Devens, with a population of about 6,500, is a 4,400-acre mixed-use community in north central Massachusetts adjacent to the towns of Ayer, Shirley, and Harvard. It occupies part of Fort Devens, a former US Army base that was deemed surplus and closed in 1996 after serving as the US Army’s New England headquarters for 79 years. The Massachusetts Development Finance Agency (MassDevelopment), the state’s development finance agency and land bank, subsequently purchased the property and, with financial support from the Commonwealth, has been redeveloping Devens into a sustainable and diverse community.
Although portions of the sprawling former base are still used as a training facility for military personnel, it has also become home to more than 400 residents and close to 100 businesses, nonprofits, and government organizations1. More than 6,000 workers are employed at various commercial enterprises such as a pharmaceutical research and development facility, a Hollywood movie sound stage, hotels, restaurants, and others.
Potential impacts from decades of military training and waste disposal activities at Fort Devens resulted in groundwater resources in the area becoming impacted by a variety of compounds, including several PFAS compounds. PFAS are a family of compounds that have been used for several decades in a wide variety of industrial and military uses, and because of their resistance to break down in the environment, they are sometimes referred to as "forever chemicals."
The Devens community is served by three drinking water wells that are all named after Army generals: the Shabokin Well, Patton Well, and MacPherson Well. Table 1 outlines the characteristics of each well, while Figure 1 shows a map of the well locations.
As shown in Table 1, the three wells typically have a total design flow rate of about 4,000 gallons per minute (gpm), while the maximum system demand is typically at or below 1,200 gpm. The system also helps support the Massachusetts Correctional Institution in Shirley, which often purchases water from Devens. With three essential supplies, Devens was quick to respond when PFAS impacts indicated the sources were at risk of being out of compliance.
Figure 1. Locations of Devens Water Supply Wells.
Discovering the Issue
In 2016, the US Environmental Protection Agency (EPA) released their first Health Advisory (HA) for two PFAS compounds – perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS). During this timeframe, water utilities in Massachusetts received letters from MassDEP to sample their water sources for PFAS, leading to Devens’ first sample event for PFAS at their wells in July 2016. The Patton and Shabokin Well sample results indicated that PFOA and PFOS combined concentrations were below the EPA’s 70 parts per trillion (ppt) HA, and the MacPherson well was approaching it at 69 ppt.
In June 2018, the Massachusetts Office of Research and Standards set their standard for the sum of five of the most critical PFAS compounds, including PFOA and PFOS, to not exceed 70 ppt. The MacPherson Well was then taken offline while the Shabokin and Patton wells were able to continue supplying drinking water. Although those two wells were still able to supply water, Devens needed to take action to enable use of the MacPherson Well, meeting their system redundancy agreements.
Prior to PFAS becoming the forefront of drinking water issues, Devens had been working toward implementing greensand filtration for the Shabokin and Patton wells. However, those efforts were put on hold while feasible PFAS treatment options were being considered. Emergency full-scale demonstration systems were selected as the most viable option and in February 2019, the efforts to mitigate the issue of PFAS in Devens began.
Potential impacts from decades of military training and waste disposal activities at Fort Devens resulted in groundwater resources in the area becoming impacted by a variety of compounds, including several PFAS compounds.
As with many water quality concerns, it is important to keep the public informed, which Devens did by releasing a public statement on their website in April 2019. Open, honest, and transparent messaging was key, and Devens clearly explained the reason for taking the MacPherson Well offline, in addition to the concentrations of PFAS compounds at the Patton and Shabokin Wells. They offered bottled water to those who were concerned and assured the public that they were working as quickly as possible to take the necessary steps to resolve the situation. With the issuance of a letter from MassDEP highlighting their proposed PFAS6 standard of 20 ppt, the team continued to work toward a game plan for each well source.
