United States
                    Environmental Protection
                    Agency
Water Engineering Research
Laboratory
Cincinnati OH 45268
                    Research and Development
&ERA         Project  Summary
 EPA/600/S2-85/079  Aug. 1985	™  *
 	Vyf »v
                    Corrosion and Calcium
                    Carbonate  Saturation  Index in
                    Water  Distribution Systems
                    J. Edward Singley, Rodolfo A. Pisigan, Jr., Abdul Ahmadi, Portia 0. Pisigan, and
                    Ting-Ye Lee
                      Corrosion in water distribution sys-
                    tems was studied to gain a better
                    understanding of the processes needed
                    in developing control strategies. Equa-
                    tions and calculation methods for de-
                    termining the pH, (or the theoretical pH
                    at which the water is saturated with
                    CaCO3) were developed using a chemi-
                    cal model with and without ionic speci-
                    ation. Several calculation procedures
                    for determining pH, were analyzed and
                    compared. Laboratory experiments
                    using weight loss and electronic tech-
                    niques for measuring corrosion rates
                    were conducted in laboratory batch and
                    continuously circulating systems using
                    waters with different saturation indices
                    (SI), pH, alkalinity, chlorides, sulfates,
                    chlorine  residual,  dissolved oxygen,
                    organic color, calcium, buffer capacity,
                    and other  water quality  parameters.
                    Changes in water quality were moni-
                    tored in experiments with mild steel,
                    galvanized steel, and copper coupons.
                    Corrosion products obtained from cou-
                    pons exposed to continuously circulated
                    solution were analyzed by wet chemical
                    techniques and X-ray diffraction. In-
                    creases in dissolved oxygen, chlorides,
                    sulfates. chlorine residual, and hydrogen
                    ion concentration accelerated corrosion
                    of metal specimens. Higher pH, calcium
                    content, buffer capacity, and organic
                    color decreased the corrosion rate. A
                    model was formulated that expresses
                    corrosion rate in terms of chlorides,
                    sulfates. alkalinity, dissolved oxygen,
                    calcium,  buffer capacity, saturation
                    index, and immersion time. A four-
                    variable corrosion rate model was de-
veloped from the multivariate analysis
of corrosion data for mild steel in the
batch studies. The model included total
dissolved solids (TDS), dissolved oxy-
gen, saturation index, and exposure
time. Results indicate saturation index
cannot be used alone to indicate the
extent of corrosion or to measure the
corrosivity of a water.
  This Project Summary was developed
by EPA's Water Engineering Research
Laboratory, Cincinnati, OH, to announce
key findings of the research project that
is fully documented in a separate report
of the same title (see Project Report
ordering information at back).
Introduction
  Corrosion of pipes in community water
supply systems has been recognized as a
serious problem since the beginning of
the 20th century. Economic estimates of
corrosion costs in the water industry have
increased from $40 million in 1947 to as
high as $700 million in 1980. In addition
to the economic consequences,  as the
metallic structure deteriorates, drinking
water quality is degraded. The resulting
corrosion products also  pose potential
health  risks.  Consequently, the U.S.
Environmental Protection Agency (EPA)
has amended the National Interim Pri-
mary Drinking  Water Regulations to
require that public water utilities carry
out a corrosion  control  program. This
program includes identifying the presence
and sources of corrosion products  and
subsequently implementing effective
corrosion control measures.

-------
  CaCOa deposition on the metal surface
has been one of the better control strate-
gies for combatting corrosion. Using the
parameter pH8 (or the  pH of CaCO3
saturation) is  commonly employed  to
ensure that the processed or treated
water has sufficient protective property.
Because most of  the pHs calculation
methods are not accurate  enough and
provide only an approximate value, more
precise methods are needed for deter-
mining pH8. Likewise, further studies are
still needed to assess the effects  of
important water quality parameters on
corrosion of metals used in the distribu-
tion systems.

