Heterotrophic bacteria control in a residential reverse-osmosis drinking-water filter.

Introduction

In 1992, a project involving drinking-water treatment for over 50 homes and a commercial office was begun in a subdivision of Suffolk, Virginia. The primary goal was to reduce the naturally occurring fluoride levels in the municipal well water source from 6 milligrams per liter (mg/L) to below 2 mg/L. The federally established maximum contaminant level (MCL) for fluoride in drinking water is 4 mg/ L, and the U.S. Environmental Protection Agency (U.S. EPA) has established a secondary MCL of 2 mg/L. The primary MCL for fluoride is set at a level at which no known adverse health effects occur or are anticipated. The secondary MCL has been set to assure good aesthetic water quality. The municipal water utility considered three options for reducing the fluoride to an acceptable level:

* a pipeline from the city central system,

* central treatment at the subdivision, and

* treatment in each home using a point-of-use filter.

After cost, waste disposal, maintenance, and timeliness factors were studied, the point-of- use filters were chosen as the preferred solution.

When this project began, point-of-use treatment was not acceptable for permanent remediation of high fluoride levels, but could be used as an interim method of treatment if the utility retained control and consistently demonstrated conformance to all required regulatory conditions (1). The local water utility for the City of Suffolk, the Virginia Department of Health, U.S. EPA, and three manufacturers of point-of-use reverse osmosis filtration products were involved (2). These groups jointly determined that the performance of the filters should be evaluated by measuring conductivity, fluoride, total dissolved solids (TDS), pH, sodium, residual chlorine, total coliform, and heterotrophic bacteria as measured by heterotrophic plate count (HPC). The municipal water supplier retained responsibility and control of the treatment systems, even though the equipment was installed, serviced, and maintained by local representatives of the equipment manufacturers.

From the start, the filters performed satisfactorily except for variable HPC results, with some units testing higher than desired. These results brought all interested parties together to determine the future direction of the project and to define an upper limit for HPC in potable water. Since there was no drinking-water MCL for HPC, U.S. EPA's formulation process for other drinking water microbiological requirements and the history of that process were studied. During the development of the drinking-water Total Coliform Rule, U.S. EPA considered using HPC levels of less than 500 colony-forming units per milliliter (CFU/mL) as a "negative" acceptance criteria for questionable total-coliform test results. Levels above 500 CFU/mL were considered a "positive" for questionable total-coliform test results (3). In other words, when a total-coliform test for a water sample had questionable results, an HPC of less than 500 CFU/mL was sufficient reason to record the coliform test as negative and the water was considered safe. If levels above 500 CFU/mL were found, then the test was positive and the water was not acceptable for consumption. In addition, U.S. EPA, in developing the Surface Water Treatment Rule, decided that an HPC of less than 500 CFU/mL was acceptable as evidence of adequate disinfection of drinking water (4,5).

The above considerations relate to municipal and small drinking-water systems only. For these systems, the Safe Drinking Water Act permits use of bottled water as a substitute for municipal water when the water supply does not conform to any of the health requirements (6). Bottled water is monitored by the Food and Drug Administration (FDA), and FDA has established no HPC standards for bottled water. This inconsistency between governmental agencies leaves open the question of the health effects of heterotrophic bacteria in drinking water. It is not clear if an upper limit for HPC is necessary.

Conversely, there may be benefits to higher levels of heterotrophic bacteria in drinking water. Rollinger and Dott present coincidental evidence that heterotrophic bacteria suppresses growth of coliforms and other enteric pathogens (7). In another study, inoculation of filters with three different pathogens resulted in no vigorous growth, and even an eventual die- off when dechlorinated tap water was being filtered (8). Thus, some evidence indicated there may be benefits to having heterotrophic bacteria populate filters.

