EPA/600/A-95/100
AN EVALUATION OF ION EXCHANGE SOFTENING
ON THE LEACHING OF METALS FROM
HOUSEHOLD PLUMBING MATERIALS
Thomas J, Sorg
Michael R. Schock
Darren A. Lytle
Drinking Water Research Division
United States Environmental Protection Agency
Cincinnati, Ohio 45268
INTRODUCTION
The leaching of lead, copper and other metals from metallic
components of household plumbing systems is impacted by the
corrosivity of the distribution water.1,2 Lead is present in a
variety of plumbing materials such as lead service 1 ines,
galvanized steel pipe, solders, and faucets. Copper is the main
component of copper pipe and also a major component of the brass
in faucets. Zinc occurs in galvanized steel pipe and is also a
component of the brass in faucets.
Water quality plays a significant role in the corrosion
process. The most significant factors that influence the rate of
corrosion are pH, total inorganic carbon (TIC), dissolved oxygen,
chlorine, and temperature.1 Other factors that also play a role
in corrosivity are calcium, silicate, organic material, ammonia,
chloride, sulfate, phosphate, nitrate, and fluoride. The relative
significance of each factor varies by the material and the nature
of corrosion process.
Naturally soft, low mineralized (hardness <25 mg/L as CaC03
and total dissolved solids (TDS) <50 mg/L) waters with low pH's
have been widely demonstrated to have a corrosive effect on most
household plumbing materials. These types of waters are common in
surface waters of the Pacific northwest, New England area, and the
southeastern United States.1,3"5 Naturally soft waters contain low
levels of calcium and magnesium ions, and are generally low in pH,
(TIC and TDS). Primarily because of their pH and paucity of
dissolved minerals, such waters are often inherently corrosive and
frequently result in the leaching of excessive levels of lead and
copper from common plumbing materials.
Moderately hard mineralized waters (hardness = 25-125 mg/L
CaC03 and TDS - 50-300 mg/L) and hard mineralized water (hardness
= >125 mg/L> CaC03 and TDS >300 mg/L) usually having pH above 7 is
generally considered to be nonaggressive, although this concept
does not always hold true1. Hard mineralized waters are
frequently scale forming, and municipalities will reduce the
1
-------
hardness to a moderate level by either lime softening or ion
exchange. Where municipal softening is not practiced, or lower
hardness water is desired, consumers commonly install home ion
exchange water softeners.
Ion exchange softening removes essentially all of the
calcium and magnesium ions by exchanging them with sodium ions.
Municipalities using ion exchange treatment will blend a portion
of the source water with the treated water to provide a low or
moderate level of hardness to their consumers. Because blending
is not'a design feature of household water softener systems, the
hardness level of this water distributed through the home is near
zero. The pH, alkalinity and TDS, three key corrosion parameters,
remain approximately the same.
Although naturally soft waters are known to have corrosive
effects on plumbing materials, the effect of ion exchanged
softened waters have jiot been extensively evaluated. Furthermore,
little data exist on the fate of various water quality
constituents that play an integral role in protective surface film
development or the corrosion of metals, as they pass through
domestic water softeners.
STUDY OBJECTIVES
The potential effects of ion exchange softening on the
corrosivity of household plumbing materials can be divided into
three possible areas. First is the potential effect of a direct
increase in the metallic corrosion by-products. The second
potential effect is a change in the chemical characteristics of
the water that would either remove or alter film forming
constituents in the water. The last one, that is very difficult
to achieve with short term testing, is the potential for producing
reversible changes in passivation surface films of aged plumbing
systems.
The Drinking Water Research Division (DWRD) began to address
these issues by conducting a controlled pilot plant study designed
to evaluate the impact of home water softeners on the leaching of
metals on new plumbing materials. This project was conducted with
the cooperation and financial assistance of the Water Quality
Association (WQA) through a Cooperative Research and Development
Agreement under the Federal Technology Transfer Act.6
A research plan was developed to conduct two studies using
two different water qualities at the City of Cincinnati Bolton
Water Treatment Plant. One study was conducted on a finished tap
water that had a total hardness near 160 mg/L as CaC03 and a pH of
9.1-9.3. The second study was to be conducted on untreated
groundwater (well water) that had a hardness of around 330 mg/L as
CaC03, about twice the hardness as the finished tap water, and a
2
-------
lower pH of 7.1-7.3. This report presents the test results of the
first study (Phase I) on the finished tap water.
TEST PROCEDURE
The research plan consisted of conducting a pilot plant
study using two pipe loop systems: one fed with the control water
(finished tap water) and the second one fed with the test water
(ion exchanged softened water). The goal was to operate the
systems for a minimum of nine months and to measure and compare
the metal leaching levels from the household plumbing materials
that made up the pipe loop systems. After the nine months of
planned operation, the data would be evaluated and the study
extended to a year or longer as time and resources permitted.
SOURCE WATER
The source water was finished tap water from the City of
Cincinnati's Bolton Water Treatment Plant (BWTP). The BWTP treats
groundwater with about 330 mg/L (as CaC03) of hardness by lime
softening reducing the hardness to around 160 mg/L (as CaC03).
The lime softening process also increases the pH of the
groundwater to about 9. Treatment also includes the addition of a
low level of polyphosphate (approximately 0.4-0.5 mg PO4/L) for
protection against CaC03 scaling of their filters, chlorination,
and fluoridation, A summary of the major chemical constituents in
the source water (control water) is shown in Table 1.
