United States
Environmental Protection
Agency
Water Engineering
Research Laboratory
Cincinnati OH 45268
Research and Development
EPA/600/S2-87/036  July 1987
Project  Summary
Relationships  Between Water
Quality  and  Corrosion  of
Plumbing  Materials in
Buildings

Chester H. Neff, Michael R. Schock, and John I. Marden
  A study was  conducted on the
interrelationships of corrosion  rates,
metal concentrations, and water qual-
ity occurring within galvanized steel
and copper plumbing systems. A com-
prehensive water sampling  program
was implemented to quantify the total
metal  concentrations and the  major
inorganic constituents found in  stand-
ing and  running water samples. Cor-
rosion rates were measured by using
the ASTM D2688 corrosion tester.
  Lead, zinc, copper, iron, and manga-
nese exceeded the MCL in 10.6% to
25.6% of the standing samples  and in
2.2% to  16.0% of the running samples
collected. Chrome-plated brass sam-
pling valves were found to contribute
significantly  to the lead, zinc, and
copper contents of the samples. Cad-
mium did not exceed 0.0005 mg/L in
any sample. During the two years the
sites were operating, significant water
quality variations were noted in the six
public water supplies investigated.
Multiple linear regression analyses of
the data were unsuccessful because of
the exceedingly  large numbers of
variables encountered under uncontrol-
lable field conditions.
  This Project Summary was  devel-
oped by EPA's Water Engineering
Research Laboratory, Cincinnati, OH,
to announce key findings of the
research project that  is fully  docu-
mented in two separate volumes of the
same title (see Project Report ordering
information at back).
Introduction
  Corrosion has been a  long-standing
and serious problem  in  public  water
supply distribution systems. The poten-
tial health  effects  of  corrosion  led to
federal  regulations  establishing  maxi-
mum contaminant  levels (MCL's) for
certain metal concentrations in drinking
water. The regulations state that water
supplies should be  noncorrosive to all
plumbing  materials. Though  some
researchers have attempted to identify
or predict which waters  are corrosive,
none have found a universally acceptable
identification method that considers all
piping materials and conditions of expo-
sure. Some of the controversy in  estab-
lishing the  Secondary Drinking  Water
Regulations was in defining noncorro-
sive water and acceptable methods for
determining the corrosivity of water.
  Two materials that have found wides-
pread use  in  plumbing  systems are
copper and  galvanized steel pipe. Expe-
rience has  shown  that both materials
offer good corrosion resistance to drink-
ing water when the materials are prop-
erly selected  and  installed. However,
many corrosion failures have been
documented for both copper and galvan-
ized steel piping. The impact of these
corroded materials on water quality has
not been adequately investigated. Brass
valves,  lead-based solders, bronze
meters, and other fittings associated with
copper and galvanized steel also make
significant contributions  to soluble or

-------
particulate  metal concentrations in
drinking water.
  The ASTM D2688, Corrosivity of Water
in the Absence of Heat Transfer (Weight
Loss Methods), Method C was developed
by the Illinois State Water Survey (ISWS)
in 1955 and has been used extensively
in state facilities. The method has been
a reliable indicator for corrosion and/or
scale deposition, and it has been used
primarily to indicate the effectiveness of
various water treatment programs.
Attempts were made by the ISWS to
correlate the  mineral  content  of  the
water supplies with the observed corro-
sion rates  for  copper  and galvanized
steel. The  correlations met  with  only
limited success because important
chemical parameters were not deter-
mined or evaluated.
  Complete  water  chemistries  are
required  for many water supplies  if
significant correlations are to be found
between the water quality parameters
and the corrosion rates of metals. Both
corrosion-inhibiting and aggressive
influences have been reported for  var-
ious constituents of drinking water.  The
chemical constituents commonly cited as
influencing the corrosivity of water are
calcium, alkalinity, pH, carbon dioxide,
sulfate, chloride, dissolved oxygen, silica,
temperature, and dissolved solids. Other
constituents, such as chlorine, organics,
and polyphosphates, are also suspected
to influence the  corrosivity of water.
Interaction among the various influences
can  be observed  only  by prolonged,
complicated  laboratory testing or by
determining the corrosivity of water in
real distribution systems.

