Practical Application of NSF/ANSI 53 Lead Certified Filters: Investigating Lead Removal, Clogging and Consumer Experience

NSF/ANSI 53 lead-certified point-of-use filters (POUs) have been distributed to consumers in many cities facing water lead crises, including Washington D.C., Flint, MI, Newark, NJ, and University Park, IL. It is expected that these filters would reduce water lead to levels that are safe for consumption as residents wait for municipalities to provide more permanent solutions (e.g., corrosion control, lead service line replacement). These filters are certified by the National Sanitation Foundation (NSF) after meeting the challenges of treating two lab synthesized waters with 150 μg/L of soluble and particulate lead. In Flint, as in Washington, there were initial concerns that the filters would not be effective when exposed to lead levels far above the NSF/ANSI 53 150 μg/L Pb level used for certification. However, the EPA conducted a 2016 study in Flint, MI, with over 240 homes with lead up to 4080 μg/L, revealing that all POUs reduced lead levels below 1 μg/L.

Newark, NJ, in response to Lead and Copper Rule (LCR) violations, distributed over 40,000 NSF/ANSI 53 lead-certified pitcher and faucet POUs to protect consumers from high water lead levels. In the summer of 2019, preliminary tests in some homes with the highest lead in water concentrations revealed that 2 of 3 POUs used in Newark had effluent lead levels above 15 μg/L. The publication of these results caused citywide angst, distrust, and EPA mandated a switch to bottled water. However, a later and more extensive study revealed that 97.5% of homes (n=198) with properly used filters had effluent lead levels below 10 μg/L. As a result, the EPA approved Newark's request to discontinue bottled water distribution and only provide POUs to residents. Nevertheless, the experience indicated that it is vital to understand the limitations of POUs. This dissertation comprises three manuscripts that examine the efficacy of POUs under laboratory and field conditions.

The first manuscript sought to provide perspective into potential causes of the filter failures observed in the field. We conducted an extensive laboratory investigation that examined the performance of 10 pitcher and faucet POU brands under extreme conditions (e.g., up to 200% of rated capacity, influent lead levels ≈ 1000 μg/L). Our tests confirmed successful performance documented in some field testing and replicated underperformance observed in others. In this investigation, we observed structural failures due to poor manufacturing (i.e., leaking units, a filter with a large hole in the media) and performance failures (filtered water >10 μg/L Pb). Some of the performance failures occurred when we tested particulate lead waters, which we created, proving to be very difficult to treat relative to those used for NSF/ANSI testing. While the POUs almost always reduce consumer lead exposure, even when operated beyond their rated capacity, this study highlights instances where treated water could far exceed 10 μg/L lead.

High particulate iron (Fe) and manganese (Mn) concentrations often co-occur with high lead in many low-income, rural communities with small community water systems (CWS) or in homes with private wells. These communities are more likely to depend on POUs for protection from waterborne lead as they typically do not have the funds to maintain and upgrade infrastructure, improve corrosion control, or replace service lines. Waters with high levels of Fe and Mn could potentially impact the performance of the POU lead filters. However, such problems would not be detected in NSF/ANSI certification testing because these constituents are not included within the test water.

The second manuscript validated anecdotal reports of premature POU failure due to clogging in rural communities with high iron concentrations in their water. POU pitcher filters were tested with waters containing high lead and iron up to 100% of their rated capacity, or until they clogged as defined by a 75% reduction in initial flowrate. Iron levels above the 0.3 mg/L Secondary Maximum Contaminant Level (SMCL) resulted in rapid clogging, markedly increasing treatment costs, and decreasing consumer satisfaction. At 0.3 mg/L Fe, half of the 6 POU filters tested were clogged at between 38-68% of their rated capacity. When considering the cost of using POU filters vs. purchasing bottled water, the POU devices were often more cost-effective at iron levels at or below 0.3 mg/L. However, as iron concentrations increased, bottled water often became cost-effective depending on the circumstance. The presence of iron did not have an adverse effect on lead removal but significantly affected the cost and reduced flow rates in treating water.

The third manuscript presents a two-phase field study that sought to monitor the long-term filter performance in residential homes in New Orleans and Enterprise, LA. Previous field studies have captured POU removal efficiencies in single event (grab) samples; however, this study quantified filter performance for all the water treated up to POU practical capacity (i.e., filter life) based on consumer judgment regarding acceptable flow rate. The first phase was a rigorously controlled study that tested the POUs (100-gal capacity) at up to 200% of their rated capacity in two New Orleans unoccupied homes. Historically, the first home had consistently high lead levels (10-25 μg/L) even after flushing for > 8 min. Duplicate POUs treated that water to below 5 μg/L at up to 100% capacity, with only two exceptional samples with 12 μg/L Pb in 10-gallon batches of the treated water. The second home had a disturbed lead service line (LSL), resulting in varying concentrations of influent particulate lead ranging from 9-3000 μg/L. The duplicate POUs had difficulty producing water lead levels <10 μg/L before reaching filter capacity, with eight exceedances prior to 100% capacity. This work demonstrated that flushing alone for extended periods (>8 minutes) is not guaranteed to reduce lead levels in all homes with LSLs and highlights some limitations of POU filters in treating water with high levels of particulate lead.

The second phase of the field study monitored POU faucet filter performance in the homes of 21 residents in New Orleans (8) and Enterprise (13), LA. New Orleans is a large urban area with low to moderate water lead levels with many partial LSL replacements. Enterprise (population <300) is a rural, low-income community with an unincorporated water system with moderate to high water lead, iron, and manganese levels. Overall, the POUs consistently reduced lead to <1 μg/L, iron <171 μg/L, and manganese <180 μg/L. Enterprise's high influent concentrations of iron significantly impacted filter capacity due to reduced flow and clogging. Enterprise homes saw an average 62% flowrate reduction, and most of the homes did not reach 50% of the filter's rated capacity before consumers decided the filters were clogged. Most New Orleans residents did not experience clogging, and the homes that did saw only a 16% flow rate reduction. Overall, the New Orleans POUs were 2.3X faster in treating water by the study's end than Enterprise. There was no simple correlation between average iron concentration and days of filter life amongst residents in Enterprise as would be expected given variations in the volume of water used daily and consumer subjectivity in deciding when to end the study due to clogging. However, residents in Enterprise and similar communities would likely need to purchase 2-4 times as many filter cartridges due to clogging when compared to cities like New Orleans with lower iron concentrations. This study shows how POUs have promise for the removal of Pb and Fe in residential homes, but clogging has emerged as an important practical limitation to widespread successful POU deployment.

This dissertation highlighted the multifaceted nature of the question: "How well do POU filters work and under what conditions?" Overall, the POUs have shown their ability to reduce water lead levels effectively <5 μg/L, with a few exceptions primarily attributed to particulate lead and manufacturing quality control issues. However, when treating waters with high levels of iron and other contaminants, POU clogging can cause consumer dissatisfaction and make purchasing bottled water a more favorable solution than POU filters.