing with decreasing treatmentgoal which is unrealistic result. This was caused by too few data points available for the model topredict reliable results.