PIPE LOOP SYSTEMS
Two pipe loop systems similar, but not identical, to the
system outlined in the American Water Works Association Research
Foundation's Lead Control Strategies manual were designed and
constructed (Figure 1). One system used for the control was fed
BWTP tap water. The second system was fed the ion exchange
softened tap water having only a trace of hardness. Each pipe
loop system contained two loops of lead pipe, copper tubing,
copper pipe connected with 50:50 lead-tin solder joints,
galvanized pipe, and two brass faucets. Each pipe loop was 50
feet in length. Faucet selection was based upon the DWRD study
metal leaching study of 1987.8 A summary of the plumbing
materials used in the pipe loop system is shown in Table 2.
All connecting pipe and other materials, such as flow meters
and sampling valves to support the system, were made of plastic
(PVC) or stainless steel. Sample taps were installed before and
after the water softener and after each individual pipe loop. A
flowmeter was installed before each pipe loop for flow control and
the entire system was pressurized by using a solenoid valve at the
end of the discharge line of the loops. Check valves were not
installed before and after each loop as given in AWWARF design.
3
-------
To achieve isolation of the loops during sampling, manual valves
were placed before and after each loop. The faucets themselves
maintained water pressure and a electrically controlled hydraulic
arm attached to the faucet handle automatically opened and closed
the faucets. A water meter was also installed in the main line
before each system to measure the total amount of water that
passed through each system.
All piping materials and faucets were purchased new from
local plumbing supply firms. The piping material was formed into
50 ft loops or rectangular sections. The copper soldered pipe
loops were joined with 50:50 lead-tin solid-core solder by a
skilled technician. Each loop contained 18 elbows and 36 soldered
joints. The galvanized steel pipe loops were joined by threaded
90 degree elbows. All the loops were connected to the PVC system
with PVC reinforced flexible hose and clamps. The faucets were
all of the same model having a single handle, brass interior and
an internal volume of approximately 100 mL.
WATER SOFTENER
The water softener was provided by Culligan International
Company, Northbrook, IL. The softening system, MARK 512, Model
3526-46, is listed by the National Sanitation Foundation (NSF)
under Standard 44. The softener was installed by a local
Cincinnati dealer who programmed the system to regenerate every
other day at noon time. The regeneration cycle consisted of three
operational cycles: (a) backwash (15 min), (b) brine (64 min),
and (c) rinse (4 min).
SYSTEM OPERATION
The pipe loop test systems were connected to a tap water
source having a line pressure of over 100 psi. A pressure
regulator provided approximately 54 psi to the pilot systems.
Each pipe had an inline flow controller set to provide a flow rate
of near 1 gpm through each loop and about 0.5 gpm through each
faucet. The programmable on/off timing system was set to open and
close the solenoid valves and the hydraulic faucet handle system
to provide 105 minutes of flow time each day through each pipe
loop and faucet. A water meter was installed before each system
to provide a rough check of the total flow.
The flow pattern consisted of 5-15 minute and 1-30 minute
flow times with to Z-k hour of off times between the flowing
periods (Table 3). The flowing/standing periods were provided to
simulate daily water usage in a home. A 8.5 hour standing period
occurred from each morning to late afternoon to represent over
night standing.
4
-------
WATER SAMPLING
Water samples from both the control and the test systems
were collected twice a week on Monday and Wednesday, except on a
few occasions because of holidays, plant or pipe-rig operational
problems, or other special problems. Background water samples
were also collected before each system during the last flowing
period prior to the 8.5 hour standing period. These samples were
collected to characterize the difference in the control and test
waters and to document variations of the source water during the
study period.
One liter standing water samples were collected from the
pipe loops and 500 mL samples from the faucets. To isolate the
loop during sampling, all valves of the test system were closed
during sampling, except for the influent line valve of the loop
being sampled. Becau-se of system design and sample tap location,
250 mL's of water was wasted from each loop prior to collecting
the liter sample. This two step procedure was done to assure that
the 1-liter sample was in direct contact with the pipe loop. The
1-liter sample was selected because it was consistent with
compliance sampling required by the Lead and Copper Rule.8
A 500 Ml sample size was selected for the faucets because
the contact volume of the faucet was approximately 100 Ml and the
DWRD Faucet Study suggested that about four to five bed volumes of
water are required to collect 95-99 percent of the lead and other
metals leached from the brass surfaces in a faucet.7
CHEMICAL ANALYSES
Because of sample instability, pH, chlorine, and dissolved
oxygen analyses were conducted immediately on-site by CWW
personnel. All other analyses were conducted by DWRD personnel at
the EPA Research Center.
Atomic absorption (AA) flame and furnace methods were used
to analyze for metals on water samples collected prior to June 30,
1994. After this date an inductively coupled argon plasma
spectrometer (ICAP) was used for the metals, except leads, plus
silicon, sulfur and phosphorus and a number of other elements.
Because of lower detection limits, lead continued to be analyzed
by the graphite furnace AAS method and potassium by the flame AAS
method throughout the study.
Laboratory quality assurance procedures for analytical
precision and accuracy for all samples followed the procedures
established by the I&PCB, DWRD for all research studies. These
procedures are documented in a QA laboratory operation plan and
included analytical duplicates, spikes, and external quality
controls (USEPA, USGS reference standards).9 Approximately 15
5
-------
percent of the analytical measurements performed by the I&PCB
laboratory consisted xif spikes, duplicates, and QC standards.9
RESULTS
Water Usage and Sampling
The pipe loop corrosion study began on November 2, 1992 and
officially ended on February 24, 1994, the last day for bi-weekly
routine sampling (483 days). The number of sampling days during
the 16-month study period ranged from 6 to 10 per month as shown
in Table 4 for a total of 119 sampling days.
The total flow through the pipe loop systems obtained from
the water meter readings are shown in Figure 2. Over the 483 day
study period, the total flow through the control and test pipe
loop systems was 458,173 gallons (948.6 gpd) and 440,822 gallons
(912.8 gpd), respectively.