Emergency Actions
With all three source water wells impacted by PFAS, full-scale emergency treatment efforts began. The MacPherson Well was the first to utilize PFAS treatment, as concentrations were over the MassDEP’s 20 ppt maximum contaminant level (MCL) and the EPA’s 70 ppt HA. In addition, the water quality was not the only factor in determining the treatment technology to be used at the wells. General groundwater chemistry, available space, and environmental constraints also played a significant role in choosing treatment technologies at the different wells.
Figure 2. Timeline of Regulations and Events.
Raw Water Quality - All Wells
Although the wells are all located within three miles of one another, their water quality characteristics vary. This led to the need to test full scale treatment processes and analyze their effectiveness at each well. Table 2 shows the raw water quality at each of the well sites.
Pressure vessels with adsorption media such as granular activated carbon (GAC) or ion exchange (IX) resin were considered the best available technology for emergency PFAS treatment. Site constraints dictated the size and number of pressure vessels for each system. Reverse osmosis was considered but was quickly dismissed as an option due to the expense of the process, both capital and operational, along with the need to handle concentrated PFAS levels in the reject water. The raw water quality data from all three wells indicated the presence of iron and manganese. Although potentially fouling due to oxidation and precipitation within the media bed was anticipated, actual data on the rate and extent of fouling was unavailable from manufacturers and vendors. Demonstration piloting of various media types would prove useful in determining long-term remedies. In addition to water quality challenges, marketplace availability of large high-volume vessels impacted the ability to deploy emergency, full scale treatment.
Treating PFAS at MacPherson
Figure 3. The Original MacPherson Well Site.
With the highest PFAS concentrations, the MacPherson Well was the first of the three wells to be targeted for PFAS treatment. Pressure vessels are typically arranged in a lead/lag fashion and have significant space requirements. The MacPherson Well did not allow for such space, as the majority of the site is owned by the U.S. Fish and Wildlife Service. The land was deemed to be in a flood zone, so the treatment system would need to be on top of the slope east of the existing pump station. Figure 3 shows the well site and the available 12-foot by 16-foot space for PFAS treatment infrastructure. The land to the east and west of the fence surrounding the area slopes downward toward the flood zone. High volume sources with PFAS concentrations below 100 ppt and low Total Organic Carbon (TOC) concentrations had shown promise for the cost-effective removal of PFAS with GAC filtration. For these reasons, Norit GAC 400 was the chosen technology for the MacPherson Well. Additionally, elevated chloride levels discounted the long-term effectiveness of ion exchange resin media. A readily available 10-foot diameter pressure vessel was sourced from TIGG Corporation. Using 20,000 pounds of Norit GAC 400, a single vessel system with multiple side ports for performance monitoring was employed on a 12-foot square poured concrete slab.
With the location in the back corner of the site, surrounded by steep side slopes, two cranes were required to bring the vessel up and over the chemical feed building and the pump station to place it on the new concrete pad. The vessel was delivered laying on it’s side on a flatbed, so the two cranes worked together to rotate it vertically. Figure 4 shows the vessel being rotated by the cranes.
The system was brought online 108 days after the initial conversations with MassDEP. This process required considerable cooperation from MassDEP, US Fish and Wildlife, and the operations team (SUEZ employees now Veolia).
Figure 5. The Original Shabokin Well Site.
Treating Shabokin
Concurrently, PFAS treatment was developed at the Shabokin Well site. With the presence of iron and manganese at levels higher than those at the MacPherson Well, GAC vessels were used in a lead-lag configuration.
Space restrictions did not pose an issue when choosing GAC vessel configuration at the Shabokin Well. A large, cleared space adjacent to the well building was utilized for additional water treatment infrastructure (Figure 5). Two pairs of 12-foot diameter refurbished GAC pressure vessels were available from Calgon Corporation. The lead vessel was designated as "sacrificial" removing iron and manganese while protecting the lag vessel from these water quality impacts. The chosen vessels, though larger in diameter than the 10-foot vessel at MacPherson (12-foot diameter vs. 10-foot diameter), were significantly shorter, so they would readily fit in the final greensand water treatment plant (WTP) footprint design. Using the emergency filters in the final remedy provided approximately $1 million in cost savings.
