Testing UF Systems for Non-Community Surface Water Supplies

From 2003 to 2005, the Minnesota Department of Health conducted a comprehensive evaluation of membrane filtration to meet the needs of non-community surface water systems that provide water to the public in places other than homes. In Minnesota, these include vacation lodges, resorts, campgrounds and other recreational facilities. The Minnesota Department of Health identified about 80 non-community water systems using surface water as their source of drinking water. The suppliers are typically subject to limited budgets, limited operational personnel, remote locations and seasonal use. Despite their small size, these suppliers must comply with the requirements of the U.S. EPA’s Surface Water Treatment Rule (SWTR), as well as the subsequent Long Term 1 and Long Term 2 Enhanced Surface Water Treatment Rules (LT1ESWTR and LT2ESWTR).

Study Objectives

The objective of the Minnesota Department of Health study was to compare the operational performance of several membrane technologies through pilot testing to determine applicability to small public water systems. The removal of particulate and microbial contaminants was the primary filtration goal of the study. In addition, the Minnesota Department of Health wanted to identify surface water treatment options that had a realistic chance of being affordable to small suppliers and could be relatively easy to maintain despite their remote locations.

Many of the identified suppliers were using bag filtration to meet the filtration requirements of the SWTR. Bag filtration can be used to achieve greater than 99% (2-log) removal of Giardia cysts when operated properly on relatively clean water sources. Some water sources in Minnesota are not suitable for bag filtration and have difficulty meeting finished water turbidity standards. It was also difficult to find a bag filter that could achieve 2-log removal of Cryptosporidium oocysts to satisfy LT1ESWTR requirements and provide a reasonable filter life.

To conduct the study, the Minnesota Department of Health chose to test membrane technologies based on cost, operational simplicity and availability. The ideal system was an “off-the-shelf” system with good product support, priced for the residential or small system market, with third-party particle and/or microbe removal data available.

The study involved two stages: initial testing and full-scale beta-site testing. Only five technologies were selected for initial testing, and as of 2005 (when the results of the study were first presented formally), only one of the five technologies had advanced to full-scale beta testing.


Initial testing involved placing the technologies inside a utility trailer along with analytical equipment and pretreatment equipment. Systems were tested concurrently using the same feedwater. The technologies that performed well during initial testing advanced to beta testing at various locations subject to differing water qualities. Beta testing was also used to help determine operator compatibility and fit with existing pretreatment systems.

The Minnesota Department of Health determined that for membrane technology to be applicable to non-community water supplies, the system must be easy to operate and understand, the chemical cleaning interval must be reasonable, the system should integrate well with existing systems, and particulate removal goals must be met. Ease of operation is a subjective parameter, but an attempt was made to classify systems into low, medium or high degrees of operational difficulty during initial testing.

The source water supply for initial testing was drawn from Rainy Lake, located on the border between Minnesota and Canada. Turbidity levels varied from 1 to 25 NTU throughout the year and averaged 6.1 NTU. Total coliform levels ranged from 12 to 800 (/100 mL ) and averaged 264.

Water quality sampling and testing included seven influent parameters that were tested on site either online or daily; four effluent parameters that were tested on site either online, hourly or daily; and 14 influent parameters that were tested at offsite laboratories either monthly or biweekly.

Beta-site testing for the technology that advanced to this stage was conducted at three locations. The first was adjacent to the initial testing site on Rainy Lake. The second was at Lake of the Woods, and the third was at Lake Vermilion. Raw water turbidity at Lake of the Woods was 6.5 NTU and at Lake Vermilion was 1.4 NTU.

The beta-tested system consisted of a matched pair of hollow fiber ultrafiltration (UF) units operating in parallel, each containing a membrane module of approximately 20,000 strands of hollow membrane fibers and an internal stainless steel mesh prefilter. Flow was provided by a source water pump. Each system included a programmable controller to regulate backwashing according to influent water feed quality. Also contained within each system was a small accumulator tank that collects filtrate during service and performs a back-pulsing membrane cleaning sequence during backwash. Initiation of backwash and back-pulse cycles is based on elapsed time.


After approximately 268 hours of initial testing in 2003, with favorable results, the first beta-site installation took place in March 2004. This system ran until Aug. 9, 2004, and treated 156,000 gal of water before requiring chemical cleaning (normally performed with a one-hour soak in a 5% solution of household bleach or other chemical formulations in accordance with manufacturer’s instructions). On Sept. 2, 2004, an advanced version of the system controller was installed to allow more frequent backwashing, and the system did not require additional chemical cleaning in 2004.

The second system was installed on July 8, 2004. That system ran until the facility closed on Aug. 8, 2004, and did not require chemical cleaning. Due to relatively low water usage, the system treated only 12,000 gal of water, averaging between 4 and 5 NTU.

The third system was placed in service on July 28, 2004, and ran until the facility closed on Sept. 30, 2004. The system produced 125,000 gal of water during this period and required two chemical cleanings. The first chemical cleaning was performed after eight days of operation and was likely due to problems with the pressure sand filter pretreatment.

Comments & Conclusions

Study comments advised that these systems are packaged nicely, and operators were reasonably comfortable with the technology and impressed with the finished water quality. Chemical cleaning restored flow and reduced trans-membrane pressure to initial conditions for all installations. The membrane systems integrated well with existing pretreatment, with pressure sand filters and bag filters providing adequate pretreatment to the membranes.

The study concluded that, based on pilot and full-scale testing, membrane filtration appears to offer a viable means for small systems to achieve compliance with existing and future EPA drinking water regulations. The variety of membrane systems available allows treatment of specific water quality issues previously requiring chemical treatment. The movement of membrane technologies into the residential and point-of-entry market is beneficial to non-community public water supplies. Based on commercial pricing, the capital costs of equipment packages using membrane technologies are now within the budgets of very small systems.

Note: Since the presentation of these results in 2005, the manufacturer of the systems that were beta tested (GE Zenon) has introduced the next generation of UF point-of-entry systems. The Minnesota Department of Health’s testing used an earlier version of the Homespring-branded systems. The next generation series has incorporated more features and benefits, including independent third party testing and certification to EPA Microbiological Purifier Guide Protocol. Further information about the systems is available at www.homespring.com.