Although the total flows and daily rates are not identical
for each system, they were within 4 percent of each other. Water
meters were not installed on each loop and faucet, therefore, the
amount of water that flowed through loop and faucet could not be
exactly determined. -Any differences in total flow from one loop
to another are believed to be small, however, considering the
manifold design and the closeness of the total flows for the two
systems.
Water Quality Monitoring
The control and test waters to the pipe loop systems were
monitored to meet two objectives. One objective was to determine
any water quality difference between the control (unsoftened)
water and the test {softened} water, and the second objective was
to monitor any temporal changes of the source water that might
impact the study results.
Source Water Quality
Because the source water was BWTP finished water, the
quality did not undergo significant changes during the study
period. Variation in several water quality parameters did occur,
however, because of seasonal temperature changes, subsurface
hydrochemical changes, and variations in plant operation. The
water quality parameters affected the most significantly by these
factors were water temperature, pH, hardness, chlorine, dissolved
oxygen, and total ortho-phosphate.
The pH of the source water did not vary substantially but
some temporal variation was noted (Figure 3). The pH of the
source water ranged between 8.8 and 9.4 (except for two
6
-------
unexplained outliers), with a mean of 9.1. The temperature of the
source water varied from a high of near 18°C to a low of near 130C
during the winter periods (Figure 4). Dissolved oxygen
concentration varied from a high of near 8.5 mg/L to a low of near
4.5 mg/L (Figure 5). The variation in dissolved oxygen was due to
operational changes of the treatment plant and dissolved oxygen
solubi1i ty.
Total hardness -of the source water ranged from about 150 to
189 mg/L (as CaC03). The average hardness level was calculated at
159 mg/L (as CaC03) with a standard deviation (SD) was 22.6. The
results of the hardness tests are plotted in Figure 6.
Chlorine is added to the finished tap water to provide a
concentration of near 1 mg/L (Figure 7). Total chlorine averaged
1.12 mg/L and the standard deviation was 0.09 mg/L. Free
available chlorine was only slightly less at a mean of 1.04 mg/L
with a SD of 0.09 mg/L.
Sodium hexametaphosphate (~.5mg/L as PO 4) was added to the
partially treated water prior to filtration to prevent the
encrustation of the sand filters with calcium carbonate. Total
phosphate averaged 0.4 mg/L and orthophosphate 0.09 mg/L. Little
change of either parameter occurred during the study (Figure 8).
The TIC of the source water averaged approximately 16 mg/L
with a range of about 13 to 18 mg/L during the first 150 days of
the study (Figure 9). Because of laboratory analytical problems,
the TIC was not measured during the latter part of the study.
Softened--Unsoftened -Water Quality
A comparison of the mean values of measured water parameters
of unsoftened (control) and ion exchanged softened waters (test)
is shown in Table 1. The most significant differences between the
two waters occurred with five measurements, calcium, magnesium,
potassium, sodium and pH. As expected, the ion exchange softener
reduced the calcium (26 mg/L) and magnesium (24 mg/L)
concentrations in the control water to near zero while increasing
the sodium (25 mg/L) level to 101 mg/L. Potassium, a monovalent
cation, (3.6 mg/L) was reduced by 68 percent to 1.1 mg/L and the
pH was increased slightly from 9.1 to 9.3. Polyphosphate and
orthophosphate species passed through the exchange resin
indicating that they were in anionic or uncharged complex forms.
METAL LEACHING DATA--DATA EVALUATION
The metal leaching data for all of the lead pipe loops are
shown in Figures 10-16. All of the metal leaching data were
examined visually and statistically to determine whether the ion
exchange softened water produced higher metal levels. The first
7
-------
step of the statistical process consisted of determining whether
metal leaching from the loops and faucets reach "stabilization".
This was determined by identifying the point in time at which the
slopes of all of the four curves for each loop type achieved zero
within 95% confidence using 1inear regression tests. Using this
procedure, only two complete sets were determined to reach
stabilization; lead leaching of the lead loops (69 days) and zinc
leaching of the faucets (153 days). For all other sets, at least
one of the four curves did not test for zero slope.
True "stabilization" times may range from weeks to years,
depending upon the nature of the surficial passivation films
formed for a given material. Recent research has indicated that
with copper, for example, initial Cu(0H)2 films will "age" into
less soluble Cu,(0H)2C03 or CuO over periods of perhaps more than
15 to 20 years. Stabilization is also difficult to achieve for
materials such as soldered joints and brass, where physical
depletion of source material, such as lead, occurs. Long term
slow drift may be present even for the two sets of loops that had
apparently stabilized. Another factor that likely interfered with
the stabilization tests was the variation in the leaching data
caused by the changes in water quality.
For the sets of loops and faucets that did not stabilize
based upon the zero slope test, a stabilization time was selected
from a visual inspection of the curves in order to complete the
statistical analysis. Using the metal leaching data from the
stabilization time to the end of the study, statistical tests
(paired t-test or non-parametric equivalent tests) were conducted
on each duplicate loop to determine whether the two data sets
could be paired (averaged). The results of these tests showed
that 10 sets were considered statistically different and six sets
were not different. Visual examination of the data suggests that
pair testing of many of the loops failed because of short periods
of metal leaching variation (increases) of one loop. These short
period increases cannot be explained, but are typical of pipe loop
testing.
The results of the tests for reproducibility of duplicate
loops/faucets are shown in Table 5. If the duplicates agreed
statistically, the results were paired (averaged) for statistical
tests to determine whether the metal leaching levels between the
non-softened and softened conditions were statistically different.
