Lead-leaching characteristics of submersible residential water pumps.

Introduction

Although various health effects of high lead exposure have been recognized for several decades, recent medical studies show that lead causes irreversible neurological damage in infants and young children at far lower exposure levels than previously believed (1). Medical studies have shown that blood lead levels as low as 10 micrograms per deciliter ([[micro]gram]/dL) can pose significant health risks for children (2,3). Additionally, it has been shown that, while adults absorb 35 percent to 50 percent of the lead they drink, the absorption rate for children may be greater than 50 percent (4). These medical findings, along with statistical evidence that one in five homes nationally has unhealthy tap water levels of lead, prompted the U.S. Environmental Protection Agency (U.S. EPA) to greatly lower permissible levels for lead in drinking water. In June 1991, as a step toward primary prevention [TABULAR DATA FOR TABLE 1OMITTED] (i.e., elimination of lead hazards before populations are poisoned), U.S. EPA promulgated the Lead and Copper Rule, which lowered the existing drinking-water standard of 50 [[micro]gram]/L lead at point of use to a 15-[[micro]gram]/L "action level" (5). Because medical evidence identified no threshold for the harmful effects of lead, this new ruling also issued a maximum-contaminant- level goal of 0 and required that corrosion control be implemented at water plants for water systems that exceeded the action levels for more than 10 percent of a representative sampling population.

Lead in tap water represents a health hazard that is particularly difficult to control through traditional regulatory measures. While other pollutants can be monitored and potentially regulated at the municipal treatment plant, nearly all lead in tap water originates from the residential plumbing system - after the water has left the municipal distribution system. For private residents served by systems that deliver well water, the total lead concentration is influenced not only by the components of the residential plumbing system, but also by the chemistry of the well water. In particular, the amount of lead leached from lead-based solder, lead pipes, lead-alloy faucet fixtures, and other plumbing system components has been shown to be significantly influenced by water corrosivity, which cannot be easily controlled in private well systems (6).

Current research is providing answers about the extent to which lead leaches into water from plumbing components such as faucet fixtures, drinking water fountains, ice makers, instant hot-water dispensers, residential water meters, and submersible well pumps (7-12). It should be noted that all U.S. residential submersible pump manufacturers agreed to discontinue the use of leaded-brass alloys in the manufacturing and production of these pumps as of December 1994.

Before that date, the metallurgical composition of many U.S. residential submersible pump housings contained up to eight percent lead, which was added to improve machinability. Currently, many U.S. plumbing components still contain up to eight percent lead, as allowed by the 1986 Safe Drinking Water Act Amendments (13).

The amount of lead leached from leaded-brass pumps and other plumbing components appears to be related to the manufacturing process employed; sand-cast fixtures show much greater lead leaching than machined, fabricated, or permanent-mold units that use the same brass alloys (14). Before 1994, a significant component of most commercially available submersible water pumps was a pump housing made from sand-cast leaded brass.

After initial laboratory testing, it was determined that lead leaching from submersible well pumps could be characterized best if tested under actual residential installation conditions. In addition to initial laboratory lead-leaching experiments, this study therefore involved controlled experiments performed in the field under typical residential installation and usage conditions.

Methods

Five common models of 4-inch submersible well pumps were tested for the purpose of observing lead-leaching characteristics in both laboratory and field installation settings. The five models were selected on the basis of their large consumer market share. The same makes and models were tested in both studies.

The pump housings from four of the pumps were constructed from a leaded-brass alloy, and the fifth pump housing was composed of stainless steel. Each model was tested in triplicate.

Laboratory Study

The experimental procedure used in the laboratory portion of this study followed the procedure as outlined in the NSF-61 Section 8 protocol for testing mechanical devices, with slight modifications to the extractant water (15). Each submersible pump was immersed in a 4-inch-diameter polyethylene cylinder to 1 inch above the pump's top surface at room temperature (about 23 [degrees] C). The modified extractant water had a pH of 8.0 (plus or minus 0.5), total hardness of 100 milligrams per liter (mg/L), and alkalinity of 50 mg/L, with no added chlorine. Although the NSF 61 protocol specifies 2 mg/L free chlorine, 0 hardness, and 500 mg/k alkalinity, this extractant water composition was used because it was deemed to be more representative o f average U.S. well water.

