Jump to main content.


  Arsenic Removal From Drinking Water by Iron Removal and Adsorptive Media, U.S. EPA Demonstration Project at Stewart, MN Final Performance Evaluation Report (EPA/600/R-09/144) December 2009

This report documents the one-year EPA arsenic-removal technology demonstration project at the Stewart, Minnesota, facility. The main objective of the project was to evaluate the effectiveness of Siemens’ Type II AERALATER system for iron removal, and AdEdge Technologies’ Arsenic Package Unit-300 system for subsequent arsenic removal. The ADedge system was evaluated based on its ability to remove arsenic to below the new arsenic Maximum Contaminant Level of 10 micrograms per liter (μg/L). This project also:

  • Evaluated the reliability of the treatment system for use at small water facilities
  • Determined the required system operation and maintenance (O&M) and operator skill levels
  • Characterized process residuals generated by the treatment process
  • Determined the capital and O&M costs of the technology

The types of data collected included system operation, water quality (both across the treatment train and in the distribution system), process residuals, and capital and O&M costs.

The 250 gallons per minute (gpm) treatment system consisted of an AERALATER pretreatment unit and an arsenic package unit (APU)-300 arsenic removal unit. Used for iron removal, the 11-foot by 26-foot carbon-steel AERALATER package unit was composed of an aeration tower, a detention tank, and a four-cell gravity filter in one stacked, circular configuration. The effluent from the gravity filter was subsequently polished with AD-33 media, an iron-based adsorptive media developed by Bayer AG for arsenic removal. The APU-300 system consisted of two skid-mounted 63-inch by 86-inch fiberglass vessels configured in parallel. Each vessel contained 64 cubic feet of pelletized AD-33 media supported by gravel underbedding.

The treatment system began routine operation on February 2, 2006. Through the end of the performance evaluation study on February 28, 2007, the system treated approximately 20,441,000 gallons of water with an average run time of 4.7 hours per day. The average daily demand was 52,418 gallons. Water to the treatment system was supplied by two wells (Wells No. 3 and 4), each operating at an average flowrate of 191 and 184 gpm, respectively, on an alternating basis. These reduced flowrates resulted in longer detention times (45 to 46 minutes versus the design value of 34 minutes) within the AERALATER detention tank and lower hydraulic loading rates (2.0 to 1.9 gpm per square foot (ft2) versus the design value of 2.6 gpm per ft2) to the gravity filter. The corresponding flowrates measured through the APU-300 system also resulted in longer empty bed contact time (5.4 minutes compared to the design value of 3.8 minutes) in each vessel. No significant operational or mechanical issues were experienced during the one-year performance evaluation study period. However, four months after the end of the performance evaluation study, the operator reported biofouling of the AERALATER filter that necessitated the use of chlorine to clean the filter media and re-injection of a previously selected, but later abandoned, oxidant (sodium permanganate [NaMnO4]), to oxidize soluble arsenic(III).

The source water contained 31.4 to 56.4 μg/L of total arsenic, with soluble arsenic(III) at an average concentration of 35.3 μg/L as the predominant species. To oxidize soluble arsenic(III), NaMnO4 was selected due to the presence of elevated total organic carbon (TOC) (6.4 milligrams per liter [mg/L] on average) and ammonia levels (1.6 mg/L [as nitrogen] on average) in raw water. Based on February 2, 2006, data, 90 percent of soluble arsenic(III) was oxidized to soluble arsenic(V) when NaMnO4 was added prior to aeration. Soluble arsenic(V) was then adsorbed onto and/or co-precipitated with iron solids, resulting in 57 percent soluble arsenic(V) removal. The arsenic-laden iron solids were effectively removed by the gravity filter, achieving approximately 60 percent total arsenic and 100 percent total iron removal. The remaining arsenic was present mostly as soluble arsenic(V) at 26.4 μg/L (on average), which was subsequently removed by AD-33 media. The elevated soluble arsenic(V) in the gravity filter effluent was most likely caused by the relatively high levels of pH (7.9 on average), competing anions (such as phosphorous [301 μg/L (as phosphorus) on average] and silica [25.1 mg/L (as silica) on average]), and TOC in source water.

After one week of operation, NaMnO4 addition was inadvertently discontinued due to problems with the chemical feed pump. It was subsequently decided to operate the system without NaMnO4 addition due to the discovery of microbial-mediated arsenic(III) oxidation processes and elevated manganese levels (e.g., 127 μg/L on February 2, 2006) in the gravity filter effluent. The elevated manganese concentrations in the gravity filter effluent were attributed to the formation of colloidal manganese dioxide in the presence of TOC. Elevated manganese levels have been shown to be detrimental to AD-33 media, based on studies at other EPA demonstration sites where high manganese loadings were found to coat and/or foul AD-33 media in the presence of chlorine.

