United States
Environmental Protection Office of Water EPA 815-B-97-004
Agency , 4607 December 1997
&EPA The Effect of Temperature
on Corrosion Control
-------
-------
"The Effect of Temperature on Corrosion Control"
EPA 815-B-97-004
December 1997
This Publication Contains the Following Documents:
Colling, J.H., Croll, B.T., Whincup, P.A.E., and Harward, C. June 1992. Plumbosolvency
Effects and Control in Hard Waters. JIWEM, 6(3):259-268.
Dodrill, D.M., and Edwards, M. July 1995. Corrosion Control on the Basis of Utility
Experience. JournalAWWA, 74-85.
Edwards, M., Schock, M.R., and Meyer, T.E. March 1996. Alkalinity, pH, and Copper
Corrosion By-Products Release. Journal A WWA, 81-94.
Rezania, L.W. and Anderl, W.H. 1996. Copper Corrosion and Iron Removal Plants.
Conference Paper. Section of Drinking Water Protection, Minnesota Department of Health.
U.S. Environmental Protection Agency. An Evaluation of the Secondardy Effects of Enhanced
Coagulation, With Emphasis on Corrosion Control. Conference Paper prepared by D.A. Lytle,
M.R. Schock, and RJ. Miltner, Treatment and Technology Evaluation Branch, Water Supply and
Water Resources Division, National Risk Manageent Research Laboratory.
U.S. Environmental Protection Agency. December 19, 1996. Memo from Michael R. Schock,
Treatment and Technology Branch, Water Supply and Water Resources Division, National Risk
Management Research Laboratory, to Jeffrey B. Kempic of the Office of Ground Water and
Drinking Water, entitled: Seasonal Monitoring Revision. December 19, 1996. (Note:
References 5, 6, and 7 cited in the memo are not provided for public review and comment. The'
Agency is not factoring the data contained hi these studies into its decision making.)
Wagner, I. June 18-22, 1988. Effects of Inhibitors on Corrosion Rate and Metal Uptake.
Proceedings of American Water Works Conference. Memo from Michael R. Schock to Judith
M. Lebowich, February 13, 1997, regarding Requested References.
-------
FE?-06-1997 12:37
USEPA
5135597172 P.02
•-•nt of new Hltcr
•/. IVX7.86. (II.
v from sludno.
!. /. E. Ilallj.
Plumbosolvency .Effects and Control in Hard Waters
By J. H. COLLING. BSc'. B. T. CROLL. BSc. PhD (Fellow)**. P. A. E. WHINCUP.
PhD", and C. HARWARD. BSc (Member)**
ABSTRACT
A laboratory lead-pipe rig has been used to support
Anglian Water's successful orthophosphate dosing
programme to reduce plumhosolvency. The hard
waters in the region generally fall into low or high
plumfaosolvency categories according to the types of
crystalline deposit formed. To improve the cost-
effectiveness of plumbosolvency control, the effects
of temperature, phosphate doses, blending and
alternation of these waters were investigated.
Initial phosphate concentrations must be above
0.6 mg P/I (as phosphorus) to establish plumbo-
solvency control. Subsequently, phosphate doses
may be reduced, provided that dosing is continuous
and sufficient phosphate reaches the extremities of
the distribution system. When high and low piumbo-
solvency waters are blended before distribution.
both (or the mixture) must be phosphate dosed.
However, where waters alternate in distribution.
laboratory studies have shown that low plumbo-
solvency deposits are more stable, resulting in low
lead concentrations. The high plumbsolvency of
some hard waters may be due to the presence of low
concentrations of humic substances.
AVy words: Hard water: loath phosphate: plumhosolvency:
treatment: vvurcr.
INTRODUCTION
Up to the mid-1970s hard waters, such as exist in.the
Anglian Water region, were not normally con-
sidered to be plumbosolvent. the problem normally
being seen as one of soft waters. The increased
water supply monitoring in the Anglian region
following the reorganization of the water industry in
1974 showed that some hard waters were plumbo-
solvent. although not to the >ame degree as soft
waters. This information wu> continued by the
survey organized by the Department of the Environ-
ment'(DoE) from'1979 to IVSI1" which identified
priority areas for action to reduce lead concen-
trations at the tap.
This paper is an updated version 01 .1 paper presented 1" the
Institution's Scientific Suction Symp>v-mm tin l.twl in Vt'uier held in
London on 12 April lixy.
•Research Fellow and Senior Lecturer iHumberside I'nlytcchnici.
"Proecv. Science Manager and r.-tncipiil IVocc^ Scientist.
Anglian Water Services. Ltd.
0. 6. June.
J.IWKM. l«M2. 6. June.
. 3;
Concurrent with the DoE survey, experimental
work at the Water Research Centre (WRc) identi-
fied that the addition of orthophosphate to waters at
a concentration of approximately 1.0 mg P/l (as
phosphorus) was able to reduce lead solvency in
hard waters1-'. This treatment was successfully
installed in two triai areas in Anglian Water (Boston
and Sleaford)1-" and was subsequently recom-
mended for plumbosolvency control in hard waters
by the \VRc'a' and DoE'-"."
The 1979-S1 survey identified those areas where
the incidence of lead solvency was highest, based on
the proportion of random samples givina elevated
lead concentrations. Therefore, if an area had very
few lead pipes but a plumbosolvent water, it was
possible that it would not be designated high
priority. The former Anglian Water Authority
decided that the small number of customers in such
areas should be protected against high lead concen-
trations, and therefore adopted a policy of treating
all high plumbosolvency waters, unless it .--mid be
demonstrated -h.it lead pipes were abseii: in the
distribution s\stem.
In order to classify its waters, a contract was
placed with Humberside College of High Education
(now Humberside Polytechnic) to develop a test to
classify waters us high-plumbsolvency or low-
plumbosolvency and subsequently to test all Anglian
Water treated waters1'". On the basis of this work.
phosphate dosing to control plumbosolvency has
been successfully installed at S5 water-treatment
works in Anglian Water.
Following the survey performed by Humberside
College, further information on the topics listed
below was required in order to ensure cost-effective
control of plumbosolvency:
Effects nf temperature on Icatj solubility:
The minimization «t' phosphate do>e to eMuhlish
and maintain plumliOsoHency control:
The effects of breaks it! phosphate dosing:
The mixing of waters in distribution: and
The effect of ustnu exhumed old lead pipes in the
lest apparatus.
During the experimental programme on the
above topics, it was discovered that low concen-
trations of organic compounds, probably naturally
occurring, can influence plumbosolvency. This work
is also described later in this paper.
• 259
-------
FEB-06-1997
12:38 USEPft
WHINCUP AND HARWARD ON
LEAD CONCENTRATION
5135697172 P.03'
HIGH
PLUMBOSOLVENCY
LOW
PLUMBOSOLVENCY
PHOSPHATE DOSED (1.0mg P/i)
5 10 15
TEMPERATURE (°C)
Fig. 1. Effects of temperature on lead solvency
PLUMBOSOLVENCY TESTING EXPERIMENTS
Full details of the standard test have been pub-
lished' '. The test water is continuously pumped at
30 ml/h through ISO mm sections of new 12 mm lead
pipe at 25°C for up to 25 days. The measured lead
concentrations in the exit water from each pipe
normally begin to be stable after 7-10 days. The
stable lead concentration is quoted as the test result.
Most waters give a 'high piumbosolvency' result
(100 ± 20 ug/l), or a Mow plumbosolvency' result {30
± 10 ug/l).
These differences are thought to be due to the
formation of either the more soluble basic lead
carbonate or the less soluble normal lead carbonate.
as the corrosion product on the lead surface'-'. This
is despite the prediction that normal lead carbonate
should be the thermodynamicaHy stable product
over the pH and alkalinity range involved1--71.
Scanning electron microscopy (SEM) reveals char-
acteristic differences in the crystal deposits formed
in pipes during the test. Thin hexagonal plates are
formed from high piumbosolvency waters, and
260
relatively smooth solid surfaces from low piumbo-
solvency waters. Typical micrographs have been
published earlier"".
It should be noted that distinct crystals cannot be
identified in old pipes. Pipes that have been in
service are generally coated with relatively thick.
stratified and contaminated deposits. However, the
presence of basic lead carbonate in some such
deposits has been confirmed by x-ray diffraction and
infra-red absorption spectroscopy1-'.
PHOSPHATE-DOSED RESULTS AND SEM
Waters dosed with other phosphates (1.0 me P/l)
give low lead levels, frequently less than I0"ug/l.
With continued running, lead values as low as 3 ug/I
can be obtained. The lead phosphate compound
formed is clearly of extremely low solubility, and its
exact composition is not known. It has a very
characteristic appearance by SEM (rounded
granules and surface features, as evident in the
J.IWEM. 1992. 6, June.
published photograp
phorus and some calc
dispersive x-ray spec
EFFECT OF TI
S(
The monitoring rei
solvency Boston dist
lead concentrations
both untreated and j
effect was thought t
water temperature.
accurately quantify i
piumbosolvency wa'
vency water dosed
phate. were examine
and 25°C (in randor
are plotted against
the different lead
waters follow a simi
be expected for simp
(Pb) versus tTemp
Deposits on a :
examined by SEM.
crystals are generall
LEAD
110-
100-
90-
80-
70-
60-
50-
40-
30-
20-
10-
o-
R9-
J.IWEM. ISN2. 6. June.
-------
USEPA
FEB-06-1997 12:38
published photograph(6), and the presence of phos-
phorus and some calcium can be detected by energy-
dispersive x-ray spectroscopy (EDS).
EFFECT OF TEMPERATURE ON LEAD
SOLUBILITY
The monitoring results from the high piumbo-
solvency Boston distribution system showed higher
lead concentrations in summer than in winter in
both untreated and phosphate-dosed waters'6'. This
effect was thought to be due to the differences in
water temperature. In order to confirm and more
accurately quantify the effect, typical high and low
plumbosolvency waters, and the high plumbosol-
veney water dosed to 1.0 mg P/l with orthophos-
phate. were examined at 0°C. 5°C. 10°C. 15°C, 20°C
and 25°C (in random order); the mean lead results
are plotted against temperature in Fig. 1. Despite
the different lead concentrations, the undosed
waters follow a similar relationship, and, as would
be expected for simple solubility effects, plots of log
(Pb) versus. l/Temp (°K) are effectively linear.
Deposits on a selection of pipes were also
examined by SEM. At the lower temperatures the
crystals are generally similar to those described for
LEAD CONCENTRATION
5135697172 P.04
un bo-
been
lot be
en in
thick.
r, the
such
nand
M
ound
id its
very
nded
the
June.
25°C(6). although smaller in size. These examin-
ations therefore support the water lead concen-
tration results in indicating that the: temperature
effects in undosed waters are largely due to crystal
solubility. Since no change in mechanism is
involved, it is clear that the accelerated results
obtained at 25°C in the standard test are also
representative of effects at lower temperatures.
The phosphate-dosed high plumbosolvency water
is different, showing little temperature dependence,
and apparently a small decrease in lead level with
increasing temperature. However, all these lead
values are low, and the results may simply reflect a
more rapid approach to ultimate coverage of phos-
phated deposit at the higher temperatures.
SEM/EDS examinations show the expected round
nodules containing phosphorus and calcium.
INITIAL PHOSPHATE CONCENTRATION
The WRc investigations suggested that for high
alkalinity waters phosphate concentrations of
0.7-1.0 mg P/l(4) would be effective in establishing
plumbosolvency control within a distribution
system. In order to optimize the phosphate concen-
tration required for a particular high plumbo-
0.2 mg P/l
PHOSPHATE DOSED
20
T
0 5 10 15
TIME (DAYS)
Fig. 2. Effects of initial orthophosphate concentration on high
plumbosolvency water at 25°C
-------
FEB-06-1997 12:39 USEPfl
COLLING. CROLL. WH/NCUP AND HARWARD ON
LEAD CONCENTRATION
(pg Pb/i)
100
PHOSPHATE DOSED
(mg P/I)
10 15
TEMPERATURE (=
solvency water, experiments were performed in
concentrations of 0.2-1.6 mg P/I. The results are
shown m fig. 2. They demonstrate that the phos-
phate dose affects not only the finai lead concen-
trauon. but also the rate at which the phosphate is
effective. It IS clear that 0.2 mg/! has little effect. For
the water tested. 0.4 mg/1 has some effect, but at a
slow rate, and this dose is less effective with some
other waters. It is only at doses of 0.6 ma/1 and
above that the major effect of phosphate dosing is
apparent. However, the faster rate of establishment
of control at 1.0 mg P/I and above is sienificam.
particularly when distribution system conditions are
considered.
Theoretically, as noted by staff of the WRC'7)
phosphate dose requirements should be tower at
lower calcium concentrations. This was confirmed
by similar experiments using water containing 50
mg/l of calcium compared with about 150 me/I for
the water used in Fig. 2. The effect, however
appeared to be relatively small, the final lead value
262
The effect of a range of initial phosphate concen-
rranons was also examined at a series of different
temperatures (0-25°C). The results (Fie. 3) indicate
that the 0.2 and 0.4 mg p/| waters beh^ ^^
undo*d waters, wh.lst the concentrations of 0.6me
P/I and above behave as phosphate-dosed waters "
The above results indicate that an initial phos-
Sr?se (!'6 mrP/I or above is ne™'»
«tabhsh control of phimbosolvency in these waters
The higher the phosphate concentration above 0 6
mg P/l. the more rapidly is control established.
BREAKS IN PHOSPHATE DOSING
The effect of breaks in phosphate dosing is import-
ant tor two reasons, (i) to assess the~ impact of
equipment failure at the treatment works, and (ii) to
investigate intermittent dosing as a means of achiev-
J.IWEM. IW2. 6. Juno.
5135697172 P.05
ing cost-effec
dence expert
trations in ui
winter than si
a given lead <
policy of sum
In order u
phosphate do
were establisi
weeks with d.
low plumboi
stopped and i
period. Mean
ing the chans
It is clear tl
differently. 1
there was a
concentrator
undosed lea-
cessation of •
water howevi
and lead vak
occurred witf
a slow deer
values.
The pipes l
lead values w
from the reU
of basic and
2-\A
100 -
90 -
80 -
70 -
60 - [j
50 -
40 -
30 -
20 -
10 -
0 -*=
D>
Fig.-
J.JWEM. 1W2.
-------
FEJ3-06-1997 12:40
USEPfl 5135697172 ' P. 06
PLUMBOSOLVENCY EFFECTS AND CONTROL IN HARD WATERS
ing cost-effective control. The temperature depen-
dence experiments indicated that lead concen-
trations in untreated waters were much lower in
winter than summer, and therefore control to below
a given lead concentration might be achieved by a
policy of summer-only phosphate dosing.
In order to investigate the effect of breaks in
phosphate dosing, appreciable phosphated deposits
were established in pipes run continuously for 33
weeks with doses of 1.0 mg P/l on typical high and
low plumbosolvency waters. Dosing was then
stopped and undosed water run for a further lengthy
period. Mean results from two-week periods follow-
ing the change are presented in Fig. 4.
It is clear that the two types of water behave very
differently. For the high plumbosolvency water
there was a slow but steady increase 'in lead
concentration, eventually reaching the normal
undosed lead level about 30 weeks after the
cessation of dosing. With the low plumbosoivency
water however, there was a relatively rapid effect.
and lead values higher than normal undosed water.
occurred within four weeks. Subsequently there was
a slow decrease to typical low plumbosolvency
values.
The pipes were then examined by SEM. The final
lead values were similar to those normally obtained
from the relevant waters indicative of the presence
of basic and normal lead carbonates respectively.
2-WEEK MEAN LEAD
CONCENTRATION
• fog Pfa/o
However, both the surfaces consisted of small
smooth. round 'phosphate-type" granules, and bv
EDS analysis, phosphorus and calcium were still
present. The well-phosphated surface, which has
very low solubility, had apparently not been des-
troyed on cessation of phosphate dosing but must
have been encapsulated within a carbonate coating.
The results in Fig. 4 illustrate that short breaks in
phosphate dosing of. say. a few days are unlikely to
cause large increases in plumbosoivency. It appears
that the phosphated pipe deposits are reasonably
stable over short periods, resulting in only small
rises in lead concentrations. Over~Ionger periods
(several weeks) the pipe deposits gradually change.
The rate of increase in lead concentrations can vary
between different pipes, and the prospects for
plumbosolvency control by summer-only phosphate
dosing appear unpromising.
LONGER-TERM REDUCTION IN PHOSPHATE
DOSE
Although phosphate concentrations of at least 0.6
mg P/l are necessary to establish plumbosolvency
control it is possible that once established it will be
maintained by low or continuous phosphate doses.
In order to test this hypothesis, pipes which had
been exposed to high plumbosoivency water dosed
HIGH
PLUMBOSOLVENCY
PHOSPHATE
DOSING
CEASED
LOW
PLUMBOSOLVENCY
~1 1 1 1 1 1 1 T
8 10 12 14 16 18 20 22 24 26 28 30 32
UNDOSED TIME (WEEKS)
fig. 4. Effect of cessation of orthophosphate dosing (after 33 weeks at 1.0 mg P/l)
(32) ,
DOSED-
J.1WEM. IW2. 6. June.
-------
FeB-06-1997 12:40
USEPft
5135697172 P.07
-COLO-lNG. CROU-. WH1NCUP AND HARWARD ON
- at 0.8 and 1.0 mg P/l for eight weeks, and which had
established steady lead concentrations in the water
from the test rig, were exposed to reduced phos-
phate concentrations of either 0.2 or 0.4 mg P/l. The
reduced phosphate concentrations were run for a
total period of 44 weeks at various temperatures
finishing at 25°C. During this period lead concen-
trations remained low and stable.
The phosphate concentration in two of the pipes
was then reduced to 0.1 mg P/l. After several weeks"
operation the lead concentrations from these pipes
became more variable and slightly higher than those
operating at 0.2 and 0.4 mg P/l. However, the lead
concentrations remained low and similar to those
from the other pipes.
Subsequently, the phosphate dosing to all the
pipes was ceased. Unlike the pattern of the pipe in
high plumbosolvency water dose'd at 1.0 mg P/I
shown in Fig. 4. the lead concentrations from all the
pipes increased rapidly, reaching typical high plum-
bosolvency values within two weeks.
These results illustrate that plumbosolvency
control can be maintained by phosphate doses of 0.4
mg P/l or less, but that breaks in dosing are then
more important, presumably due to the less robust
nature of the pipe deposits.
MIXING AND ALTERNATION OF WATERS IN
DISTRIBUTION
BLENDED UNDOSED WATERS
The effects of mixing waters, equivalent to blending
them before distribution, 'have been previously
published'61. It was shown that only 5-10% of a high
plumbosolvency water mixed with a low plumbo-
solvency water was sufficient to impart a. high
plumbosolvency to the mixture. Further mixtures
have since been tested, confirming the original
conclusion.
ALTERNATING UNDOSED WATERS
A more usual situation in distribution is that pan
of a distribution zone will be exposed to different
waters for periods determined by demand and
pumping regimes, i.e. alternating exposure to differ-
ent waters. In order to investigate the effect of
alternating high and low plumbosolvencv waters.
various periods of alternation were investigated on
the pipe rig. Several pairs of waters were alternated
daily for a number of weeks in new lead pipes with
some commencing the experiment on high plumbo-
solvency waters and some on low. In alf cases only
low plumbosolvency results were recorded.
After weekly alternation in new lead pipes, low
plumbosolvency results were obtained after one
week on the pipes beginning with low plumbo-
solvency water (as expected), and after two weeks
on the pipes beginning with high plumbosolvency
264
water. Subsequently, stable, low plumbosolvency
results were obtained from both sets of pipes.
In practice, waters may be alternated after long
periods of pipe exposure to one type of water where
pipe deposits are well-established.. In order to
investigate this situation, pipes were exposed to high
or low plumbosolvency waters for a period of eight
weeks in order to establish longer-term pipe
deposits. At the end of this period~ lead concen-
trations were typical of the waters used. The waters
in some of the pipes were then alternated weekly.
Within two weeks of alternation only low plumbo-
solvency results were obtained. In other pipes the
waters were alternated on a 6 day/1 day basis, with
the same result;
In order to investigate even longer-term
changes, pipe deposits from low and high plumbo-
solvency waters were established over 33 weeks and
then the waters interchanged in some of the pipes.
In the pipes originally exposed to high plumbosol-
vency. low plumbosolvency results were obtained
within two weeks. Contrastingly, the pipes originally
exposed to low plumbosolvency water showed only
slight increases in lead concentrations over the 27-
week period of the experiment.
It can therefore be concluded that low plumbo-
soivency pipe deposits are much more stable than
high piumbosolvency pipe deposits, and that in
areas where high and low plumbosolvency waters
alternate only low plumbosolvency lead concen-
trations can be expected.
MIXING AND ALTERNATION OF PHOSPHATE DOSED A.ND
UNDOSED WATERS
When a high plumbosoivency water dosed with
phosphate is blended with an undosed low plumbo-
solvency water before distribution, the phosphate is
diluted and the concentration may drop below that
necessary to establish phosphated pipe deposits1"'.
Due to the dominance of high plumbosolvency in
mixtures, control may not be established in such a
situation and it will be necessary also to dose the low
plumbosolvency water to raise the phosphate con-
centration in the mixture to above 0.6 me P/l in
order to effect control.
When such waters were alternated weekly, stable
lead values were not obtained quickly, only steady-
ing after about 16 weeks. Lead concentrations
remained low and. on average, decreased throuch-
out the experiment. However, each time that the
water was changed to an undosed low plumbosol-
vency water, the lead concentrations became erratic
and early in the experiment showed some values
higher than those normally obtained for low plum-
bosolveney waters. This effect reduced with con-
tinued running. It should therefore be possible to
alternate phosphate-dosed high plumbosolvency
and undosed low plumbosolvencv waters in distribu-
tion and still obtain low plumbosolvency results.
J.IWKM. CM2. 6. June.
Howevei
be careft.
is mainu
The exp*
were pei
that the
different
of their
extraneo
experimt
exhumec
minimun
relevant
ments ii-
concentr
lead pip«
using ne
reach st<
twenty t
phospha
several \
The resi
using un
potable
high plui
phospha
essential
plumbos
time tak
INFI
As part
process.
exchangi
tested fc
Isleham
dosed at"
J.IWEM. I
-------
FEB-06-1997 12:41 USEPfi
PLUMBOSOLVENCY EFFECTS AND CONTROL IN HARD WATERS
^TABLE I. EFFECT OF ION-EXCHANGE NITRATE REMOVAL ON ISLEHAM WATER
5135697172 P.08
Water
Combined raw
Raw - added chloride
IEX treated water (mean mix)
IEX treated water with added nitrate
IEX treated water with added sulphate
IEX treated water with hoth additions
IEX treated water (end of run)
Nitrate
(mg/l NO,)
97
97
2U
97
20
97
35
Sulphate
(me/I)
X7
X7
,X
X
X7
X7
58
Chloride
(me/I)
ftX
177
177
177
177
177
121
Plumbosolvcncy test results
(UR/I Pb)
KW. KG : HIGH
KXJ. 97 : HIGH
:s. 31 : LOW
2X. 29 : LOW
31. 31 : LOW
27. 29 : LOW
3X. 33 : LOW
However, the experiments indicate that this should
be carefully investigated to ensure that good control
is maintained.
OLD LEAD PIPES
The experiments described in the preceding sections
were performed using new lead pipe. It is possible
that the old lead pipes in distribution will behave
differently, due to the more heterogeneous nature
of their old surface deposits which may include
extraneous and organic matter. Several of the
experiments were therefore repeated using carefully
exhumed old lead pipes which were subjected to
minimum disturbance and kept filled with the
relevant water. From the results of these experi-
ments it can be concluded that the final lead
concentrations obtained on various waters using old
lead pipes are not significantly different from those
using new lead pipes. However, the time taken to
reach stable values is generally much longer, up to
twenty times as long. With ol'd pipes the effect of
phosphate-dosing can continue to increase for
several years.
SURFACE WATERS
The results illustrated in this paper were obtained
using underground waters. AH the surface-derived
potable waters in the Anglian Water region have
high plumbosolvency. The final test results from the
phosphate dosing of surface-derived waters are
essentially the same as those derived from hieh
plumbosolvency underground water. However, the
time taken to reach similar lead levels is longer.