Project Objectives
  The principal objective of the corrosion
study was to determine the influence of
various water quality characteristics on
the corrosion rates of galvanized steel
and  copper pipe specimens installed in
the potable water systems of residences
and  large buildings. A second objective
was to monitor the trace metal concen-
trations contributed by  galvanized steel
and  copper  plumbing materials to the
drinking water.
Methods
  Because of cost limitations, the inves-
tigation was limited to six public water
supplies in  Illinois representing both
groundwater and surface water sources.
Each  water supply used  various water
treatment methods that  included lime
softening, ion-exchange softening, pol-
yphosphate, and silicate programs.
  The 19 corrosion test sites selected for
the study  were  located  in  different
sections  of the distribution  system of
each water supply.  A planned-interval
corrosion test was considered the best
procedure for evaluating the  effects of
time and water corrosivity on the cor-
rosion of metals.  Corrosion specimens
prepared according to ASTM D2688,
Method  C, are easily  adapted  to  the
planned-interval method of evaluation.
Installing two of the ASTM corrosion test
assemblies (each containing two  corro-
sion specimens) at each test site, and
replacing each of the specimens in turn
with another specimen  at  6-month
intervals during  the  corrosion  study
provided  seven weight  loss  measure-
ments, each  representing  a different
period  during the study.  The use  of
duplicate specimens was considered and
would have provided useful information,
but the additional plumbing and handling
costs prohibited their installation.
  Total zinc, copper, lead, iron, manga-
nese, and cadmium concentrations were
determined for both flowing  and non-
flowing samples collected  from taps in
the buildings where corrosion test sites
were located.  Water  samples were
collected for a complete chemical anal-
ysis at 2-week intervals throughout the
corrosion study. To evaluate the reliabil-
ity  of the data  and  to improve  the
accuracy  of  interpretation,  additional
studies were  performed at some  of the
sites to assess the rates of polyphosphate
reversion,  the contribution  of  metals
associated with suspended particulates
to the total metal load, and the contri-
bution of the chrome-plated brass sam-
pling taps to the observed metal levels.
  On completion of the data  collection
phase of the corrosion study,  the water
quality, corrosion, and trace metal data
were evaluated  by a  multiple  linear
regression method in an  attempt to
identify  significant factors influencing
the corrosivity of water.

Sample Collection and
Analysis
  Sampling procedures were demon-
strated,  documented, and  standardized
before the study began because ISWS
field personnel  were  responsible for
sample collection at only five sites. These
experts visited the remaining sites every
6 weeks to monitor compliance with the
specified procedures.
  At corrosion test sites located in homes
or buildings,  water samples were col-
lected from the  cold water tap before
early morning use.  The simulated cor-
rosion test loops were  regulated with
times and solenoid valves  to enable
sample  collection at convenient times
and to control the running and standing
intervals.
  Water samples were first taken from
the sampling  tap immediately following
the galvanized pipe loop to preserve the
nonflowing conditions  in the copper
section.
  At all  corrosion test sites, water
samples were collected from a chrome-
plated brass faucet or service tap. Screen
filters were removed from the faucets
before sampling. The first 50 to 100 mL
of water was collected from the tap for
a  temperature  measurement of the
nonflowing sample.  The next 125 mL of
water drawn from the tap was collected
in a 125-mL polyethylene bottle contain-
ing 25 ml of 1:1 nitric acid for use in
determining the metal  concentrations of
the standing water in the test loop.
  The sampling valve was then opened
to allow approximately 0.5 gpm of water
to flow to waste until the water temper-
ature appeared stable and representative
of the distribution water. The tempera-
ture of the running sample was recorded,
and a  second  125 mL of water  was
collected in a polyethylene bottle con-
taining nitric  acid. This second sample
was for use  in determining the  metal
concentrations of the flowing water. Two
polyethylene sample bottles (one 125-mL
and one 500-mL, each containing  1 mL
of concentrated sulf uric acid) and a glass
vial for TOC analyses were filled with the
flowing water from the tap
   Free and combined  residual chlorine
and pH measurements were made on the
running water after the samples  were
collected.  Every 6 weeks,  dissolved
oxygen was analyzed in the field as well.
The electrometric  pH  value, residual
chlorine concentrations, temperature,
and meter reading were then recorded
on a prepared form for enclosure with
the sample bottles.
   Styrofoam-msulated  cartons contain-
ing two ice packs were used to ship water
samplestothe laboratory. Water samples
from  some corrosion  test sites  were
shipped by 1-day express delivery ser-
vices. Samples from the remaining test
sites  were delivered  directly to the
laboratory by ISWS or  utility  sampling
personnel. The water  samples  were

