Jump to main content.


  Arsenic Removal From Drinking Water by Ion Exchange, U.S. EPA Demonstration Project at Fruitland, ID, Six-Month Evaluation Report (PDF) (81 pp, 1.86 MB) (EPA/600/R-07/017) April 2007

This report documents the activities performed during and the results obtained from the first six months of the performance evaluation of a Kinetico ion exchange (IX) system to remove arsenic and nitrate from source water at the City of Fruitland, Idaho. The 250-gallons-per-minute (gpm) IX system consisted of a bank of five sediment filters and two 48-inch-diameter by 72-inch-tall pressure vessels configured in parallel. Each resin vessel contained 50 cubic feet of a strong base anionic exchange resin, i.e., A300E manufactured by Purolite. The system installation first began in March 2004; however, the commencement of the system operation was repeatedly delayed until June 2005 due to a series of problems. The problems started with excessive sediment production from the original supply well, followed by the failure of a replacement well to pass bacterial testing even after repeated sanitation efforts. The problems were further compounded by the need to replace the resin that was erroneously installed in the resin vessels and a broken well pump that was salvaged from the original supply well into the replacement well.

During this reporting period, June 14 through December 16, 2005, the IX system operated for a total of 3,635 hours, averaging 20 hours per day. The system treated 35.9 million gallons of water with an average daily production of 194,000 gallons per day (gpd). The average flow rate was 165 gpm, which was equivalent to 66 percent of the 250-gpm design flow rate. This average flow rate yielded a 4.5-minute empty bed contact time (EBCT) and a 6.6-gallon-per-minute-per-square-foot hydraulic loading rate to each resin vessel. The IX resin was regenerated in a downflow, co-current mode using a sodium chloride brine solution at a target salt level of 10 pounds per cubic foot (lb/ft3) of resin. Triggered automatically by a pre-set throughput in the programmable logic controller, the two IX vessels were regenerated sequentially, each cycling through the steps of brine draw, slow rinse, and fast rinse before returning to service. A total of 110 regeneration cycles took place during this reporting period, consuming approximately 172,390 pounds (or 86 tons) of salt. Therefore, each regeneration cycle used an average of 1,567 pounds of salt, or 15.7 lb/ft3 of resin, which was 57 percent higher than the design value. Close examination of the regeneration steps revealed that this unexpectedly high salt usage was the result of a higher brine draw rate caused by improper flow control.

Total arsenic concentrations in raw water ranged from 33.6 to 60.8 micrograms per liter (µg/L) and averaged 42.1 µg/L, which existed primarily as arsenic (V). Nitrate concentrations in raw water ranged from 6.9 to 11.2 milligrams per liter (mg/L) (as nitrogen) and averaged 9.5 mg/L (as nitrogen). After treatment, total arsenic and nitrate concentrations were reduced to below the respective maximum contaminant levels, except when the system was freshly regenerated or experiencing mechanical problems. Removal of uranium, vanadium, and molybdenum by the IX system also was observed.

Sulfate, the most preferred anion by the resin, was removed from an average of 58 mg/L in raw water to less than 1 mg/L in the treated water, except when the system was experiencing mechanical problems. Raw water pH values ranged from 7.3 to 7.9. A significant reduction in pH in the treated water was observed immediately after resin regeneration, presumably due to the removal of bicarbonate ions by the freshly regenerated IX resin, as evidenced by the corresponding decrease in total alkalinity.

Resin run length studies were conducted over the course of three separate service runs. The purpose of the studies was to delineate the arsenic and nitrate breakthrough curves and determine the resin run length between two consecutive regeneration cycles. Based on the results of these studies, the resin run length was upwardly adjusted from the initial factory setting of 214,000 gallons (or 286 bed volume [BV]) to 335,000 gallons (or 448 BV), then downwardly adjusted to 316,000 gallons (or 422 BV) to reach an optimal service run length. Effluent samples collected from the IX vessels indicated arsenic and nitrate leakage during the first 50,000 to 60,000 gallons (or 67 to 80 BV) of throughput, which was consistent with the observations made during the treatment plant sampling in the six-month period. As expected, total alkalinity and pH values were significantly reduced during the early stage of all service runs.