High bacteria counts have been found in normal tap water, particularly when water stagnates (9,10). Other researchers confirm what was found in domestic water treatment units, including domestic reverse osmosis treatment units (11,12). Thus, even though high heterotrophic bacteria populations seem acceptable in some situations, there is considerable reluctance among all involved to accept any number above 500 CFU/mL, because there is no preponderance of evidence to support a higher number.

After the results of this literature review were considered and interested parties were consulted, an HPC of 500 CFU/mL was determined to be the upper-limit goal. This level was chosen primarily on the basis of the desire to "err on the safe side." No clear evidence supports changing the goal, nor are there data to support a new number if a change from an upper limit of 500 CFU/mL is desired.

Methodology

Because the HPC limit could not be raised, the reverse-osmosis (RO) units required modification. An operation and maintenance procedure was designed to reduce the counts to, and maintain them at, less than 500 CFU/mL. Previous experience by the filter unit manufacturers [TABULAR DATA FOR TABLE 1 OMITTED] pointed to the carbon-post filter elements as the probable source of the elevated HPC levels. Therefore, the design was changed by removing the carbon-post filter elements which were part of the total RO system at all sites. The municipal system also had a chlorinator installed to provide a residual disinfectant to the system. Thin-film polyamide RO membranes were replaced with cellulose-based membranes which were more tolerant of chlorine. At the inlet to the RO system, a level of 1.2 to 1.5 mg/L residual chlorine resulted in a post-RO residual chlorine level of 0.1 to 0.2 mg/L. A routine maintenance sanitization schedule of the systems and replacement of the sediment filters every three months was established. The sanitization procedure included brushing and rinsing the interiors of the units, followed by a 30-minute exposure to 500 mg/L chlorine made from a dilute solution of household bleach. Service technicians wore latex gloves when handling sanitized parts of the unit during reassembly and took precautions not to contaminate the filter elements when handling them. After sanitization, water was passed through the units until the residual chlorine was less than 1 mg/L. Even after these changes, some sites retained HPC that progressively climbed above 500 CFU/mL. These counts dropped when the residents were encouraged to use the filters regularly so that stagnation of the water did not occur.

With the carbon-post filter elements removed, several residents then complained of taste problems caused by residual chlorine coming through the filters. This led to an experiment at a commercial office site, where a carbon-post filter was installed after the RO system.

At the commercial site, three occupants historically had used bottled water and chose not to use the water produced by the RO system. The site was remotely located at the end of the municipal distribution line. The offices were used an average of five days a week, allowing stagnation in the city distribution and building water lines during weekends and nights. Test results from the RO unit (even without carbon-post filters) showed consistently high HPC. A timer-controlled solenoid was added to the RO filter, so that one gallon of water could be flushed each night at 1:00 a.m. This schedule minimized any interruptions for the user. The addition of the solenoid-timer components were for experimental purposes only. It is not being suggested that these are necessary to control heterotrophic bacteria in a RO filter. The intent was to determine whether, if a resident consistently used the system (preventing stagnation of water for no more than one day), the HPC would remain below 500 CFU/mL, even with carbon-post filter elements present.

Results

Table 1 lists the data obtained from the commercial site for a one-year period. All analyses were performed by the city of Suffolk water-testing laboratory. Test results were obtained in two stages. First, from December 16, 1993, to June 28, 1994, the flush mechanism was operated without a carbon-post filter element in the unit. The data show that with the daily flush operating, the HPC in the system slowly dropped to nine CFU/mL over five months. It is believed the high initial HPC level of 15,300 CFU/mL can be attributed to insufficient sanitizing of the timer and solenoid system installed to control the flush process. On June 28, 1994, the start of the second stage, a carbon-post filter element was introduced. Except for the 520 CFU/mL count on September 14, 1994, the test results remained well below the 500 CFU/mL limit. These results supported the belief that a filtration system containing carbon-post filter elements will maintain reasonable HPC levels when there is regular flow through the unit. This conclusion has since been verified by Snyder, et al. after field testing activated carbon filters in residences supplied by rural groundwater (13). The modifications to the RO systems were made without adversely affecting the fluoride removal performance.