-------
refrigerated until time  of shipment  if
delays were experienced.
  Upon receipt of the water samples in
the laboratory, the 500-mL unpreserved
samples were filtered and immediately
analyzed for alkalinity.  When two sites
were located in tandem, only one of the
unpreserved samples  was  taken for
analysis  because  of  time  and cost
constraints.  Every  tenth sample (as
received) was split into duplicates to use
as precision quality controls.
  Analyses for trace metals (iron, zinc,
manganese,  copper, lead, cadmium),
major cations (calcium,  magnesium,
sodium, potassium), and major noncar-
bonate anions (chloride, sulfate, nitrate)
were performed by the Analytical Chem-
istry Unit of the State Water Survey.
  Analytical rechecks of  suspicious
anion values  were performed  with
somewhat different procedures to verify
freedom  from interferences. Quality
control checks included  visual inspection
of analytical reports, ion balance  com-
putations, submission of U.S. Geological
Survey (USGS) and  U.S. Environmental
Protection  Agency (EPA) reference
standards as knowns and unknowns, and
use of the aforementioned split samples.
  Some special studies were conducted
to determine whether  the trace metal
levels were attributable  to particulate
material  or to dissolved species. The
study was complicated by the inability to
separate the metal contribution from the
sampling taps, and definitive conclusions
could not be drawn

Data Evaluation Methods
  Following checks  of  analytical accu-
racy and internal  consistency, the ana-
lytical data for  the major ionic constit-
uents  (plus silica)  were put into the
WATSPEC3 computer  aqueous equili-
brium chemistry  speciation  model  to
compute the saturation  states of several
important film-forming solids, aggressiv-
ity parameters  such as Larson's  ratio
(chloride plus sulfate to bicarbonate), and
the free COz concentration. The program
makes corrections for complexation and
ion pairing of major  water constituents.
The absence of analytical speciation and
thermodynamic data  for polyphosphate
species made many of the computations
inaccurate  for the  polyphosphate-
containing systems. Thus they were not
included in the overall water chemistry
data base.
  The weight  losses  and calculated
corrosion  rates for  the ISWS tester
coupons were entered m a separate data
base to be accessed by computer pro-
grams that would find  water quality/
corrosivity relationships.
  The following three statistical evalua-
tions were  performed  by  using  the
statistical analysis system (SAS) on the
University  of  Illinois  IBM  4341*
computer:

  1.  The relationship of the trace metal
     concentrations at each site to the
     independent variables (major con-
     stituents and physical parameters
     such as temperature and velocity)
     at each site.

  2.  The relationship of the trace metal
     concentrations to the independent
     variables among all the sites.

  3.  The relationship of the measured
     weight  losses  of the test  pipe
     specimens to the measured water
     quality variables among sites.

  Various stepwise linear regression and
general linear model  analyses were
performed on  untransformed data, but
significant  correlations  were   not
observed.  Regressions  were not per-
formed for cadmium, iron, and manga-
nese. All cadmium concentrations were
at or below the analytical detection limit.
Many source  waters  contained consid-
erable iron, so its presence could not be
attributed to corrosion. Manganese is not
a significant component of any of the
piping materials.

