Arsenic in groundwater poses a risk to public health and may require treatment to meet drinking water regulations. Treatment is particularly challenging in oxic groundwaters that require simultaneous removal of arsenic with other co-occurring trace elements (i.e., chromium, vanadium, selenium, and uranium), in which case strong base ion exchange is the best-available, non-desalination technology. However, regeneration every 400–3000 bed volumes is required, because arsenic breaks through faster than other trace elements. Partial regeneration with 6 bed volumes of 0.2 N NaCl maintained arsenic and chromium treatment objectives over at least three consecutive loading-regeneration cycles by selectively recovering arsenic, (bi)carbonate, and sulfate. More than 99% of chromium was retained on the resin during partial regenerations until conventional regeneration. Compared to conventional regeneration, each cycle that uses partial regeneration simplifies the waste stream composition, decreases the relative corrosivity of the treated water, and decreases the total salt costs by up to 85% by using a lower mass of salt per regeneration. Staggered contactor operation may be necessary to manage nitrate leakage and chromatographic peaking for raw waters with high sulfate (e.g., >50 mg/L) and nitrate near regulatory limits (e.g., 10 mg/L–N). More broadly, comparisons between bench and pilot tests support the use of bench tests for evaluating process throughput but also identify a need for improved methodologies to better represent mass transfer, such as chromatographic peaking. Overall, this study demonstrates an effective method for simultaneously removing arsenic and co-occurring trace metals (e.g., chromium) that could be implemented at existing facilities. • Ion exchange is best method to remove co-occurring trace metals from drinking water • Salt use for regeneration decreases up to 85% using partial regeneration • Bench testing predicts capacity but not mass transfer (chromatographic peaking) • Current facilities can implement new method with existing infrastructure
Korak et al. (Sun,) studied this question.