T-tests (normal data sets) or non-parametric equivalent
tests (non-normal data sets) were used to determine significant
differences between softened and unsoftened conditions for the
same type of loops/faucets. As stated above, comparisons were
made using mean values of duplicate sets, if duplicates tested as
the "same". If the duplicate loops/faucets did not test as the
8
-------
"same", each loop or 'faucet was tested against the other mean
(set) or individual loop/faucet.
Lead Leaching Results
Results of the lead sampling data from the lead pipe loops
are shown in Figure 10. Statistical tests showed the lead levels
from the nonsoftened and softened water conditions to be
different. A visual interpretation of the data (Figure 10)
suggests that lead levels of the non-softened water pipe loops
were slightly higher than the lead levels of the softened water
pipe loops by about 0.2 to 0.3 mg/L for each month of the 16-month
study.
The results of the lead leaching data of the copper soldered
pipes are shown in Figure 11. Treatment comparison tests
indicated statistical differences between the individual loops of
each set. The majority of the lead levels of all loops was at the
detectable level of 0.001 mg/L except for some increase between
the 8 and 16 month period for several of the pipes. Consequently,
although the statistical tests indicated lead leaching differences
between the non-softened and softened water pipe loops, the very
low levels of all loops indicate that the softened water had no
real impact on increased lead leaching.
The results of the lead leaching from the faucets
(containing brass) are shown in Figure 12. The statistical tests
for reproducibility of the duplicate faucets showed that the
softened-water faucet data of the two could be paired, but the
non-softened water faucet data could not. Statistical tests for
treatment comparison showed statistical difference for the two
conditions. Examination of the data in Figure 12 shows that the
lead leaching from the softened water faucets were generally
higher, but because of the lead levels for both were very low,
0.003 to 0.008 mg/L, the trend probably had little meaning.
Furthermore, the results show that during the last three months,
the lead levels from the two systems were within 0.002 mg/L of
each other in the 0.003 to 0.005 mg/L range. The results
indicate, therefore, that the softened water had no effect on
increased lead levels.
Copper Leaching Results
The results of copper levels from the copper tubing loops
average are plotted in Figure 13. The reproducibility tests for
the duplicate loops showed that the softened water loops could be
paired, but the non-softened water loops could not be paired.
Tests for treatment comparison was conducted by using the paired
averages of the softened loops against each of the two nonsoftened
water loops. The statistical tests showed the copper levels of
the two to be statistically different indicating differences in
copper leaching of the non-softened and softened loops. The data
9
-------
plotted in Figure 13 showed that the copper levels from the
softened water loops to be higher than the non-softened loops
during the last six months by about 0.2 to 0.4 mg/L. Copper
levels from all loops were within a range of 0.07 to 0.13 mg/L
that significantly lower than the USEPA copper action level of 1.3
mg/L. Consequently, although the softened water had slightly
higher levels, the level of increase was small and may not be
significant.
The copper leaching data of the soldered pipes are shown in
Figure 14. The reproducibility tests were similar to the copper
tubing results. The softened water loop data could be paired, but
the nonsoftened water loop data could not be paired. Two
treatment comparison tests were, therefore, conducted using the
average of the softened water loop data and the individual loop
data for the non-softened loops. One test showed a statistical
difference and the second one did not. Examination of the data
from Figure 14 indicates little or no difference between the two
sets of loops suggesting that the softened water did not have an
impact on copper leaching.
The copper leaching data of the faucets are plotted in
Figure 15. The reproducibility tests showed that one set
(nonsoftened water) could be paired, but the other set (softened
water) could not. Treatment comparison tests, conducted with the
paired set against the individual loops of the softened water,
resulted in both being statistically different. Visual
examination of the monthly average data of Figure 15 suggests that
copper leaching levels were about the same for both the
nonsoftened and softened water. Moreover, the copper levels of
both sets were very low, 0.01 - 0.05 mg/L in comparison to the
copper action level of 1.3 mg/L. The data suggests that the
softened water did not increase the copper levels.
Zinc Leaching Results
The zinc leaching data from the galvanized loops are plotted
in Figure 16. The statistical tests for reproducibility of
duplicate loops showed that neither set could be paired and
averaged, therefore, treatment comparison tests were performed on
the four possible pairs. The test results indicated significant
differences of zinc leaching between the nonsoftened and softened
water galvanized steel test loops. Examination of the monthly
average plotted data of Figure 16 indicates that the zinc levels
may be higher in the softened water loops than nonsoftened water
loops. Because the paired loop data did not show good agreement
for either condition, and because on one nonsoftened water loop
and one softened water loop had a very comparable zinc level,
concluding that the softened water produces higher zinc levels
cannot be supported by the data.
10
-------
The zinc leaching data from the faucets averaged on a
monthly basis are plotted in Figure 17. The statistical tests of
reproducibility of duplicate faucets showed that the softened
water faucet data could be paired, but the nonsoftened data could
not. Treatment comparison tests were therefore conducted by
averaging the paired softened water levels and testing this
average against each of the two loops of the nonsoftened water.
The results of the two plots showed significant differences
between the two exposures for both tests. However, visual
examination of the monthly averaged data suggests no differences
between the leaching levels of each set. Except for the first few
months, the zinc levels from all faucets were very low in the
0.002 to .1 mg/L range.
DISCUSSION OF RESULTS
The objective of this study was to determine whether water
softened by ion exchange treatment to approximately zero hardness
is more corrosive to household plumbing materials than the
nonsoftened water. Naturally soft waters are low hardness, low
mineral content waters with pH's often at or below neutrality and
have been widely demonstrated to have a corrosive effect on most
household plumbing materials. This suggests that ion exchange
softened water would have the same type of corrosive effect on
these materials.