Additional Laboratory Studies
  Three laboratory studies were  con-
ducted during the course of this project
to determine  the significance  of some
problems that were  appearing in  the
chemical data. These involved investiga-
tions  of  (1) polyphosphate  hydrolysis
before sample  analysis at four sites, (2)
the contribution of zinc, copper, and lead
from the chrome-plated  brass sampling
valves at two sites, and  (3) the amount
of trace metal (copper,  zinc,  iron, lead,
cadmium)  leaching into solution from
chrome-plated brass taps exposed over-
night to deionized or tap water.
  To determine  whether  prior piping
contamination of new  chrome-plated
brass sampling taps could be responsible
for a significant part of the observed trace
'Mention of trade names or commercial products
 does not constitute endorsement or recommenda-
 tion for use
metal  levels, a  new test faucet was
installed before the water meter and test
loop 306.  A 125-mL  standing  water
sample was initially taken at this loca-
tion. After  97 days of operation, three
successive 125-mL unfiltered samples
were taken,  and this  procedure was
continued through the end of the oper-
ation at this site. After 130  days the
chrome-plated brass faucet was replaced
with  one   of   low-density  linear
polyethylene.
  Tof urther pursue the leaching from the
chrome-plated sampling taps, a 2-week
benchtop experiment was set up. Six new
faucets were rinsed  copiously  with
deionized water, inverted,  and filled to
capacity (approximately 20 mL)  with
either tap water or deionized water. The
faucets were sealed with new silicone
stoppers and mounted with finger clamps
on  a flexiframe  stand in  an  inverted
position until the next day. After samples
were withdrawn,  the silicone stoppers
were rinsed with 1:1 HNOaand deionized
water, the  faucets were filled,  and the
stoppers were replaced. Water samples
were taken on alternate days for  either
lead or cadmium,  or  zinc,  iron, and
copper.

Corrosion Testing Methods
  The  corrosion  test  assembly was
reduced in  size from 1  in., as specified
by the ASTM D2688 method, to 0.5  in.
to conform with the nominal pipe dimen-
sions encountered in household plumb-
ing systems. The smaller test assemblies
successfully simulated the actual surface
conditions found in a straight length of
pipe where the  flow of water is not
distorted. Visual  inspection of the cor-
rosion  specimens  and the associated
piping  of the test loop after 2 years of
exposure showed that the interior sur-
faces  were identical in appearance.
Construction details for the smaller test
assemblies will be provided to ASTM for
inclusion  in ASTM Method  D2688,
Corrosivity  of Water in the Absence of
Heat Transfer. An estimate of procedure
variability for the ASTM  D2688 method
is presented for the first time.

Results

Corrosion Data
  The  corrodibility of both copper and
galvanized  steel  decreased during the
24-month exposure in the more aggres-
sive public water supplies. This decrease
is attributed to the formation of films that
reduce the rate  of corrosion.  In less