Conclusions and Follow-Up

When water from a point-of-use filtration system attached to a microbiologically safe water supply is used regularly, the HPC can be controlled. Should stagnation of the water occur (e.g., during vacations), flushing of the system re-establishes low counts. When RO treatment systems are being considered, sizing of the system is also important. Because the RO membrane treats the water and then stores it until use, a membrane and storage tank that is not too large for the application or the site will have comparatively more frequent flows and, therefore, will provide less opportunity for bacterial growth. An oversized RO system will have low water usage rates or near stagnation in the storage tank, providing opportunity for heterotrophic bacteria populations to grow above desired levels.

The Amendments to the Safe Drinking Water Act passed and signed by the 104th congress in August of 1996 now permit point- of-use filter systems to achieve MCL compliance for nonbiological problems in need of remediation (14). Control of and responsibility for the operation and maintenance of such systems remain with the municipal water supplier to ensure conformance to all requirements of the Safe Drinking Water Act.

REFERENCES

1. "National Primary Drinking Water Regulations: Synthetic Organic Chemicals: Final Rule" (July 1987), 40 Code of Federal Regulations, [section]141 and 142.

2. Lykins, B.W., R. Astle, J.L. Schlafer, and P.E. Shanaghan (1995), "Reducing Fluoride by Managed P.O.U. Treatment, "J. Am. Water Works Ass'n., 87(11):57.

3. "Total Coliform Rule," (July 1989), 40 Code of Federal Regulations, [section]54.

4. "Total Coliform Rule," (July 1989), 40 Code of Federal Regulations, [section]54.

5. "National Primary Drinking Water Regulations: Filtration and Disinfection: Final Rule," (July 1989), 40 Code of Federal Regulations, [section]141.72.

6. "National Primary Drinking Water Regulations: Variances and Exemptions," (July 1993), 40 Code of Federal Regulations, [section]142.62.

7. Rollinger, Y., and W. Dott (1987), "Survival of Selected Bacterial Species in Sterilized Activated Carbon Filters and Biological Activated Carbon Filters," Appl. Envtl. Microbiol., 53(4):777-781.

8. Reasoner, D., J.C. Blannon and E.E. Geldreich (1987), "Microbial Characteristics of Third-Faucet Point-of-Use Devices," J. Am. Water Works Ass'n., 79(10):60-66.

9. Geldreich, E.E., and D.J. Reasoner (1989), "Home Water Treatment Devices and Water Quality," in G.A. McFeters, ed., Drinking Water Microbiology: Progress and Recent Developments, Springer-Verlag, Inc., New York, N.Y., pp. 147-167.

10. Geldreich, E.E. (1990), "Microbiological Quality Control in Distribution Systems," m Water Quality and Treatment: A Handbook of Community Water Supplies, 4th ed., Denver, CO:Am. Water Works Ass'n., pp. 1113-1158.

11. Geldreich, E.E., J.C. Blannon, D.J. Reasoner, and R.H. Taylor (1985), "Bacterial Colonization of Point-of-Use Water Treatment Devices," J. Am. Water Works Assoc., 77:72-80.

12. Payment, P. (1989), "Bacterial Colonization of Domestic Reverse Osmosis Water Filtration Units," Canadian J. of Microbiol., 35:1065-1067.

13. Snyder, J.W., R.E. Anderson, G.K. Bissonnette, and C.N. Mains (1995), "Effect of Point- of-Use, Activated Carbon Filters on the Bacteriological Quality of Rural Groundwater Supplies," Appl. Envtl. Microbiol., 61:4291-4295.

14. Amendments to the Safe Drinking Water Act (1996), S.R. 1316, [section]105.

Corresponding author: John L. Schlafer, Ph.D., P.E., Manager, Product Engineering, Ecowater Systems, P.O. Box 64420, St. Paul, MN 55164.