Ion exchange softened waters are different in chemical
characteristics than naturally softened waters, however. Ion
exchange softened waters do not necessarily have a low mineral
content, nor are they necessarily low in pH. Ion exchange
softening fundamentally removes the calcium and magnesium and
replaces these elements with sodium. The process does not
substantially alter the mineralization level, alkalinity or pH.
In other words, ion exchange softening does not produce water with
chemical characteristics commonly associated with naturally soft
waters. The process does remove calcium an element that is
commonly thought to g-ive some beneficial protective effort to
corrosion. Because of this change, ion exchange softening is
often portrayed as a treatment process that could increase the
corrosivity of the water, thereby increasing the metals level.
Little or no data exist to confirm or disprove this hypothesis,
however.
By erroneously using calcium carbonate saturation indices as
a surrogate measure for "corrosivity", much misinformation has
been generated in the past. High pH and high alkalinity also
produce higher (less corrosive) values for the CaC03 saturation
state. However, that combination has been demonstrated
theoretically, 017 by controlled experiments 1°-12-141718) ancj by
field data 0 to be detrimental in some substances to either lead
or copper solubility.
11
-------
Water quality plays a very significant role in causing
corrosion and calcium is only one of many factors that influence
the rate of corrosion. Moreover, calcium is normally not
considered one of the major ones. The AWWARF Lead Control
Strategies Manual states "In spite of the fact that there is
little evidence in the research literature that adherent,
continuous CaC03 films actually form to seal lead pipe against
leaching, calcium carbonate deposition has gained wide acceptance
as a viable lead control strategy." The Manual further states
that the water quality factors that have the greatest influence on
lead mobilization are pH, alkalinity, and dissolved inorganic
carbon (DIC). The treatment strategies stressed, therefore, are
pH, alkalinity and DIC adjustment and the use of corrosion
inhibitors such as orthophosphate and silicate compounds. The
addition of calcium is not considered a corrosion control
treatment strategy. It can be reasoned, therefore, that the
removal of calcium may not necessarily increase the corrosivity of
water.
The water softening study was designed to evaluate the
effect of ion exchanged softened water of one water quality on
household plumbing materials under controlled conditions. The
study was conducted to measure and compare metal leaching of the
materials exposure to nonsoftened and softened water, and also to
examine the potential effect of changes of key water quality
factors that might remove or alter the film forming constituents
in the water. The key factors would certainly include pH,
alkalinity, and DIC.
Statistical evaluation of the metal leaching, generally data
followed the recommendations outlined in a recent AWWARF
document.21 The problem in applying these procedures centers on
the variable nature of the metal leaching data and inclusion of
high or low levels that may be considered outliers. Occasional
high concentration of metal levels are common in home tap samples
and to pipe loop studies. They generally result from chemical and
physical factors of the systems and not sampling or analytical
error. They reflect the variable nature of corrosion and metal
solubility. They also present problems in evaluating and
interpreting the results.
The statistical tests to determine metal level stabilization
indicated that many of the loops and faucets did not reach
stabilization. Stati-stical tests to determine whether data from
duplicates could be paired (averaged) also showed that many of
duplicates could not be paired although visual examination
suggests otherwise. And, finally, based upon an arbitrarily
selected stabilization time, statistically testing to determine
whether the metal leaching levels of the two systems were the same
or different indicate that most results were different. Visual
examination of the data (Figures 10 - 19) suggest little
12
-------
difference in metal levels between the two systems and where
differences were apparent, there was no pattern of the softened
water metal levels being higher than the nonsoftened water levels.
The only observed data to have a consistent difference in metals
levels throughout the study was the lead levels from the lead
pipe. In this case, the levels of the nonsoftened water loops
were always 0.2 to 0.*3 mg/L higher than the softened levels, which
is contrary to the hypothesis that softened water is more
corrosive than nonsoftened water.
The lead levels from the faucets are observed to be slightly
higher in the softened water system, but the absolute levels were
below 0.003 to 0.007 mg/L suggesting little or no difference. The
lead levels from the copper solder pipe were both near the minimum
detectable limit of 0.001 mg/L except for several spikes during
the last eight months of the study.
During the last six months of the study, the copper levels
of the copper tubing loops were about 0.01 to 0.02 mg/L higher in
the softened water loops while there was essentially no difference
between the copper levels of copper pipes or faucets of the two
systems.
The zinc levels of one galvanized steel loop from each
system were essentially the same. The second loop of the softened
water had zinc levels higher than the three other ones while the
zinc level from the second nonsoftened water system was lower than
the other three. If the duplicate sets are averaged, the zinc
levels for the softened system show higher levels throughout the
entire study. The zinc leaching levels from the faucets were
essentially the same after the fifth month.
The lead levels observed for the lead pipe are much higher
(even approximately by a factor of 3) than would be expected for a
water of the same pH and DIC. This model was also verified
experimentally. 10 Therefore, the behavior of lead in this
study suggests an interferant to normal passivation film
formation. X-Ray diffraction analyses of pipe specimens showed
almost no film formation, and not significant basic lead carbonate
as would be expected.
At this OIC/pH combination, the difference in lead
solubility for the small pH difference between the softened and
unsoftened water should be minimal. Therefore, some surface
reaction not directly resulting from hydroxide or carbonate ion is
possibly occurring.
Additionally, copper levels from soft and hard copper pipe
are higher than would be expected based on current experimental
and theoretical work.15"17 X-ray diffraction analyses of pipe
13
-------
specimens from this project did not show normal film formation of
Cu(II) solids, such as CuO or Cu(0H)2 under controlled conditions.
Abnormally higher copper levels were previously observed in
presence of 60-120 mg/l Sulfate at pH 8.1-8.8. However, this
study water also contains, in addition to significant sulfate,
appreciable polyphosphate known to increase lead solubility at
least at pH 8.2 ' 4 Similar effects on copper would be expected.