INFLUENCE OF ORGANIC MATTER ON
PLUMBOSOLVENCY
As part of the commissioning procedure for a new
process, water from the "Anglian Water ion-
exchange nitrate-removal planf at Isleham was
tested for plumbosolvency. The untreated water at
Isleham has high plumbosolvency and is phosphate-
dosed after nitrate removal. The water from the ion-
J.tWEM. 1
-------
FEB-06-1997 12:43
USEPft
5135S97172 P. 10
PLUMBOSOLVtNCY trr-ECTS AM) CONTROL IN HARD WATtKS
TABLE III. EFFECT OF REMOVAL OF ORGANICS FROM HIGH
Pl.UMBOSOLVENCY WATERS WITH FRESH GRANULAR ACTIVATED C\RBON
Highplumbo$oivcn<:y water
islchum (raw)
Typical groundwaier
Typical surface water
TOC (mg C/l)
Umrcutcd
1 .5
1.2
4.2 '
GAC filtered
'!.?
(1.2
C.3
Test result (ug/1 Pbl
Untreated
I»5
107
IIS
GAC tillered
5X
30
:s
a real inverse relationship. Results at Boston over a
much longer period indicated higher summer lead
concentrations {20 ug/I compared to 10 ug/1 in the
winter).
Short breaks in phosphate dosing do not lead to
significantly increased lead concentrations in waters
dosed at 1.0 mg P/l; however, long breaks lead to a
re-establishment of plumbosolvent pipe deposits.
Waters dosed at.concentrations lower than 1.0 mg
P/l show a more rapid return to plumbosolvent
conditions.
Once phosphated pipe deposits have been estab-
lished, they can be maintained in the pipe rig by
phosphate concentrations as low as 0.1 mg P/l.
However, breaks in phosphate dosing at these lower
maintenance concentrations result in a rapid rise in
lead solvency, returning to the original high plumbo-
solvency results within two weeks. Thus if mainten-
ance dosing is adopted, it becomes important to
ensure that interruptions to dosing are not longer
than a day or two. Longer periods will lead to a
return, to plumbosolvent conditions necessitating at
least a one-year period of dosing to 1.0 mg P/l to re-
establish phosphated deposits before maintenance
dosing can be reintroduced. Phosphate doses as low
as 0.1 to 0.2 mg P/l may not penetrate to the ends of
a distribution system, leading to loss of control at
the system extremities. In practice, therefore, ir is
expected that in the longer term phosphate doses
may be reduced by up to 50%. thereby lowering
costs whilst maintaining effective plumbosolvency
control. Lower doses may be possible in some
systems. However, care must be taken to maintain
. an adequate phosphate concentration at the ex-
tremities of the distribution system and to avoid
breaks in phosphate dosing.
• Where high and low plumbosolvencv waters are
blended before distribution, even a small proportion
of high plumbosolvencv water will give a plumbo-
solvent mixture. It will therefore normally be
necessary to phosphate dose both waters (or the
mixture) in these circumstances. In the more normal
situation of waters mixing in distribution in an
undefined manner and some parts of the system
receiving alternating waters, it has been shown that
when high and low plumbosolvencv waters alternate
the low plumbosolvencv pipe deposits are the more
stable. In such a situation low plumbosolvencv
results ure obtained whichever water is present.
These results explain why some parts of certain
J.JWEM. IW2. 6. June.
distribution systems give low plumbosolvencv
results, even though they more often contain high
plumbosolvency water than low. In practice, how-
ever, there will always be part of the system
exclusively supplied by high plumbosolvency water
and therefore this water will be-phosphate dosed.
Alternation of this water with the low plumbosol-
vency water may give erratic initial lead concen-
trations but should enable control to be maintained
in the long term.
The general introduction of phosphate into plum-
bosolvent zones in Anglian Water has given success-
ful control of lead solvency. The mean reductions
(thirty-minute stagnation lead levels) at fixed points
are entirely consistent with the trials and experimen-
tal data. 45% for 3-6 months dosing, increasing to
• 63% for 9-12 months. The lead concentrations now
achieved easily comply with the requirements of the
EC drinking water Directive1'" and the more
stringent requirements of the Water Supply (Water
Quality) Regulations 1989'"".
Experimental evidence from high plumbosol-
vency waters, where passage through ion exchange
or GAC columns ,gave low plumbosolvency. and
from the addition of peat extract to low plumbosoi-
vency waters, making them highly plumbosolvent.
indicates that low concentrations of humic acids may
be the cause of high plumbosolvency in hard waters.
In practice this rinding is unlikely to alter piumbo-
soivency control measures in Anglian Water in the
foreseeable future as phosphate dosing is much less
expensive, at present, than the removal of organic
compounds.
CONCLUSIONS
1. In order to establish effective piumbosoivency
control, initial doses of phosphate of at least 0.6
mg P/l {as phosphorus) are required.
2. Lower continuous phosphate doses can maintain
Control.
3. Short breaks in phosphate dosing of 1 or 2 days
should not affect plumbosolvency control.
4. The effects of longer breaks in phosphate dosing
depend on the phosphate dose. High plumbo~-
solvency conditions can return within two weeks
at low phosphate doses.
5. Where high and low piumbosoivency water are
blended before distribution both waters, or the
mixture, must be phosphate dosed.
267
-------
FEP-06-1997 12:42
USEPfl
5135697172 P.09
COtCING. CROLL. WHINCUP AND HARWARD ON
water. Increasing quantities of this extract were
added to 25 1 of low plumbosolvency water and the
resulting solutions tested for their plumbosolvency.
It will be seen that addition of sufficient quantities of
the peat extract gave higher plumbosolvency.
During these experiments the TOC of the water
tested ranged from 0.6 mg C/l in the control water to
1.7 mg C/l in the water with 1000 ml of peat-extract
added. Addition of the peat extract to the ion-
exchange treated water from Isleham gave high
plumbosolvency results (Table II).
TABLE II. EFFECT OF ORGANIC ADDITIONS TO Low
PLUMBOSOLVENCY WATERS USING A STANDARD PEAT
EXTRACT
Addition to 25 1 low
plumbosolvency waier
no addition
+• 10 ml peat water
+• 50 ml peat water
* 75 ml peat water
+ 1(X) ml peat water
+ 250 ml peat water
* ItXKI ml peat water
Isleham IEX treated water
•>• HXI ml peat water
Plumbosolvencv test result
f[jg Ph/n
30 \ LOW
26 f Plumbosolvcncy
5!
S2 %
V7 1 HIGH
1 14 [ Plumhosolvencv
tIC }
29 LOW
X2 HIGH
REMOVAL OF ORGANIC COMPOUNDS FROM HIGH
PLUMBOSOLVENCY WATER
The organic content of three typical high plumbo-
solvency waters was reduced by passage at a slow-
rate (1 1/h) through a column (500 mm x 400 mm
dia.) of fresh granular activated carbon (GAC).
TOC analysis confirmed that a substantial propor-
tion of the organics present had been removed. The
results of testing these waters for plumbosolvencv
are shown in Table III. It will be seen that in ail
cases the high plumbosolvency waters had been
reduced to low plumbosolvency. Thus two different
methods of removing organic compounds from
water have been shown to reduce plumbosolvency.
If humic matter contains the factor promising high
plumbosolvency. it would be expected that~GAC
columns would have a short lifetime for plumbo-
solvency reduction, as they have only a short
lifetime for the removal of "humic material. Con-
tinued use of the GAC column confirmed that its
capacity for plumbosolvency reduction was rupidlv
exhausted.
Although the above initial experiments are not
conclusive, and further experiments to investigate
the influence of organic compounds on plumbo-
solvency are being carried out. they give a strong
indication that the main factor causing high pfumbo~-
solvency in hard waters is their content of humic
substances.
266
SURVEY OF ANGLIAN WATER POTABLE WATERS
The TOC content of Anglian Water potable
waters was compared with their plumbosolvency. It
was found that waters with the highest TOC
concentrations (>3 mg C/l) were all pluinbosolvent.
This category includes all the Anglian Water sur-
face-derived water supplies. Waters with very low
TOC («J.6 mg C/l) were mostly of low plumbo-
solvency. However, for the majority of Anglian
Water supplies TOC content is not a reliable ,
indicator of plumbosolvency. This is not surprising if
humic substances or some fraction or component"of
them are responsible for the effect, as the contri-
bution of humic matter to the TOC will vary with
the water source.
DISCUSSION
The plumbosolvency of hard waters is determined
by the type of crystalline deposit formed on the
inside of the lead pipe. These deposits can be
identified using electron microscopy. The solubility
of the deposits is temperature dependent. The
addition of orthophosphate to high plumbosolvencv
waters causes a phosphated pipe deposit of low
solubility to be formed. In order to form an effective
deposit, it has been shown that phosphate concen-
trations of at least 0.6 mg/l are required. The higher
the phosphate concentration (up to 1.6 mg P/lf the
more rapidly the deposits are formed.
Old lead pipes contain thick surface deposits
which have been shown to change more slowly than
the younger deposit in the new lead pipes normally
used in the pipe rig tests; however, the long-term
effects of treatment are the same. Thus" in a
distribution system where the phosphate takes some
time to penetrate to the ends of the system and only
old lead pipes are present, it is expected that initia!
phosphate concentrations of at least 1.0 mg P/l will
need to be maintained in order to establisrTplumbo-
splvency control within a reasonably short period of
time. Indeed in most distribution "systems, where
phosphate may be reduced by adsorption or absorp-
tion on deposits in non-lead pipes, it may not be
possible to achieve the 0.6 mz P/l. the lowest
concentration required to establish plumbosolvency
control, at the ends of the system unless phosphate
concentrations above this value are closed.
_ The minimum period to establish control in the
held will depend upon the situation, but experience
at Boston and Sleaford"" indicates that a year is not
unreasonable. !n subsequent years further small
reductions in lead solvency will be achieved. In test-
rig results. lead concentrations from phosphated
lead pipes appeared to decrease with increasing
temperature. This apparent effect was thought to be
due to the more rapid establishment of highly
phosphated deposits at the higher temperatures in
the limited time-scale of the experiments rather than
J.IWF.M. 1W2. 6. Juno.
a real
much
conce
winte
She
signiti
dosed
re-est
Watei
P/l s|-
condi'
On<
lished
phosp
Howe
maint
leads
solver
ance
ensur-
than
returr
least ;
establ
asO.l'
a dist
the s\
ex pec
may 1
costs
contrt
syster
an nt.
tremii
break
Wh
hlend
of his
solver
neces:
mixtu
situat
undet
receK
when
the lo
stable
result
These
-------
FEg-06-1997 12:44
USEPfl
- PLUMBOSOLVENCY EFFECTS AND CONTROL IN HARD WATERS
6. Where high and low plumbosolvency waters
alternate in distribution, low plumbosolvency
results are obtained.
7. Where phosphate-dosed high plumhosolvency
and low plumbosolvency waters alternate in
distribution, low plumbosolvency results are
obtained after a short period of erratic behav-
iour.
8. Initial results indicate that humic substances (at
low concentrations) may be the cause of high
plumbosolvency in hard waters.
5135697172
t
P. 11
REFERENCES
ACKNOWLEDGEMENTS
The authors wish to thank the Director of Quality of
Anglian Water for permission to publish this paper.
Grateful thanks are also expressed to Mr C. R. Hayes. Mr
D. N. Harris and all the-other Anglian Water and former
Anglian Water staff who provided assistance, to Mr R.
Gregory of the WRc and to Mr J. G. Watson of SCM
Chemicals Limited for scanning electron microscopy.
DISCLAIMER
The views expressed in this paper are those of the authors
and do not necessarily represent the views of Anglian
Water Services Ltd.
(I) DEPAXiMKNr OF tut ENVIRONMENT. Report of the Expert
Advisory Group on identification and Monitoring. Lead in
Poiable Water. Technical Note No. I. 19X0 (official use
only).
(2) SHHIHAM. I.. AND JACKSON. P. ). The scientific basis for
control of lead in drinking water. J. last. Wat. Eng. Sci
mi. 35. (6). 491-515.
(3( COLUNO, J. H.. AND HARMS. D. N. The use of orthophos-
phatc for plumbosolvency control of hard graundwatcrs.
Paper presented to Scicmiric Section of the Institution of
Water Engineers and Scientists. Huntingdon. 19K5.
(4) GREGORY. R.. AND JACKSON. P. J. Reducing Lead in Drinking
Water. WRc Regional Seminars. May-June 19X"» WRc
Report 2JV-S. 19X3.
(5) DEPARTMENT or run ENVIRONMENT. Report of the Expert
Advisory Group on Remedial Water Treatment for Reduc-
ing Lead Concentrations in Tap Water. Lead in Potable
Water. Ttrhaical \ate No. 5. 19K4 (official use onlv).
(ft) COLLING. J. H.. WHINCLT. P. A. £.. AND HAVES. C. R. The
measurement of piumbosolvency propensity to guide the
control of lead in lapwaters. J. Irani Wai. &' Envir Manet
19X7. |. (3). 263-;69.
(7) JACKSON. P. J.. AND SHEIHAM. I. Calculation of Lead
Solubility in Water. WRc Technical Report TR 152. [980.
<*W7»EEC). Official Journal L229. August I9KO.
(10) DEPARTMENT OF THE ENVIIIONMENT. The Water Supply (Water
Quality). Regulations 19S9. Siauaory Instruments tVf». No.
11-17. Her Majesty's Stationery- Office. 1989.
IV
T
T
si
al
«
tc
tf-
N
K
rr
268
J.IWEM. 1992. 6. June.
*i'-c ••'•• • ^'"^^-'---^^^
-------
DISTRIBUTION SYSTEMS
Although selection of corrosion control optimization strategies is more art than
science, utility experiences can provide a basis for rational decision-making.
Donna IVI. Dodrill
and Marc Edwards
";Utility experience'under the Lead and Copper Rule was
'examined to provide improved insight into corrosion control.
, Average 90th percentile lead concentrations were highest in very-
' low-alkalinity waters (<30 mg/L as CaCO3) at utilities that did not
use inhibitors; lead release was significantly reduced at higher
. alkalinities. Average lead releases were 20-90 percent lower for
' utilities using phosphate inhibitors (orthophosphates,
polyphosphates, and blended phosphates) in very-low-alkalinity
["waters than for utilities not using inhibitors. At alkalinities of
• 30-74 mg/L as CaCO3 and at pH values > 7.40, it appeared that
. P^YPljosphate inhibitors had adverse effects on average lead
^release.ptffities with pH < 7.40 and high-alkalinity waters had
{/the highest copper concentrations. Phosphate inhibitors were
' 'usually beneficial in mitigating copper release; however, most
benefits were at utilities with pH < 7.80 and alkalinity > 90 mg/L
il.a?
-------
Average 90th percentlle lead release and percentage
of utilities exceeding the lead action level for utilities
not adding phosphate Inhibitors
<7.40 -.--»;-;?«;F*^
-^jSsfa^fe^i
considered by assigning the utilities to specific pH
and alkalinity categories. pH was divided into four
ranges: (1) pH < 7.40, (2) pH 7.40-7.80, (3) pH
7.81-8.40, and (4) pH > 8.40. Utilities in each pH
category were further subdivided according to alka-
linity: (1) < 30, (2) 30-74, (3) 75-174, and (4) > 174
mg/L as CaCO3. These divisions maintained an
approximately even distribution of the data, ensured
an ample number of points within most pH-alkalin-
ity categories, and allowed for qualitative classifications
of potable water according to pH and alkalinity (i.e.,
very low, low, moderate, and high).
After sorting the data according to these cate-
gories, average 90th percentile lead and copper con-
centrations were calculated using data from all util-
ities in each category. In addition, the
percentage of utilities in each pH and
alkalinity category that exceeded the lead
or copper action level (i.e., 0.015 mg/L Pb
or 1.3 mg/L Cu) was also examined. Para-
metric statistics were used to rigorously
evaluate the significance of observed
trends for selected data at the 95, 90, and
85 percent confidence levels (i.e., the 5,
10, and 15 percent significance levels,
respectively) using a standard comparison
of means test.2 Given the lognormal dis-
tribution of the data, it was first necessary
to normalize the data using the log-trans-
formation of Benjamin and Cornell.3
Details of the statistical analysis are pro-
vided elsewhere.4
It is important to qualify the work by
noting several points.
• Given that this analysis compares
aggregate data of utilities with differing
water qualities, results will not apply
quantitatively to a specific utility. Indeed,
even the qualitative trends must be
viewed with caution when viewed from
the perspective of a single utility.
• Terms such as "increase" and
"decrease" refer to comparative changes
in the pH and alkalinity categories for the
aggregate data. For example, the phrase
"increasing the pH from 7.0-7.4 to pH >
8.4" does not mean that a given utility
or utilities made such changes and
observed the cited effect. Rather, it is a
comparison of averaged corrosion by-
product release for utilities having pH val-
ues of 7.0-7.4 with those utilities having
pH values > 8.4. The cited trends might
apply if a given utility did make such
changes, but such conclusions cannot be
made from this work unambiguously.
• pH and alkalinity values are six-
month averages calculated from water
released to the distribution system. Some
utilities reported pH and alkalinity data
averaged from a number of distribution
sites or point-of-entry (POE) values. In some
instances, pH and alkalinity values were derived from
samples collected over several months or days.
Regardless of how the data were taken, variations in
pH and alkalinity are expected in any distribution
system water, and this analysis cannot account for
these effects.
Lead corrosion
Of the 397 total survey respondents, 365 utili-
ties reported their first-round 90th percentile lead
concentrations. The effects of pH, alkalinity,
inhibitors, calcium, and temperature on 90th per-
centile lead values are based on their responses. Use
of phosphate inhibitors was given special consider-
JULY 1995 75
-------
Percent reduction In lead release as a result of the indicated
increase In alkalinity for utilities not adding phosphate inhibitors
Alkalinity Chango—tng/L 03 CaCO,
<30 to 30-74
30-74 to 75-174
75-174 to >174
••21*
'•51*.
74*
..8 r
Reduction In Lead Release—percent
•'18':'.vr-
'—'--' atittie 95 percent corifidenceTnte'n4l"
ation in this analysis—the term "inhibitors" refers
only to phosphate-based inhibitors (e.g., orthophos-
phate, zinc orthophosphate, hexametaphosphate,
polyphosphates, and various blends) in subsequent
discussion.
Utilities not using phosphate-based inhibitors.
For utilities not adding inhibitors, average 90th per-
centile lead levels were highest in very-low-alkalin-
ity (< 30 rng/L as CaCO3) waters (Figure 1). Raising
pH to < 8.40 did not appear to strongly affect lead
release in very-low-alkalinity waters, although
increasing pH above 8.40 did appear to reduce lead
release.
Lower lead levels (at a given pH) were associated
with higher alkalinities in each pH category with only
one exception (Table 1). Differences in lead levels in
each category were represented by calculating percent
reductions. In this study,
percent reduction was
defined as [(Initial Pb -
Final Pb)/Initial Pb x 100
percent]. Thus, in Table 1,
the initial Pb was the lower
of the two alkalinity cate-
gories being compared in a
given alkalinity change. As
shown in Figure 1, increased alkalinity of 30-74 mg/L
as CaCO3 (as compared with alkalinity < 30 mg/L as
CaCO3) yielded statistically significant lower lead
release a{ the 95 percent confidence level when pH
was < 8.4 (Table 1).
For utilities that did not add inhibitors, the per-
centage exceeding the lead action level seemed to be
correlated with differences in both alkalinity and pH
(Figure 1). In the lowest pH and alkalinity category,
utilities had an 80 percent likelihood of exceeding
the lead action level. This likelihood was less in higher-
alkalinity or higher-pH categories almost without
exception. As observed for lead release, a large reduc-
tion in the percentage of utilties exceeding the lead
action level was realized at the transition from < 30
to 30-74 mg/L as CaCO3.
Utilities adding phosphate-based inhibitors.
The use of phosphate inhibitors may reduce lead
concentrations in drinking waters.5-9 The initial analy-
sis of the data did not distinguish between the vari-
ous types of phosphate
inhibitors or doses used. That
is, all utilities that added
orthophosphate, blended
orthophosphate, or polyphos-
phate were classified as
"adding inhibitors." In the
1992 survey, 116 utilities
reported using phosphate
inhibitors; 261 utilities
reported they did not use
phosphate inhibitors. With-
out considering the effects of
either pH or alkalinity, differ-
ences between average 90th
percentile lead were insignificant between utilities
regardless of whether they added inhibitors.
Any comparison of utilities that add inhibitors
with those that do not add inhibitors may be subject
to systematic biases that are impossible to quantify. At
one extreme, most utilities might be adding phos-
phate-based corrosion inhibitors to control problems
with iron corrosion. Consequently, the data might
be overweighted in pH-alkalinity categories that prop-
agate red water problems. Nonetheless, there is no
reason to suspect that this bias would invalidate use
of the database for the examination of lead and cop-
per corrosion problems.
At another extreme, utilities that add inhibitors
might be doing so to mitigate serious problems with
lead or copper corrosion. In. this case, the utilities that
add inhibitors might represent only the most prob-
ncreased lead release appeared to occur
at higher alkalinity when inhibitors
were used.
lematic waters with respect to lead and copper corro-
sion. Given the relative lack of concern regarding lead
and copper corrosion by-product release at the tap
before enactment of the Lead and Copper Rule, how-
ever, there is no strong reason to suspect this bias
exists in the data set used from the first round of sam-
pling. Also, the database cannot quantify the effect
of time on the development of passivating films or
decreasing lead leaching (by physical depletion).
Keeping these potential problems in mind, it is
still illustrative to make comparisons between utilities
that currently add inhibitors and those that do not.
Particular attention was paid to identifying pH-alka-
linity regimes in which inhibitors appeared to have
beneficial, detrimental, or no effect on lead and cop-
per corrosion by-product release. The comparison is
used to highlight water qualities for subsequent lab-
oratory confirmation studies that unambiguously test
these identified trends in inhibitor performance.
Trends in the data were also examined for consis-
76 JOURNAL AWWA
-------
tency with theoretical predictions that
guide inhibitor use.5-7-10
For utilities that added inhibitors
between pH 7.40 and 8.40, lead release
was adversely affected by increasing alka-
linity from < 30 to 30-74 mg/L as CaCO3
(Figure 2; Table 2, negative numbers).
These trends were statistically significant
at the 85 percent confidence interval. In
contrast, at pH < 7.40 increasing alka-
linity from < 30 to 30-74 mg/L as CaCO3
had the opposite effect, leading to a 53
percent reduction in lead release. This
observation was not supported at the 85
percent confidence level, however, given
the high standard deviation of the data in
this category.
Trends in the percentage of utilities
exceeding the lead action level were com-
plicated by inhibitor addition. For
instance, the highest percentage of util-
ities (47 percent) exceeded the lead
action level at 30-74 mg/L as CaCO3 and
at pH 7.40-7.80 (Figure 2). Lead con-
centrations did not exceed the action
level in systems using inhibitors at very
low alkalinities (<30 mg/L as CaCO3)
above pH 7.80, and the use of inhibitors
generally reduced lead leaching through-
out the low-alkalinity range.
Inhibitor effectiveness. An exami-
nation of the difference between systems
with and without inhibitors is instruc-
tive, as shown by the following equation:
Reduction in lead release associated
with inhibitors = Average lead release
without inhibitors - Average lead
release with inhibitors
In this equation a positive number
means an improvement (reduced lead :•
release), whereas a negative number ;
means ah adverse effect (increased lead
release) associated with inhibitors. This
general formula was used to evaluate the effect of
phosphate-based inhibitors on average 90th perccntile
lead release as well as on the percentage of utilities
exceeding the lead action level.
Beneficial effects of inhibitors were observed only
in the lowest alkalinity category (<30 mg/L as CaCO3)
at all pH values (Figure 3). Conversely, in several
instances in the 30-74-mg/L as CaCO3 alkalinity cat-
egory and at pH > 7.40, total lead release and the
percentage of utilities exceeding the lead action level
were apparently higher. Differences between systems
that add inhibitors and those that do not add inhibitors
were inconsequential at alkalinities > 74 mg/L as
CaCO3.