-------
aggressive water supplies, the corrodi-
bility of the  metals remained relatively
constant  at  low corrosion  rates,
indicating a slow development of surface
films.
  The weight  loss data for  galvanized
steel specimens were more erratic than
similar data  for copper  specimens. This
result was attributed to the spotty nature
of the surface film on the galvanized steel
and to  the uniform, continuous surface
film  on copper. The  use of  multiple
specimens for each interval of exposure
would have  been beneficial in reducing
the variance in the galvanized  steel
corrosion data.
  For a single corrosion  measurement to
provide reliable  information on the
corrodibility of copper or galvanized steel
in public water supplies, the minimum
exposure period should equal or exceed
12 months for copper and 18  months for
galvanized steel.
  The weight loss  data for copper cor-
rosion  specimens indicated that the
Dwight Correctional Center water supply
was the most aggressive. Here corrosion
rates ranged from 3.6 to 1.4  milligrams
per square decimeter per day (mdd)  for
6- and 24-month exposure intervals,
respectively. The distinguishing charac-
teristics of this water supply were very
high concentrations of chloride, sulfate,
sodium, and (consequently)  total  dis-
solved solids.
  Water from  Springfield, Illinois, was
the least aggressive  toward copper
materials,  with corrosion rates ranging
from 0.4 to 0.2 mdd for 6- and 24-month
exposure  intervals, respectively.  This
supply  was characterized by  a high pH,
low  alkalinity, high dissolved oxygen,
high chlorine, and high  nitrate content.
  Champaign,  Illinois,  had  the  water
supply  that was least aggressive toward
galvanized steel materials, with galvan-
ized steel corrosion rates ranging from
0.6 to  0.9 mdd for all  seven corrosion
specimens. This water supply is a  lime-
soda-softened supply in which the
Langelier index is maintained near +0.4
and  mineral content  is moderate in
concentration.
  The  weight  loss data for  galvanized
steel corrosion specimens indicated that
the City of Carbondale, Dwight Correc-
tional Center, and  City  of Dwight water
supplies were equally aggressive toward
galvanized steel. Corrosion rates ranged
from 6.3 to 2.8 mdd for 6- and 24-month
exposure  intervals,  respectively.  The
galvanized steel weight loss data  must
be evaluated with caution, as the cor-
rosion of the zinc  surface  was  not
uniform  and  accounted  for  a  large
variance  in  weight  loss  data  for
galvanized specimens.
  The corrosivity of the water  supplies
(evaluated by the planned interval  test
method) did  not change significantly as
a result of major fluctuations in  water
quality of short-term or cyclic nature. The
varying concentrations of several chem-
ical constituents known to influence the
corrosivity of  water were apparently
averaged out  over  the long  exposure
interval.
  A change  in the corrosion treatment
program in the Carbondale, Illinois, water
supply from  a  zinc polyphosphate pro-
gram  to a pH control program  resulted
in a pH increase from 7.0 to  8.3  and
reduced the corrosivity of the water. The
change  in corrosivity was  more signif-
icant for galvanized steel than for copper
plumbing.
  The corrosivity of the City of Dwight
water  supply  decreased  significantly
during the last 18 months  of the  study.
Total copper concentrations of samples
collected from a copper plumbing system
installed  in this  water  supply also
decreased, which confirmed the reduc-
tion  in corrosivity. The reason for the
decrease in corrosivity was not identified,
and apparently it was due to one of the
unknown factors influencing corrosion in
public water supplies.
  Copper  tubing was susceptible to
increased corrosion at water  velocities
exceeding 1.22  m/s. The increase in
corrosion because of velocity was
observed before  the actual erosion of
surface films and the exposure of bare
metal occurred.
  Copper corrosion rates were not influ-
enced by stagnant water conditions in
the Springfield, Illinois, water supply, but
galvanized steel  experienced an appre-
ciably higher corrosion rate for the same
exposure. Very low corrosion rates were
observed for both metals under flowing
conditions in this system.

Quality of Public Water
Supplies
  Significant variations in  water  chem-
istry occurred for the six water supplies
studied, although  each  supply was
originally conceived of as being stable in
quality.  A comprehensive chemical
analysis of water samples collected every
15 days  over a  2-year  period was
required to observe the real  stability of
the water supplies.
  The  variation  in  water quality was
caused by changes  in water treatment
programs, changes in operational proce-
dures, and  unanticipated equipment
failures. Excursions in water quality were
observed to be both random and cyclic,
with cyclic variations occurring at 3-day,
monthly, or seasonal intervals.  Random
changes in water quality occurred when
equipment failed or when a major change
was made in the treatment program.
  The  surface  water  supplies expe-
rienced normal  seasonal changes  in
chemical content  and  temperature.
However,  major fluctuations  in water
quality were  observed  in groundwater
supplies using  multiple wells when
various wells were put into or taken out
of operation. An increased sampling
frequency is required  to observe the
cyclic fluctuations in water quality.
  Water supplies with multiple sampling
sites showed no evidence of a significant
change  of water  chemistry within the
distribution systems, although corrosion
tests indicated a difference in corrosivity.
At the Fox Developmental Center water
supply,  a microbiologically mediated
change  was presumed to  slightly
increase the  nitrate concentration and
reduce  the  phosphate  concentration
during  the filtration  and  softening
process.
  Variations in  water quality appear  to
be commonplace  in public water supp-
lies. Change in water  chemistry may
adversely influence the corrosivity of the
water supply and therefore should be
controlled whenever possible.