Field observations have shown the only lead action level
exceediences for major utilities at pH >8.4 were in 13 systems
dosing polyphosphates.
CONCLUSIONS
Considering all of the lead, copper, and zinc leaching data
from all loops and faucets, there is no clear evidence that the
ion exchange softened water produced higher metal levels than the
nonsoftened water. Furthermore, the water quality data show
little difference between the two water qualities for alkalinity,
DIC, two significant corrosion factors.
The ion exchange softening process increased the pH of the
control water by 0.2 to 0.3 units which could have a beneficial
effect on the metal level. Consequently, except for the decrease
in calcium levels, the softened water did not change any of the
significant water quality corrosion parameters that would cause a
prediction of higher metals leaching in the softened water system.
This study involved one water quality test of a non-
aggressive water and, therefore, the results cannot be
extrapolated to all water qualities. Currently, a second study is
being conducted by DWRD using a more aggressive water with a pH of
about 7.3 and a hardness of near 300 mg/L as CaC03.
ACKNOWLEDGMENTS
The authors wish to extend their appreciation to the
Cincinnati Water Works Utility for their assistance in the study,
especially the operation, maintenance,and water quality personnel
of the Bolton Treatment Plant. The EPA, DWRD Laboratory support
staff are acknowledged for the hundreds of water sample analyses
conducted during the study. The authors also extend their
gratitude to WQA's Science Advisory Committee for their technical
advice and comments during the planning and operation phases of
the study, and for their review comments on this report.
14
-------
REFERENCES
1. Lead Control Strategies, AWWA Research Foundation, Denver,
CO. (1990).
2. Schock, Michael R. Causes of Temporal Variability of Lead
in Domestic Plumbing Systems. Environmental Monitory and
Assessment. 15:59 (1990).
3. Karalekas, P.C., Jr. Lead and Other Trace Metals in
Drinking Water in the Boston Metropolitan Area, Jour. NEWWA
90:150 (1976).
4. Murrell, N. E. Impact of Lead and Other Metallic Solders on
Water Quality. EPA/600/S2-90/056, ORD, USEPA, Cincinnati,
OH (Feb. 1991).
5. Reiber, S.H., et al. Corrosion in Water Distribution
Systems of the Pacific Northwest. EPA/600/S2-87/042, ORD
USEPA, Cincinnati, OH (Nov. 1987).
6. Title 15, United States Code 3710a-3710d, Federal Technology
Transfer Act.
7. Gardels, M.C. and Sorg, T.J. A Laboratory Study of the
Leaching of Lead from Water Faucets. JAWWA.81:7:101 (July
1989).
8. Drinking Water Regulations: Maximum Contaminant Level Goals
and National Primary Drinking Water Regulations for Lead and
Copper. Final Rule. Fed. Reo.. 56:110:26460 (June 1991).
9. Quality Assurance Plan: Laboratory Operation and Standard
Operating Procedures for Analysis and Quality Assurance.
Inorganics and Particulates Control Branch, DWRD, RREL,
USEPA, Cincinnati, OH (1994 draft).
10. Schock, M. R. Response of Lead solubility to Dissolved
Carbonate in Drinking Water. JAWWA 72:12 1965 (December
1980).
11. Schock, M. R. Response of Lead Solubility to Dissolved
Carbonate in drinking water. JAWWA,73:3:36 (March 1981).
12. Schock, M. R. and Gardels, M. C. Plumbsolvency Reduction by
High and Low Carbonate-Solubility Relationships. JAWWA.
75:2:87 (February 1983).
13. Schock, M. R. Understanding Corrosion Control Strategies
for Lead. JAWWA, 81:7:88 (July 1989).
15
-------
14. Edwards, M., Meyer, T.E, & Schock, M. R. Alkalinity, pH and
Copper Corrosion By-Product Release. (Submitted manuscript
1995).
15. Schock, M. R., Lytle, D. A. & Clement, J. A. Modeling
Issues of Copper Solubility in Drinking Water. Proceedings
ASCE National Conference on Environmental Engineering,
Boulder, CO (July 11-13, 1994).
16. Schock, M. R., Lytle, D. A. & Clement, J. A. Effects of pH
Carbonate, Orthophosphate and Redox Potential on
Cuprosolvency. • NACE Corrosion/95, Orlando, Ft. (March 20-
24, 1995).
17. Schock, M. R., Lytle, D. A. & Clement, J. A. Effect of pH,
DIC, Orthophosphate and Sulphate on Drinking Water
Cuprosolvency". Office of Research and Development,
Cincinnati, OH. (In Press 1995).
18. Meyer, T. E. & Edwards, M. Effect of Alkalinity on Copper
Corrosion. Proceedings ASCE National Conference on
Environmental Engineering, Boulder, CO. (July 11-13, 1994).
19. Oodril1, D. M. & Edwards, M. "Corrosion Control on the
Basis of Utility Experience". JAWWA, 87:7:In Press 1995.
20. Marshall, W. Copper in Drinking Water: What the Lead and
Copper Rule Tells US and What it Doesn't Tell Us.
Proceedings AWWA Water Quality Technology Conference, San
Francisco, CA. (November 6-10, 1994).
21. Holm, T. R. & Smothers, S. H. "Characterizing the Lead-
Complexing Properties of Polyphosphate Water Treatment
Products by Competing-Ligand Spectrophotometry Using 4-(2-
Pyridylazo) Resorcinol". Int.Environ.Anal.Chem, 41:71.
(1990).
22. Holm, T. R. & Schock, M. R. "Potential Effects of
Polyphosphate Products on Lead Solubility in Plumbing
Systems". JAWWA, (1991).