In the lowest alkalinity category (<30 mg/L as
CaCO3) average lead release was 20-90 percent lower
Average 90th percentite tead release and percentage
of utilities exceeding the lead action level for utilities
adding phosphate inhibitors
r.46-^8b;%M^p^i?«*a:;
&&&^&£@g%&S3Sg%*
for utilities using inhibitors compared with those not
using inhibitors (Table 3). These results were all sig-
nificant at the 95 percent confidence interval, indi-
cating that addition of inhibitors may be very effective
in reducing lead release in very-low-alkalinity waters
regardless of pH (Table 3). Inhibitors did not produce
a statistically significant reduction in lead release (at
the 95 percent confidence level) in any other pH-alka-
linity category tested. In fact, in the alkalinity range of
30-74 mg/L as CaCO3 and at pH > 7.40, inhibitors
had higher average lead release (negative numbers
in Table 3). These increases were all significant at
either the 95 or 85 percent confidence interval. Thus,
the indication is that inhibitors exacerbate 90th per-
centile lead release when used in waters with pH >
7.40 and alkalinity 30-74 mg/L as CaCO3.
JULY 1995 77
-------
Percent reduction in lead release as a result of the indicated
Increase In alkalinity for utilities that add phosphate inhibitors
Alkalinity Change—mg/L tat CaCO3 ' /,-'
75-174 to >174
"~ic*.-.rv
Percent reduction in lead release for utilities adding phosphate
inhibitors compared with those not adding phosphate inhibitors
>174
Reduction In Lead Release—percent
Explanations for increasing lead levels at moder-
ate alkalinities for utilities adding phosphate-based
inhibitors include the possibility of inadequate phos-
phate dosing to form protective films. That is,
orthophosphate is known to inhibit the growth of
basic carbonate and oxide films on copper pipe.11 It
is not known whether it may also inhibit the growth
of cerrusite (PbCO3) or hydrocerrusite [Pb3(CO3)2OH].
For example, utilities may not have the optimal
inhibitor dosage to form a protective layer of hydroxy-
pyromorphite [Pbs(PO4)3OH] or lead phosphate
[Pb3(PO4)2]. Phosphate may also be interfering with
other film formation.
Detailed examination of inhibitor use. Of the
116 utilities using phosphate inhibitors, there was an
approximately even split
between utilities using
orthophosphate and those
Phosphate-based inhibitors
had significant effects only in
the two lowest alkalinity cat-
egories (<30 and 30-74 mg/L
as CaCO3). In the lowest pH
region (<7.40), average 90th
percentile lead release for util-
ities with alkalinities <30
mg/L as CaCO3 and using
polyphosphates was more
than double that for utilities
applying orthophosphate
alone (Figure 4). For poly-
phosphates, average lead re-
lease was even higher than
was observed for utilities not
adding inhibitors. Thus, in this
pH-alkalinity category at least,
it seems reasonable to at-
tribute beneficial effects of
inhibitors to orthophosphates
and either no or adverse ef-
fects to polyphosphates.
Interestingly, in the same
alkalinity category but at pH
7.4-7.8, the opposite effect
was observed (Figure 4); util-
ities using polyphosphates had
an average corrosion by-prod-
uct release about three to four
times lower than utilities
using orthophosphates. However, the average lead
release was lower at utilities using either type of
inhibitor than at utilities not using inhibitors. For
utilities with alkalinities of 30-74 mg/L as CaCO3 at
pH 7.4-7.8, there did not appear to be a significant dif-
ference between the two types of inhibitors in per-
formance (but inhibitors did have a statistically sig-
nificant adverse effect in this pH-alkalinity category).
In the two highest pH categories (7.81-8.40 and
>8.40) there was either insufficient data or a highly
uneven division between inhibitor types, so an
inhibitor-to-inhibitor comparison is not possible. Nev-
ertheless, this overweighting is still noteworthy. For
instance, at pH >8.40 and alkalinity 30-74 mg/L as '.
phates or did not specify the
type of inhibitor added
(Table 4). For the inhibitor
effects cited as statistically significant for lead in the pre-
ceding section, the types of inhibitors used in each
identified pH and alkalinity category were then care-
fully scrutinized. That is, the information on inhibitor
types was used to clarify whether beneficial or adverse
inhibitor effects could be attributed to orthophosphate
alone, polyphosphates alone, or both in combination.
release is reduced by higher pH
presence and absence of inhibitors.
CaCO3 when inhibitors increased average lead release
by 91 percent (Table 3), all 13 utilities used polyphos-
phate inhibitors. Thus, it seems reasonable to attribute
the significant adverse effects observed for inhibitors
in this category to polyphosphate exclusively. The
finding that utilities are using polyphosphate chem-
icals above pH 8 apparently contradicts the notion
78 JOURNAL AWWA
-------
that phosphate inhibitors are used only
near neutral pH.12
In the lowest alkalinity range (<30
mg/L as CaCOj) and pH values > 7.81,
where significant beneficial effects of
inhibitor were observed, all of the utili-
ties used polyphosphate inhibitors. Lead
release was significantly lower for utilities
in these two pH categories; however, only
three utilities used phosphate chemicals
(two in the pH 7.81-8.40 region and one
in the pH > 8.40 region). In the alkalin-
ity category 30-74 mg/L as CaCO3 at pH
7.81-8.40 in which lead release was
increased by inhibitors, five utilities used
orthophosphate and only one utility used
polyphosphate.
Potentially adverse effects of poly-
phosphate addition on lead release have
been noted by others. Holm and Schock,
using a discrete two-ligand model, pre-
dicted the effects of polyphosphate addi-
tion at pH 8.0 and 40 mg/L Ca as a func-
tion of alkalinity.13 The authors modeled
solutions in equilibrium with either
hydrocerrusite or hydroxypyromorphite.
In both situations, waters treated with a
proprietary liquid formulation of
polyphosphate (consisting of 40 percent
orthophosphate) dosed at 1 mg P/L were
predicted to have significantly higher
plumbosolvency than untreated waters
(or waters dosed with 0.4 mg/L
orthophosphate in the case of hydrox-
ypyromorphite precipitation). Similarly,
Sheiham and Jackson demonstrated that
lead release was higher in new and old
lead pipes (at pH 6.5 and 7.5, respec-
tively) for two polyphosphate chemicals
than for orthophosphate.8 All three
inhibitors were dosed at 1 mg/L in a low-
alkalinity (< 10 mg/L as CaCOj) water.
Although the preceding discussion f^.,':V ,
considerably clarifies matters with respect £i • g
to the practical effects of inhibitors, { >
inherent shortcomings of using the re- :
suits to guide phosphate inhibitor use ?::• - :; -•• = '•
remain. Nevertheless, in the absence of j~- : •:"-::.';
systematic laboratory studies to confirm
trends, the preceding discussion pro- '. •
vides a good starting point for bench or
pipe-loop investigations at individual
utilities. Future work and experimentation should
examine the inhibitor question in greater detail.
Effects of temperature. Average 90th percentile
lead as a function of alkalinity and cold (<15°C) or
warm (> 15°C) water was calculated in the presence
and absence of inhibitors (Figure 5). Temperature
was taken by utilities at several distribution system
water quality sampling locations (data were collected
between January and June 1992). In general, lead
Improvement In percentage of utilities exceeding
the lead action level as a result of adding phosphate
inhibitors
I Comparison of average 90th percentile lead release
' for utilities adding orthophosphate and polyphosphate
; and utilities not adding inhibitors
t§£ Orthophosphate HI Polyphosphate .:..P| No phosphate^: jS'i_
0.050
y&,-.-f& PH . 'V-r?"£v'.'*:Vv-:??!'.-:"*?•-':
aCp3;mimber- -•--— ^—•*--»---••. •-
many utilities were using that Inhibitor
levels decreased with increasing alkalinity for both
cold and warm distribution waters regardless of
inhibitor use. In contrast to the conventional wis-
dom, warm waters did not appear to induce higher
lead release than did cold waters at a given alkalin-
ity.14-16 If anything, the opposite effect was observed
in the lowest alkalinity category. In retrospect this
trend probably should have been expected, given
that at-the-tap monitoring precludes wide variations
JULY 1995 79
-------
Utilities using specified type
of phosphate inhibitor
->"S--cv ft"?*-:; -,:
•LargeTs defined as utilities sening 50.000 peocterj^SSSS- •£>•..
in temperature (either seasonal or geographical) after
an overnight stagnation period.
Effects of calcium. Similar to the effect of tem-
perature on lead release, the effect of calcium on
90th percentile lead was considered as a function of
alkalinity. As with the pH and alkalinity data, cal-
cium concentrations were determined either at point-
of-entry locations or in the distribution systems. Two
divisions for calcium content, low (<50 mg/L as Ca)
and high (>50 mg/L as Ca) were made (data not
shown). Overall, there was not a strong trend between
lead release and calcium concentration in any sys-
tem studied.
Unity < 74 mg/L as CaCO3, raising pH from < 7.40 to
7.40-7.80 resulted in a 43-68 percent reduction in
average 90th percentile copper release—changes that
are significant at the 95 percent confidence level.
Raising pH from 7.81-8.40 to >8.40 always led to
significant reductions in copper release without regard
to alkalinity, whereas increasing pH from 7.40-7.80
to 7.81-8.40 significantly decreased copper release
only if alkalinity was >75 mg/L as CaCO3.
The described trends are magnified when viewed
in the context of the percentage of utilities exceed-
ing the copper action level in each pH and alkalinity
category (Figure 6). No utilities with pH > 7.80
exceeded the copper action level (Figure 7). The
highest percentage of utilities exceeding the action
level was at pH < 7.40 and alkalinity <30 mg/L as
CaCO3. However, all utilities exceeding the cop-
per action level in the very-low-alkalinity cate-
gory also had pH < 7.0. Excluding those utilities
with pH < 7.0, all other utilities exceeding the cop-
per action level had alkalinity > 75 mg/L as CaCO3
and pH < 7.80. A closer examination of that data
(not shown) demonstrated that all utilities exceed-
ing the action level had alkalinity > 90 mg/L as
CaCO3 and many had pH > 7.0. Copper corrosion
by-product release is likely worse at higher alka-
linity as a result of the formation of soluble cupric-
bicarbonate complexes.18-19
Copper corrosion
In-the 1992 AWWA sur-
vey, 361 utilities reported their
90th percentile copper levels
from the first sampling round.
The same strategies previously
described for lead release were
used for copper.
Utilities not using phos-
phate-based inhibitors. For
utilities not adding inhibitors,
average 90th percentile cop-
per levels were highest in
waters with pH < 7.40 (Fig-
ure 6). A cpmbination of low
pH (<7.80) and high alkalin-
ity (>74 mg/L as CaCO3) pro-
duced the worst-case 90th
percentile copper levels. This
agrees with recent research
hypotheses that Cu(OH)2
solids control copper solubility
under some circumstances in
distribution systems.10-17-19
In the absence of inhib-
itors, average 90th percentile
copper release clearly tended
to decrease at higher pH (Fig-
ure 6). Percent reductions
ranged from I to 100 percent
with only one exception
(Table 5). In waters with alka-
.--T'- ':•' ~
?iS 3 Alkalinity '.':•££
':\.\'mg/L as CaCO3
^.T.- •;-~"!C'-.^--r-.r •--r'--^- :i
•.. i-:.-,--i<30 • •-;•'., •
.;.; '',:-/3O-74 .>•.'-•••-
•" '''•
. •
•Significant at the 95 percent confidence interval
tSignificant at the 85 percent confidence Interval
, {Significant at the 90 percent confidence interval
Percent reduction in copper release as a result of the indicated
increase in pH for utilities not adding phosphate inhibitors
. '., Reduction .In CppporReleaso^p«e^_,>;^;^^;^;
43* • : "'•'•- "30 ' •.}^^--'^SJ^Sf'i^^i-rV
39 :" si* "..*"'^ff^:'^..'.
51*
53f
Percent reduction in copper release for utilities adding phosphate
inhibitors compared with those not adding phosphate inhibitors
• ...
<7.4 : 7.4O-7.80
pH Cliango
s-8.40
Alkalinity
mg/L as CaCOt
Reduction In Copper Release — percent
m®,
sC?fr--'«;-
^S#r:
Sw-ia>5'
s?-??-XW
•S?H?"S.';
56*:' .-.-.-•: 13
. 11 ..:,;•:..•: -,-2
51. ;.,;:^V^.,.34t
V23 - •" =••' ''^^V" 4 =
V:.SNo data or IrisufBctent data for the category "."'"' ~'
f^'^::-'-~i^tJ^'i.'^^^^^J^"'^-^'-- -;. --:
80 JOURNAL AWWA
-------
Average 90th percentile lead in cold (<15°C) or warm
(al5°C) waters for utilities not adding inhibitors
Inhibitor effectiveness. As was observed for
lead corrosion, when the effects of pH and alkalinity
were ignored, overall differences in copper release
attributable to inhibitor addition were not signifi-
cant. Thus, performance of inhibitors is clearly depen-
dent (at a minimum) on the pH and alkalinity of the
specific system.
Although few systematic studies have evaluated
the effects of phosphate inhibitors on copper release,
most work completed to date demonstrates improve-
ments to copper corrosion at higher pH values (7.0-8.0)
and no effects at lower pH (< 7.0). For instance, Reiber
found that corrosion rates were reduced by a factor
of 3-5 for utilities having waters at pH 7.5 dosed with
1-5 mg/L orthophosphate compared with utilities hav-
ing waters at the same pH that did not add inhibitors.20
In another study, weight loss of copper pipes in the
presence of 1 or 5 mg/L orthophosphate was dramat-
ically reduced by orthophosphate addition at pH 8.O.21
However, at pH < 6.0, orthophosphate inhibitors did not
affect corrosion rates in the Reiber study. Similarly,
when dosed with 3.2-4.5 mg/L of zinc orthophos-
phate) Boston water at pH 6.7 exhibited no change in
copper release when compared with the same water
without inhibitors.15
As was the case for utili- T—; ——
ties that did not add in- s :•=;';.
hibitors, copper release de- '"••• •*$
creased with increasing pH !.'; '•;
for utilities that added in- =»
hibitors (data not shown).
Percent reductions as a result
of increasing pH category
ranged from 0 to 80 percent
with two exceptions. Interestingly, the apparent bene-
fits of inhibitor addition are mostly confined to the
lower pH ranges for the copper release data (Figure 7),
although the percentage of utilities exceeding the cop-
per action level was reduced by inhibitors in all pH and
alkalinity categories (Figure 7).
At pH < 7.80 utilities using inhibitors generally
had lower average 90th percentile copper release
when compared with utilities that did
not add inhibitors, with percentage
improvements as high as 56 percent
(Table 6). However, only two of these
trends were significant at greater than
90 percent confidence. Above pH 7.80,
addition of inhibitors had disparate
effects on copper release, depending on
the specific alkalinity category, with per-
centage increases attributable to
inhibitors (negative numbers in table)
as high as 45 percent. The result at pH >
8.40 and alkalinity < 30 mg/L as CaCO3
was significant at the 95 percent confi-
dence interval, suggesting that inhibitors
can have adverse effects for copper cor-
rosion, at least under, some circum-
stances. No other adverse effects were
significant at or above 85 percent confi-
dence. On the basis of this analysis, it seems safe to
state that the effects of inhibitor use above pH 7.80 are
highly variable for copper corrosion.
Effects of temperature. Average 90th percentile
copper as a function of pH and cold or warm water
was quantified in both the presence and absence of
inhibitors. In general, copper levels decrease with
increasing pH for both cold and warm waters whether
or not the utilities added inhibitors (data not shown).
There does not appear to be a significant trend in
terms of temperature (i.e., copper concentration is
dependent on pH and other water quality parameters
and not on whether the water is warm or cold). In
addition, inhibitors are most effective in reducing
average 90th percentile copper at pH below 7.80 and
did not seem affected by temperature.
Effects of calcium. Similar to the trend with
temperature, calcium content did not have much of
an effect on average 90th percentile copper release.
Comparison to solubility modeling theory
It is instructive to compare the trends described
earlier to current frameworks of corrosion control
based on solubility modeling.
utilities with pH <7.40 and high-alkalinity
waters had the highest copper
concentrations.
Lead. The chemical equilibrium modeling pro-
gram MINEQL+ (version 3.01)22 and associated
database were used to make equilibrium predic-
tions of lead solubility using various models. The
basis for the prediction was to model the approxi-
mate midpoint of the pH-alkalinity categories used
previously in segregating the practical utility data.
That is, pH 7.0, 7.6, 8.1, and 8.8 were modeled at
JULY 1995 81
-------
Average 90th percentile copper release for utilities not
adding phosphate inhibitors
--
alkalinities of 15, 50, 75, and 175 mg/L as CaCO3.
The different modeling approaches are based on
distinct assumptions as to solids that control lead sol-
ubility—in this work solids considered included
cerrusite (PbCO3), hydrocerrusite [Pb3(CO3)2(OH)2],
and both in combination.
The best agreement between the utility data and
solubility models was observed when both cerrusite
and hydrocerrusite formation were considered. In
this model lead levels are predicted to decrease with
higher alkalinity at pH < 7.6 (Figure 8). This predic-
tion is consistent with the significant trends in the
utility data (Figure 1 and Table 1). The solubility mod-
els also predict nearly insignificant reductions in lead
solubility at alkalinity > 50 mg/L as CaCO3 at pH < 7.6,
results that are again consistent with significant trends
in the practical data (Table 1).
82 JOURNAL AWWA
At pH £: 8.1 there was general lack
of agreement between the solubility mod-
els and the practical data. That is, the
model invariably predicted higher lead
solubility at higher alkalinity, whereas
the practical data very clearly demon-
strate the opposite trend (Figure 8 versus
Figure 1 and Table 1). Some of these
trends in the practical data (at the low-
.est alkalinity category) are significant at
the 95 percent confidence level.
In addition to the solids PbCO3 and
Pb3(CO3)2(OH)2, hydroxypyromorphite
[Pb5(PO4)3 OH] was included in the sol-
ubility model for lead in the presence of
phosphate-based inhibitors. A dose of
0.95 mg PO42'-/L orthophosphate was
used in the model. In general, lead re-
lease showed some qualitative agree-
ment between solubility modeling and
the practical data. For instance, in the
lowest alkalinity category, formation of
Pb5(PO4)3OH is predicted to greatly re-
duce lead solubility regardless of pH, a
trend observed in the practical data as
discussed previously (Figure 2 and Table
3). Predicted lead release approximately
doubled when alkalinity increased from
15 to 50 mg/L as CaCO3 in the pres-
ence of orthophosphate at all pH values
tested (Figure 8). Once again, this trend
is qualitatively consistent with some of
the practical observations defined in
Figure 2.
Despite the noted agreements
between the model and practical data,
in general the solubility models tended to
overpredict beneficial effects of inhibitors
at higher alkalinities. That is, in the pres-
ence of orthophosphate inhibitors, lead
levels are predicted to be reduced in
nearly all pH-alkalinity categories (Figure
8). This was not observed in the utility
data. Much of this disagreement is
because utilities were probably dosing polyphosphate
inhibitors or dosing orthophosphate at concentra-
tions too low to observe significant effects. Never-
theless, some observable beneficial effects would be
expected at higher alkalinity if the model was perfectly
accurate. There are also many reasons why predictions
based on solubility models are not expected to agree
with the observed trends In the practical first-draw
lead data, as discussed by Schock.5 This work does
not provide any additional basis for speculation as to
which of those effects is dominant in the practical
data collected for this work.
Copper. There is good correlation between
Cu(OH)2 solubility models10-17-19 and the previously
described utility database (utilities without inhibitors).
As pH increases, copper solubility decreases for both
theory and practical data. In addition, high copper
-------
release is observed at high alkalinities
when pH is below about 7.8-8.
Comparisons between theory and prac-
tice for copper release with inhibitor addi-
tion are problematic because it has been
difficult to determine which copper-phos-
phate solids are controlling copper solu-
bility. Recently, assuming the formation
,of Cu3(PO4)2 -2H2O solid, Schock18 pre-
dicted the effects of orthophosphate on
Cu(OH)2 solubility. Below pH 8.0,
orthophosphate addition reduces copper
release according to these predictions,
whereas above pH 8.0 copper release does
not decrease when orthophosphate is
added. Although solubility modeling can-
not currently predict adverse effects attrib-
utable to inhibitor addition (because there
are no complexation constants for
polyphosphates and because interference
with the formation of protective oxide,
hydroxide, or hydroxycarbonate scale
cannot be predicted), the general trend of
the solubility model provides insight into
the disparate effects of phosphate
inhibitors at pH above 7.8.
Recommendations to utilities
Although other water quality parameters (e.g.,
dissolved oxygen, temperature, calcium, sulfate, and
chloride) can influence lead and copper corrosion,
this treatment indicates these factors are of secondary
importance when compared with pH and alkalinity.
That is, if any other water quality factor was "of pri-
mary importance in lead or copper corrosion con-
Improvement in percentage of utilities exceeding the
copper action level as a result of adding phosphate
inhibitors
hate inhibitors were usually
beneficial in mitigating copper release;
however, most benefits were at utilities
with pH < 7.80.
trol, no significant trends would have been observed
when this data set was sorted on the basis of pH and
alkalinity. Trends in by-product release, when viewed
in the context of pH and alkalinity categories, were
highly significant in some cases.
On the basis of the graphical and statistical findings,
a variety of simple measures have been identified that
utilities may follow to mitigate high lead or copper
release. Given the variation in corrosion phenomena
from system to system, however, these predictions may
not be universally applicable. Further research is needed
to better define the types of inhibitors and chemical
dosages that are effective in specific water quality types,
and more information is needed regarding desirable
target pH values or alkalinities. Laboratory experiments
performed thus far are encouraging in that they gen-
erally support the trends identified in this work regard-
ing lead and copper release as a function of pH, alka-
linity, and phosphate inhibitor addition.23
Lead. Most problems with lead release were in
low-alkalinity (< 30 mg/L as CaCO3) waters. In such
waters, if pH is < 8.40, increasing the alkalinity to
30-74 mg/L as CaCO3 is likely to significantly reduce
lead release. Likewise, adding inhibitors to low-alka-
linity waters may also be
effective in reducing lead
release regardless of pH.
Utilities with very-low-alka-
linity waters (<30 mg/L as
CaCO3) that already add
inhibitors are not advised to
increase alkalinity because
of potential aggravation of
lead by-product release.
Whether considering
alkalinity-pH adjustment or
addition of orthophosphate inhibitors, maintaining
pH is absolutely critical for corrosion control in poorly
buffered waters. For example, adding acidic
orthophosphate formulations (most commercial zinc
orthophosphate liquids are pH 1 or 2) can inadver-
tently lower the pH below the optimal level. Utili-
ties must ascertain and maintain the proper pH.
For utilities in all other pH-alkalinity categories not
adding inhibitors, either increasing alkalinity, pH, or
both is predicted to reduce lead release and the likeli-
hood of exceeding the lead action level. Inhibitors were
generally not useful in reducing lead release in waters
with alkalinity > 30 mg/L as CaCO3, although, once
again, it would be useful to test such conditions if
JULY 1995 83
-------
Model-predicted lead solubilities in tha absence
of orthophosphate (A) and in the presence of
orthophosphate (B)
Model assumes no orthophosphate (A) and 0.95 mg/L'
.orthophosphate as PO4-3(B) ' .. .•
inhibitor use was desired at a given utility. At pH >
7.40 and alkalinity of 30-74 mg/L as CaCO3, adding
inhibitors can apparently increase lead release. Many of
these adverse effects of inhibitors appeared attribut-
able to polyphosphates, so stopping polyphosphate dos-
ing if it is used for sequestering might be a viable opti-
mization strategy to be tested further at affected utilities.
In high-alkalinity waters near or exceeding the
action level, pH adjustment often is not feasible
because of possible calcite precipitation. Orthophos-
phate or blended phosphate chemicals are probably
the only viable alternatives in these
waters, and they may have to be added
at high dosages. Overall, the use of phos-
phate inhibitors alone did not signifi-
cantly correlate with low lead release.
The qualitative effect observed depends,
at a minimum, on the specific pH and
alkalinity of the water of interest.
Copper. The following strategies are
suggested as effective for control of cop-
per corrosion. At alkalinities < 90 mg/L
as CaCO3, no utilities exceeded the cop-
per action level unless pH was < 7.0.
