Formatted Title

Sorptive Media Selection for PFAS Treatment in Drinking Water Using Rapid Small-Scale Column Tests: A Case Study

Background/Objectives

Removal of per- and polyfluoroalkyl substances (PFAS) from drinking water sources is of significant interest as many regulatory jurisdictions have established advisory or regulatory limits. In particular, Health Canada has recently published a draft objective of 30 ng/L for the sum of total PFAS detected in drinking water and United States Environemntal Protection Agency (EPA) proposed a National Primary Drinking Water Regulation (NPDWR) to establish legally enforceable levels, called Maximum Contaminant Levels (MCLs), for six PFAS in drinking water. Currently, the most effective treatment technologies for PFAS removal from drinking water sources are granular activated carbon (GAC), reverse osmosis (RO), nanofiltration (NF) and anion exchange (IX). Recently, new adsorptive media (i.e., clay-based sorbents or polymeric sorbents) have also been tested for PFAS removal.
 

Approach/Activities

Generally, the key consideration when selecting PFAS drinking water treatment technologies is the presence of competing anions and co-contaminants, organic matter and the frequency of regeneration or changeout required for the sorptive medium used. However, effectiveness of different media for PFAS removal can take months to years at pilot scale to achieve breakthrough. As a result, rapid small-scale column tests (RSSCTs) are being used to develop potential design criteria and to estimate the replacement frequency of an adsorbent at a full-scale treatment plant. Two GAC media, two IX resins and one specialty adsorbent were evaluated in this study. All selected adsorptive media are certified through National Sanitary Foundation (NSF) 61 rating. The design of the RSSCT column was completed by selecting the media size to use and the operating conditions that are hydraulically similar to the anticipated full-scale system's design and operation using the constant diffusivity scale-up model.

Results/Lessons Learned

The sum of the PFAS measured above the minimum reporting limit (MRL) of 2 ng/L using EPA Method 537.1 in the source water was 71.05 ng/L. Based on the results of RSSCT, Calgon F400-1 and Evoqua UC1240LD had the poorest removal performance among all media tested. Calgon CalRes 2301 had better PFAS removal performance than both GAC adsorptive media, but it achieved PFAS breakthrough quicker than Evoqua PSR2+ IX media during the operation of the RSSCT. Evoqua PSR2+ and Fluoro-Sorb® (FS) 200 appeared to have the best PFAS removal performance with no clear differentiation between them. The secondary goal of this study was to provide a relative range of expected media life with the understanding that RSSCT does not account for real-world operating conditions including inorganic or organic fouling of the media and varying raw water quality. For this study, the volume of water treated to reach 60% exhaustion using Yoon-Nelson empirical model was used to forecast the full-scale breakthrough data and media changeout projection.

Based on RSSCT results, the best performing media (FS) is projected to take approximately 22 months to reach 60% lead-bed exhaustion on the full scale. For IX, the PSR2+ and CalRes 2301 are projected to take approximately 17 and 11 months to reach 60% lead-bed exhaustion on the full scale, respectively.

Overall, the RSSCT and breakthrough modeling were coupled to provide insight on adsorbent performance of low-concentration short- and long-chain PFAS in a drinking water source. This work gives a detailed insight in designing a RSSCT model on a laboratory scale, scaling adsorption capacity of PFAS by various adsorbents and to predict a full-scale breakthrough. Further, evaluation of adsorption was extended beyond PFAS breakthrough with the inclusion of unit adsorbent cost as a decision metric influencing adsorbent selection. Given the media performance projections and conceptual-level cost estimates, treatment using IX or FS are found to be the most cost effective.