Research Objectives
  This investigation was conducted:

 (a) to derive a more accurate theoretical
    expression for  evaluating pHs;
 (b) to make a comparative analysis  of
    other methods for calculating pHB;
 (c) to develop a simplified method for
    accurately predicting  the exact pH,;
 (d) to evaluate the corrosive behaviors of
    waters with  different saturation
    indices;
 (e) to study the effects of several water
    quality factors  on the  corrosion rates
    of mild steel, black steel, galvanized
    steel, and  copper; and
 (f) to  analyze the corrosion  data and
    develop empirical and statistical cor-
    rosion rate models.

Experimental Section
  Different synthetic waters with differ-
ent saturation  indices were prepared  by
mixing appropriate quantities of the re-
quired reagents. A  24-hour  equilibration
test in which analytical reagent CaCO3
was added to each water determined the
actual saturation state with respect  to
CaCOa. The changes  in  pH, Ca,  and
alkalinity were determined before and
after the equilibration period.
  Two sets of corrosion  evaluations  at
room  temperature  (20+3°C) were  con-
ducted: jar tests and loop studies. In the
jar  tests (non-flowing system), several
cleaned metal  coupons were suspended
and immersed by  nylon  cords in the
plastic jars containing water with differ-
ent chemical compositions and saturation
indices. The jars  were  placed over  a
stirring apparatus (Figure 1), which can
be used simultaneously for aerated and
deaerated systems. The weight loss was
used  to calculate the corrosion rate:
                          - Compressed Air -

Figure 1.    Stir ing apparatus (D = deionized water; F = glass wool filter; J = jars).
before and after a specified immersion
period, the oven  dried weight  of each
coupon was determined. In addition to
the weight-loss method, the corrosion
rate in some experiments was electron-
ically monitored at specific time intervals
using a Magna Corrater Model 1120,
which employed three-electrode probes.
  Figure  2 shows the loop set-up con-
structed with  nonmetallic  materials to
study corrosion of metals in contact with
circulating water  of different saturation
indices. A Manostat Varistaltic* pump
(advanced model) was connected to the
water reservoir to feed the pipe sections
and these regulate the water flow. Bypass
tubes were provided to prevent disrupting
the entire system operation when remov-
ing metal specimens for examination and
analysis.  Each  PVC tube section in  the
loop system held four metal inserts. The
corrosion  rate  of the exposed  internal
surface of each pipe coupon was deter-
mined  by the weight-loss method. After
each corrosion  test, the loop set-up was
flushed with  acidified water and then
rinsed  twice  with deionized water  to
remove any adhering corrosion products.
An aliquot of the last wash water was
analyzed for the presence of metal ions to
ensure that no metallic substances still
remained in the system.

  Dried corrosion products obtained from
loop studies of mild steel, black steel, and
galvanized steel were analyzed by wet
chemical methods and X-ray diffraction
analysis. X-ray diffractograms of standard
compounds were developed simultane-
ously during the sample analysis.
Results

Theoretical A spects
  After presenting and reviewing several
CaCOaSaturation concepts and corrosivity
indices, two alternative  calculation me-
thods for pH8 were derived. The first
equation was a quadratic pHs  formula
(QUAPHS)  developed from a chemical
model of CaCO3 solubility equilibrium in
which several ionic association equilibria
were considered.  The final form of the
equation for hydrogen activity at CaC03
saturation was a function of the first and
second thermodynamic dissociation con-
stants of H2CO3 (Ki and K2), thermody-
namic CaCOa solubility product constant
(Ks), free calcium ion concentration, total
equilibrium carbonate concentration, and
activity  coefficients of Ca+2, COa2, and
HCOa.
  A computer program, constructed dur-
ing the study, calculates QUAPHS through
iterative calculation procedure whereby
values of the variables, except equilibrium
constants at a given temperature, initially
are given approximate values. Solving for
pHs  continues  until both sides of the
equation are almost numerically equal.
  Another  simplified pHs (SIMPHS) ex-
pression similar to Langelier's equation
was derived  from equilibrium relation-
ships in which ion pairing was neglected
and a new salinity correction factor was
added. The resulting pHs equation was
SIMPHS = -log
      [Ca+2]
Z.5S/T
                        + 3.63   I
'Mention of commercial products or trade names
 does not constitute endorsement or recommenda-
 tion for use.
               1.0+3.30N/T+2.61  I