[TABULAR DATA FOR TABLE 3 OMITTED]

The extractant water was changed at 24-hour intervals from Monday through Saturday of each week over a total of 30 exposure days. A 250-mL sample was taken daily, following which, the cylinder received a deionized rinse. The cylinders were then refilled with fresh extractant water. The internal volume of each pump was approximately 0.5 liters, and the leachate volume from each cylinder was approximately 2.2 liters. A blank sample of extractant water was also collected each day. On Monday mornings after the weekend exposure period (a 64-hour exposure period), the sampling protocol was followed, although no actual sample was taken. Thus, five 24-hour dwell-time samples were obtained each week for each pump, and six weeks were required to obtain 30 samples. After collection, each sample was acidified with concentrated nitric acid to a final concentration of 0.5 percent, producing a final pH of less than 2.0. These acid-preserved samples were held for at least 28 hours before lead analysis to ensure complete dissolution of any adsorbed or micro-precipitated lead. Samples were then analyzed for total lead with a Thermo-Jarrell-Ash Video 11 and Video 12 atomic absorption spectrophotometer equipped with Thermo-Jarrell-Ash Model

188 Controlled Temperature Furnace Atomizers with a detection limit of approximately 0.3 [[micro]gram]/L.

Field Study

Five wells were drilled on the property of cooperative landowners approximately 18 miles north from the University of North Carolina at Asheville. The wells were outfitted with pumps of the same makes and models as those previously tested under laboratory conditions. Well characteristics such as depth, suspended-pump depth, and water quality are summarized in Table 1. Well water temperature for all wells throughout the study period ranged between 15 [degrees] and 16.5 [degrees] C.

The five wells were spaced approximately 100 feet apart, and, as shown in Table 1, the water yield of each well was easily large enough to preclude the possibility of overlapping draw-down zones at the pumping rate of 180 gallons per day. Each pump was connected to the above- ground pressure tank by a 1-inch-diameter pipe that contained approximately 1 gallon of water per 25 feet of pipe. The wells fitted with the Aeromotor and Goulds pumps showed somewhat lower pH, hardness, conductivity, and alkalinity than the other three wells, although these differences were considered too minor to have a significant effect on the lead- leaching results.

All wells were drilled and connected with a 20-gallon above-ground pressure tank exactly as the local well-drilling company would have installed a system for a typical residence in the area. One exception was that the leaded-brass T-valve typically used at the pressure tank junction was replaced with a PVC version to eliminate the possibility of additional lead contamination from this source.

Each pressure tank contained a 5-gallon-capacity bladder that expanded and contracted as a function of pressure. Approximately 400 gallons of water were purged through each well over a three-day period immediately before measurement of lead leaching began. A plastic spigot was mounted about 2 meters beyond the pressure tank outlet for purging of the wells and for water sample collection.

A survey of local water department billing records indicated that average residential usage in the area is approximately 180 gallons per day per dwelling; this amount of water was run through each well daily to "age" the system at a rate comparable to that of typical household usage. The above-ground pressure tanks were set to commence pumping when the water pressure in the system dropped to 30 pounds per square inch (psi) and to cut off when the pressure reached 50 psi. This setting is typical for residential installation; the pumps came on and recharged the pressure tanks after every approximately 5 gallons of water use.

Each of the pumps delivered water at a rate of approximately 20 gallons per minute; thus, the pumps came on for about 15 seconds over each 5-gallon pump cycle. With 180 gallons per day of usage and a pump cycle of 5 gallons, each pump came on 36 times per day.

An attempt was made to apportion these 36 pump cycles to produce a range of dwell times approximately representative of typical household use. At the start of each day, the first 5 gallons of water out of the spigot were collected in a 5-gallon Nalgene bucket and labeled. A composite sample of the second 5 gallons was collected in the same way. These were the overnight dwell samples. The flow rate was then set at 0.5 gallon s per minute, and the faucet was run continuously for 80 minutes. The last 5 gallons (taken over 10 minutes) were collected in the 5-gallon bucket, they were mixed thoroughly, and a 250-mL sample was poured off. This sample represented a composite of eight successive 10-minute dwell times. At this point, the pressure tank was emptied and the water turned off for one hour. At the end of that hour, 5 gallons were run from the faucet. Following yet another one-hour dwell time, 5 more gallons were run from the faucet and were collected in the 5-gallon bucket. After each collection, a 250-mL sample was poured off to capture a slug of water with a one-hour dwell time.

This sampling protocol was initially run for a total of 40 sampling days (or eight weeks) with overnight dwell times of approximately 14 hours on weekdays and weekend dwell times of 63 hours. In addition to the 400-gallon initial purge, a test run of the experimental design was implemented on a Friday, and lead concentrations were measured but not reported. After eight weeks of testing, the three pumps showing detectable lead concentrations were equipped with timers set so that the pumps would come on for three hours each day to deliver 180 gallons per day, thus approximating a continued aging of the system.

Samples of 14-hour dwell times were taken once a week for 14 additional weeks.