Thus, increasing pH to > 7.0 would seem
to be an effective and simple mitigation
strategy for utilities in the low-pH and
very-low-alkalinity regime. At alkalini-
ties > 90 mg/L as CaCO3, two strategies
seem viable. Because utilities with pH >
7.80 in this category did not exceed the
copper action level, increasing pH seems
to be promising.
For utilities with high-alkalinity and
high-calcium waters that cannot in-
crease pH above 7.80 because of con-
cerns regarding calcite precipitation,,
any increase in pH would likely
improve matters substantially. Many
small systems using groundwater sup-
plies are likely to have such waters.
Some trends in the data suggest that
inhibitors were successful in mitigat-
ing copper release in such systems;
however, these trends were variable
and not of high significance. Above pH
7.80 there is some evidence indicating
that inhibitors can adversely affect cop-
per corrosion by-product release. In
general, inhibitors caused highly vari-
able effects above pH 7.80.
Conclusions
The following conclusions were
reached regarding lead.
'•!=.:"-': ' -^ • For utilities not dosing inhibitors
•i":" ":>v?.i,;-.; at pH < 8.4> lead release was significantly
:".:._:L lower if alkalinity was 30-74 mg/L com-
pared with alkalinity < 30 mg/L CaCO3.
• For utilities dosing inhibitors at
alkalinity < 30 mg/L as CaCO3, inhibitors
appeared to improve lead corrosion by-product release
compared with utilities not dosing inhibitors. At pH
< 7.4 these benefits were attributable to orthophos-
phate and not polyphosphates. No significant
improvement in lead release attributable to inhibitors
was observed at other pH-alkalinity categories,
although this might be the result of improper inhibitor
dosing with respect to corrosion control.
• At alkalinities 30-74 rng/L as CaCO3 and at pH
> 7.40, inhibitors seemed to adversely affect lead
release. Many of these effects were directly attrib-
84 JOURNAL AWWA
-------
uted to polyphosphates and not orthophosphates.
Increased lead release seemed to be significantly cor-
related with higher alkalinity (<30 to 30-74 mg/L as
CaCO3) when inhibitors were used.
The following conclusions were reached regarding
copper.
• Copper release is reduced by higher pH in the
presence and absence of inhibitors.
• Problems with meeting the copper action level
are confined to two water quality regimes: (1) pH <
7.0 at alkalinity < 30 mg/L as CaCO3 and (2) pH < 7.8
at alkalinity > 90 mg/L as CaCO3.
• The optimal inhibitor dosages are probably
dependent on pH. Inhibitors appear to reduce aver-
age 90th percentile copper release at pH < 7.80. Above
pH 7.80, inhibitors had highly variable and some
adverse effects on copper corrosion by-product release.
Acknowledgment
This work was supported by a grant from the
AWWA Research Foundation (AWWARF). The
authors appreciate the constructive criticism provided
by their project advisory committee members: Joel
Catlin, Leonard Graham, Preston Luitweiler, and
Tiffany Iran. The insightful comments provided by
Michael Schock and Steve Reiber were of immea-
surable value to this work. Special thanks to all util-
ity personnel and project participants in the 1992
AWWA Lead and Copper Survey—without their effort
this work would not have been possible.
References
1. Final Report: Initial Monitoring Experiences of
Large Water Utilities Under USEPA's Lead and
Copper Rule. Prepared by Montgomery Watson,
Economic & Engineering Services, Inc., and Peter
Karalekas, Consulting Engineer. AWWA, Den-
ver, Colo. (1993).
2. GIBRA, I.N. Probability and Statistical Inference for Sci-
entists and Engineers. Prentice Hall, Englewood
Cliffs, N.J. (1973).
3. BENJAMIN, J.R. &• CORNELL, C.A. Probability, Sta-
tistics, and Decision for Civil Engineers. McGraw Hill,
New York (1970).
4. DODRILL, D.M. Lead and Copper Corrosion Con-
trol Based on Utility Experience. Master's thesis,
Univ. of Colorado, Boulder (1995).
5. SCHOCK, M.R. Understanding Corrosion Control
Strategies for Lead. Jour. AWWA, 81:7:88 (1989).
6. SCHOCK, M.R. & WAGNER, I. The Corrosion and
Solubility of Lead in Drinking Water. Internal
Corrosion of Water Distribution Systems,
AWWARF/DVGW-Forschungsstelle Cooperative
Res. Rept. (1985).
7. Lead and Copper Rule Guidance Manual. Cor-
rosion Control Treatment, vol.2. AWWA, Denver,
Colo. (1992).
8. SHEIHAM, I. & JACKSON, P.J. The Scientific Basis
for Control of Lead in Drinking Water by Water
Treatment. Jour. Inst. Water Engrs. & Scientists,
35:6:491 (1981).
9. LEE, R.G.; BECKER, W.C.; & COLLINS, D.W. Lead
at the Tap: Sources and Control. Jour. AWWA,
81:7:52 (Aug. 1989).
10. SCHOCK, M.R. & LYTLE, D.R. The Importance of
Stringent Control of DIG and pH in Laboratory
Corrosion Studies: Theory and Practice. Proc.
1994 WQTC, San Francisco, Calif. Nov. 6-10.
11. HOLM, T.R. ET AL. Polyphopshate Water Treat-
ment Products: Their Effects on the Chemistry
and Solubility of Lead in Potable Water Systems.
Proc. 1989 WQTC, Philadelphia, Pa., Nov. 12-16.
12. BOFFARDI, B.P. Polyphosphate Debate. Jour.
AWWA, 83:12:10 (Dec. 1991).
13. HOLM, T.R & SCHOCK, M.R. Potential Effects of
Polyphosphate Products on Lead Solubility in
Plumbing Systems. Jour. AWWA, 83:7:76 (July
1991).
14. COLLING, J.H. ET AL. Plumbosolvency Effects and
Control in Hard Waters. Jour. Inst. Water &Envir.
Mngmnt., 6:3:259 (1992).
15. KARALEKAS, P.C. JR.; RYAN, C.R.; & TAYLOR, F.B.
Control of Lead, Copper, and Iron Pipe Corrosion
in Boston. Jour. AWWA, 75:2:92 (Feb. 1983).
16. GREGORY, R. & JACKSON, P.J. Central Water Treat-
ment to Reduce Lead Solubility. Proc. 1984
AWWA Annual Conf., Dallas, Texas.
17. SCHOCK, M.R.; LYTLE, D.A.; & CLEMENT, J.A. Mod-
eling Issues of Copper Solubility in Drinking
Water. Proc. 1994 Nad. Conf. on Environ. Engrg.,
Boulder, Colo., July 11-13, 1994.
18. SCHOCK, M.R.; LYTLE, D.A.; & CLEMENT, J.A. Effect
of pH, DIG, Orthophosphate, and Sulfate on
Drinking Water Cuprosolvency. (Unpubl.)
19. MEYER, T.E. & EDWARDS, M. Effect of Alkalinity on
Copper Corrosion. Proc. 1994 Natl. Conf. on
Environ. Engrg., Boulder, Colo., July 11-13,
1994.
20. REIBER, S.H. Copper Plumbing Surfaces: An Elec-
trochemical Study. Jour. AWWA, 81:7:114 (July
1989).
21. BENJAMIN, M.M. ET AL. Corrosion of Copper Pipes
in Chemistry of Corrosion Inhibitors in Potable
Water. AWWA, Denver, Colo. (1990).
22. MINEQL+, Version 3.01: A Chemical Equilib-
rium Modeling System for Personal Computers.
Environmental Research Software (1994).
23. HIDMI, L.; GLADWELL, D.; & EDWARDS, M. Effect
of Phosphates on Lead and Copper Corrosion.
Poster session, 1994 WQTC, San Francisco, Calif.,
Nov. 6-10.
About the authors: Donna M.
Dodrill is an engineer with Black &
Veatch, 8400 Ward Pkwy., Kansas
City, MO 64114. At the time this work
was conducted, Dodrill was a research
assistant at the University of Colorado
at Boulder. Marc Edwards is an assis-
tant professor in the Department of .
Civil Engineering, Box 428, University of Colorado at Boul-
der, Boulder, CO 80309-0428.
JULY 1995 85
-------
REGULATORY GRIDLOCK
The conventional Langelier index or
Larson's ratio approach to controlling
copper corrosion by-product release is
demonstrably inaccurate.
Marc Edwards,
Michael R. Schock,
and Travis E. Meyer ' Ithough bicarbonate was previously
"• ' - ^^^^b^r^i^B^is^^tiaj to pj-eyen^g copper c6rr8sibn
work has conclusively demonstrated
bicarbonate under certain condi-
.. ....-, _-;__- —ities exceeding the US Environ-
•--=;. mental Protection Agency (USEPA) 'action limits" for
Corrosion by-products must consider pH and bicar-
bonate (alkalinity) adjust-
.ments to optimize corro-
sion,3 understanding the
role of bicarbonate in cop-
per corrosion is of critical
importance. That is, if the
basic effects of pH and bi-
carbonate are not under-
stood, utilities that follow
JuuCl^X aoDlYiarfWa-ri ooforr-oxT-'-^-f-^'^^i-r^W^;-^ Wic-i^ii^s,;.-—-—£:-;T-i -\:L-: * -:- • • "-.' -'• *^ -* "•<
aggravate copper corrosion
problems.
This work was de-
ter pH signed to provide practical
insight into the effects of
bicarbonate on copper
*
55®?^
.-jixqsic
--—tyaS-contrdl
•^^mts&zg&tt-
^m
SVSKSS
SS»
!^si
oi^m
ftS^Snfc
:COt
attractive sti
*. 1 —™-r*>-«-»'»-^".di.»_gj ut\-&uac UCJULCULS jLjrujtn mgne
. are realized without adverse effects from higher alkalinity.
MARCH 1996 81
-------
°°PPer corros'on by-product release from pipe rigs
^fl^^ro^^S^c^Vi^^^l^'^bS'':'-''-'/:.!-"
82 JOURNAL AV^/WA
-------
excessive concentrate
corrosion by-prodiii
corrosion and
unambiguous!
mechanism by
acts. An in-depth analysis of mon-
itoring data from large utilities,
basic solubility experiments, and
pipe-rig testing allowed rigorous
evaluation of accepted corrosion
mitigation strategies. A special
focus of this investigation was the
range of pH 7.0-8.5, over which a
critical transition in bicarbonate
effects was observed in previous
research.2
Materials and methods
Tap water test solutions. Because.combined
lirne-CO2 processes are the least expensive meth-
ods of pH and alkalinity adjustment for many utili-
ties, experiments were conducted on tap water .from
Boulder, Colo., modified with lime and CO2. Dur-
ing these experiments, unaltered Boulder tap water
had an average pH of 7.2 and alkalinity of 15 nWL
as CaCOj (Table 1). CO2 was always bubbled to the
solution before the addition of lime to prevent pre-
cipitation of CaCOj.
ihoug'h bicarbonate was previously
nsidered essential to preventing
copper corrosion problems, recent work
has conclusively demonstrated adverse
effects from bicarbonate under certain
conditions.
stituents in addition to pH and alkalinity. Thus, these
tests were conducted with "synthetic waters" consti-
tuted in the laboratory. Copper solubility was exam-
ined in 1-L solutions at a predetermined alkalinity
and pH in the presence of a 0.0 1 -M NaClO4 swamp-'
mg electrolyte. A 4 x 4 experimental matrix was eval-
uated, including alkalinities of 10, 50, 100, and 250
mg/L as CaCO3 at levels of pH including 7.0, 7.5, 8.0
and 8.5. Experiments examining bicarbonate and
chloride interactions also utilized synthetic solutions
In these experiments alkalinity and pH were adjusted
to the target value using an
appropriate combination of
C02 and NaHC03. There-
after, if desired, 1 mM of
NaCl was added. Finally,
solution ionic strength was
adjusted to 0.01 M by addi-
tion of NaClO4 (an inert
electrolyte).
Soluble and total cop-
per. Soluble copper was
measured at 20°C after
membrane filtration (0.2-
um-pore-size cellulose ace-
acid solutions const mtom ldt
bubbled with CO, to pH 4 O^re dosed 3?hnw
controller. This
T - Approximately 25
°r S3mple Were filtered for ^-analysis, and no
essary) without ah kmv buhhhn rn 1 acidification to pH 2.0. for inductively coupled
Synthetic test solutions. A few experiments
required precise control over other inorganic con-
TOPCON Co.. Paramus. N.J.
fHach Co., Lovcland. Colo.
MARCH 1996 83
-------
_ . • ••3:->&.tw&msS!££*s&%-. ^•X'
Typical base water quality durir-" ~;--" - "" *" at(=
period
short-
by-
. *
Electrochemical erosion rate measurement
g Procedure. Specifics of stan-
tests and the apparatus have
Y^ la^ the key parameter
eiiaf3l testing, is the total cor-
by the exposed macroscopic
corroding metal; i.e., the instan-
ate of the sample. 7^ was deter-
^S^an -electroanalysis system.f This article
Sfe-results from duplicate analysis
ses/determinations were quantitatively
: uA/on2).6
tests, the general goal was
on-copper pipe representative of
of conventional exposure.
• T^asp^^^Ji^er years o conventional exposure.
~~" '"' ~'"~ " "•' "CoPPe..r was exp'osed to the solution of interest for
using the colorimetric test. For all waters in which the TP*-P ***- f^n a c°frosion-a^elerating potential
colorimetric test was used-possible maS irSriS £~< co? ~ ™Y W3S appIiecL At the end of
samples.
Scanning Electron Microscope (SEM) analy-
sis. Samples were prepared for SEM analysis by cut-
ting the pipes lengthwise. The pipe surfaces were not
coated because the conductivity of the copper itself
was sufficient, thereby guar-
anteeing that surfaces
would be unaffected by a
coating procedure. The sam-
ples were examined in a
vacuum at 30 kV on an
electron microscope.*
Pipe-rig tests. Copper
corrosion by-product release
was examined in standard-
copper surfaces, as reported in the previous s°ec-
tion. The flow rate (0.5 gpm [0.031 L/s] ± 0.02 °pm
[0-0013 L/s]), pH (± 0.03 pH units), and alkalinity
(± 3 mg/L CaCO3) were rigidly controlled during
experiments.
concentrations are a surrogate
copper carbonate complexation
capacity in many waters.
ized "pipe rigs." Type M copper pipe measuring 24
m (609 mm) x Win: (19 mm) was purchased from
a local hardware store and was washed in 0.1 M
NaOH for 2 min to remove organic deposits" There-
after, pipes were rinsed five times with deionized
water before immediate exposure to target solutions
Pipes were filled with the target solutions three times
a week (Monday, Wednesday, Friday) and maintained
in a horizontal position at all other times A rubber
stopper tightly sealed each end.
Filtered and unfiltered samples were taken weekly
to determine soluble and total copper concentrations
after 72 h of exposure (Friday to Monday stagna-
tion). After three months of exposure, representa-
tive coupons were cut from each pipe for electro-
chemical and SEM examination. Prior to the
electrochemical analysis, pipe coupons were exposed
to the target solutions at a flow rate of 0.2 gpm (0.013
L/s) for,30 min. Each water quality was tested using
replicate pipe rigs. The only difference between repli-
cates was that exposure to test solutions was initi-
Monitoring experience of large utilities. Data
collected during a national survey of 435 large utili-
ties were used in this investigation.7 The analysis of
that database proceeded as follows.
• Because inhibitor additions might hopelessly
complicate the role of pH and alkalinity in copper
corrosion, any utilities using corrosion control other
than pH, alkalinity adjustment, or both were deleted
from the database.
• Any utility not reporting essential information
including pH, alkalinity, or 90th percentile copper
release was also deleted.
The 151 remaining utilities were sorted into pH
and alkalinity categories for analysis. pH and alka-
linity values in the database, typically six-month aver-
ages for water released to the distribution syst<
possible variations within the i
L or between different sources.
84 JOURNAL AWWA
-------
Bicarbonate, alkalinity,
and dissolved inorganic
carbon. Over the pH range
of most natural waters (pH
5.3-8.7) and at alkalinities
>10 mg/L as CaCO3, bicar-
bonate concentrations are
directly proportional to alka-
linity within ± 3 percent error
(i.e., 50 mg/L alkalinity as
CaCO3 = 1 mAf HCO3-). Ac-
cordingly, this work will
apply the terms alkalinity and
bicarbonate interchangeably
according to utility conven-
tion. Approximate units of
dissolved inorganic carbon
(DIG) are provided for reader
convenience using die equa-
tion DIG (mg C/L) == 0.24
[alkalinity (mg/L as CaCO3)].
At pH values between 7.5
and 8.7 this approximation is
valid to within ± 5 percent,
but the approximation un-
derpredicts actual DIC by 17
percent at pH 7.0.
Experimental results
The experimental results
are divided into three sec-
tions, with the first two
addressing the effect of bi-
carbonate on copper corro-
sion by-product release and
solubility. Thereafter, the
combined effects,.of bicar-
bonate and chloride are
examined in detail.
Effect of bicarbonate
on by-product release and
solubility. Bicarbonate and
by-product release from pipe
rigs. Copper corrosion by-
product release was. more
dependent on water .quality
than on time of exposure
(Figure 1). That is, although
the exposure of replicate
(•s^ssssr-""-"1-----*.---'-
endar date rather than versus time of exposure Th s
was particularly evident on July 5 1 993 when each
pipe exhibited a dramatic reduction in coppIrSeas?
If differences in exposure time caused th? observed
decrease only the pipe exposed for the extra 12-day
time penod would have exhibited this drop IntS
estmgly, the decrease in by-product release corre
sponded to a period of highsLrmwater runoS *
Copper release was fairly constant at each condition
over the last four weeks of'exposure^reTS
'
W3S Very g°°d' avera§e c°PPer corr°-
* "*** funai°n (Figure
sion C°™ati°n)- The «»«.
data SDH 7 ?nri? ^ ^ ^ f°r the averaged
nerrenr 2 rh resPecuvely. Because more than 95
C°P?f r m the samples Passed
/' ^ "^ W3
3nd not Particulate-
*****
MARCH 1996 85
-------
Measured soluble copper after solubility
solubility
experiments and predicted soluble copper based on Cu(OH)
('•JL*r
-------
conducted. In these experi-
ments soluble cupric ion was
added to solutions at fixed
pH, ionic strength, and alka-
linity. After allowing 20 min
for precipitation to occur, sol-
uble copper was then mea-
sured in filtered samples.
Consistent with previous
trends, soluble copper was
approximately a linear func-
tion of alkalinity (Figure 4).
The experimentally observed
slope at pH 7.0 and 7.5 was
about 0.0116 mg soluble cop-
per/mg alkalinity, whereas
the slope at pH 8.0 and 8.5
was about 0.004 mg soluble . .
copper/mg alkalinity. Thus, at pH 7.0-7.5 -soluble
copper was about 2.9 times more sensitive to the pres-
ence of alkalinity than at pH 8.0-8.5. This was fairly
consistent with the 2.0-2.7-fold slope differential in
the by-product release data (Figure 2) and utility mon-
itoring data (Figure 3).
Given the previous experimental findings, various
Model constants from MINEQL+ database12'1*
complexes at higher pH, because the Cu(OH)2
model predicts a weak -dependency on alkalinity
whereas the solubility data (Figure 4), by-product
release experiments (Figure 2), and utility data
(Figure 3) indicate a much stronger dependence
on alkalinity.
By comparison, predictions of soluble copper
pH and alkalinity ranges,
whereas that based on
bronchantite overpre-
dicted copper solubility by
a factor of three to four.
Moreover, malachite sol-
ubility predicts less solu-
ble copper at higher alka-
linities if pH was >7.4, a
tendency that was obvi-
ously inconsistent with
the pH range of afaout 7.0-8.0,
adverse effects from bicarbonate are
significantly reduced by even small
increases in pH.
evaluation, though an internally consistent critique
has been presented elsewhere.10 Four cupric solids
known to form on copper pipe were examined for
consistency with experimental data including tenorite
[CuO], cupric hydroxide [Cu(OH)2], malachite
[Cu2(OH)2C03], and bfonchantite [Cu4OH6(SO4)]."
The model prediction'was based on the assumption
that soluble copper was controlled by equilibrium with
one of these solid phases (Table 2).
Predictions of soluble copper based on cupric
hydroxide solubility were both qualitatively and
quantitatively consistent with the previous data
(Figure 4). Quantitative predictions always agreed
to within 1.5 mg/L copper, a very reasonable result
given uncertainties in the required equilibrium
constants. The agreement between the Cu(OH)2
model and observed solubility was excellent at pH
7.0, with weaker predictive capability at the higher
pH values. The discrepancy between the model
prediction and the experimental data appears to
be due to an underestimate in copper carbonate
tance of cupric hydroxide has recentfy been put
forward by Schock et al1" using data from USEPA
pipe rigs.
Effect of alkalinity on corrosion rates and
scale morphology. Corrosion rates of copper in
pipe-rig tests. At the end of the by-product release
experiments, coupons were cut from the pipes, and
instantaneous corrosion rates were determined elec-
trochemically under flow conditions. Although the
trends in corrosion rate are not a linear function of
alkalinity, the copper corrosion rates did increase
consistently with higher alkalinity at both pH 7.0
and 8.0 (Figure 5).
The corrosion rate, /„„., represents total copper
oxidized at the pipe surface per unit of time. This
oxidized copper may either be incorporated into solids
on the pipe surface, forming a scale, or be released to
solution as a corrosion by-product. Assuming that
corrosion proceeds through a one-e- transfer, the fol-
lowing conversion was derived relating corrosion rate
to copper mass corroded:
MARCH 1996 87
-------
Average copper release and measured corrosion current in naturally
aged samples showing the same trend
Corrosion currents in the presence and absence of 1 mM chloride
-«3§$t5;=£-• ^yprSSl.-1 r.~-Z-'^^.:3.-"!*£-'S->-'=p?xy:v? •--*• -w-w ••,•*. » " •
[^•SlSrr^::"?'^^^^"^'^^^"*"-^"'*^-^"^
IpA 1A .
cm2 . 106 jzA X coupon
20 cm2 1 C/s
x
le-
_ n
~ °'018
(3)
mmolCii
day
Two major factors might contribute to differ-
ences m corrosion-rate estimates based on copper
mass lost to solution (i.e., the by-product release
data) versus the direct electrochemical measure-
ment. The first factor was that by-product release
measurements were made under stagnant con^-
tions, whereas corrosion rates were determined
under flow conditions. This difference would tend
to increase the corrosion rate' during flow versus
stagnation because of improved mass transport The
second factor was that much of the oxidized copper
could be incorporated into a growing scale layer
and not released to solution. Any copper incorpo-
rated into scale will not be quantified by measure-
ment of released corrosion by-products, thereby
reducing the estimate of cor-
rosion rate based on by-prod-
uct measurement.
Despite these difficulties,
it was instructive to examine
the actual measured corrosion
rate and compare it with that
.estimated on the basis of by-
product release to stagnant
| waters. Assuming that corro-
: sion rate was not influenced
by differences in flow, calcu-
lations indicate that about 3
percent of copper that was
corroded was actually released
to solution; in this case, about
97 percent of the corroded
copper must have been incor-
porated into scale formed on
the pipe surface. At another
extreme, if it was assumed
that all the corroded copper
was actually released to solu-
tion as a corrosion by-prod-
uct and none was incorpo-
rated into scale, the corrosion
rates under conditions of
stagnation must have been
more than 33 times slower
during stagnation than under
conditions of. flow. Reality
probably reflects a compro-
mise between these two ex-
treme assumptions.
Dual ion interactions:
chloride and bicarbonate.
Although the previous exper-
iments clearly isolated the
effects of pH and bicarbonate
on copper corrosion rates, other water quality con-
stituents such as natural organic matter, chloride
and sulfate can also strongly influence copper cor-
rosion.s.ii Because both chloride and bicarbonate
appear to be critical to controlling copper corrosion
rates, their combined effect was examined.
The accelerated aging technique was used to
examine copper corrosion in well-defined synthetic
waters at pH 7.0 and 7.3 and constant ionic strength
At a given level of alkalinity, solutions were consti-
tuted with or without the addition of 1 mM chloride
at all alkalinity values (Figure 6). Chloride addition
decreased corrosion rates between 84 and 95 per-
cent in aH cases. Thus, from this perspective.at least
the presence of chloride can counter adverse effects
of bicarbonate.