Treating Patton
Figure 6. The Original Patton Well Site.
At the time treatment was being discussed for the Patton Well, GAC was successfully removing PFAS from the MacPherson and Shabokin well water and supplying sufficient flow to the system. Testing IX resins for effective PFAS removal was deemed a priority. This technology had proven to be successful at other sites in New England (including the Pease International Tradeport in Portsmouth, New Hampshire), and offered a treatment with a shorter empty bed contact time (EBCT) than GAC. As a result, the pressure vessels required for IX resin are much smaller than GAC contactors. Again, site constraints including an extremely steep driveway and a narrow area constrained by wetlands limited the deployment of large vessels. Multiple 4-foot diameter, 8-foot tall vessels could be placed at the site. Figure 6 shows the available space at the Patton Well.
The three separate IX vessels operating in parallel would allow treatment. Each vessel contained a different IX resin to compare removal efficiency: Purolite’s Purofine PFA694E, DOWEX PSR-2 Plus, and Calgon’s CalRes. These resins had been placed on the Massachusetts List of Approved Technologies for PFAS treatment in October 2018, March 2019, and October 2019 respectively. In addition, having vessels treat in parallel rather than in series would minimize system headloss while staying within the small footprint. The contactors and two bag filter assemblies were placed within steel shipping containers and piped together on site.
Implementation
Each PFAS treatment system was on a fast track to be online due to the emergency scenario. The MacPherson Well treatment system was first to go online, followed by the Shabokin Well, and finally the Patton Well. Figure 7 shows the series of events required to bring each system online. With all three wells up and running, the requirements of the Redundancy Agreement were met, and PFAS treatment was active.
Results and Ramifications
Each of the emergency treatment systems also acted as a pilot or demonstration study where information was gathered to inform operational optimization methods and the permanent full-scale design. As with most pilot studies, there were challenges requiring adjustments to the operations at each well. For example, detailed conversations were conducted with the treatment plant operators to ensure that any challenges they faced during the pilot study period were resolved in the final design.
Figure 7. Timeline of submitting design to DEP, implementation, and startup date for each emergency.
MacPherson
Combating Manganese Fouling
Manganese levels in the raw water at the MacPherson Well proved to be the main issue that affected operations. Due to elevated or rising differential pressures in the pressure vessel, a unique backwashing method was developed. The use of clean system water, combined with an extended soaking period prior to backwash, effectively reduced PFAS discharge within the backwash waters. The operators were forced to backwash the system every two months on average, depending on the flow rate and seasonal demand, as backwashing affects the mass transfer zone (MTZ). In a perfect media bed, the MTZ begins at the top of the bed and slowly moves down. However, when the bed is backwashed, the media gets mixed, disturbing the MTZ. Adsorption and mass removal of PFAS are still able to take place even when the MTZ is affected, but it is much more difficult to track PFAS removal effectiveness through the media bed.
Pilot Study Results
By monitoring PFAS levels in ports installed at 25, 50, and 75 percent of the media bed depth, general performance of the GAC filter was obtained. When PFAS levels in the 75 percent port exceeded the 20 ppt concentration, the media was replaced. Media was only replaced once in 2.5 years. Hydraulic performance of the filter vessel was returned to normal or original differential pressure following backwash cycles.
The filter showed continued PFAS mass removal even after multiple backwash cycles. This indicates that while optimum performance of a single filter vessel may not be possible if backwashed, manganese impacts and continued PFAS mass removal could be managed using a lead or primary filter as a sacrificial or protective filter in a process train.