23. Holm, T.R. & Schock, M. R. "Polyphosphate Debate (Reply to
Comment)". JAWWA 83:12:10, (1991).
16
-------
TABLE 1. COMPARISON OF THE NON-SOFTENED AND SOFTENED WATER
ANAYTE, UNITS
CONTROL WATER
Nori-softened*
TEST WATER
Softened
Lead, nq/L
Calcium, mg/ L
Copper, mg/L
Iron, mg/L
Potassium, mg/L
Magnesium, mg/L
Manganese, mg/L
Sodium, mg/L
Aluminum, mg/L
Sulfur, mg/L
Zinc, mg/L
Mean
Std Dev
Mean
Std Dev
<0.002
0.001
0.002
0.001
25.6
1.183
<0.002
0.001
<0.003
0.003
<0.003
0.003
<0.05
0.016
<0.05
0.009
3.55
0.54
1.13
0.41
23.7
1.17
<2.0
0.19
<0.02
0.003
<0.02
0.002
26.4
3.13
101
6.28
<0.025
0.01
0.01
0.01
22.4
1.11
22.39
1.03
<0.01
0.007
<0.01
0.006
Alkalinity, mg CaC03/L
76.7
4.39
76.8
4.
29
Hardness, mg CaC03/L
159.8
22.63
0.13
1.
22
Sulfate, mg S04/L
67.3
5.72
67.7
6.
32
Chloride, mg/L
45.9
4.30
46.0
4.
33
Silica, mg Si02/L
10.0
0.69
10.0
0.
66
Nitrate, mg N/L
2.19
0.68
2.05
0.
56
Ammonia, mg NH3/L
<0.03
0.02
<0.03
0.
03
Orthophate, mg PO4/L
0.09
0.06
0.09
0.
04
Total Phosphate,mg P04/L
0.42
0.11
0.45
0.
19
Dissolved oxygen, mg/L
6.86
0.97
6.82
0.
98
Total inorganic carbon, mg/Cl
15.9
1.41
16.0
1.
34
Free chlorine, mg C12/L
1.04
0.09
1.02
0.
10
Total chlorine, mg CI2/L
1.12
0.09
1.09
0,
10
pH, pH units
9.09
0.19
9.31
0.
15
Temperature, degrees Celsius
15.1
0.98
15.2
0.
96
* Source water, City of Cincinnati, Bolton Water Treatment Plant, finished
water.
17
-------
TABLE 2. PLUMBING MATERIALS FOR PIPE LOOP SYSTEMS
PIumbing
Material
Identification
(nominal size)
Dimens ion Vol time
IO(in) OD(in) mL foot
Lead Pipe
Copper tubing
Copper pipe
Galvanized pipe
Solder
Faucet
1/2 in ID
1/2 in ID type L Drawn
3/8 in ID type M,
schedule 40
1/2 in ID, schedule 40
50:50 Pb:Sn
single handle
0.47
0.53
0.37
0.56
0.94
0.625
0.45
0.75
87
46
60
59
TABLE 3. DAILY FLOW PATTERN OF PIPE LOOP SYSTEMS
Control System
(Unsoftened)
Time'
(Minutes)
Test Sytstem
(Softened)
Time
(Minutes)
5:00 -
5
15
pm
15
5
15 - 5
30
pm
15
8:00 -
8
15
pm
15
8
15 - 8
30
pm
15
11:00 -
11
15
pm
15
11
15 - 11
30
pm
15
2:00 -
2
15
am
15
2
15 - 2
30
am
15
5:00 -
5
15
am
15
5
00 - 5
15
am
15
8:30 -
9
00
am
30
9
00 - 9
30
am
30
TOTAL
105
TOTAL
105
18
-------
TABLE 4. NUMBER OF SAMPLE COLLECTION DAYS PER MONTH
Sample Collection
Year and Month Number of Days
1992
November 6
December 9
1993
January 7
February 6
March * 10
April 8
May 8
June 7
July 7
August 9
September 8
October 7
November 7
December 8
1994
January 7
February 5
TOTAL 119
19
-------
TABLE 5. RESULTS OF STATISTICALLY TESTS TO DETERMINE REPRODUCIBILITY
OF DUFLICATE PIPES/FAUCETS
Pipe Loops/Faucets
(Metal Leached)
Normality
Tests
Reproducibility
Tests
Lead Loops (lead)
Non-Softened pair Non-normal
Softened pair Normal
Copper/Soldered pipes (lead)
Non-softened Non-normal
Softened Non-normal
Passed
Passed
Failed
Failed
irass faucets (lead)
Non-softened
Softened
Non-normal
Non-normal
Failed
Passed
Copper Tubing (copper)
Non-softened
Softened
Non-normal
Non-normal
Failed
Passed
Copper/Soldered pipes (copper)
Non-softened Non-normal
Softened Non-normal
Failed
Passed
Faucets (copper)
Non-softened
Softened
Non-normal
Non-normal
Passed
Failed
Galvanized pipes
Non-softened
Softened
(zinc)
Non-normal
Non-normal
Failed
Failed
Faucets (zinc)
Non-softened
Softened
Non-normal
Normal
Failed
Passed
20
-------
NJ
PnaMirc
«««*
/
Nk•*
cootrol
valte
Sample port
Water ¦
Softener
C.
Copper Copper Pipe Galvinized
Lead Pipe Tubing w/ Solder Pipe
ffti /T^s. /7^
...L
i
I
5
3
vy
i
v
m.
\J
Brass
Faucet
£
£Tl
DRAIN
Watermeter
Pressure Sensor
Ofi-i
Identical loop
Valve
Figure I. Pipe Loop System.