A comparison of scales formed in the presence
and absence of chloride verified the dominant effect
. of chloride on scale appearance. Chloride supported
formation of a reddish scale that was nearly identical
to that observed at pH 7.0 in the presence of chloride
only. 6 in contrast, scales formed in the absence of
** JOURNAL
-------
chloride were quite smooth in appearance under the
SEM, giving the surface-a nonporous appearance.
Thus, chloride induced very significant visual and
electrochemical changes to copper corrosion.
Synthesis: role of bicarbonate
in copper corrosion
This section critically evaluates key results of this
investigation in light of previous research. The effects
of bicarbonate on corrosion by-products and corrosion
rates are discussed separately. The section concludes
with a short discussion of copper carbonate com-
plexation versus dissolved CO2 as a source of copper
corrosion problems.
Effect of bicarbonate on copper corrosion by-
product release. Evidence from the pipe-rig exper-
iments and the monitoring experience of large utili-
ties is clear and unambiguous: soluble copper
corrosion by-product release increases as a linear
function of alkalinity. Because these trends were con-
using conventional units with copper in mg/L W.as- -~
pH, and alkalinity (= HCO3-) in mg/L as CaCO,^" *"*T
constant pH of 7.0 and 8.0, this equation simplifies to *L -
a linear form: ' . • •
Soluble Cu (mg/L) = 0.83 + 0.015 [alk] at pH 7.0
(7)-
ut considering inhibitor dosing
Soluble Cu (mg/L)Wo.58;*p;boi3 {aft] atpHK6$?(8)'
Qualitatively, this predicted linear relationship
between soluble copper and alkalinity (at constant
pH) is in excellent agreement with the experimental
results. .The quantitative agreement is ab'd quif<|pod,:
especially at lower pH values. Indeed, a comparison
of Eqs 1 and 7 demonstrates remarkable agreement
between the model at pH 7.0'and-theactaal by-prod--
uct release data, consistent with the excellent pre-
dictive capability of the model at the lower pH for
the solubility data (Figure 4). A comparison of Eqs 2
and 8 reiterates the fact
• that'the model underpre-
dicts sensitivity of copper
:emperature, the best advice
is to raise pH.
The slope of the equa-
tion, equal to the increase
in.soluble copper (mg/L)
per incremental increase
in alkalinity (mg/L as
Cu(OH)2 (solid) were quantitatively and qualitatively
consistent with the available data.
In this model, the predominant soluble copper
species over the pH range 7.0-8.5 include
cu(OH)2(aq), CuC03,aq), and CuHCO3+. Thus, the
concentration of soluble Cu can be calculated as-
(4)
(5)
-]2 K -
in which Kj = [Cu(OE)2
], jr = [cu+2]
w = [OH-] pff+], and Cu is in mol/L
- Because Klf K^ K3, 1^, and ^ are constant, cal-
culating results at 25°C without activity corrections
yields:
Soluble Cu (mg/L) = io(i3.4-2PHj + 0.58 + io(5.i-pH)[alkj
+ 10 (U.4-2pH)
.
pH by 0.5 units (i.e., from 8.5. to 8.0 or from 8;0..to .
7.5) increases the slope, by. about, a factor of three, •
indicating a threefold increase in copper carbonate
complexes at the lower pH.: •.'•; ....' •-;.;.- ..•:!-.\-:; :-..-;.-•.-,
If enthalpy values for the Cu(OH)2 solid in'the ••--'•
MINEQL+ database are considered (Figure.7), cop-,
per-carbonate cpmplexatiori is predicted to be a strong
function of temperature, Tyith each 10°C increase
halving the slope. In other-words, because Cu(OH)2- - -
dissolution is exothermic,,a given concentration of"
bicarbonate (alkalinity) is predicted to complex about
twice as much copper at 5?C as it would aM5°C;-
rnterestingly, this predicted temperature-dependency
can explain recent findings that copper corrosion by--
product release is lower in household hot-water taps
than it is in cold-water taps.? This temperature depen^
dency also.'calls into question recent proposals:to-
require utilities to monitor: only in the summer
because the presumption of higher copper release at- .
higher temperature is at odds, with predictions based
on solubility and is not supported by utility moni-
toring data.17 . . . :
The Cu (OH)2 model is also consistent with empir-
ical linear relationships developed at KIWA to predict
the maximum concentration of copper released to
water during stagnation:18-19
MARCH 1996 89
-------
rfubiltty to alkalinity and temperature
Old pipe:
New pipe:
(mg/L) = 0.01 10 [Alkalinity]
- 1 -37 pH + 0.02 1 [S04-2]
• ' + 10.2
onstrated that bicarbonate has
a dual nature dependent on
the solution pH. Above about
pH 8.1, the presence of bicar-
bonate tends to passivate cop-
per surfaces and decrease cor-
rosion rates.6 However, below
about pH 8.1, bicarbonate is
increasingly aggressive at
higher alkalinities (>100
mg/L as CaCO3). A closer
look at the literature reveals
an interesting trend; i.e., the
solution pH is greater than 7.7
in nearly all case's in'which
bicarbonate is reported to
have beneficial effects.20-22
In cases in which the cur-
rent findings seem to be con-
tradictory to the conclusions of previous research a
closer look reveals just the opposite. For instance
bicarbonate was studied at pH 6 and pH 8 by Matts-
son and Fredriksson.23 Though the authors conclude
that bicarbonate forms passivating scales, the data
showed that at pH 6 anodic corrosion currents did
increase with bicarbonate concentration and a very
large increase was observed at the highest alkalinity
tested (Figure 8). At pH 8, however, the currents
and adverse effects of bicarbonate were *reatlv
reduced.23 J .
Cohen and Myers24 examined treatment alterna-
tives for an outbreak of copper pitting in Fort
Shawnee, Ohio. Prior to treatment, the water had an
alkalinity of 296 mg/L as CaCO3 and an average pH
concentrations of particulate- .-v - - .
copper not removed by a
tion and actual experience.
Nevertheless, it is encourag-
• ing that trends in the experi-
mental data and the Cu(OH)2
predictive model are consis-
tent with those observed in
aggregated monitoring data
of large utilities.
Effect of bicarbonate on
copper corrosion rates and
scale morphology. At first
glance, findings that bicar-
bonate has an adverse effect
on corrosion rates appear con-
trary to results of previous
researchers. The authors' pre-
vious work, however,6 dem-
(mg/L) = 0.0104 [Alkalinity]
-2.26 pH + 18.1
in which alkalinity is in mg/L as CaCO3 and SO4-2 is
in mg/L (units have been changed for consistency
with US utility convention). The empirically deter-
mined slope of 0.0104-0.0110 mg Cu/mg alkalinity
is within the range of those determined in this work
corresponding approximately to the slope predicted
at pH 7.3 at 15°C by the Cu(OH)2 model (Figure 7)
Predictions based on the Cu(OH)2 model, even
ill l(~\T-f\ oy»l»/4*-> *<) A ««._*._ _1 __ *_ 1 -•*• . .
90 JOURNAL AWWA
-------
disappeared. Results from the current work indicate canacitv That fc thV £m£i>£«X " * ^SKS^S^y-C^f 'S
^SiSSSHS" pi&s^^?P^^iilf
f°f ^d.^°£er?^^fat.ed 18 ^te1? I11 New_, -whichjna^beiewritteii as'^i -" ;
dand and: althoue-h thevakn mr,,*,,*,* th** x;~,.' :•.-. . . .. ....^ * •. „. ," -.n~:;.'S-: :.-.: .:••
most aggressive waters had 411-"and 178-mg/L bicar-
bonate concenttatipns'krpH 7.3rarid 7.1, respec-
tively.2^ Both of these waters are within the aggres-
sive region for bicarbonate' denned in recent work
done by the auttb'fs of tffis'article. In addition, the
beneficial effects attributed to bicarbonate in Moss •
and Potter's work was partly based on anodic polar-
ization experiments/and the.cathodic oxygen reduc-
tion reaction was ignored; Thus, these investigators
would not have observed an increase in the cathodic
oxygen reduction reaction rate as was observed in'
previous work.2'5
v^^^yj^i^^^-i^gl^^^^r^Xp^lii
CCCu (mg/L) = [C0j (moW.)] .x {3j09 x 10? + IO^H [5.79^ upfi "
usini
K&i.i
•' Ce,-:. 'A?/-
e described benefits of aeration are
ixpected to apply to many high-alkalinity,
low-pH waters.
copper-carbonate-'cbmplexes become increasingly
important at higher alkalinity;'tending to reduce the
free copper (Cu-*-2) actiyity'in solution. At a fixed pH
and dissolved oxygen' concentration, the driving force
for the anodic corrosion reaction increases as free
copper activity decreases:
Cu 4- &02 + H2O =* 2OH- + Cu+2
Thus, the presence of coinplexihg-carbonate
speaes could reduce iihe free copper concentration
thereby increasing the corrosion rate. Alternatively,
PTi S)f^trf*rcf* ii-»fIii£»*-»^^ nf V;«._i ^.__ *•* • _
the maximum concentra-
tion of carbonate complexes
that would form or, in other
words, the' tendency of a
given solution to form car-
bonate complexes. Future
research:, should revisit
hypothesized direct roles for
CO2 in copper corrosion,
because, correlations be-
tween copper corrosion problems and CO2 may be
an artifact of copper-carbonate complex formation.
ImplicationsfoVutHities^O;';!: ','.^', "
Important implications of this research fall mto
two broad categories: corrosion control strategies to
avoid and recommended strategies. Given the regu-
latory framework of the Lead and Copper Rule, these
.strategies apply .spedficaUy t6:mitigation of copper
(i.e., not lead) corrosion by-product release.
Corrosion control strategies to avoid. Utili-
^^^copperco^^^n^teses
rates would be forniitrti:iP:^° - friendly corrosion optimization software sold
^^^ J-V/A.I.UJ.LWU..S. • , triFono'Ti ATATXA/A 27 *T*J-»* 1 J
soIvTcarb^dToxSe6°(CO^SZ^T^n ^ ^?*tbeaefa"feMS^outbSmSa^unS
been correlated to ma^y SS'o?^^^ SSTl C°nditir th^ Provide the ^vice most
P^^^u^gp^g^^^°^0^0^ ^lyo exacerbate copper corrosion by-product
^example.the^gelierindex^approachis
MARCH 3996 91
-------
At a given pH, calcite saturation indexes tend to increase at higher
alkalinity (data shown for pH 7.2-7.4)
calcite supersaturation are correlated with reduced
corrosivity:28
Calcite saturation index (SI) = [Ca
in which [Ca+2] = calcium concentration, [CO3-2] =
carbonate concentration, and K = calcite solubility
product.
To connect the SI expression with the traditional
LI approaches, a positive II corresponds to SI values
above 1.0, whereas a negative II corresponds to SI
values below 1.0. The SI and II both increase as Ca+2
and CO3~2 concentrations increase. Thus, in the sim-
plest test of this model's validity, at constant pH cop-
per corrosion problems should decrease as Ca"1"2 and
CO3~2 increase.
The work described here directly repudiates this
hypothesis. In experiments in which both Ca+2 and
.CO3-2 were increased at pH 7.0 or 8.0 (Figure 2), lead-
ing to a higher SI and presumably a better condition for
copper according to Langelier theory, problems with '
copper corrosion by-product release worsened. Simi-
larly, as evidenced by the costly experience of large
utilities under the Lead and Copper Rule (Figure 3), in
a given pH category copper corrosion by-product release
increases at higher alkalinity. Because higher-alkalin-
ity waters have much higher calcite SI values at a given
pH (Figure 9), the II predicts reduced problems with
copper by-product release at higher alkalinity. In sum,
both findings are completely inconsistent with the Lan-
gelier approach but are consistent with the Cu(OH)2
model developed by Schock et al and in this work.10-29
With respect to Larson's ratio (LR) :30
LR = [Cl-]/[HCO3-]
the higher the LR ratio, the higher the corrosivity of
the water supply. With respect to copper corrosion,
this work and others have demonstrated that chloride
has long-term beneficial effects (on copper corrosion
rates), whereas bicarbonate has adverse effects on
both by-product release and corrosion rates.6 Thus, if
anything, the exact opposite
of Larson's prediction is likely
to be valid for copper.
The preceding discussion
should not be interpreted as a
criticism of the index origi-
nators. They, more than those
who followed, understood
the limitations.of their work
For instance, the Larson's
index was not derived for
copper, it was derived for steel
and cast iron. Likewise, with
respect to the. Langelier the-
ory, it has long been known
that calcite does not actually
precipitate on pipes in most
distribution systems; thus, the
theory might'actually work
quite well for the few cases in which precipitative
calcite scales do form.
The question arises, however, as to why the Lan-
gelier model has gained such widespread and appar-
ently misguided acceptance. The answer, in the
authors' opinion, is that it provides an easy-to-use
guide to solving a complex problem. Moreover,
because the II is typically applied to answer the ques-
tion "How high should we raise the pH?," it will work
quite well under some circumstances. Because higher
pH typically reduces copper corrosion by-product
release, the index occasionally works by accident.
Thus, when used in the context of calculating the
maximum pH allowable in a given system to avoid
problems with calcite precipitation, the LI still has
considerable value.
Recommended corrosion control strategies.
Without considering inhibitor dosing .(viable but
beyond the scope of this work) or temperature (which
is uncontrollable), the best advice is.;to.raise pELIn
particular, over the pH range of about 7JO-8.0, adverse -
effects from bicarbonate are significantly reduced by
even small (+ 0.2 pH units) increases in pH.
Interestingly, the authors' analysis indicates that
the method of increasing pH is important. Consider
three different options including NaOH (caustic),
CaOH2 (lime), or aeration-CO2 stripping. For the
caustic and the lime up to a pH of about 8.5, the OH~
added will increase alkalinity:
OH- + CO2 -* HCO3-
In contrast, raising pH by stripping CO2 increases
pH but does not alter alkalinity, because the reaction
producing OH~ consumes bicarbonate (which is sub-
sequently regenerated):
HCO3- -* CO2 (stripped) -^ + OSr
The significance of this effect on predicted copper
solubility depends on the initial alkalinity and pH of
the water. Consider a hypothetical water with initial
92 JOURNAL AWWA
-------
alkalinity of 250 mg/L as
CaCO3, initial calcium hard-
ness of 100 mg/L as CaCO3,
and temperature of 25°C. If
the final pH is raised to 7.2
with lime or caustic, soluble
copper is predicted to be 3.8
and 3.4 mg/L if the initial pH
is 6.5 or 6.7, respectively (Fig-
ure 10). If pH is raised to 7.2
using aeration, soluble copper
is predicted to be only 2.8
mg/L regardless of initial pH
(because alkalinity does not
change upon aeration). This
represents a 17-26 percent
enhancement in copper cor-
rosion control (at a given final -
pH) using aeration compared
with lime or caustic.
More important, however, is that pH may be
increased to higher values without precipitating cal-
cite if aeration is used. In the waters modeled in Fig-
ure 10 and for lime addition, calcite is supersatu-
rated at only pH 7.2 or 7.3, depending on the initial
pH. If pH is raised using aeration, calcite is not super-
saturated until pH 7.6. Thus, aeration provides flex-
ibility to raise pH to higher values without precipi-
tating calcite.
Quantitatively, increasing pH to the point of cal-
cite supersaturation (pH 7.6) using aeration is pre-
dicted to yield soluble copper of 1.3 mg/L (Figure
10). If pH is also raised to the point of calcite super.-
saturation using lime, soluble copper is predicted "to
be 3.9 and 2.8 mg/L at an initial pH of 6.5 and 6.7,
respectively. Therefore, aeration offers a predicted
improvement of 53-66 percent compared with lime
if pH is raised to the point of calcite saturation. This
improvement may be attributed to the higher possi-
ble pH as well as the lower final alkalinity. Because the
use of caustic would increase alkalinity without con-
comitant increases in calcium (as with lime dosing),
the pH at which calcite becomes saturated using caus-
tic is between that observed for lime (lowest pH) and
aeration (highest pH). ,
Although the described benefits of aeration are
expected to apply to many high-alkalinity, low-pH
Predicted equilibrium soluble copper as a.function of final pH
r"-—"- ~"2' "•* "umai (.else acicUIUn Will 3CtU-
aliy reduce pH and thereby worsen copper corrosion
problems. Nevertheless, aeration is an attractive, low-
cost option deserving of testing at many utilities.
Conclusions
• At constant pH, soluble copper corrosion by-
product release in relatively new copper plumbing
is a linear function of bicarbonate concentration
(alkalinity). ' .
• Qualitative and quantitative trends in soluble
copper corrosion by-product release are consistent
with predictions based on Cu(OH)2 equilibrium.
• The sensitivity of copper solubility to alkalin-
ity (bicarbonate), expressed as mg additional soluble
copper/mg alkalinity as CaCO3 added, depends on
pH and temperature. Copper carbonate complexes
are most significant in cold and low-pH waters.
• CO2 concentrations are a surrogate for cop-
per-carbonate complexation capacity in many
waters, providing a possible mechanistic basis to
correlations between dissolved CO2 and copper cor-
rosion problems.
• The conventional Langelier index or Larson's
ratio approach to. controlling copper corrosion by-
product release is demonstrably inaccurate. :
• Compared with lime or caustic addition, raising
pH by aeration (CO2 stripping) has inherent advan-
tages for copper corrosion control in low-pH, high-
alkalinity waters. These advantages are attributed to
lower final alkalinity and a reduced likelihood of cal-
cite precipitation.
Acknowledgment
The authors acknowledge the financial assistance.
of the AWWA Research Foundation (AWWARF) and
the National Science Foundation (NSF) under Grant
No. BCS-9309078. Any opinions, findings, and con-
clusions or recommendations expressed in this mate-
rial-are those of the authors and do not necessarily
reflect the views of the NSF or AWWARF. Portions of
this work were based on a survey by the Water Indus-
try Technical Action Fund. Special thanks to Gregory
Kirmeyer, Tim Chinn, Donna Dodrill, and the many
utility personnel responsible for development of this
database. The authors acknowledge insightful con-
versations with Werner Wolf, Tom Holm, Eilen Vik,
Jonathan Clement, Darren Lytle, Russell Taylor, and
David Nicholas that helped shape this work.
MARCH 1996 93
-------
References
1. CRUSE,
VJWJ.AvsyAV'Uj.v/j., yjy/j*jk*wj.vuui j. vyi.ciLn.v- rvaLv-j. ova-,- .. •.; v
t&^.fintfflatCorrosiqtiofW^ ?* '; *• (1988); '
^w- /-™™,~,t,,r~ Boo^^-b Rept., AWWARF 19. VAN DEN HO'VEN, T.J.J. Personal comm. (1993).
-.,-;:,.,...r _ . 1^ain-?n€er"?lfflffe; £.. ;?{l-; ^n-osiEV, I.JETAL. Breakdown of Passive Film on
l--Universi^t-,KarJsra £«;u;*.v>,,.,v,., o-i,,*:— ^—^_r_- „ -,
ffS!iiEkrSa'-i^-.-:Cj-i& BJ */•*;* V*%rfc-ii8£3y5:-§-'ri%«' ,5£"1~ "• '•£
^J^ai^^ife^^iisi^SSisii^^ifei^sSgssss -J
and:
Institut
WlS
2. EDWARD'S^;
^^f^ll^s^ntt^LriibMAs; JlopT. &.TILLER, A.K. Formation and
AKKKK^oo^;/5 ii^^yy4):;;.v;X7::.U.;;'.; •. ,.;.:r Breakdown^bf Surface Films on Copper in
3- ^^^i^Jf^^R^P^ft^:'^6^ & 26547 r:;. f; Sodium Hydrogen Carbonate and-Sodium Chlo-
^"^43i^^^;£^ii:pvg^>£Jv^^;?::i. i?; ^v^:i i vnde;Solutions L Effea of Anion Concentrations.
4. Standard M&nofeJOTtn&Examin^
^^^^l?^®?^^^!^ BROSSARD, L.; & MENARD, H.
ton, D.Cr (18th ed:;-1992):' • - - _.-•-.-.
Copper Dissolution in NaHCO3 and NaHCO3 .
NaCl Aqueous Solutions at pH 8. Jour. Electrochem.
Sac., 139:1:39 (1992).
MATTSSON, E. & FREDRKSSON, A.M. Pitting Cor-
rosion in Copper Tubes—Cause of Corrosion and
Countermeasures. British Corrosion Jour., 3:246
(1968),
COHEN, A. & MYERS, J.R. Mitigating Copper Pit-
ting Through Water Treatment. Jour. AWWA,
79:2:58 (Feb. 1987).
Moss, G. 6- POTTER, B.C. An Investigation of the
Green-jWater Problem in Auckland, New Zea-
land, and A Discussion of Possible Remedies.
CSIRO Restricted Internal Rept., Victoria, Aus-
tralia (1984).
. REIBER, S. Personal communication (1994).
27. ROTHBERG, M.R. ET AL. Computerized Corrosion
; Control. Jour. AWWA, 85:7:26 (July 1993).
28. LANGELIER, W. The Analytical Control of Anti-
Corrosion Water Treatment. Jour. AWWA,
28:1500 (1936).
'29. SCHOCK, M.R. S- LYTLE, D.A. Effects of pH, Carr
. ' bonate, Orthophosphate,.and Redox Potential. .
, on Cuprosolvency. Proo NACE Corrosion/95, ,
. Mar. 26-31/1995, Orlando^ Fla.
30. LARSON, T. Laboratory Studies Relating Mineral •
, . Quality of Water to Corrosion of Steel and Cast
. _^...... _, ' Iron. Corrosion, 41:285 (1958)!
Cations. Wiley-Interscience, New York (1976)'. 31.' WERNER, W. ETAL. Untersechuzen zur Flachekor-
13. VUCETA, J, &--MORGAN, J.J. Hydrolysis of Cu(n). . ,rion in Trinkwasserteitijen aus Kupler. Wasser.
T Twwnl f*}- f)rf0sft*stst *">-*7yt T /1 O*7T\ " :" • * "+ *t -. ~^ ~ *, ' ..
10.
11
12
5. EDWARDS, M; S- FERGUSON; J.F. Accelerated Test^
ing of Copper Corrosion. Jour. AWWA, 85:10:105
(Oct. 1993).
6. EDWARDS, M. ET AL. Role of Inorganic Anions,
NOM,':and ;Water"-lreatment Processes in Cop-
per Corrosion. Final Proj. Kept., AWWA, Denver,"
Colo, (in'press): '•-'' ' :
7. Database for AWWA Water'Industry Technical"
Action Fund Project: Initial Monitoring Expe-
riences of Large Utilities Under USEPA's Lead
and Copper .Rule: AWWA, Denver, Colo.
(1993)'.! •' '^•--•••.' . -
8. Brad Segal, Bbulder Water Department. Personal
communication (1993); '
9. SCHECHER/W.D'. MINEQL+: A Chemical Equilib-
rium Program for Personal Computers. The Proc-
tor and Gamble Co:, Cincinnati, Ohio (19.94).
SCHOCK, M~R:/--LYTLE; D.A.;''S- CLEMENT, J.A:
Effects of pH, DIG, Qrthophosphate, and Sul-
fate on Drinking Water Cuprosolvency. Res.
Kept., USEPA;'ORD,' Risk Reduction Engrg.
Lab., EPA/ 600/R-95/085, Cincinnati, Ohio
(June 1995)i--^-:*^-..--., :,;•; •.,... . . •:<••. •:: ..- • •
EDWARDS^ :M.;: FERGUSON, J.F.; &• REEBER; S. On the
Pitting Corrosion of Copper. Jour. AWWA; 86:7:74
(July 19.93)vJ-"r::'-:1.'::''-;.-•'•''• :'"' •'• '-"•• '"
BAES, C.F..jR\ &•" MESMER, R.E.' The Hydrolysis of
23.
24.
25
26
Limnol. &Oceanog., 2:742 (1977).
14. ScHiNDiERi:p;;'"REiNERii'M!; & GAMSJUGER, H. Lus-
lichkeitskonstahten iind reie Bildungsenthalpien
von Cu2(OH)2CO3 (Malachit) und Cu3(OH)2
~'Abwassen, 135:2:92 (1994).
15.
16.
17.