Total Metal Concentrations of
Samples
  The maximum contaminant level
(MCL) for iron was  exceeded in 25.6%
of the standing water  samples  and  in
16.0% of the running  water samples.
Manganese exceeded the MCL in 14.5%
of the standing samples and 11.6%  of
the running samples.
  The naturally occurring iron  or man-
ganese content of the water source was
observed to  be  more  significant than
corrosion processes or water quality in
influencing the total iron or manganese
concentrations in samples. The presence
of polyphosphate in the water supplies
was observed to complex both  iron and
manganese, maintaining the concentra-
tions above the MCL in  some supplies
  Cadmium concentrations  were  very
low in  all  samples, both running and
standing. Most samples were below the

-------
minimum detection limit of 0.3 /ug/L for
cadmium. The MCLfor cadmium was not
exceeded by any sample collected during
the study, with the maximum observed
concentration being 4.8 fjg/L cadmium.
  The zinc, copper, and lead concentra-
tions of the water samples were found
to be  associated with  corrosion in
systems containing  either  copper or
galvanized steel  plumbing  materials,
although multiple  linear regression
models were  unable to identify  any
significant relationships between metal
concentrations and water quality.
  The  MCL for lead was exceeded in
17.1% of the standing  samples and in
3.1% of the running samples (Figures 1
and 2). The running samples exceeding
the MCLfor lead were collected from one
galvanized and five copper  plumbing
systems. The corrosion of lead-tin solder
and  the brass sampling  valves was
responsible for the lead content of
samples.
  Copper concentrations  exceeded the
MCL in 10.6% of the standing samples
and 4.5% of the running samples. Zinc
exceeded the MCL in 11.8% of standing
samples  and 2.2% of  the running
samples.
  The concentrations of copper, zinc, and
lead  in  both  standing and running
samples  generally decreased over the
24-month  sampling   period    and
approached an apparent mean concen-
tration plateau around which the metal
concentrations fluctuated. Before reach-
ing the concentration plateau, the metal
content fluctuated over a wide concen-
tration range from sample to sample. The
overall distribution of the trace  metal
concentrations appeared to be approxi-
mately log-normal rather than Gaussian.
  In one water supply, the  metal con-
centrations sharply increased during the
latter stages of the study after apparently
attaining stable concentrations. Similar
observations have been made in Seattle,
Washington, although the possible metal
contribution by sampling valves was not
considered.
  The  total metal  concentrations in
samples were influenced by other factors
such as piping configurations before the
sampling  valve,  quantity and rate of
flushing  before sample collection,  and
water quality.

Additional Laboratory Studies
  In the polyphosphate hydrolysis study,
substantial changes were observed in
the concentration of orthophosphate
over time at two of the sites  examined.
  60 r
  55 -
 Figure 1.    Percent of standing water samples exceeding the MCL for lead.
  25 r
  20-
  15-

-------
The  conclusion reached  was that the
orthophosphate data from these sites are
probably not very accurate. No quantita-
tive correction could be applied to the
data collected  before or after discovery
of the problem.
  In the study of new faucets to deter-
mine whether old piping or new chrome-
plated brass sampling taps were respon-
sible for the observed trace metal levels,
the data (Table 1} clearly  show that the
chrome-plated sampling tap contributed
large amounts of lead and zinc and a
somewhat smaller amount of copper and
iron to the standing samples. The con-
tamination decreased with time but was
still  considerable  after  more than  5
months. The elevated zinc values for the
first 125-mL samples with the polyeth-
ylene faucet represent  a contribution
from the relatively new, short, galvanized
section.  These data  are particularly
enlightening because  the low  metal
levels shown with the plastic tap corrob-
orate the low corrosion rates measured
in the weight-loss  coupon  tests. The
metal levels obtained during the  main
portion of the study seemed  unusually
high compared with the measured cor-
rosion rates. The low lead  levels indicate
insignificant contamination  from the
galvanized pipe coating.
  Results from the  2-week  benchtop
experiment with  new faucets showed
that extremely high levels of lead  were
leached from the tap, even after 2 weeks.
The  MCL was exceeded by all samples
taken. A small amount of cadmium was
removed in the first equilbration from the
taps containing Champaign  tap water,
but  it  was  not detectable afterwards.
More cadmium was removed by  deio-
nized water, and  leaching was contin-
uing at detectable levels after 2 weeks.
  Iron leaching was odd in that more was
generally taken into the tap water than
the deionized water. As expected,  there
was considerable zinc leaching with both
water  types, although there was less
from the faucets  containing tap water.
The leaching of copper was similar in the
two water types  and showed a  fairly
consistent level after about 4 days.
  These experiments conclusively dem-
onstrated  that  chrome-plated  brass
sampling  taps can be a considerable
source of readily leachable lead and zinc,
when  lead solder, lead pipes, and gal-
vanized pipes  are not present. The use
of  such  materials  might constitute a
health risk under some conditions, and
they  can render data from  corrosion
studies such as this difficult or impos-
sible  to  interpret  in  a framework of
relating analyzed metal levels to corro-
sion  rates,  mechanisms,  or  pipe
solubility.