Results and Discussion

Laboratory Study

Table 2 presents a condensed version of the extract-water lead concentrations from each set of submersible pumps tested under laboratory conditions. For all of the pumps constructed of leaded brass, extractant lead levels decreased over time during the six-week sampling period. Such a decrease has been observed previously in similar studies of leaded- brass faucet fixtures. By contrast, the Grundfos stainless steel pumps leached nearly undetectable levels of lead over the same period. Given the fact that these pumps are constructed of stainless steel and supposedly contain no lead, it is not particularly surprising that only background levels of lead were found in their extractant waters and that no changes in leachate lead levels were recorded over time. The Myers brand pump produced much lower lead concentrations than did the other three leaded- brass models; this result was consistent with the visual observation that less leaded-brass surface area was exposed on this model. For all pump models, it is noteworthy that the concentration means for days 21 to 30 were not very different from the means for days 11 to 20.

Field Study

The lead concentration data obtained from the field experiments are summarized in Table 3. Only the Sta-Rite, Aeromotor, and Goulds pump results are shown, because the other two pumps exhibited essentially undetectable lead concentrations. Data obtained from this study indicate that, while substantial decreases in overall lead leachate concentrations occurred initially, by about Week 9, lead concentrations of all three pumps appeared to level out. Over the next 14 weeks, all three pumps continued to show detectable levels of lead. It is also interesting to note that, on average, about 32 percent of the overnight lead leaching occurred in the first 10 minutes of running the faucet.

The data in Table 3 were developed by taking the arithmetic means of the lead concentrations in the first and second 5-gallon composite samples collected each sampling morning. These experiments did not adequately address the following question: How long does significant lead leaching from the leaded-brass alloy systems continue? One possible scenario is that lead concentrations would continue to decrease, reaching undetectable levels within five to six months.

Alternatively, the results summarized in Table 3 may represent a situation of initial rapid concentration decrease followed by a relative leveling out of concentrations in the range observed over weeks 9 to 23. Most likely, the actuality lies somewhere between these extremes.

Conclusions

Of the three leaded-brass models, the field studies showed more rapid and continued concentration decreases over time for the Sta-Rite and Goulds models than for the Aeromotor pump, which, over the extended field usage period, appeared to stabilize at overnight dwell concentrations of approximately 3 [[micro]gram]/L. It should be noted that these field experiments, although carefully controlled and closely representative of typical household conditions, tested only one individual pump of each model and only one set of water quality conditions. Previous laboratory work with leaded-brass faucet fixtures, submersible well pumps, and water meters indicates that variations of a factor of three or more are commonly noted between individual pieces of plumbing of the same brand, model, and even manufacturing lot. It is also well known and well documented that water corrosivity can have a strong effect on the extent and duration of lead leaching from leaded-brass alloys. The well water used in these field experiments would be characterized as low to moderately corrosive.

Although the lead concentrations observed in actual field usage were much lower than those observed in earlier laboratory experiments involving just the standing water within the submersible pumps, the results appear to be generally consistent when the operative dilution effects are considered. In the laboratory', the leaded-brass alloy pumps Sta- Rite, Goulds, and Aeromotor exhibited average overnight internal water concentrations of about 300 [[micro]gram]/L over days 10 to 30. Thus, given that the internal storage volume of these pumps is approximately 500 mL (0.134 gallons), a dilution factor of about 75 would be expected for a 10-gallon composite sample, which translates to a lead concentration of about 4 [[micro]gram]/L. This calculated value is very close to what was actually observed for the Sta-Rite, Goulds, and Aeromotor pumps over days 10 to 30 of these field experiments. Although the field sampling protocol was purposely designed to capture the entire slug associated with pump surface contact, this may not have occurred in every case because of dispersion factors in the supply line and hydraulic vortices created in the well casing during pumping. Because the percentage of lead actually leached from the pump exterior is unknown, as is the percentage of the lead slug that actually makes it into the composite leachate in the field study, two dilution factors were calculated to provide a range that is likely to contain the true dilution factor. Calculating the dilution (18X) on the basis of the 2.2-liter leachate volume might be appropriate if the total lead leached in the field study from both the exterior and interior of the pump' s surface were included in the composite sample. Table 4 shows the relationship between the average overnight lead concentrations for the laboratory study and the field installation study. The adjusted laboratory values are numerically close to the concentration levels found in the field installation study.

Nevertheless, concentration variations should be expected because of differences in temperatures, aging processes, and water chemistry and because of inherent differences among individual pumps, even of the same model. For example, water temperatures in the field study were approximately 7 [degrees] C colder than in the laboratory study, which would tend to decrease lead leachate concentrations.

Similarly, pumps in the field study were initially aged to a greater extent by the larger volume of water pumped through them, resulting in lower observed lead concentrations. The overnight dwelt time for the laboratory study was 24 hours, while the dwell time for the field installation study was 16 hours; thus, the laboratory study had longer leaching times.