(CO3)2 (Azurit) bei 25°C. Helvetica Chim.Acta,
1:1845 (1968);}-' " -'••:"' ; ' ;
ZIRINO; Av^yAMAMOTO, S. A pH-Dep'endeni"
Model for'jhe Chemical Spedation of Copper,'
Zinc, Cadmium, and Lead in Sea Water. Limnol.
& Oceanoijr.'; 17:661 (1972).
DUBY, P. The thermodyhamic Properties of Aque-
ous Inorganic Copper Systems. INCRA Mono-
graph IV, IhtL .Copper Res. Assn., New York (1977).
DODRILL, D. & EDWARDS; M. Corrosion Control
Based on Utility Experience. Jour. AWWA, 87:7:74
(July 1995).
About the authors: Marc Ed-
wards is an assistant professor in
the Department .of Civil Engineer-
ing, University of Colorado (CU),
Boulder, CO 80309. He is a gradu-
ate of the University ofWashington
(Seattle) with PhD and MS degrees
and of the University of Buffalo
(New York) with a BS in biophysics.
Michael R. Schock is a research chemist with the US Envi-
ronmental Protection Agency, 26 W. Martin Luther King
Dr., Cincinnati, OH 45268. At the time of this research,
Travis E. Meyer was a research assistant at CU.
94 JOURNAL AWWA
-------
Copper Corrosion and Iron Removal Plants
Lih-in W. Rezania, P.E. and William H. Anderl, P.E.
Public Health Engineers
Section of Drinking Water Protection
Minnesota Department of Health
121 East Seventh Place
St. Paul, MM 55164-097
ABSTRACT
INTRODUCTION
ssrir^sTr^^^
a corrosion control
*.
COPPER CORROSION IN MINNESOTA
All copper exceedances in Minnesota came from groundwater s^
5SS5Ssr:^^S£SSMr^
-------
Scaling Water
tanglier Index (LI) and calcium carbonate precipitation potential (CCPP) are two
indexes commonly used to predict the scaling tendency or corrosivity of a water
Scaling water conceptually is non-corrosive, unfortunately, it is no so in Minnesota
as far as copper corrosion is concerned. Seventy-five percent of the water
systems with a copper exceedance are having a positive CCPP in their treated
waters.
High Dissolved Inorganic Carbon
Dissolved inorganic carbon (DIG) and PH are the two key parameters associated
with cuprosolvency. The high dissolved inorganic carbon content in finished water
is blamed for causing the copper exceedances in Minnesota. Theoretical dissolved'
inorganic carbon contents ranging from 200 mg/L to 900 mg/L as CaCO3 were
predicted for all systems using the Rothberg, Tamburini & Winsor (RTW) Model
The majority of them have DIG between 500 and 800 mg/L as CaCO, or between
60 mg C/L and 95 mg C/L in their treated water.
Iron Removal Treatment Plants
Iron removal treatment is used across the State of Minnesota by about one-third of
the municipal public water systems in Minnesota. In most cases/the treatment
process consists of aeration/oxidation, filtration, chlorination, and fluoridation It is
notable that among those 142 public water systems exceeding the copper action
level, 83% of the medium-size (Figure 1) and 57% of the small-size systems
(Figure 2) are iron removal plants. It is believed that the aeration/oxidation step in
the iron/manganese removal process results in a more corrosive finished water due
to the higher dissolved oxygen levels and/or the content of oxidizing agents such
as potassium permanganate, and chlorine in the water.
IRON REMOVAL PLANTS D.O. STUDY
A study on copper corrosion of iron removal treatment systems in Minnesota was
conducted ,n summer 1993, by the Public Water Supply Unit. Nineteen (19) iron
removal filtration plants were studied for their finished water dissolved oxygen
levels associated with the aeration/oxidation process they use and their reported
90th percentile copper values. This study focused on the way oxygen is
introduced into the water and demonstrated a strong association of 90th percentile
copper levels with the dissolved oxygen levels (figure 3)
-------
Gravity Plants
f' t /em°V flltratl°n Pl3ntS are the ones wlth c°PPer Problems. Most
f.ltratjon systems use a combination of steps such as tray aerator, spray
aerator, grav.ty drop, flow splatter or splitter to aerate the water. Some also use
oxidizing agents such as Chlorine and KMnO4 oxidizing agents in this
0mananeS!, °XLdatlon. steP and followed by a gravity filtration process to
,OX, . Prec'P'tates. Finished water dissolved oxygen levels of gravity
l lants are
irorp ,, . . r ssove oxygen le
iron removal plants are generally at or near the D.O. saturation point.
In winter 1993, eight more systems were added to the study. This study
concluded that gravity systems are most susceptible to copper corrosion The
average , finished water dissolved oxygen level from fourteen (14) gravi™ plants
was 6.5 mg/L and the average 90th percentile copper level was 2 35 rna/L rL
01"" r93™ Carb°n W3S 71 m9 C/L Thetrge dissolved oxygen
erCent'le C?Per l6Ve!S' and diss°lved inor9anic carb°n »evel from 9
pressure plants are 1.52, 0.74, and 76 mg C/L, respectively.
Pressure Plants
Pressure systems use compressed air and/or oxidizing agent(s) to precipitate iron
and manganese and a pressure filtration process to remove the precipitates
exceed" 0° moT T^ "l *?* ***** *"** ^ ^^ #™ S* do not
fivaromn™ °3" • noted that among the 27 iron removal plants studied, the
five compressed air aerated plants resulted with the highest average dissolved
Carb0neVeI °f87
henedaTH . e in
the injected a.r. The eight non-aerated pressure plants on the other hand have the
owest average DIG of 65 mg C/L. Average 90th percentile copper valCe for these
two groups are .88 mg/L and 0.60 mg/L, respectively. '
Phosphate Treatment Survey
°f the ir°n rem°Val SVStems in Minnesota also treat with
t C°rrOSf0n 3nd SCale C°ntroL ln 1994' « Phosphate
treatment survey study conducted by the Public Water Supply Unit confirmed that
iron removal systems are susceptible to copper corrosion even with the presence of
phosphate inhibitors. This survey study found the only exceptions "are svstems
-------
SEASONAL VARIATION IN COPPER LEVEL
In winter 1 994, 35 water systems with no first round problems exceeded the
3A lVeL ,H!?her COPPSr tap levels were tested in a significant number of
Z" £ * 6m exPerienC6d inc^ases from a less than detection limit
<0 05 r * s an eecton mt
(<0.05 mg/L) in the f.rst round monitoring to a greater than the 1.3 mg/L copper
ae&on level m the second testing. Winter heating has been blamed for causing the
elevated copper levels. The Minnesota Department of Health, Public Water Supply
Unrt, has investigated the seasonal variation issue in following efforts.
Demonstration Tests
w taps were collected seasonally and analyzed for copper concentrations
homes for an e.ghteen months period. This testing program found six
xhibrted a clear seasonal pattern in tap copper levels, peak in midwinter
and big drops for the rest of the year with the lowest levels tested in summer and
round mo ^ demons,trated the on»Y elevated copper level tested was the second
round monitonng sample, collected in January and February 1994. One site shows
e peaks occurred in summer months. These tests strongly demonstrate the
conrprnfS th °°PPer ^^ '" *e f'rSt draw taPs dur|ng heating season and raised
concerns of the corrosion control treatment issue for water systems caught up with
a copper exceedance due to this seasonal phenomenon.
Copper Tap Level Reproducibility
The lead/copper monitoring results of 5311 sites were analyzed for their copper tap
DatlT hUC I'- S'teS W6re SeleCted °Ut of tne Lead/Copper Monitoring P
Database, by matching up each location with the 1st and 2nd samples taken
approximately s,x months apart, so that a well-defined "season" can be referred to
and the seasonal variation issue can be examined
Produced and Analyzed as presented in Table 1. The average
« H W!re hi9her '" the SGCOnd r°Und testin9- About 8-5% of the sites
Pvtrpm K 3 ^ 9e m C°PPer l6Vel 9re&ter than °'5 m9/L in the two tests, with
to 7 1/1 Sn™'ngtH°PPerrJ.eV8lS I6SS than detectlon Iimits at one samP'ing and up
°ther; Th'S percenta9e <8-5%> is significant due to its potential to
R.PPer eXCeedance' Since 10% is ^ed as the trigger in the Lead and
Copper Rule.
-------
Data Set
No. of Sites
2603 Sites
2652 sites
55 sites
Overall Reprod
|Cu1 - Cu2
0.1 rng/L <
|Cu1 -Cu2
• ! '"' ' " TT"
1st Round Monitoring
Sampled in
JuI/Aug 93
Sep/Oct 93
Nov/Dec 93
Ave. Cu
.18 mg/L
.29 mg/L
.21 mg/L
.
2nd Round Monitoring
Sampled in
Jan/Feb 94
Mar/Apr 94
May/Jun 94
ucibility in Two Tests:
| <• O.1 mg/L «i QOA
|Cu1 - Cu2| ^ C
> 0.5 mg/L
).5 mg/L 30 2%
„ R C;OA
==============^^
Ave. Cu
.21 mg/L
.35 mg/L
.28 mg/L
• — i
Table 1
TREATMENT EVALUATION
The Lead and Copper Rule suggested that four treatment approaches be evaluated
by every public water system exceeding an action level. Evaluations, based on the
S?6d.Water Ch![a°terJ??LCAS for systems with a c°PPer exceedance using both the
EPA guidance and the AWWA's RTW Model, show that phosphate inhibi?or is the
most prominent alternative for copper corrosion control in Minnesota.
Carbonate Passivation
Optimal water characteristics for employing this treatment approach for lead
control are low dissolved inorganic carbon (DIG) and high PH. This treatment
application is limited to few lime softening water systems for lead control
Calcium Carbonate Precipitation
The RTW Model was used to obtain the LI, DIG, and the CCPP values These
values were used to assist public water systems in the evaluation of the suggested
corrosion control treatment alternatives required by the lead and copper rule The
calcium carbonate precipitation corrosion control treatment approach was
eliminated by the majority of water systems due to the exhibited high CCPP values
-------
corrosion contrnf "h the trea!fd W3ter PH appears t0 be beneflcial 1™ copper
Silicate Inhibitor
Phosphate Inhibitor
Phosphate treatment is well received by most of the public water systems and
es^^
corrosion control for lead and copper (Figure 4). optimal
TREATMENT EXPERIENCE FROM IRON REMOVAL PLANTS
°f ^ ^ -mova,
Utility #1 - Treatment With Poly-Orthophosphate Blend
Population Served: 7,500
90th Percentile. Copper Level: 2.25 mg/L
-------
Water Treatment Used: induced aeration, gravity filtration, fluoridation, chlorination
Treatment Modification After Exceeding Cu A.L.: blended phosphate treatment
Corrosion Control Treatment: using a blended phosphate inhibitor with a poly/ortho
ratio of 1 at 1.5 - 2.0 mg/L as total phosphates; phosphate feed rate is constrained
by costs and local wastewater discharge limits
Treatment Outcome (Figure 5): immediate reduction in tap copper levels was
appreciated, the effectiveness started to level off after achieving 50% to 65%
reduction.
Water Quality Data:
Parameters Finished Water
PH 7.6 - 7.9
Calcium as CaCO3 150 - 180 mg/L
Alkalinity as CaCO3 320 - 340 mg/L
Dissolved Solids 350 mg/L
Orthophosphate as PO4 0.8 - 0.9 mg/L
Utility #1 demonstrates corrosion control treatment experience pf the majority of
the iron removal systems in Minnesota who have started or switched to feeding
/ blended phosphates for corrosion control purposes. Copper tap level reduction of
50% or more can be easily achieved by feeding phosphate at 1 to 2 mg/L as total
phosphate (orthophosphate ranging from 0.7 to 1.0 mg/L). However, the reduced
copper levels remain at 1.0 to 1.5 mg/L, indicating the likelihood of exceeding
copper action level in future monitoring effort.
Utility #2 - Calcium Carbonate Precipitation
Population Served: 22,000
90th Percentile Copper Level: 4.08 mg/L
Water Treatment Used Prior to Pb/Cu Monitoring: chlorination and fluoridation
Water Treatment After Exceeding Cu A.L.: two new pressure iron removal plants
have been installed, treatment involving strip aeration, pre-chlorination, pressure
filtration, and fluoridation
Corrosion Control Treatment: practicing calcium carbonate precipitation treatment
approach; as a result of the aeration, water pH has increased from 7.5 to 8.1;
-------
theoretical CCPP for treated waters are 40 mg/L and 80 mg/L as CaC03.
Treatment Outcome (Figure 6): copper reduction above 50% has been achieved
contributing to the higher finished water pHs and CCPP values produced by the
new treatment plants; system remains to face the challenge in meeting the 1 3
mg/L copper action level •
Water Quality Data:
Parameters Plant #1 pjant #2
Raw Water pH: 7.5 7 7
Finished Water pH: 8.1 8.1
Alkalinity as CaC03: 430 mg/L 390 mg/L
Calcium as CaC03: 260 mg/L 150 mg/L
Dissolved Solids: 630 mg/L 490 mg/L
Free Ammonia mg/L as N: 2.8 mg/L 3.0 mg/L
Utility #2 demonstrates that regardless of the positive (high) calcium carbonate
precipitation potentials exhibited in the finished water, copper levels remain above
the action level. The appreciated copper reduction can be credited to the aeration
used that has increased pH from 7.5 to 8.1. This utility also demonstrated a very
rare case in that copper remains as a concern at pH of 8.1. The high ammonia
concentrations in the treated water may be the reason for their continuing copper
problem. K
Utility #3 - Treatment With Zinc-Orthophosphate
Population Served: 49,000
90th Percentile Copper Level: 3.36 mg/L
Water Treatment Used: utility has one pressure plant and one gravity plant; copper
problems are located in areas supplied by the gravity plant; the treatment involves
cascade aeration followed by gravity filtration, chlorination and fluoridation;
potassium permanganate and chlorine are added as oxidizing agent
Treatment Modification After Exceeding Cu A.L.: zSnc-orthophosphate treatment
Corrosion Control Treatment: zinc-orthophosphate at 0.8 - 1.3 mg/L as ortho-PO4
Treatment Outcome (Figure 7): treatment successfully reduced copper levels to
below 10 mg/L at four test sites, achieved greater than 7O% reduction in copper
tap levels showing promise in meeting the copper action level in future monitoring
8
-------
Water Quality Data: ;
. Parameters Gravity Plant Pressure Rant
Finished Water pH 7.1-7.5 6.8-7.0
Finished Water D.O. 8.2 mg/L 0.8 mg/L
Alkalinity as CaCO3 270 - 380 mg/L 200 - 310 mg/L
Calcium as CaCO3 190 - 240 mg/L 170 - 200 mg/L
Ortho-phosphate as PO4 0.8 - 1.3 mg/L None ••*'•
Dlc as C 87 mg/L 77 mg/L
Utility #3 demonstrate a successful copper corrosion control treatment experience
using zinc-orthophosphate, achieved by a gravity iron removal system. Copper tap
levels were reduced below 1.0 mg/L at a relatively low orthophosphate feed rate
(0.8-1.3 mg/L) with respect to the high level of dissolved inorganic carbon and
dissolved oxygen in the finished water. This study echoed the conclusion from the
small systems phosphate treatment survey study that orthophosphate treatment is
the most prominent corrosion control treatment option for Minnesota's iron removal
systems.
SUMMARY AND CONCLUSIONS
1. High dissolved inorganic carbons in groundwater sources and high dissolved
oxygen in treated water from iron removal plants are causes for copper
exceedances in Minnesota.
2. Iron removal plants using gravity filtration are most susceptible to copper
corrosion due to the high levels of dissolved oxygen introduced fay aeration."
3. Calcium carbonate precipitation appears to be ineffective for copper corrosion
control. This treatment approach has very limited application in Minnesota due
to the,high hardness and alkalinity in groundwater
4. Seasonal variation in tap copper levels was demonstrated in Minnesota, raising
concerns of requiring corrosion control treatment for systems exceeding the
copper action level due to this seasonal phenomenon.
5. The common experience using blended phosphates in Minnesota is that a 50%
reduction in copper levels can be easily achieved. However, the lowered
copper levels remain near the 1.3 mg/L copper action level, leaving water
systems with continuing challenge of meeting the copper action level.
-------
6. Ortho-phosphate levels at 0.8 - 2.0 mg/L have successfully reduced copper tap
level below 1.0 mg/L for five gravity iron removal systems. A minimum ortho-
phosphate level of 1.0 mg/L is recommended to be maintained throughout the
distribution for iron removal systems treating with a phosphate inhibitor for
copper corrosion control.
REFERENCES
1. "Copper Corrosion Study of Iron Removal Treatment Systems in Minnesota"
Minnesota Department of Health, Section of Drinking Water Protection, Public
Water Supply Unit, November 1993.
2. "Corrosion Control Treatment Survey for Small-Size Community Public Water
Systems - Summary and Conclusion", Minnesota Department of Health, Section
of Drinking Water Protection, Public Water Supply Unit, June 1994.
3. Rezania, L.W. & Anderl, W.H., "Copper Corrosion and Iron Removal Plants the
Minnesota Experience", Proc. of AWWA WQTC, 1995, New Orleans, LA
10
-------
<*-•
I
en
-------
I
II
O)
-------
o
==
O
O
o
I
ji>
o
in
O)
3
cn
(V6n) uoiiDJ|U8Ouoo jeddoo MDJQ isj
-------
o
O
CM
• ^^"w»
O
P
a
I
o
.2
a
(V6n)
(spuosnoqi)
<£>
S-
3
cn
-------
^__ BOMB
DO.
3
05
-------
An Evaluation of the Secondary Effects of Enhanced Coagulation, With Emphasis
on Corrosion Control
Darren A. Lytle, Michael R. Schock, Richard J. Miltner
U.S. Environmental Protection Agency
NRMRL, WSWRD, TTEB
Cincinnati, OH 45268
Abstract
The proposed Dbinfectant-Disinfection By-product (D-DBP) Rule will require many
water purveyors to meet enhanced coagulation or softening objectives for total organic
carbon (TOC) removal as a means of controlling DBF precursors. Enhanced coagulation can
be achieved by performing chemical coagulation at lower pH values by increasing the
coagulant dose, adding acid, or a combination of the two. Pilot studies using enhanced alum
coagulation revealed a number of secondary effects, many of which directly impact lead and
copper corrosion. Aluminum, sulfate, pH, and dissolved inorganic carbon were significantly
altered during enhanced coagulation. These parameters can directly affect metal solubility
and surficial pipe film stability.
Introduction
Under the proposed Disinfectant-Disinfection By-product (D-DBP) Rule, many utilities
will be required to meet enhanced coagulation or softening objectives for total organic carbon
(TOC) removal as a means of controlling DBF precursors.1 A recent survey found that 64 percent
of responding coagulation plants would comply with the specified TOC removal requirements of
the proposed rule.2 Several researchers have investigated natural organic matter (NOM) control
by coagulation and discussed the importance of pH in removal efficiency. Qasim et al.3 found that
TOC removal in natural waters by coagulation and softening was strongly pH dependent and
Randtke found that removal of NOM by coagulation was optimum at pH 5.0 to 6.0.
Most coagulation plants that will be required to meet TOC removal requirements under the
proposed D-DBP Rule will likely do so by either increasing coagulant dosage, lowering the pH
by acidification during coagulation, or a combination of the two. Enhanced coagulation will
provide improved removal of DBF precursor material, as well as secondary benefits such as lower
chlorine demand, increased disinfectant stability in the distribution system, lower color and
reduced substrate for microbiological growth.3 However, negative secondary impacts of enhanced
treatment must also be addressed to insure that other water quality parameters and treatment plant
operations are not compromised. Some of the secondary impacts that must be considered include
sludge production, filter run time, inorganic water quality changes, turbidity and particulate
removal efficiency, and distribution system corrosion and corrosion control.
The U.S. Environmental Protection Agency's (USEPA's) National Risk Management
Research Laboratory (NRMRL), in Cincinnati, Ohio, in conjunction with the University of
Cincinnati, conducted bench- and pilot-scale tests to evaluate enhanced coagulation for optimal
NOM or DBF precursor removal from a surface water using aluminum sulfate, more commonly
referred to as "alum". Results of those studies directly related to the goals of the D-DBP Rule
have been previously discussed.5-6 This paper explores several secondary impacts relative to
corrosion control of enhanced alum coagulation based on observed water parameter changes
-------
during pilot-scale studies. Specifically, the effect of pH, dissolved inorganic carbon (DIG) and
alkalinity, sulfate and chloride (during ferric chloride coagulation), total organic carbon (TOC),
and aluminum on the corrosion and corrosion control will be discussed.
Experimental
Based on preliminary jar tests studies6, East Fork Lake water (Cincinnati, Ohio) was
selected for pilot testing. Two parallel pilot-scale treatment plants consisting of rapid mix,
flocculation, and sedimentation basins were used to compare conventional and enhanced
coagulation as shown in Figure 1. Settled water was chlorinated and sand filtered. The
adjustment of pH for corrosion control was made after filtration in the conventional plant. The
enhanced plant was split into two parallel filters where pH was adjusted before one filter and after
the other. Final pH in all cases was adjusted to approximately 8 using NaOH. Chlorine was fed
as NaOCl. A number of water quality parameters were monitored at a variety of locations
throughout the plant and are given in Table 1. Operational data including chemical doses,
temperature, head loss development and filter run time were also regularly monitored. The
sampling described in Table 1 took place on six days between September 7 and September 13,
1994. East Fork Lake water was trucked to the USEPA research facility daily during pilot plant
operation.
Results
Because the results of the pilot study have been presented elsewhere5'6 and page constraints,
data will not be presented or analyzed in great detail. A number of secondary impacts (both
positive and negative) beyond the corrosion impacts addressed in this report have also been
previously identified in some detail.7 Positive secondary impacts of enhanced coagulation
included: reduction in turbidity, particle counts, coliform and heterotrophic bacterial densities,
and chlorine residual. Increased cost and sludge production, and potential future regulatory
conflicts such as the proposed sulfate standard were identified among negative secondary impacts.
This document lists the differences in water quality between conventional and enhanced
coagulation treatment modes that are considered important in lead and copper corrosion control.
Table 2 shows the major average water quality parameter measurements;'throughout the
treatment trains. It should be pointed out that in the case of East Fork Lake water, only 35
percent TOC removal would be required given the TOC (4.8 mg/L) and alkalinity (99 mg
CaCO3/L) of this water.8 It is important to note that in these pilot studies, conventional
coagulation (29 percent removal of TOC) was compared to optimum coagulation (54 percent
removal of TOC). Enhanced coagulation in the regulatory sense (35 percent removal of TOC)
was not studied. Thus, findings regarding secondary effects must be viewed as examining the
extreme shift from conventional to optimum coagulation rather than the expected shift from
conventional to enhanced coagulation.
Alum doses were 44 mg/L (3.7 mg/L as aluminum) and 152 mg/L (13.3 mg/L as
aluminum) for conventional and optimized coagulation, respectively, as shown in Table 2.
Aluminum concentrations in settled waters were lower after optimized coagulation than after
conventional coagulation (0.47 mg/L versus 0.65 mg/L) despite the higher alum dose. This
observation follows the established pH-aluminum hydroxide solubility relationship.9"13 The
location of pH adjustment was clearly the most important factor impacting aluminum residuals
after .the filters, as shown in Figure 2. When pH adjustment was practiced following filtration
-------
(filters Fl and F2), the lowest clearwell aluminum residuals were observed. This results from the
ower solubility of aluminum at lower pH values, and retention of unsettled residual alum floe by
Ac filters. However^ when the pH was raised to 7.95 prior to filtration (filter F3), aluminum
solubility was increased, enabling the passage of more dissolved aluminum into the clearwell The
trade off was that filter run times were shorter in the lower pH waters
fiir«f • Alun™umrle;;els were ** lowest 0« than 0.025 mg/L) following optimized coagulation,
filtmion and pH adjustment in clearwell 2. This can be explained by the fact that the pH of
settied water entering filter 2 was 6.90, which is closest to the minimum solubility of aluminum
hydroxide. The highest aluminum levels were observed following optimized coagulation, PH
adjustment and filtration because the PH of the water (pH=7.95) was the highest, resulting in
the most soluble aluminum entering the filters. Aluminum concentrations in filter effluents were
used to develop the empirical pH-aluminum solubility relationship shown by Figure 3 This
Sulfate increased after both coagulation approaches as a result of the addition of alum as
seen m Table 2. Optimally coagulated water contained at least 40 mg SO42VL more than
conventionally treated water. The point of pH adjustment had no significant impact on sulfate
concentration.