Without NaMnO4 addition, the total arsenic removal rate averaged 34 percent and the iron removal rate was 100 percent across the gravity filter. The oxidation of iron(II) was accomplished through aeration. It also was observed that the oxidation of soluble arsenic(III) to soluble arsenic(V) was occurring at a rate of over 94 percent across the gravity filter via naturally occurring microbial mediated processes, with only 1.6 μg/L of soluble arsenic(III) in the filter effluent (on average). Nitrification also was observed within the gravity filter, but was not related to the microbially mediated processes as noted. The soluble arsenic(V) concentration averaged 26.4 μg/L after the gravity filter, which is comparable to the vendor’s design estimate of 20 to 27 μg/L of arsenic after the gravity filter and before the AD-33 adsorption system. Therefore, the arsenic removal rate without NaMnO4 was within the vendor’s design basis of 30 percent to 50 percent across the gravity filter.

With or without the addition of NaMnO4, soluble arsenic(V) remained above 10 μg/L in the gravity filter effluent, thus requiring further treatment with the APU-300 unit. The arsenic concentration in the APU-300 system effluent was below 10 μg/L during the one-year performance study. Based on compliance samples collected after the end of the study and average daily production values, the AD-33 media run length was estimated at 25,300 bed volumes (BV) of water, which was only 31 percent of the vendor-projected APU-300 capacity of 82,500 BV. As discussed above, the total arsenic-removal efficiency of the gravity filter was reduced from approximately 60 percent to 34 percent after discontinuing NaMnO4 addition, which shifted the burden of arsenic removal from the gravity filter to the downstream adsorption vessels. However, the average concentration of soluble arsenic(V) (26.4 μg/L) in the gravity filter effluent (without NaMnO4 addition) was close to the design basis of 20 to 27 μg/L in the influent to the APU-300 system. Therefore, the reason for the discrepancy in run length was attributed, in part, to competition from elevated total phosphorous in the source water, which was not accounted for in the vendor’s run-length estimate. Biofouling in the adsorption vessels also might have contributed to the short run length.

AERALATER backwash was manually initiated weekly by the operator. The APU-300 system was backwashed manually four times during the one-year performance evaluation study. Approximately 406,400 gallons of wastewater, or 2 percent of the quantity of the treated water, was generated during the one-year performance study from the AERALATER. The AERALATER backwash wastewater contained, on average, 87 mg/L of total suspended solids, 38 mg/L of iron, 343 μg/L of arsenic, and 57 μg/L of manganese, with the majority existing as particulate. The average amount of solids discharged per backwash cycle was approximately 5.5 pounds, which was composed of 2.4 pounds of elemental iron, 0.002 pounds of elemental manganese, and 0.02 pounds of elemental arsenic. In addition, 25,415 gallons of wastewater were generated by the APU-300 unit, or 0.1 percent of the quantity of treated water.

Comparison of the distribution system sampling results before and after system startup showed a significant decrease in arsenic concentration from an average of 31.2 to 6.1 μg/L. The arsenic concentrations in the distribution system, however, were generally higher than those following the AD-33 adsorption vessels. Desorption and resuspension of arsenic that previously accumulated on the distribution pipe surfaces was the probable reason for the higher concentration in the distribution system. Iron concentration in the distribution system was significantly reduced, while manganese levels appeared to remain the same after system startup. Both lead and copper concentrations in the distribution system were significantly lower than their action levels.

The capital investment for the system was $367,838, including $273,873 for equipment, $16,520 for site engineering, and $77,445 for installation, shakedown, and startup. Using the system’s rated capacity of 250 gpm or 360,000 gallons per day (gpd), the capital cost was $1,471 per gpm of design capacity ($1.02 per gpd). This calculation did not include the cost of the building to house the treatment system. The O&M cost consisted primarily of the media replacement cost, which was estimated by the vendor at $41,370, to change out the AD-33 media. Media change-out did not occur during the performance evaluation period. The O&M cost is presented as a function of potential media run length in this report.


Tom Sorg

You will need Adobe Reader to view some of the files on this page.
See EPA's PDF page to learn more.

Office of Research & Development | National Risk Management Research Laboratory

Local Navigation

Jump to main content.