-------
500000
400000
300000
200000
100000
0
0 100 200 300 400 500
Time- days
Figure 2. Water flow (total) through pipe loop systems.
non-softened system
softened system
22
-------
10.0
9.5
£ 9.0
8.5
8.0
0 100 200 300 400 500
Time- days
Figure 3. pH of the unsoftened and softened water.
u
o non-softened
• softened
i . i i i l
23
-------
20
18
C
u
4)
Ut
B
V-
«
cx
E
«
H
O non- softened
• softened
100 200 300 400 500
Time- days
Figure 4. Temperature of the unsoftened and softened water.
24
-------
o non-softened
• softened
4.0
100 200 300
Time- days
400
500
Figure 5. Dissolved oxygen concentration of
the unsoftened and softened water.
25
-------
o
u
cs
u
60
B
I
C/3
CO
OJ
c
"O
u-
CO
K
O-
r
160 h
140 -
120
100
80 r
i
60 h
r
i
40 h
L
20 h
I"
Qm
¦ oo,
0
'J
o non-softened
• softened
-I
100 200 300
Time- days
400
500
Figure 6. Hardness concentration (as CaC03) of the
unsoftened and softened water.
26
-------
o non-softened
• softened
q 4 1 1 1 1 1 1 1 1 1
0 100 200 300 400 500
Time- days
Figure 7. Chlorine (total) concentration of the unsoftened
and softened water.
27
-------
1.0
0.6 -
o noil-softened
• softened
total phosphate
0.4 *
ortho-phosphate
0 100 200 300 400 500
Time- days
Figure 8. Total and ortho-phosphate concentration of
the unsoftened and softened water.
28
-------
20.0
m
i
2.0
0.0
0
O non-softened 4
• softened
100 200 300
Time- days
400
500
Figure 9. Total inorganic carbon of the
unsoftened and softened water.
29
-------
0.00
0
non-softened
softened
o o
J3 0.20
100 200 300 400 500
Time- days
Figure 10. Lead leached from lead-pipe loops.
30
-------
non-softened
• ¦ softened
200 300
Time- days
500
Figure 11. Lead leaching from copper/solder pipe loops.
31
-------
0.04
0.03
0.02
0.01
0.00
0 100 200 300 400 500
Time- days
Figure 12. Lead leached from brass faucet loops.
Q ~ non-softened
• ¦ softened
32
-------
0.25
0.20
0.15
0.10
0.05
0.00
0 100 200 300 400 500
Time- days
Figure 13. Copper leached from copper tubing loops.
O - non-softened
• ¦ softened
o
~
33
-------
0.40
0.35
0.30
d 0.25
to
S
a.
o.
U 0.15
0.10
0.05
0.00
0 100 200 300 400 500
Time, days
- non-softened
• softened
Figure 14. Copper leached from copper/solder pipe loops.
34
-------
0.25
O ~ non-softened
softened
Q 0.10
0.00
0
100 200 300
Time- days
400
500
Figure 15. Copper leached from brass faucet loops.
35
-------
50
~ non-softened
¦ softened
H 30
N 20
100 200 300 400 5CX
Time- days
Figure 16. Zinc leached from galvanized pipe loops.
36
-------
2.0
1.5
J
"So
6 1.0
i
o
c
N
0.5
T
6
0
non-softened
softened
100 200 300 400 500
Time- days
Figure 17. Zinc leached from brass faucet loops.
37
-------
TECHNICAL REPORT DATA
(Mease read Instructions on the reverse before completingj
1. REPORT NO.
EPA/600/A-95/100
2.
3. REC
4. title and subtitle
An Evaluation of Ion Exchange Softening on the
5. REPORT DATE
Leaching of Metals from Household Plumbing Systems
6. PERFORMING ORGANIZATION CODE
7. AUTHOHfS!
Thomas J. Sorg
Michael R. Schock
Darren A. Lytle
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME ANQ AOORESS
Drinking Water Research Division
NRMRL, ORD, USEPA
26 W. I-iartin Luther King Dr.
Cinti, OH 45268
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Drinking Water Research Division
13. TYPE OF REPORT AND PERIOD COVERED
NRMRL, ORD, USEPA
26 W. Martin Luther King Dr.
Cincinnati, OH 45268
14. SPONSORING AGENCY CODE
EPA/600/14
15. SUPPLEMENTARY NOTES
Project Officer: Thomas J. Sorg, (513) 569-7370
Proceedings: Annual AWWA Conference in Anaheim, CA., June 19-22, 1995
16. ABSTRACT
A 16 month pilot plant study was conducted to determine the effect
of ion exchange softening on the leaching of metals from household plumb-
ing materials. Two pipe loop pilot plant systems were assembled. Each
system consisted of duplicate loops of lead pipe, copper pipe with 50:50
lead-tin soldered joints, copper tubing, galvanized pipe, and brass
faucets. The source water had a pH of about 9 and a hardness of about
160 mg/L as CaC03. One system (control) was fed non-softened water and
the second system (test) ion exchange softened water. Water samples were
collected from each loop, twice a week, for 16 months. The metal
leaching results of lead, copper, and zinc indicated that there was no
consistent pattern of higher metal levels from the softened water leading
to the general conclusion that the softened water was not more corrosive
to the plumbing materials than the non-softened water with the water
quality used for this study.
17.
KEY WORDS ANO DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c, COSAT1 Field/Group
Corrosion
Ion Exchange
Softening
Water Supply
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS iThis Report)
Unclassified
21. NO. OF PAGES
Release to Public
20. SECURITY CLASS (This page!
Unclassified
22, PRICE
EPA Form 2220-1 (R«*. 4-77! previous edition u obsolete
------- |