In this study, the pH needed to meet optimum enhanced coagulation conditions was 0 7
I "u M6' *? PH dUring conventional coagulation. Without pH readjustment, the drop
urf^Z T V" -KCr^Sed ^ md C°Pper ****** ** Stabilization of existing pip^
surface films in the distribution system. Thus, pilot plant finished water pH values were adjusted
to near 8 to represent reasonable conditions under the Lead and Copper Rule and to drive DBF
reactions under representative conditions. .
imnnrt '" ^^f * the&ffQCi of the location of pH adjustment to aluminum residual, another
important consideration is the impact of the pH adjustment chemical on finished water quality
eS
issuesnR , ' ng
rnnT ; ? ^Jjustment chenucals such as caustic soda (NaOH) or soda ash (Na2CO3) wiU
contnbute sodium to the finished water. Although lime was not used in this study for pH
im uritiel mCreaS£ ^^ ^^^ ^ may contribute *"™™™ as a result of
d^r, ^ ^^f Cfb°n CnQ W3S n0t ^^^ ^ «»vw»tional coagulation; however, TIC
decreased significantly from optimized coagulation (see Table 2). The amount of TIC decrease
mg C/L) is clearly far beyond
Corrosion impacts
hnth ^ effeCt f PH Snd DIC °D Iead and C0pper corr<*ion. DIG has been shown to have
both a positive and negative impact on corrosion control.15'17 DIC serves to control the buffer
£™,!L* ri -^f SyStemS' ^ theref°re' SUfflcient DIC is necessary to Provid^ ^ stable pH
throughout the distnbubon system for corrosion control of copper and lead.17'19 However larger
amounts can result in increased lead and copper solubility.17'20
The importance of pH to lead and copper solubility is well-established. Therefore,' the
reduction of pH during enhanced coagulation will likely have to be counteracted by PH adjustment
following treatment to maintain corrosion control objectives outlined under the Lead and Copper
-------
Rule21"23 or state or local wastewater discharge requirements for copper. Without pH
readjustment, the drop in pH from coagulation would result in increased lead and copper solubility
and destabilization of existing pipe surface films in the distribution system, if corrosion control
is currently being practiced through pH/DIC adjustment. Even the effective use of orthophosphate
or blended phosphate chemicals for lead or copper control may require a higher pH than that
achieved during enhanced coagulation.15'17
Considerable research has been devoted to the study of the aqueous speciation of copper
in natural waters and seawater, in which the role of carbonate species was extremely important.24"30
The concept of the significance of carbonate complexation was then applied to drinking waters
with indications that high DIG levels would aggravate cuprosolvency.31'32 Recently, a variety of
new research has helped define the complicated interrelationships of pH, DIC, orthophosphate,
the formation of metastable solids, and plumbing material age on cuprosolvency and observed
copper levels in drinking waters. 19-20-33JW
Figure 4 illustrates the effect of DIC and pH on cupric hydroxide solubility, as would be
the case with relatively young plumbing systems. An increase in copper solubility with lower pH
and higher DIC is evident from the figure. Above a pH of approximately 9.5, an upturn in
solubility is predicted, caused by carbonate and hydroxide complexes increasing Cu(OH)2(s)
solubility. In the pH range of approximately 6.5 to 9, significant increases in copper solubility
are predicted from the addition of even small amounts of carbonate, although maximum solubility
remains less than about 0.3 mg/L.
Figure 5 shows a solubility diagram for copper(II), corresponding to equilibrium with
either Cu2(OH)2CO3(s) (malachite) or CuO(s) (tenorite), whichever is thermodynamically stable
at a given pH/DIC point. This kind of situation represents the case with aged plumbing, though
the number of years of exposure needed to achieve these stable solids likely depends greatly on
the water chemistry of the system.19-34 Several important contrasts in cuprosolvency behavior
between the case represented by Figure 4 as opposed to the assumptions behind Figure 5 should
be noted. If Cu2(OH)2CO3 (malachite) is present and capable of forming, below a pH of about
6.5 the addition of DIC is predicted to decrease cuprosolvency, but increase cuprosolvency above
about pH 7. There is a small transition zone between these values where the first approximately
5 mg/L of DIC should slightly reduce copper solubility, but additional carbonate would decrease
it or have essentially no effect. In an aged system, below a pH of about 8, cuprosolvency
becomes essentially insensitive to DIC above approximately 30 mg C/L. Malachite formation
would enable attainment of 1.3 mg/L after stagnation below approximately pH 6.5 for all DIC
levels. This is in stark contrast to the effect of pH and DIC when only cupric hydroxide is
formed, where a pH of over 7 would be necessary to stay under 1.3 mg/L for long stagnation
times at very low DIC levels, and over 7.5 for systems with high DIC.
In a simple system of metallic lead immersed in water containing dissolved carbonate
species, the solubility will be controlled by either the simple lead carbonate (PbCO3, cerussite)
or one of the two basic lead carbonates, Pb3(CO3)2(OH)2 (hydrocerussite) or Pb10(CO3)6(OH)6O
(plumbonacrite). Above approximately pH 12.5, lead hydroxide (PbtCH)^ may form . The
conventional solubility constants for lead hydroxide reported in the literature vary over about five
orders of magnitude (log K,,, from -19.85 to -14.9). Samples of deposits from potable water
systems and accompanying dissolved lead concentrations do not indicate either lead oxide (PbO,
massicot or litharge) or lead hydroxide to be active in governing lead solubility, though PbO is
frequently found in underlayers of surface films. Under very oxidizing conditions, the lead(TV)
-------
solid PbO2 may form, and it has occasionally been found in pipe deposits. PbO2 is less soluble
than any of the Pb(II) carbonates or hydroxycarbonates.
When only pH and carbonate concentration effects are considered, lead solubility is shown
by Figure 6. Detailed computations show that the minimum lead solubility is at a pH of
approximately 9.8, with a DIG concentration of between 3.6 and 4.8 mg C/L (0.0003 to 0.0004
M).15'17-31 Above approximately pH 8, the closeness of the lines and steepness of the slopes
represent the high sensitivity to both pH and DIG. These properties, plus the reversal of solubility
trends above about 5 mg C/L DIG result from the stability of the solid Pb3(CO3)2(OH)2
(hydrocerussite) plus strong hydroxide and carbonate complexation. The downward-trending lines
for pH 6 and 7 are caused by the formation of PbCO3 (cerussite), and an absence of carbonate
complexation. In the lead carbonate stability region, below approximately pH 8 and above
approximately 25 mg C/L, lead solubility is not very sensitive to either pH or DIG, although the
trend is toward slightly lower levels as DIG is increased below about pH 7.5. Detailed discussions
of lead solubility controls for drinking water are readily available elsewhere. 15-l7il8'31>41>42
The effect of chloride, sulfate, and aluminum on lead and copper corrosion. The
effect of chloride, sulfate and aluminum on copper corrosion has recently come under renewed
investigation, as a result of the interest in controlling cuprosolvency for drinking water or waste
water regulations. Interesting results were obtained from X-Ray diffraction and Energy-
Dispersive X-Ray (EDXA) analyses of the deposits formed on experimental pipes used in DWRD
pipe loop studies.19-34 Significant diffraction peaks for Cu4(OH)6SO4-H2O(posnjakite) were found
in pipes from experiments at pH 8 and pH 9 with 5 mg C/L DIG, even though sulfate levels were
only approximately 30 mg/L. This mineral has also been reported in some copper pipes in a
German study of pipe corrosion in a hospital.43 Diffraction peaks likely, corresponding to the solid
Cu(Cl,OH)2-2H2O (calumetite) were found, particularly at pH 8 and 7. Qualitative elemental
analysis confirmed the presence of S, Cl and also Al on the pipe surfaces. A large peak and a
secondary peak apparently consistent with the solid CuAl4SO4(Oir)12-3H2O (chalcoalumite) were
found on the pH 7 specimen, but only a corresponding minor peak was found on the pH 8 sample.
Additional peaks for Cu2(OH)2CO3, CuO, and Ci^O were also identified. From the qualitative
elemental analysis, the Al concentration on the pipe appeared higher at pH 7, consistent with the
general trend in solubility of many aluminum minerals, and a statistically-significant decrease in
aluminum in the water during the pH 7 experiment. The presence of aluminum on the pipe is also
noteworthy because of the low Al concentration in the water in all three experimental systems (<
0.1 mg/L), which suggests possibly a strong role for aluminum in the formation of natural
diffusion barriers in plumbing and distribution systems.
Examination of copper leaching results from DWRD studies at pH's > 8 show the copper
levels were consistently above the solubility of copper predicted by cupric hydroxide or oxide
models when elevated levels of sulfate (> 30 mg SO42VL) were present.19-34 An example of this
effect is shown in Figure 7, representing data from 72-hour stagnation samples from pH-adjusted
Cincinnati tap water containing approximately 70-120 mg/L of sulfate.19'34 A detailed literature
investigation suggested that above some threshold combination of pH and sulfate concentration,
metastable hydroxysulfate solids may form, rather than the more protective cupric hydroxide or
tenorite.19-34 Edwards, et al.33>4° showed that sulfate increased copper corrosion rates in water.
Rehring and Edwards44 attributed higher copper corrosion rates in enhanced coagulated waters to
additional sulfate carryover from the alum coagulant and also showed lower copper corrosion rates
from chloride carryover from ferric chloride enhanced coagulation. The effect of sulfate on
-------
copper corrosion is significant; however, the degree sulfate enhances copper solubility and effects
copper action level exceedences is not fully known. The greatest negative impact of sulfate on
cuprosolvency seems to be for waters with a pH over 8, based on early USEPA experiments at
low DIG. Increases in copper solubility resulting from increased sulfate levels may more likely
be a concern for wastewater discharge exceedences of copper.
To add to the confusion about sulfate effects, equilibrium chemical modeling shows that
there is some potential that the same cupric hydroxysulfate solids that cause increased
cuprosolvency above pH 8 may actually reduce cuprosolvency below approximately pH 7, relative
to the solubility of cupric hydroxide on young plumbing.19 This hypothesis needs experimental
verification.
Several researchers have considered possible effects on lead solubility of simple anions,
such as nitrate, chloride, and sulfate. Beccaria et al.45 used a variety of electrochemical and x-ray
techniques to study the corrosion of lead in seawater, which provided an extreme case for effects
of chloride and sulfate. During the initial passivation stage, Pb(OH)Cl(s) and Pb3(CO3)2(OH)2(s)
were found to be constituents in the deposit. After long immersion periods, PbCO3-PbCl2(s),
Pb2O3(s), PbO(s), and PbCl2(s) were also found. The presence of sulfate ions did not interfere
with the film formation, and sulfate ions did not precipitate to form a lead sulfate solid on the
surface. In solutions with extremely high sulfate concentrations, some basic lead sulfate solids
were found, notably Pb3(SC>4)2(OH)2 and an unidentified compound with a 1:1 oxide:sulfate ratio.
Lead(E[) forms somewhat weaker complexes with sulfate and chloride than with carbonate,
either of which might occasionally be found in water supplies at a sufficiently high level to impact
lead solubility. The greatest impact, if it happened, would be at relatively low pH and low DIG
levels, where less lead is complexed by hydroxide and carbonate species. Calculations were
performed for chloride, sulfate, and for both sulfate and chloride at DIG levels of 3-30. The ionic
strength of the waters for these modeling calculations was set to 0.05 mol/L, to allow higher
chloride and sulfate concentrations more plausibly. However, there appears to be no significant
impact of practical proportions is decided based on solubility.
Ironically, reducing aluminum residuals could cause the destabilization of Al-containing
films built up in domestic plumbing over many years of normal plant operation. These films have
been suggested as having been beneficial to lead and copper leaching from plumbing.17>42>447 One
study has observed an increase in lead mobilization apparently resulting from sloughing-off of Al-
based pipe scales.47
Considerable recent evidence for a corrosion-reducing aluminum or aluminosilicate film
has been described in a study conducted by the Denver Water Department, where significant
precipitated coatings were inhibiting lead and copper release from corrosion control pipe loop
study test rigs.46 For utilities adding sodium silicate as a corrosion inhibitor, residual aluminum
from conventional coagulation may enable additional protective films to form. The solubility and
complexation chemistry of aluminosilicates is highly complex and somewhat controversial in its
details.48"52 However, aluminosilicate minerals clearly are important naturally-forming solids, and
are geochemically plausible for many drinking waters with a near-neutral to slightly acidic pH.
An adverse affect of aluminum on copper piping in hot water lines was reported by
Tunturi,53 who found pitting failures associated with deposition of aluminum hydroxide from a
drinking water with 0.3 to 0.7 mg/L.
As described earlier, aluminum has a tendency to form several highly-insoluble
orthophosphate compounds under certain chemistry conditions. Whether or not the formation of
-------
such solids would occur in the distribution system if a water system was dosing orthophosphate
is not known, and requires more investigation. Depending upon the surface interaction properties
of any of these solids formed, the result could be an increased protection from additional surface
film formed, increased turbidity in hot or cold water, or depletion of orthophosphate dosed and
consequent lessening of passivation of lead or copper.
The effects of NOM on lead and copper corrosion. A goal of the D-DBP Rule is lower
TOC concentrations prior to chlorination, so that lower DBF concentrations of regulated and
unregulated DBFs will result.1 The impact of TOC on lead and copper solubility and in the
structure of protective films, however, may be significant.
Many investigators have looked at interactions between Cu2"1" and dissolved natural organic
matter. Some studies suggest that while copper can form complexes with NOM ligands
(sometimes called DOM for "dissolved organic matter"), at concentrations of copper typical of
drinking waters, copper speciation is more likely dominated by hydrolysis or carbonate
complexes.54'56.
Cu2+ ion may also bind with adsorbed organic material containing appropriate functional
groups. The binding with adsorbed organic matter seems to be stronger than direct binding with
surface sites on several materials tested.57 Copper present as an organic complex may bind
preferentially with adsorbed organic material.57-58 These studies suggest that some reduction in
copper concentration may be caused by adsorbed organic material acting as either a diffusion
barrier, or as a less-soluble corrosion film.
Some other studies suggest, however, that NOM may play a major role in the aqueous
speciation of cupric ion, particularly when carbonate concentrations are low.59-60 Organic ligands
produced by marine diatoms and during diatom blooms have been shown to strongly complex
copper, though usually at low copper concentration.61-62 Unsaturated organic ligands were also
shown in experiments at pH's generally lower than drinking water to increase the dissolution rate
of copper metal in the presence of cupric ion, by a complicated interaction affecting the electron
transfer rate between Cu(s) and Cu2+, and by stabilizing the Cu(I) state by complexation.63 '
The significance of NOM to cuprosolvency relative to drinking water concentrations of
copper and competing non-metals and ligands has not been conclusively determined, though it is
an area under active investigation by some research groups in the United States. Research into
copper plumbing pitting has indicated that some NOM may actually alleviate the propensity of a
water to cause pitting attack, and possibly alter some scale formation characteristics of uniform
copper corrosion.64 Any effect on cuprosolvency will likely be stronger in untreated surface water
supplies than in ground waters having very low TOC, or historically coagulated and filtered
surface waters.65-66
In many potable water systems, the behavior of lead is complicated by its interaction with
organic ligands, or colloidal material, or both.67 Some other studies have found lead associated
with flakes of nonadherent corrosion products related to iron and organic material.68
Samuels and Meranger69 found concentrated "fulvic acid" solutions of pH 6.2 to be
aggressive toward lead in solder and brass, and Thresh70 believed that organic material in the
water could prevent the deposition of films on lead surfaces by enhancing the solubility of lead.
Miles found that the organic content of moorland waters appeared to accelerate the initial attack
on the lead surface.71 Harrison and Laxen67 analyzed water from three localities in the United
Kingdom and found that significant quantities of dissolved lead (over 50 percent) in some samples
appeared to be found in organic complexes. Therefore, simple solubility computations that
-------
include only inorganic species may underestimate the levels of dissolved, lead attainable in a
distribution system. Including the problem with associations with colloids or nonadherent scales,
even though the inorganic dissolved lead equilibria may be accurately predicted given enough
information, the accurate prediction of observed levels of total lead may not be possible.
In contrast to NOM species that may enhance metal solubility, organic material that can
adsorb or otherwise adhere to the inside surfaces of the pipe could serve to decrease lead
solubility. Theoretical interpretations are hindered by the lack of characterization of the numerous
types of organic materials involved. Some simple experiments by Sheiham and Jackson41 on two
waters with varying degrees of natural color suggested that only highly colored waters might have
a significant impact on lead corrosion. They emphasized that this conclusion must be treated with
extreme caution because only two waters' were tested and because the organic component
characteristics might vary from place to place. Campbell has unambiguously expressed the
opinion that at least some NOM species provide an effective natural inhibitor for copper
dissolution,65'66 and that flocculation can remove the protective substance.72
In drinking water systems, the presence of extensive amounts of metal piping as a source
creates a considerably different environment than natural water or aquifer systems where
NOM/copper complexation has been studied most extensively. In addition to pH and ionic
strength, which have been widely acknowledged to be important in complexation studies with
NOM, consideration must also be given to the considerable competitive role that cupric carbonate,
hydroxide, and hydroxide/carbonate complexes must play, particularly as the pH increases above
7. Further, depending on source water and disinfection conditions, the redox potential of drinking
waters can vary over a wider range than natural systems, and a variety and quantity of solids may
exist that are also not present in ground waters, lakes and sea water.
The behavior of organic ligands can be modified by their association with or adsorption
on particle surfaces, and some natural ligands, such as fulvic acid, may enhance metal removal
by precipitation.73'74 NOM species interactions with surfaces and dissolved metal ions are greatly
affected by the concentrations of other metals and ligands in the ionic medium, with their own
competitive complexation and surface-binding effects, and the relative rates of reactions of the
NOM ligand species with trace metals as opposed to major ions such as calcium.74'75 Aluminum
in conjunction with sulfate may be an important agent in assisting particle formation and possible
surface adsorption of NOM.76 Adsorption films on actual pipe surfaces may take dozens of years
to form, so they are not amenable to laboratory study. Studies are further complicated by the fact
that the natural ligand or adsorbing molecules are difficult to identify and quantify, and may be
altered by isolation from their natural water chemistry matrix.
Conclusions
A number potential secondary impacts on lead and copper corrosion have been identified.
Specific negative impacts on lead and copper corrosion include: (1) Possibly reduced protection
from NOM and Al-containing diffusion barrier or passivation films, (2) Increased corrosion
control costs from additional pH adjustment and possible need for corrosion inhibitor additions,
and (3) Increased sulfate levels causing increased cuprosolvency under some pH/DIC conditions.
In this study, the use of ferric chloride or other iron-based coagulants was not investigated
in similar detail to alum. A general prediction about its effect is that chloride effects on copper(II)
or lead(II) corrosion, at the likely chloride levels, should be insignificant.
The removal of NOM and absence of aluminum will cause the potential for passivation or
-------
diffusion barrier film destabilization as noted above. The long-term practical impact of this
potential is unknown, as it depends on the unknown tendency for reversibility of the different
existing pipe deposits in the distribution systems.
In contrast to the generally negative anticipated impacts of enhanced coagulation there are
two anticipated benefits. First, if a water supply has particular NOM species that are effective
complexants for lead and copper, reduced metal levels might ultimately result after complete
implementation of the combination of coagulation changes and corrosion control. Second, there
may be reduced aluminum residuals in the distribution system, if those levels are of concern for
reasons of health, turbidity control, or possible hot water pitting enhancement.
Acknowledgments
The authors thank the managers of East Fork Lake State Park for their cooperation in
providing water and information for the pilot plant investigation. They also acknowledge the
University of Cincinnati undergraduate and graduate students, USEPA co-op students, technicians,
and chemists, and DYNCORP/TAI contract staff for their contributions to pilot plant operation,
data collection, data analysis, and report preparation. This work was funded through a
combination of USEPA in-house funds and cooperative agreement CR 816700 between USEPA's
Drinking Water Research Division and the University of Cincinnati.
Disclaimers
Mention of commercial names does not constitute endorsement or recommendation by the
agency. The views expressed in this paper are those of the authors, and do not necessarily reflect
USEPA policy.
References
1. Pontius, F. W. "D/DBP Rule to Set Tight Standards". Journal of the American Water Works
Association, 85:11:22 (1993).
2. Randtke, S. J. A Comprehensive Assessment of DBP Precursor Removal by Enhanced
Coagulation and Softening. AWWA Annual Conference, New York, NY, June 19-23, 1994.
3,. Qasim, S. R. "TOC Removal by Coagulation and Softening". Journal of Environmental
Engineering, 118:3:432(1992).
4. Randtke, S. J. "Organic Carbon Removal by Coagulation and Related Process Combinations"
Jour..AWWA, 80:5:40 (1988).
5. Tryby, M. E., Miltner, R. J. & Summers, R. S. TOC Removal as a Predictor of DBP Control with
Enhanced Coagulation. Proc. AWWA Water Qualtity Technology Conference, Miami, FL,
November 7-11, 1993.
6. Miltner, R. J., et al. The Control of DBPs by Enhanced Coagulation. Proc. AWWA Annual
Conference, New York, NY, June 19-23, 1994.
7. Lytle, D. A., et al. An Evaluation of the Secondary Effects of Enhanced Coagulation. Proc.
AWWA Specialty Conference on Enhanced Coagulation, Charleston, S.C., Dec. 5-6, 1994.
8. USEPA. "Draft Guidance Manual for Enhanced Coagulation and Enhanced Precipitative
Softening", Office of Ground Water and Drinking Water, Washington, D.C., 1993).
9. Qureshi, N. & Malmberg, R. H. "Reducing Aluminum Residuals in Finished Water". Journal of
the American Water Works Association, 77:10:101 (1985).
-------
10. Letterman, R. D. & Driscoll, C. T. "Survey of Residual Aluminum in Filtered Water". Journal
of the American Water Works Association, 80:4:154 (1988).
11. Simpson, A, M. "Aluminum, Its Use and Control in Potable Water". Environmental Technology
Letters, 9:907 (1988).
12. Jekel, M R. & Heinzmann, B. "Residual Aluminum in Drinking Water". AQUA, 38:281(1989).
13. Letterman, R. D. & Driscoll, C. T. Control of Residual Aluminum in Filtered Water. AWWA
Research Foundation, Denver, CO, (1994).
14. Schock, M. R. & George, G. K. Comparison of Methods for Determination of Dissolved
Inorganic Carbonate (DIG). Proc. AWWA Water Quality Technology Conference, Orlando,
FL, November 10-14, 1991.
15. AWWARF. Lead Control Strategies. AWWA Research Foundation and AWWA, Denver, CO,
(1990).
16. Gregory, R. & Jackson, P. J. Central Water Treatment to Reduce Lead Solubility. Proc. AWWA
Annual Conference, Dallas, TX, June 10-14, 1984.
17. Schock, M. R. "Understanding Corrosion Control Strategies for Lead". AWWA, 81:7:88 (1989).
18. Schock, M. R. "Internal Corrosion and Deposition Control". Ch. 17 In: Water Quality and
Treatment: A Handbook of Community Water Supplies. McGraw-Hill, Inc., New York,
(Fourth ed., 1990).
19. Schock, M. R., Lytle, D. A. & Clement, J. A. "Effect of pH, DIG, Orthophosphate and Sulfate
on Drinking Water Cuprosolvency", U. S. EPA Office of Research and Development,
Cincinnati, OH, EPA/600/R-95/085.1995).
20. Schock, M. R. & Lytle, D. A. The Importance of Stringent Control of DIG and pH in Laboratory
Corrosion Studies: Theory and Practice. Proc. AWWA Water Quality Technology
Conference, San Francisco, CA, November 6-10, 1994.
21. Lead and Copper. Final Rule. Fed. Reg. 56:110:26460 (June 7, 1991).
22. Lead and Copper. Final Rule Correction. Fed, Reg. 56:135:32112 (July 15, 1991).
23. Lead and Copper. Final Rule Correction. Fed. Reg. 57:125:28785 (June 29, 1992).
24. Paulson, A. J. & Kester, D. R. "Copper(II) Ion Hydrolysis in Aqueous Solution". J. Solution
Chem., 9:4:269 (1980).