In conclusion, even though U.S. submersible pumps are no longer manufactured with leaded-brass alloys, a potential contamination problem still exists.

Approximately 11 million residential well pumps in current operation in the United States were installed before 1995. A nationwide study is currently investigating what percentage of these pre- 1995 well pumps are leaching elevated levels of lead into residential drinking water. Preliminary and as yet unpublished results indicate that well pumps of all ages can pose a contamination problem, with the risk generally decreasing with pump age.

Approximately five percent of pumps tested to date are delivering overnight dwell concentrations in excess of the U.S. EPA action level of 15 [[micro]gram]/L, and approximately 11 percent are exhibiting overnight dwell concentrations of 5 [[micro]gram]/L or greater.

Acknowledgements

The authors and the staff at the University of North Carolina-Asheville

Environmental Quality Institute would like to acknowledge and thank the following individuals for their assistance with the experimental design of the field installation study: Peter Lassovsky, U.S. EPA, Office of Groundwater and Drinking Water; Tom Sorg, U.S. EPA Office of Research and Development-Drinking Water Research Development (ORD- DWRD); Michael Schock, U.S. EPA ORD-DWRD; Ann Marie Gephart, National Science Foundation; and Steve Rust, Battelle Corporation.

Funding for this research was provided by U.S. EPA, the Environmental Defense Fund, and the Office of the California Attorney General.

Corresponding Author: Leslee Thornton, Environmental Quality Institute, University of North Carolina at Asheville, One University Heights, Asheville, NC 28804-3299.

E-mail: patch@unca.edu

REFERENCES:

1. The Centers for Disease Control (1991), Preventing Lead Poisoning in Young Children, Atlanta: U.S. Department of Health and Human Services.

2. Needleman, H., and D. Bellinger (1991), "The Health Effects of Low Level Exposure to Lead," Annual Review of Public Health, 12:111- 140.

3. Sciarillo, W.G., G. Alexander, and K.P. Farrell (1992), "Lead Exposure and Child Behavior," American Journal of Public Health, 82:1356-1360.

4. The Nature and Extent of Lead Poisoning in Children in the United States: A Report to Congress (1988), Atlanta: Agency for Toxic Substances and Disease Registry.

5. U.S. EPA (1991), "Drinking Water Regulations: Maximum Contaminant Level Goals and National Primary Drinking Water Regulations for Lead and Copper," Federal Register, 53:110.

6. Economic and Engineering Services, Inc. (1990), Lead Control Strategies, Denver, Colo.: American Water Works Association Research Foundation.

7. Maas, R.P., and S.C. Patch (1990), "Single-Family Residence Tap Water Lead Contamination," Journal of the North Carolina Section of l he American Water Works Association, 66:115-160.

8. Maas, R.P., G.M. Brown, D.M. Morgan, S.C. Patch, and E.R. Rainey (1993), The Rate and Extent of Lead Leaching from Faucet Fixtures and Hot Water Dispensers Under Laboratory Conditions, Technical Report #93-009, Asheville, N.C.: University of North Carolina-Asheville Environmental Quality Institute.

9. Lee, R.G., W.C. Becker, and D.W. Collins (1989), "Lead at the Tap: Sources and Control," Journal of American Water Works, 81(7):52-62.

10. Gnaedinger, RH. (1993), "Lead in School Drinking Water," Journal of Environmental Health, 55(6):15-18.

11. Maas, R.P., A.M. Gagnon, and S.C. Patch (1994), "The Dynamics of Lead in Drinking Water in U.S. Workplaces and Schools," American Industrial Hygiene Journal, 55(9):829-832.

12. Maas, R.P., S.C. Patch, J.P. Pope, and E.R. Rainey (1994), The Effects of Acidity, Pre-Cleaning and Surface Coating on Lead Leaching from Standard Leaded-Brass Water Meters., Technical Report #94-002, Asheville, N.C.: University of North Carolina-Asheville Environmental Quality Institute.

13. U.S. Congress (1986), Safe Drinking Water Act Amendments, Public Law 99-339, [section]109, [section]1417.

14. Maas, R. P., G. M. Brown, B.T. Peek, and ER. Rainey (1993), Sixty- One Day Lead Leaching Characteristics of Low Lead Content Brass Alloys, Technical Report #93-005, Asheville, N.C.: University of North Carolina- Asheville Environmental Quality Institute.

15. American National Standards Institute and NSF International, Joint Committee on Drinking Water Additives (1994),Drinking Water System Components - Health Effects,, Sections 1-9, Ann Arbor, Mich.: NSF International.

Maas, Richard P.; Patch, Steven C.; Pope, Jason; Thornton, Leslee, Lead-leaching characteristics of submersible residential water pumps.., Vol. 60, Journal of Environmental Health, 01-11-1998, pp 8(6).