Conclusions
  Public water supplies experience
variations in water quality that may affect
the corrosion of plumbing materials and
the concentration of trace metals found
in drinking water. Statistical analyses of
the water  quality  and corrosion  data
failed to identify any meaningful relation-
ships because of the variation in  quality.
There are more factors and interrelation-
ships affecting  corrosion than  can be
taken into account by simple statistical
regression models. Controlled laboratory
studies  are necessary  for pursuing
correlations between water  quality,
corrosion  rates,  and  trace  metal
concentrations.
  Corrosion of plumbing materials was
responsible for the lead, copper, and zinc
concentrations found in drinking water.
Iron and manganese concentrations
were primarily derived from naturally
occurring substances in the  source.
Application of polyphosphate  to the
water supply complexed  both iron and
manganese and may have partially
solubilized  corrosion products of  lead,
copper, and zinc. Cadmium concentra-
tions were not significantly influenced by
the  corrosion  of galvanized steel or
copper  plumbing  materials.  Chrome-
plated brass valves, commonly used for
sampling, contributed significant quan-
tities of lead, copper, and zinc to water,
particularly  in  standing  samples. The
corrosion of lead-tin  solder in  copper
plumbing also  contributed  to the  lead
content of drinking water as reported in
previous studies; however, the contribu-
tion of  lead by common  household
faucets may not have  been recognized
or differentiated in  past  studies.  Lead,
copper, and zinc concentrations exceed
the MCL in samples collected from new
plumbing systems at a greater frequency
than MCL's are exceeded in older sys-
tems.  In more aggressive  waters, the
metal concentrations of water from new
plumbing decrease gradually for several
months  before reaching a minimum
concentration plateau. In a  few instan-
ces, the  metal concentrations exhibited
an initial decrease, followed by a gradual
increase which eventually exceeded the
initial concentrations. Proper flushing of
plumbing prior to  use will reduce the
occurrence and concentration  of trace
metals in drinking water.
  Corrosion rate measurements, mad<
by using  the  ASTM  D2688  corrosioi
testers  and the planned  interval  tes
method, were effective in assessing thi
corrosivity of the water and the corrod
ibility of exposed  plumbing materials
Under  controlled  field conditions, th<
procedures provided reliable informatior
on the effects of treatment and change;
in water quality on corrosion rates durmc
long-term  (>6-month)  test intervals
Short-term variations in treatment oi
water quality  are  not detected by the
testing procedures  used in this study.

Recommendations
  1.  A  standard procedure  should be
     adopted   for  determining  anc
     reporting corrosion measurements
     in public water  supplies.  The
     planned  interval test method
     which compares the weight loss ol
     specimens   over  various  time
     frames, is the ideal  approach foi
     detecting  significant changes in
     the corrosivity of  potable water.
     The  conversion  of  weight  loss
     values to corrosion rates assumes
     a  linear  relationship with time.
     Such conversions are applicable in
     controlled environments where the
     corrodibilityof a metal is of interest.
     Satisfactory weight loss measure-
     ments are obtained by the ASTM
     D2688 method for either purpose,
     but multiple  corrosion  specimens
     are  recommended for each  test
     interval.