25. Symes, J. L. & Kester, D. R. "Thermodynamic Stability Studies of the Basic Copper Carbonate
Mineral, Malachite". Geochim. Cosmochim. Acta, 48:2219 (1984).
26. Rickard, D. T. The Chemistry of Copper in Natural Aqueous Solutions. Acta Universitatis
Stockholmiensis, Contr. Geology 23:1:64 (1970).
27. Mann & Deutscher, R. L. "Solution Geochemistry of Copper in'Water Containing Carbonate,
Sulfate and Chloride Ions". Chem. GeoL, 19:253 (1977).
28. Lindsay, W. L. Chemical Equilibria in Soils. John Wiley & Sons, New York, New York, (1979).
29. Byrne, R. H. & Miller, W. L. "Copper(II) Carbonate Complexation in Seawater". Geochim.
Cosmochim. Acta, 49:1837 (1985).
30. Byrne, R. H., Kump, L. R. & Cantrell, K. J. "The Influence of Temperature and pH on Trace
Metal Speciation in Seawater". Marine Chemistry, 25:163 (1988).
31. Schock, M. R. & Gardels, M. C. "Plumbosolvency Reduction by High pH and Low Carbonate--
Solubility Relationships". Journal of the American Water Works Association, 75:2:87 (1983).
32. Schock, M. R. Treatment or Water Quality Adjustment to Attain MCL's in Metallic Potable
Water Plumbing Systems. Plumbing Materials and Drinking Water Quality: Proceedings
of a Seminar, Cincinnati, OH, May 16-17,1984, 1985.
-------
33. Edwards, M, Meyer, T. & Rehring, J. P. Effect of Various Anions on Copper Corrosion Rates.
Proc. AWWA Anrnial Conference, San Antdhio, TX, June 6-10, 1993.
34. Schock, M. R., Lytle, D. A. & Clement, J. A. Modeling Issues of Copper Solubility in Drinking
Water. Proc. ASCE National Conference on Environmental Engineering Boulder CO July
11-13,1994. '
35. Meyer, T. E. & Edwards, M. Effect of Alkalinity on Copper Corrosion. Proc. ASCE National
Conference on Environmental Engineering, Boulder, CO, July 11-13, 1994.
36. Rehring, J. P. The Effects of Inorganic Anions, Natural Organic Matter, and Water Treatment
Processes on Copper Corrosion. Master of Science, University of Colorado at Boulder
. : Boulder, Colorado (1994).
37. Schock, M. R., Lytle, D. A. & Clement, J. A. Effects of pH, Carbonate, Orthophosphate and
Redox Potential on Cuprosolvency. NACE Corrosion/95, Orlando, FL, March 26-31, 1995.
38. Dodrill, D. M. & Edwards, M. A General Framework for Corrosion Control. Proc. AWWA
Water Quality Technology Conference, San Francisco, CA, November 6-10, 1994.
39. Dodrill, D. M. & Edwards, M. "Corrosion Control on the Basis of Utility Experience". Journal
of the American Water Works Association, 87:7:In Press (1995). ..
40. Edwards, M., Meyer, T. E. & Schock, M. R. "Alkalinity, pH and Copper Corrosion By-Product
Release", submitted manuscript(1995).
41. Sheiham, I. & Jackson, P. J. "Scientific Basis for Control of Lead,in Drinking Water by Water
Treatment". Journal of 'the Institute• of'Water Engineer-sandScientists, 35:6:491 (1981).
42. Schock, M. R. & Wagner, I. "The Corrosion and Solubility of Lead in Drinking Water". Ch. 4
In: Internal Corrosion of Water Distribution Systems. AWWA Research Foundation/D VGW
Forschungsstelle, Denver, CO, (1985).
43. Wagner, D., Fischer, W. & Paradies, H. H. "Copper Deterioration in a Water Distribution System
of a County Hospital in Germany Caused by Microbially Influenced Corrosion-II. Simulation
of the Corrosion Process in Two Test Rigs Installed in this Hospital" Werkstoffe und
Korrosion, 43:496 (1992).
44. Rehring, J. P. & Edwards, M. The Effects of NOM and Coagulation on Copper Corrosion.
Proc. ASCE National Conference on Environmental Engineering, Boulder, CO, July 11-13,
1994. . . .
45. Beccaria, A. M., etal. "Corrosion of Lead in Sea Water". British Corrosion Journal 17'2'87
(1982). '
46. Lauer, W. C. & Lohman, S. R. Non-Calcium Carbonate Protective Film Lowers Lead Values.
Proc. AWWA Water Quality Technology Conference, San Francisco, CA, November 6-10
1994.
47. Fuge, R. & Perkins, W. "Aluminum and Heavy Metals in Potable Waters of teh North Ceredigion
Area, Mid-Wales". Environmental Geochemistry and Health, 13:2:56 (1991).
48. Paces, T. "Reversible Control of Aqueous Aluminum and Silica During the Irreversible Evolution
of Natural Waters". Geochimica et Cosmochimica Ada, 42:1487 (1978).
49. Fanner, V. C., Fraser, A. R. & Tait, J. M. "Characterization of the Chemical Structures of Natural
and Synthetic Aluminosilicate Gels and Sols by Infrared Spectroscopy". Geochimica et
Cosmochimica Acta, 43:1417 (1979).
50. Neal, C., et al. "Aluminum Solubility Controls in Acid Waters: the Need for a Reappraisal".
Earth and Planetary Science Letters, 86:105 (1987).
-------
51. Neal, C. & Williams, R. J. "Towards Establishing Aluminum Hydroxy Silicate Solubility
Relationships for Natural Waters". Journal of Hydrology, 97:347 (1988).
52. Farmer, V. C. & Lumsdon, D. G. "An Assessment of Complex Formation between Aluminum
and Silicic Acid in Acidic Solutions". Geochimica et Cosmochimica Acta, 58:16:3331
(1994).
53. Tunturi, P. & Ylasaarl, S. A Special Case of the Pitting Corrosion of Copper in a Hot Water
System. Proc. 5th Scandanavian Corrosion Conference, Copenhagen, 1968.
54. Wilson, D. E. "An Equilibrium Model Describing the Influnce of Humic Materials on the
Speciation of Cu2*, Zn2+, and Mn2+ in Freshwaters". Limnol Oceanogr., 23:3:499 (1978).
55. Cabaniss, S. E. & Shuman, M. S. "Copper Binding by Dissolved Organic Matter: I. Suwannee
River Fulvic Acid Equilibria". Geochimica Cosmochimica Acta, 52:185 (1988).
56. Cabaniss, S. E. & Shuman, M. S. "Copper Binding by Dissolved Organic Matter: II. Variation
= in Type and Source of Organic Matter". Geochimica Cosmochimica Acta, 52:195 (1988).
57. Davis, J. A. "Complexation of Trace Metals by Adsorbed Organic Matter". Geochimica
Cosmochimica Acta, 48:679 (1984).
58. Hirose, K. "Chemical Speciation of Trace Metals in Seawater: Implication of Paniculate Trace
Metals". Marine Chemistry, 28:267 (1990).
59. Holm, T. R. "Copper Complexation by Natural Organic Matter in Contaminated and
Uncontaminated Ground Water". Chemical Speciation andBioavailability, 2:63 (1990).
60. Holm, T. R. & Curtiss, C. D., III. Copper Complexation by Natural Organic Matter in Ground
Water. Chemical Modeling of Aqueous Systems II, Los Angeles, CA, 1990.
61. Fisher, N. S. & Fabris, J. G. "Complexation of Cu, Zn, and Cd by Metabolites Excreted from
Marine Diatoms". Marine Chemistry, 11:245(1982).
62. Zhou, X., etal. "Production of Copper- Complexing Organic Ligands During a Diatom Bloom:
Tower Tank and Batch-Culture Experiments". Marine Chemistry, 27:19 (1989).
63. Brown, F. R. & Fernando, Q. "Kinetics of the Dissolution of Copper Metal in Aqueous Solutions
Containing Unsaturated Organic Ligands and Copper(II)". Talanta, 38:3:309 (1991).
64. Edwards, M., Ferguson, J. F. & Reiber, S. H. "On the Pitting corrosion of Copper". Journal of
the American Water Works Association, 86:7:74 (1994).
65. Campbell, H. S. "Corrosion, Water Composition and Water Treatment". Wat. Treat. Exam.,
20:1:11(1971).
66. Campbell, H. S. & Turner, M. E. D. "The Influence of Trace Organics on Scale Formation and
Corrosion". Journal of the Institute of Water Engineers and Scientists, 37:1:55 (1983).
67. Harrison, R. M. & Laxen, D. P. H. "Physicochemical Speciation of Lead in Drinking Water".
Nature, 286:5775:791 (1980).
68. AWWARF. Internal Corrosion of Water Distribution Systems. AWWA Research
Foundation/DVGW Forschungsstelle, Denver, CO, (Second Ed. ed., 1995).
69. Samuels, E. R. & Meranger, J. C. "Preliminary Studies on the Leaching of Some Trace Metals
from Kitchen Faucets". Water Research, 18:1:75 (1984).
70. Thresh, J. C. "Action of Natural Waters on Lead". Analyst, XLVTI:560:459 (1922).
71. Miles, G. "Action of Natural Waters on Lead". Journal of the Society of Chemical Industry,
67:1:10(1948).
72. Campbell, H. S. "The Influence of the Composition of Supply Waters, and Especially of Traces
of Natural Inhibitor, on Pitting Corrosion of Copper Water Pipes". Proc. Soc. Water Treat.
&Exam., 8': 100 (1954).
-------
73. Saar, R. A. & Weber, J. H. "Lead(II)-Fulvic Acid Complexes. Conditional Stability Constants,
Solubility, and Implications for Lead(n) Mobility". Environmental Science and Technology
14:7:877(1980).
74. Dalang, F., Buffle, J. & Haerdl, W. "Study of the Influence of Fulvic Substances on the
Adsorption of Copper(II) Ions at the Kaolmite Surface". Environ. Sci. Technol, 18:3:135
(1984).
75. Hering, J. G. & Morel, F. M. M. "Slow Coordination Reactions in Seawater". Geochimica
Cosmochimica Acta, 53:611 (1989).
76. Snodgrass, W. J., Clark, M. M. & O'Melia, C. R. "Particle Formation and Growth in Dilute
Aluminum(III) Solutions". Water Research, 18:4:479(1984).
-------
c
1
6
5
S
T* 3
•la
••S
•£
o
trt
a.
CQ
Q
g|
U
H
"to
-X
"5
ft> G
•c: G
^ «
•S
««
2
s
« *
•K '
3>
"S. e
S g-2
S? •§ n
•o 8 "S
c o &!
*s . o
« Q
o
&>
S
a
&
C.
u
/*
i" r
XX
X XXX
X XX
X XXX
X XXX
X XXX
X XXX
X XXX
X XXXX
X XX
"
XX XXXX
•o
CO _M fl^
••3 ^ ^j Jj a
« ° 13 ^ JS "2
C CM CQ ^3 "y Si
fe o o o o o "g
?>>>>> s
c2 (3 6 6 6 c3 o"
-
-------
OO — < (N •<*• o
28 -: 2 o "»
*! rr-. "H — t o\
O 0 < \0 O
• o
5
5
op! 82
O '• O *~"
V
o \q £i -* o
o pJ S 2 ^
00 —
O O\ \O <5 • O\ T—i
Sj ^ c4 o °i ^
32
o\
ov
o
i^go§
f^i C3
eo
U '
c3
CO
S
-
.S |
ll §
< CJ 2
-S
8 E
II
S o
S
af
en
O
TJ
W _
e 13
oj
.S IH
•^s C C
o & < O O 33 5 P -g.^ 5
1-S | -a £
CQ '^g J2 *•* 2J
-------
EC
O
a>
2
o
J5
*><
'i
"S
CO
"S
d>
tn
•8
CJ
O
S
I
S
"o
•g
-------
0.7
0.6
0.5
£ 0.4
c
1 0.3
0.2
0.1
0.0
-
-
8.2
i
7.4
i
7.5
8.0
6.6
i
6.9 7.8
8.0
8.1
Raw CS Fl CW1 OS F2 CW2 F3' CW3
Location
Figure 2. Aluminum residual at stages of pilot plant operation using
East Fprk Lake water (CS- conventional settled, CW- clearwell, OS-
optimized settled, F- filter). The average PH values are shown above bar
-------
0.4
0.3
0.2
0.1
0.0
90
6.0 6.5 7.0 7.5 8.0 8.5
pH
Figure 3. Residual aluminum after filtration as a
function of pH in the filter influent. Aluminum concentrations
represent all filters.
-------
W>
£
100.0000
10.0000
1.0000
0.1000
0.0100
0.0010
L /
/
pH=6.0
pH=7.0
pH=8.0
—— pH=9.0
• pH=iO.O
0 25
50 75 100 125
mgC/LDIC
150
Figure 4. Solubility of copper(II) in response to pH and DIG
assuming equilibrium with Cu(OH)2 solid at 25° C and 1=0.02.
-------
100.0000
10.0000
1.0000
0.1000
0.0100
0.0010
0.0001
pH = 6.0
-- pH = 7.0
--- pH = 8.0
---- pH = 9.0
pH=10.0
/
/ ..-••
I?. .-••'
0
25
50 75 100
mg C/L DIG
125 150
Figure 5. Solubility of copper(TT) in response to pH and DIG, assuming equilibrium with
Cu2(OH)2CO3 and CuO solid at 25° C and 1=0.02. This would represent a case of aged plumbing.
-------
100.0000
10.0000
pH = 6.0
-- - pH = 8.0
---- pH = 9.0
= 10.0
W)
£
1.0000 r
0.1000 r
0.0100 ' » « « t t
i . . . . t
0
25
50 75 100 125 150
mg C/L DIG
Figure 6. Solubility of lead(II) in response to pH and DIG, assuming equilibrium with
PbC03 or Pb3(OH)2(C03)2 solids at 25° C and 1=0.02.
-------
10.0000
W)
S
1.0000
0.1000
0.0100
0.0010
Theoretical Solubility
— Run 1 (DIC=14,pH=8.5)
- Run 2 (DIC=13,pH=7.0)
Run 5 (DIC=18, pH=7.5)
r • , , n ,
Experimental Data
Run 1
Run 2
O
« *
8
pH
10
Figure 7. Comparison of equilibrium theoretical and observed 24-hour stagnation copper levels for
coupon study, showing large deviation from expected value for system with high pH and high sulfate
Thick line is predicted "langite" solubility.
-------
1.1; AY
:tt-'H
KIV
Journal of the Institution of
WATER AND
ENVIRONMENTAL
MANAGEMENT
Vol &No 3
-------
bby ,'172
P. 13
If)
O
r
.A
^
O
>
_ r
-JO £
P
r
1
0 O. t— t- ct B
ftilS M O 3" C
•O O O 0 >1
fT >- 3 "O >-
CTO .Q S
g 3 ° »«•«•
5 ai jr o
I 3 3 C 0
1— ft _.«•"•
O H rt Q*
& O (S %
g X-o r»3
3 C, «• p. "O
C 81 H* w
IT rr D" 3 3 O
w O to Q
ire taken out from
:alled in a closed
:orrosion rate and
ermined by oxygen
latizod monitoring
the closed loop.
Because of the lack of sufficient
the connections between different
corrosion inhibition effects of the
tests have to be carried out to
for optimization.
Test methods
To obtain correct data, tests have
where the original water quality i
H/4 ) and pipe coupons are install
as a throughflow system with the
inhibitor dosage at a flow, veloc
(tutbulent flow) for several months
Ss^s
^ H* 8>
*> * < a-
o j~5 £.«
M» 3 w* ••"
a t-tj
'i:*l*
" » 2 • 8
i°sJ
3 n a,a a.
0 ® C 1* 0
- o "> 3
tJ^C-D
01 >— OJ
3 rr f-13 H-
\ 3" •-• ft rr
B (8 >< B (S
l&^
ESSS
S «a H-
c informations on
qualities and the
e chemicals, field
i the needed data
The use of inhibitor chemicals i
and potassium salts of phosphoric
additionally limited to 2,2 ppm P
cases an optimazation of the toti
dosage chemicals is wishable with
costs and a minimization of phosphi
to reduce ecological impact, mainl
tion of lakes and rivers.
2 5 * o "
8 ™ "2 = o- ?• n
2 ™ 0 0»»
= ja o. o >- a
S g g-ogJl
SSS55&S
s"|^s&
?•"«"?-
^s^:°
•o o »> ir-1 s o
o -o n£ B o
•- O 3 ^ 3 °- 0.
l— to IB B> 1—
c a s o s H- e
l D* ri- »^K; w 3
On the other hand, red water pro
such systems and are the reason
customers (1-3). Beside that, it
water is limited in Germany to 0,2
where problems arise, the use of i
to reduce or even resolve the pr
range while relining or replacement
o o o er
& O~ 3"O 3 rr, 1—
. T !-• H'^ O m
(t o O^ 3 n 3
31— a IB
^- m ri" T O •>
often arise from
complaints of the
ntent in drinking
e. In theee cases,
ors can be helpful
i in a short time
ong terra measures.
Introduction
Drinking water distribution syst
the older one, are often made fi
ductile iron without a sufficient
tection. Despite that, damages caus
attack which leads to a perforatio
ly occuring and can be linked d
to unqualified pipe-laying with re
no sufficient flushing or unfavo
tions.
c to M. a 5 x »
•i >-* §.*"§ 3
o»a.roo3*co
C* c O rn rr rf
t-* n r* .2 c^ w
iBtn^-n-^^rf^"-
^ 5 i~= IB S
o--* <*$?.«.
Us?.R»r5
2s- -o 3 o o 3
f* »~ 2 ° n ° f
>--rrS-Up-gw =
§s ^oi^^
n> a- n o J-" ^
» ° •< 2 IT
o •o -a n s 3 3
o K-'O no 2 b
3*DIBC»W'n-3h--
O. 15 3 ••( W- K 3
>— tnuroooo^-
i - - i 3 i n *<
-------
VB>
\ UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
NATIONAL RISK MANAGEMENT RESEARCH LABORATORY
CINCINNATI. OH 45268
December 19, 1996
OFFICE OF
RESEARCH AND DEVELOPMENT
MEMORANDUM . . '
SUBJECT: Seasonal Monitoring Revision
FROM: Michael R. Schock, ChemistfM£
Treatment Technology Evaluation Branch
Water Supply and Water Resources Division
TO: Jeffrey B. Keropic, Environmental Engineer
Office of Ground Water and Drinking Water
Pursuant to our prior correspondence and discussion about.
removing requirements that samples be collected in "warmest"
months/ I have collected a small set of papers for your
information and justification of the change. I agree that in
some water systems the highest lead, copper, or both levels could
be highest in the warmest months. However, I do not think that
the presumption that it is always the case is supportable by
scientific evidence. There are many cases where exactly the
opposite is true, as evidenced by both fundamental chemistry
arguments as well as data from experiments and water distribution
system sampling. Therefore, when possible, I believe that the
Lead and Copper Rule should be revised to allow sampling
flexibility to account for this phenomenon. Of course, for water
systems required to collect two or more rounds of monitoring data
that can encompass "seasonality," there is no reason to make any
change in -requirements.
Language changes may be required in some places to modify
statements that would imply guidance towards biasing of
collection times to the "warmest" months. In such cases, phrases
could simply be amended to suggest biasing towards seasons or
months "where the highest levels are most likely to occur."
For seasonally-operated water systems, a secondary
supplemental argument is that the intent of the Rule was to
reduce major exposures of sensitive populations to lead and
copper. Therefore, sampling is most appropriate when the systems
are actually being used. The question then becomes when to
sample if operation encompasses a range of months covering a
Recycled/Recyclable • Printed wBh Vegetable OI Based Inks on 100% Recycled Paper (40% Postconsumer)
-------
>
range of temperatures. There is no definitive argument on this,
because currently we can rarely predict beforehand whether a
given water chemistry coupled with physical factors, will cause
the highest values at a particular time. Therefore, I do not
think we should attempt to be too precise in proscribing exactly
when to take the samples in a confining regulatory structure.
As noted previously, I think the strongest line of argument
we should use is to say that the uncertainty of the time of
^worst" metal levels causes EPA to refrain from specifying
particular months for sampling across the country. If monitoring
data from similar systems or prior monitoring or survey
experience in that particular system is available to the States,
and if the States wish to use that information to make particular
time requirements for the systems, then I think that would be
reasonable.
There are several plausible ^theories" about why metal
levels could frequently be higher in the "winter" months, each or
some combination of which could operate in a given water system.
Some of these are:
• The intrinsic net solubility of many minerals, especially
carbonates, goes up as the temperature goes down.
• Corrosion inhibitors, especially orthophosphate, may react
more slowly at lower temperature, so passivating film
formation is less effective in colder water.
• Corrosion inhibitors and other treatment chemicals may be
more viscous at lower temperatures. Therefore, the chemical
feed rates may be lower when cold.
• Many pipes are near heating systems, and in the winter the
operation of the heating systems causes the pipes to be
hotter. Plus, the change in temperature could, also disrupt
the existing protective films in the pipes built up over the
earlier months of more stable temperatures (this is an
argument advanced by several staff members of Minnesota
DOH) .
• Dissolved oxygen levels are often higher in colder waters,
resulting in enhanced concentrations of primary oxidants in
the water (when added to chlorine species). This causes
more rapid increases in metal levels through enhanced
oxidation during short standing-times (6-16 hours).
The information I have attached is as follows, with brief
descriptions of their significance.
-------
i^.. Colling, et. al. Jour. JWEW (1992) paper showing field
data where lower lead levels are obtained at higher temperatures
in some orthophosphate-dosed hard water systems in the UK (page
262, Fig.. 3) .
-2. Edwards/Schock/Meyer Jour. AWNA (1996) paper showing
theoretical arguments why copper levels should be higher at lower
temperatures in many cases. This paper cites some data from
Boulder {not published) and utility data (see next one) .
^3. Dodrill/Edwards Jour. AWWA (1995) paper (pp. 79-80) showing
temperature trends not readily apparent in major utility
monitoring data. There was a suggestion of a trend of lower
levels at higher temperatures especially for lower alkalinities.
This paper brings up another interesting point that because of
the different structures of houses and where the first-draw liter
of water res-ides in the plumbing, major temperature trends should
not be expected. Some unpublished work Chet Neff and I did many
years ago (which could be reproduced with the assistance of some
standard heat-transfer equations) suggested that waters in lead
and copper pipes quickly (minutes to only an hour or two) take on
the temperatures of their surroundings.
^#. Rezania/Anderl conference paper (1996)on observations of
higher Cu levels in winter in Minnesota sampling (page 4) . They
have several arguments to explain this, with some in the paper,
and some noted above in my discussion.
5. Some supplemental laboratory test data by doctoral student
Loay Hidmi, University of Colorado, Dept. of Civil Engineering.
This shows higher copper release at pH 7 for lower temperature in
water of same alkalinity. Compare Cu level at most times from
Chart 1 to Chart 4. This may not be citeable.
6. Some data from a couple of water systems in MN collected by
U. Colorado. Note highest Cu levels are usually associated with
"winter" month sampling. (Probably not citeable)
7. Some other experimental data from Marc Edwards7 group at U..
Colorado (currently confidential but possibly citeable with
permission from Marc) showing soluble copper being higher in cold
water loops than in hot water loops.
Interestingly, a review of the original preamble (Page
26524-26525) shows that EPA had information showing field data
was equivocal about any particular month or period having higher
or lower metal levels. The choice was apparently made to
essentially "weight" the evidence from some unspecified studies
more highly than others, and the proposal to use July through
September was kept (after adding June). I believe, therefore,
-------
that this proposal is not really particularly new, but does help
argue that the increased inconvenience of requiring sampling in
specific months does not pay off in tangible public health
protection. It also provides an option for States that have good
reason to believe a particular time period would be preferable
for sampling to use their discretion to designate such a time
constraint.
If you need additional research to find other articles, or
if you would like me to help with some preamble language, let me
know.
Attachments
cc: Judy Lebowich
V
------- |