where: [Ca+2] = total calcium concentra-
               tion, moles/L

-------
                                       '/2-Inch Pipe Section
      Plexiglas           -.-
        Cylinder (D**"  M-1*
                                   :? ,	,  *x ,	, xx
                                1-x-L-Lx-l      l-x-LJ-x-T      fx1—lx-f
                                        3A-lnch Pipe Section
                                       / -/nc/7 P/pe Section
               C) Pomp
                Water
                Reservoir
 Figure 2.    Set-up of the loop system for evaluating metal pipe corrosion in circulating water.
         (Alk) = total alkalinity, equiv/L
            I = ionic strength, moles/L

 Several statistical predictive  pHs equa-
 tions were analyzed  and compared with
 the equations derived in this study. Two
 predictive equations, one containing 4
 variables  (STA4PHS) and the other, 5
 variables  (STA5PHS), were chosen. A
 comparative analysis was done  among
 the two statistically derived  predictive
 equations and among other calculation
 methods such as the WATSPEC2 com-
 puter program (WSP2PHS), Langelier's
 formula (LANGPHS), Larson-Buswell
 equation (LARBPHS), and  graphical ap-
 proximation  of pHB using Caldwell-
 Lawrencediagrams(CDLWPHS).ThepH8
 values of the 155 water  samples (see
 Table 1) were calculated using each
 calculation  procedure.  When the pHa
 values obtained via computer calculation
 of QUAPHS were compared with those
 derived  from other methods, pHs values
 for QUAPHS, SIMPHS, WSP2PHS,
 STA4PHS, and STA5PHS  are generally
 close to one  another.  LANGPHS and
 LARBPHS are comparatively lower than
the pHs values obtained from QUAPHS
 and the first four aforementioned calcu-
 lation methods. The  highest values are
those derived from the approximations in
Caldwell-Lawrence diagrams.
Table 1.    Mean Values and Ranges of the
           Analytical Parameters for  155
           Samples of Florida Water
Parameter*
Mean
Range
TDS
Ca
Mg
Na
Cl
sot
Fe
F
Har
-------
time) yielded a satisfactory graphical fit to
the experimental data.  Multivariate  re-
gression analysis of the correlation be-
tween the corrosion rate of mild steel and
various water  quality parameters pre-
sented in Table 2  revealed that the
corrosion rate  (CR4) is best predicted.
                              under the experimental conditions em-
                              ployed in the present study, by
                                 CR4 =
                                           (TDS)0.253(DO)0.820
                                        (10s')00876 (time, day)0373

                               The above predictive equation appears
I
I






^





T
1







T
1







T
_L





-

-

0.37
0.18^
"
2
X
J
0.12

0.06
n
Water A Water B Water D Water E Water F
SI = -0.72 SI = -0.48 SI = +0.04 SI = +0.57 SI = +0.94
to work well in waters with medium to
high DO level (5.0-9.0 mg/L). Represen-
tative plots for comparing the observed
and CR4- predicted values in three waters
are shown in Figure 6. The experimental
corrosion rates follow a decreasing trend
with time as do the corrosion-time curves
generated by the predictive equation.

Conclusion
  Two equations for determining pH,
were developed. One was a  quadratic
expression derived from a CaCOa solubil-
ity equilibrium model, which incorporated
interaction among the major  ions. The
other, derived without taking into account
ion  pairing, was similar in  form to
Langelier's equation but with a new ionic
strength correction factor. A simpler
method of computing pH9 was found in
which pHs  is defined by a  statistical
predictive equation composed of a linear
combination of the logarithmic values of
total dissolved solids, total alkalinity, and
analytical concentrations  of calcium,
magnesium, and sulfate. Although a more
accurate  method of determining the
CaCO3 saturation index was found, the
traditional interpretation of the indices
was not consistent with the experimental
corrosion data obtained from the jar and
loop studies. Analysis of X-ray diffracto-
grams of the corrosion products sugges-
ted that CaCOa would still precipitate in
waters with a negative saturation index
while in contact with a corroding metal
surface. Several water quality parameters
influenced the corrosion of metals used
in the study and would have to be taken
into  account when  assessing corrosive
behaviors of waters. An empirically con-
structed, eight-variable model indicates
that the corrosion rate of mild steel under
the experimental conditions used  is af-
fected not only by the water's tendency to
precipitate or dissolve CaCOa, but also by
dissolved  oxygen, chloride, sulfate, cal-
cium, alkalinity, buffer capacity and im-
mersion time. A four-variable model
derived from multivariate  regression
analysis defines corrosion  rate  as a
function  of  four parameters:  total dis-
solved solids; dissolved oxygen; saturation
index; and exposure time.
  The full report was submitted in fulfill-
ment of Grant No. R805400-01 -1 by the
University of Florida, under the sponsor-
ship of the U.S. Environmental Protection
Agency.
Figure 4.   Bar graph showing the corrosion rates of black steel pipes after 21 days in the loop
           system (DO = 5.0 + 0.2 mg/L) using five waters of different saturation indices.