  2.  A procedure  should be developed
     to produce  instantaneous mea-
     surement of corrosion rates and to
     detect short-term variations in the
     corrosivity of public water supplies.

  3.  The trace metal content of drinking
     water was strongly influenced by
     abnormalities in  the sampling
     process; sampling protocols should
     therefore be developed to meet the
     specific requirements of  environ-
     mental health  studies, corrosion
     studies, or water quality studies.

  4.  The  impact of brass, bronze,  and
     other copper alloy plumbing mate-
     rials on the trace metal content of
     drinking  water  should  be given
     high  priority in  future  studies.
     Chrome-plated brass faucets were
     found to be a  major source of lead,
     copper, and  zinc  concentrations
     found in the public water supplies

-------
Table 1.
Metal Concentrations (mg/L) in Successive 125-mL Unfiltered Standing Samples from the Sampling  Tap Installed Before Site
306
                                 Zinc
                                             Copper
Lead
Iron
Date
3/8/83
4/6/83
4/14/83
4/21/83
5/3/83
5/14/83
5/24/83
6/13/83
6/22/83
6/27/83
7/12/83

7/16/83
7/21/83
8/5/83
8/15/83
8/22/83
Elapsed
Days
1
30
38
45
57
68
78
98
107
112
127

131
137
151
161
168
1
1.30
0.99
2.09
2.66
2.08
1.98
0.65
1.10
1.01
0.75
0.68

0.30
0.22
0.35
0.55
0.46
2
—
—
—
—
—
—
—
0.41
0.42
0.28
0.69

0.20
0.16
0.26
0.22
0.21
3
—
—
—
—
—
—
—
0.48
0.30
0.26
0.25

0.24
0.14
0.22
0.01
0.17
1
0.73
0.31
0.18
0.77
0.06
0.07
0.29
0.04
0.06
0.08
0.03

<0.01
<0.01
<0.01
<0.01
<0.01
2
—
—
—
—
—
—
—
0.03
0.03
0.02
0.04
3
—
—
—
—
—
—
—
0.04
0.04
0.04
0.04
1
0.099
0.056
0.059
0.060
0.031
0.075
0.021
0.030
0.019
0.015
0.013
2
—
—
—
—
—
—
—
0.011
0.008
O.Q05
0.023
3
—
—
—
—
—
—
—
0.008
0.005
0.004
0.004
1
0.11
0.03
0.28
0.93
0.29
0.32
0.09
0.42
0.36
0.52
0.28
2 3
—
—
—
—
—
—
—
0.07 o.;o
0.05 <0.06
0.06 <0.07
0.72 0.05
7" // d
<0.01 <0.01
<0.01
<0.01
<0.01

-------
   Natural Resources and the University of
   Illinois under the sponsorship of the U.S.
   Environmental Protection Agency.
          Chester H. Neff and Michael P. Schock are with the Illinois Department of
            Energy  and Natural Resources. Champaign. IL 61820; and John I. Marden
            is with the University of Illinois, Urbana. IL 61801.
          Marvin C. Gardels is the EPA Project Officer (see below).
          The complete report consists of two volumes, entitled "Relationships Between
            Water Quality and Corrosion of Plumbing Materials in Buildings:"
            "Volume I. Galvanized Steel and Copper Plumbing Systems," (Order No. PB
            87-192 548/AS; Cost: $18.95)
            "Volume II. Appendices," (Order No. PB 87-192 555/AS; Cost: $18.95)
          The above reports will be available only from: (costs subject to change)
                 National Technical Information Service
                 5285 Port Royal Road
                 Springfield, VA 22161
                 Telephone: 703-487-4650
          The EPA Project Officer can be contacted at:
                 Water Engineering Research Laboratory
                 U.S. Environmental Protection Agency
                 Cincinnati.  OH 45268
United States
Environmental Protection
Agency
Official Business
Penalty for Private Use $300

EPA/.600/S2-87/036
Center for Environmental Research
Information
Cincinnati OH 45268
                                                              BULK RATE
                                                         POSTAGE & FEES
                                                                 EPA
                                                           PERMIT No G-3
CHICAGO
                                          60604

-------