Although there were some operational challenges during the pilot study, GAC proved to be successful in treating PFAS. Figure 8 shows PFAS trends over time. The green line indicates the date when the media was changed out. The spike in PFAS6 concentrations at the 75 percent port after the changeout may have been due to minor channelization in the media bed. Samples were not taken from the 25 percent port after August 2021 as concentrations were over 20 ppt in that port. Instead, the sampling location was moved to the 50 percent and 75 percent ports. The system was taken offline in the spring of 2022 when the final WTP was brought online.
Figure 8. PFAS trends over time in the MacPherson Well.
Final Design
The footprint of the final plant was constrained to the shoulder of the long narrow access road opposite the chemical feed building, with the design dictating a two-filter GAC system in series. The lead vessel could be maintained in a similar fashion to the single or demonstration pilot vessel, with periodic backwash designed to remove oxidized manganese. The backwash water was routed to the nearby sewer. Additional protection from particulate metals was realized through the installation of 10-micron bag filters prior to the pressure vessels.
The chosen vessels were two 12-foot GAC vessels operating in series, with each one containing 30,000 pounds of media. With this configuration, the full system flow could be treated. Figure 9 shows components of the full-scale MacPherson WTP.
Figure 9. MacPherson Water Treatment Plant Components.
Patton
Combating Manganese Fouling
Figure 10. Magnified image of resin media at the Patton Well treatment system fouled with manganese.
Manganese fouling was also prominent at the Patton Well site. The majority of the manganese concentrations were present as dissolved manganese with very low concentrations due to particulate manganese. However, the dissolved manganese oxidized on the resin media. Once oxidation occurred, accelerated oxidation and subsequent precipitation on the resin media occurred resulting in rapidly increasing differential pressures in the pressure vessels. Figure 10 shows a magnified photo of the resin media fouled with manganese. This media shows evidence of manganese oxidation and coating on the media material even after it had been rinsed with a mild acid.
To combat this issue, a sequestering agent was added to the pilot system and injected upstream. The sequestering agent maintained manganese in solution allowing it to pass through the resin media. Addition of the sequestering agent significantly extended the time between backwashes and diminished the oxidation occurring on the resin media. However, in order to eliminate the need to backwash the single-pass resin media, an additional modification to the pilot process was undertaken. GAC pressure vessels (3) were placed upstream of the IX vessels to act as a pre-treatment for the manganese and provide additional beneficial PFAS removal. Based on a 5-minute EBCT, manganese should oxidize in the GAC and be removed before reaching the IX resin vessels. Following the installation of the GAC filters, new resin was installed, giving the system a fresh start.
Since the installation of the GAC vessels, the differential pressures across each IX resin adsorber remain low. The new pilot configuration placing GAC upstream of IX resin to protect the resin from fouling has shown success in allowing the IX resin to focus only on PFAS removal. This data suggests that use of GAC as a pretreatment may have dual benefits of metals removal and minor PFAS adsorption while protecting the resin media and prolonging media life before premature exchange. While this application appears to be effective, it may only be appropriate at moderate to lower concentrations for iron and manganese. Media exchange costs and the viable life of the GAC are yet to be determined.
Pilot Study Results
Though there were operational issues, the pilot study results revealed that IX resin is successful in treating PFAS. The pilot system is still in operation and will continue to treat raw water from the Patton Well until the final WTP is online. Figure 11 shows PFAS results at the Patton Well site. As shown, two of the resins show similar performance trends, while the resin depicted in purple consistently treated water to non-detect. In addition, the results that are shown are representative of the system before GAC was implemented, as there is not yet sufficient data to present on the results of prefiltration using GAC.
Final Design
The Patton pilot testing revealed several important results. First, ion-exchange resins were effective at removing PFAS compounds even after backwashing and disturbance. Sequestration prior to resin medias can have a beneficial effect on oxidation within the resin media beds. However, low concentrations of iron and manganese can be removed using GAC as a sacrificial media ahead of ion-exchange media. The final system design allows for greensand for metals removal, protecting PFAS removal media.
Figure 11. PFAS6 Results at the Patton Well.