During the first six months, the resin regeneration scheme was adjusted several times to improve regeneration efficiency and minimize residual production. Originally, the factory settings for the resin regeneration consisted of 64 minutes of brine draw with a 4 percent brine, 64 min of slow rinse, and 30 minutes of fast rinse. These settings were changed to 32 minutes of brine draw with an 8 percent brine (to achieve the same salt regeneration level), 40 minutes of slow rinse, and/or 6 or 15 minutes of fast rinse. The adjustments to the regeneration settings resulted in significant reductions in wastewater production. For example, the decrease in the brine draw time from 64 to 32 minutes reduced the spent brine volume by 50 percent, from 2,304 to 1,152 gallons per regeneration cycle. The reduction in the slow rinse and fast rinse times also decreased the wastewater volume proportionally. Under a set of modified settings consisting of 25 minutes of brine draw with an 8 percent brine, 40 minutes of slow rinse, and 15 minutes of fast rinse, the amount of wastewater generated was 5,740 gallons per cycle, accounting for 1.8 percent of the total volume treated (i.e., 316,000 gallons). Because treated water was used for regeneration, the system production efficiency was 98.2 percent.

Two resin regeneration studies were conducted to evaluate the effectiveness of the resin regeneration process and characterize the residuals produced. Although the majority of arsenic and nitrate on the resin was eluted during the brine draw and slow rinse steps, arsenic concentrations as high as 35 µg/L were still measured towards the end of the fast rinse step. Therefore, it was not surprising to detect over 10 µg/L of arsenic during the early stage of the subsequent service run. Extending the fast rinse time from 6 to 15 minutes did not resolve the problem because the arsenic leakage was found to continue up to 52,000 gallons (or 70 BV) of throughput, or approximately 3 to 4 hours into the service run. The waste stream discharged to the sewer contained an average of 1.2 to 2.4 mg/L of arsenic and 0.42 to 0.5 grams per liter of nitrate, equivalent to a mass loading of 31 to 56 grams for arsenic and 8,615 to 12,649 grams for nitrate per regeneration cycle. The percent recoveries were 114 and 63 percent for arsenic, 99 and 130 percent for nitrate, and 118 and 74 percent for sulfate, in the two regeneration studies, respectively.

The capital investment cost was $286,388, which included $173,195 for equipment, $35,619 for site engineering, and $77,574 for installation. This capital cost was normalized to the system’s rated capacity of 250 gpm (360,000 gpd), which resulted in $1,146 per gpm ($0.80 per gpd). Funded separately by the City of Fruitland, the cost associated with the new building, sanitary sewer connection, and other discharge-related infrastructure was not included in the capital cost.

The operation and maintenance (O&M) cost for the IX system included the incremental cost associated with the salt supply, electricity consumption, and labor. Over the six-month operation period, the cost of the salt supply was $0.51 per 1,000 gallons of water treated based on the average salt usage of 4.80 pounds per 1,000 gallons. This salt cost could be reduced to $0.35 per 1,000 gallons if the brine draw flow was controlled properly to reach a target salt usage of 3.16 pounds per 1,000 gallons. Incremental electricity consumption associated with the IX system was not available, but assumed to be minimal. The actual power usage for operating the entire plant was approximately $0.08 per 1,000 gallons of water treated. The routine, nondemonstration related labor activities consumed about 30 minutes per day, which corresponded to a labor cost of $0.04 per 1,000 gallons. Therefore, the total O&M cost was approximately $0.63 per 1,000 gallons (actual) or $0.47 per 1,000 gallons (design), with the majority of the cost incurred by the salt supply.


Thomas Sorg

See Also

Arsenic Research

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.