-------
  500
  410
  320
  250
$180
§730
|  80
   50
   20
    10
    70.OO
  5OO
  410
  320
  250
.$• 780
g 730
I  80
-  50
    20
    10
                   18.OO
i
              26.00
34.OO
                                                            42.00
                                                        50.00
     50.00
58.OO
66.OO         74.00
         20
              82.00
                                                                          90.00
Figure 5.   X-ray pattern of corrosion products from black steel pipes in contact with water B
            (SI = -0.46).
Table 2. Means and Ranges of Analytical Variables* for the
Batch Corrosion Studies
Parameter Mean
TDS
Ca
Mg
Na
Cl
S04
Alkalinity. mg/L as CaCO3
pfmg/LasCaCOa/pH)
pH, units
pH* units
Saturation Index, units
Ionic Strength. mole/L
Dissolved Oxygen
Corrosion Kate fmpy)
Corrosion Rate (mm/yr)
473
68
16
64
119
79
116
16
7.77
7.79
-O.02
O.O081
5.1
11.9
0.301
1 7 Waters Used in the Mild Steel
Range
116-791
10-12O
2-52
7-1 3O
18-213
16-2O8
21-232
3-42
6.70-8.95
7.60-9.07
-2. 12-+1.90
O.OO01-O.O198
1.4-9.1
2.1-36.6
0.053-0.930
"All values are expressed in mg/L except as indicated.

-------
  40
  30
   20
   10
oc
I
o
O
   30
   20
    10
            Water e
            S\=-2.12
            D0=5.0±0.2
            Water e
            S\ = -2.12
            DO = 9.0±0.2
                       t     \
                                       Water k
                                       SI = +0.07
                                       DO=5.0±0.2
                                       Water k
                                       SI = +0.07
                                       DO = 9.0±0.2
Water q
SI = +1.90
DO=5.0±0.2
Water q
SI = +1.90
DO = 9.0±0.2
      02    46    8    10    2468    10   24    6    3   10

                                         Days
Figure 6.    Predicted corrosion—time  curves  and experimental values (*J for  corrosion
             of mild steel in waters e, k. and q.
                                                                  .S. GOVERNMENT PRINTING OFFICE: 1985/559-111/20666     19519F

-------
     J. E. Singley, R. A. Pisigan, Jr.. A. Ahmadi. P. 0. Pisigan. and Ting-Ye Lee are with
       University of Florida. Gainesville. FL 32611.
     Marvin Gar dels is the EPA Project Officer (see below).
     The complete report, entitled "Corrosion and Calcium Carbonate Saturation Index
       in Water Distribution Systems," (Order No. PB 85-228 112/AS; Cost: $22.00,
       subject to change) will be available only from:
            National Technical Information Service
            5285 Port Royal Road
            Springfield, VA 22161
            Telephone: 703-487-4650
     The EPA Project Officer can be contacted at:
            Water Engineering Research Laboratory
            U.S. Environmental Protection Agency
            Cincinnati, OH 45268
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use 5300

EPA/600/S2-85/079
          QCOC329   PS

          U  S  ENVIR  PROTECTION  AGENCY
          REGION 5 LIBRARY
          230  S  DEARBORN  STREET
          CHICAGO               IL   60604

-------