Shabokin
Figure 12. Bag filter fouled with iron.
Combating Iron Fouling
At the Shabokin site, issues arose due to the iron concentrations in the raw water that increased the need for maintenance at the site. Frequent backwashing due to iron fouling was not controlled even after the installation of bag filters upstream of the GAC adsorbers. Bag filter fouling and GAC fouling indicated both particulate and dissolved iron is present. Figure 12 shows the fouled bags.
Like the Patton pilot, a sequestering agent was added to the pilot process and injected into the raw water to bring the iron into dissolved form. The sequestering agent significantly decreased the need to backwash the primary GAC vessels.
Pilot Study Results
The Shabokin pilot system, and its four large diameter GAC vessels, were successful in demonstrating the use of a lead GAC vessel to manage iron and manganese. For almost three years, backwashing the lead filter allowed the lag filter to remain as an undisturbed polishing filter. Sequestrant successfully improved operational conditions with no discernible effect on GAC removal performance of PFAS compounds. Figures 13 and 14 show the progression of PFAS concentrations in the lead vessel for each of the two trains, train A and B. Once PFAS levels were elevated, monitoring moved to the 50 percent port and then to the midpoint between vessels. Despite the repeated backwash of the lead vessel, concentrations indicate continued mass removal while concentrations remain fairly stable below less than 7.5 ppt leaving the first vessel. Effluent from the lag vessel remained below detection limits for the regulated PFAS compounds.
Figure 13. Shabokin Pilot Train A PFAS6 Results.
Figure 14 reveals similar results for the second pair of filters in train B with the exception of an anomalous spike in late 2019. Train B decommissioned on October 27, 2022, and transported to the permanent Patton WTP.
Train B decommissioned on October 27, 2022, and transported to the permanent Patton WTP.
PFAS concentrations were present in the 25 percent, 50 percent, and 75 percent ports of Trains A and B lead vessels at similar points, which shows consistency in the GAC media. The graphs include breakthrough curves of a typical shape, which reflects that backwash events were not as frequent as they were at the MacPherson site. Train B was taken offline and moved to the permanent WTP site in October 2022. Consequently, Train A remained online to assist the MacPherson and Patton wells in supplying the distribution system. However, with PFAS6 concentrations trending upward in the lead vessel, the flowrate through the train was decreased to 350 gpm to avoid the need to replace the media prior to its upcoming decommission, which is scheduled to take place in the Spring of 2023.
Figure 14. Shabokin Pilot Train B PFAS6 Results
Final Design
The final design for the permanent Shabokin WTP will utilize greensand upstream of the GAC to remove iron and manganese. GAC media followed by ion-exchange resins will then remove PFAS. Pilot testing results indicate that GAC media adds flexibility and protection to the ion-exchange resin media improving bed or media life. This hybrid design provides a more robust treatment process. While GAC alone was effective, the combination of both media types in series should prove even more effective for future, anticipated reductions in regulatory limits for PFOA and PFOS.
Conclusion
When completed, three new water treatment facilities will have been completed following the short-term operation of full-scale pilot systems to mitigate PFAS contamination of Devens’ drinking water. Many public and private parties have cooperated and worked together toward the common goal of providing the population of Devens and surrounding communities with safe drinking water. The three separate full-scale pilot systems were successfully designed, constructed, and implemented to remove PFAS from three different groundwater sources in extremely short time frames. Their operation, and the experimentation with sequestrants, multiple backwashing/maintenance efforts, and the use of multiple media types allowed the delivery of acceptable drinking water while final solutions could be implemented.
Sara Francis, EIT
Engineer II
Weston & Sampson
55 Walkers Brook Drive
Reading, MA 01867
978.532.1900
francis.sara@wseinc.com
Vice President and Water
Market Leader
Weston & Sampson
55 Walkers Brook Drive
Reading, MA 01867
978.532.1900
martinb@wseinc.com
Published in Journal of The New England Water Works Association, March 2023.