ISWS

C-45
Loan c.3 Circular No. 45 1954

STATE OF ILLINOIS
WILLIAM G. STRATTON, Governor

By

T. E. LARSON and R. M. KING

Issued By

DEPARTMENT OF REGISTRATION AND EDUCATION

VERA M. BINKS, Director
State Water Survey Division
A. M. BUSWELL, Chief
Urbana, Illinois
 REPRINTED FROM AND COPYRIGHTED AS A PART OF
JOURNAL AMERICAN WATER WORKS ASSOCIATION
Vol. 46, No. 1, January 19S4
Printed in U. S. A.

Corrosion by W a t e r at Low Flow Velocity

By T. E. Larson and R. M. King
A contribution to the Journal by T. E. Larson, Head, and R. M. King,
Asst. Chemist; both of Chemistry Subdiv., State Water Survey, Ur-
bana, Ill,

U P to the present there has appar-

ently been no organized program
for studying corrosion by fresh water.
rective treatment that seems most likely
to succeed.
In 1924 Whitman, Russell, and Al-
No pattern has been established even to tieri ( J ) concluded from careful ex-
define a corrosive water under various periments with Cambridge, Mass.,
conditions of use. water that, in the pH range 4.1-10 at
The saturation index (1, 2) for cal- 22°C and 4.3-9 at 40°C, hydrogen ion
cium carbonate has been devised and concentration has no effect on the rate
applied with the intent of providing a of corrosion, and the main variable in
thin, controlled scale or film of calcium this pH region is the rate at which
carbonate on the metal surface, thereby
dissolved oxygen diffuses to the metal
reducing or preventing corrosion. The
surface. This conclusion was not in-
saturation index indicates the tendency
tended to apply to waters of other min-
toward, but not the rate of, deposition
eral character, but it has been variously
or solution of calcium carbonate in a
misquoted or enlarged upon to imply
water, and does not necessarily show
corrosivity, a fact that many water that dissolved oxygen controls the rate
works chemists have long recognized. of corrosion in natural waters. The
T h e current practice appears to be latter interpretation of the Cambridge
either: [1] to assume that a water is results was proved incorrect by Baylis
corrosive and to treat it whether neces- (4) in 1926. He demonstrated the
sary or n o t ; or [2] to assume that it practical value of calcium carbonate
is not corrosive until "red water" trou- protection by controlled pH and the
bles develop or equipment failure is low solubility of ferrous carbonate at a
experienced, and then to apply the cor- pH greater than 8.
1
2 T. E. LARSON & R. M. KING Jour. AW W A

Because, in reality, each addition of the influence of dissolved oxygen and

caustic or acid to adjust the pH of pH on corrosion rates is secondary.
Cambridge water produced a water of It is believed that the data presented
different mineral quality, the conclu- will provide a basis for comparison
sions reached by Whitman and his col- with information obtained in future
leagues need not be applicable to water studies and in practical experience.
of the same mineral quality as that at
Cambridge or to potable waters in Qualitative Studies on Ion Migration
general. An elementary investigation was
Long experience has taught water begun in 1948, with the assistance of
works personnel the value of pH con- John Grench of the Illinois W a t e r Sur-
vey Div., to obtain qualitative data on
the water composition at the cathode
TABLE 1
Tap Water Composition

Fig. 1. Experimental Corrosion Cell

The electrolysis cell is divided into nine
compartments by alundum plates between
the two iron electrodes. and the anode, as affected by waters of
different mineral composition. The
trol for corrosion protection, but it has investigation was designed to study
also shown that factors other than pH the general water quality conditions
and dissolved oxygen influence corro- that develop between two iron elec-
sion rates. Data available in the form trodes under the influence of an arti-
of mineral analyses have not been ficially impressed voltage to produce
amenable to interpretation owing to the a current density of a limited magni-
lack of experimental evaluation. tude. T h e electrodes were located at
It is the purpose of this paper to opposite ends of an electrolysis cell di-
demonstrate that water quality is a vided into nine 470-ml compartments
primary factor in corrosion (specifi- by vertical, parallel porous alundum
cally at low flow velocities) and that plates (Fig. 1). University of Illinois
Jan. 1954 LOW-VELOCITY CORROSION 3

Fig. 2. pH at Cathode
The numbers at scattered points represent the current density in milliamperes per
square foot.

tap water was permitted to flow creasing milliampere-hours of current

through the center compartment at a consumption. T h e changes were par-
rate of 300 ml per minute, while,' in ticularly significant in the end com-
the remaining compartments, the water partments. T h e increase in pH at the
was in a quasi-stagnant condition. No cathode is shown in Fig. 2, and the ac-
attempt was made to aerate the water companying reduction in calcium and
or to exclude dissolved oxygen. T h e magnesium in Fig. 3 and 4. It will
composition of the water is indicated be noted that, at low current density,
in Table 1. the pH was not affected so greatly for
Progressive quality changes oc- equivalent milliampere-hour values,
curred in each compartment with in- and, accordingly, magnesium precipita-

Fig. 3. Magnesium and Calcium Concentration

Key: A—magnesium, low current density (0.9-1.7) ma per square foot); B—mag-
nesium, high current density (3.5 ma per square foot); C—calcium.
4 T. E. LARSON & R. M. KING Jour. AW W A

tion did not occur during these

tests.
T h e general distribution of calcium,
magnesium, and alkalinity concentra-
tions in the various compartments is
shown in Fig. 5. T h e repeated loss of
alkalinity toward the anode compart-
ment was noteworthy.
In several tests, measurements were
made on polyphosphate and silica con-
centrations in each of the compart-
ments. W h e n polyphosphate or silica
was present originally in all the com-
partments, these tests showed a pro-
gressive decrease in the concentration

Fig. 4. Concentrations at Cathode

The curves show the concentrations of
calcium and magnesium in the compart-
ment adjacent to the cathode.

of these ingredients in the compart-

ments near the anode and the cathode.
W h e n all but the center compartments
were free from polyphosphate, how-
ever, the migration of polyphosphate
was definitely toward the cathode at
low current density (0.4 ma per square
foot), while, at high current density (4
ma per square foot), polyphosphate
was found to migrate toward anode Fig. 5. Concentrations in Compartments
and cathode at equal rates.
It would have been interesting to The curves show the concentrations {in
continue these tests with water con- equivalent parts per million) of alkalinity,
taining bicarbonate and more chloride, magnesium, and calcium, as well as the
but it was decided to use a different pH, in the various compartments.
Jan. 1954 LOW-VELOCITY CORROSION 5

approach to study the influence of bi- S, 0.050 m a x i m u m ; and Si, 0.010

carbonate and carbonate ions on cor- maximum.
rosion rates. The specimens were degreased in
carbon tetrachloride; placed in a 5 per
Immersion Tests cent solution of HC1 and H N O 3 for 2
As all natural waters contain bicar- min; placed in concentrated HC1 for
bonates in at least a small concentra- 1 m i n ; rinsed in acetone; and dried
tion, it was felt that consideration and weighed 48 hr before use. The
should be given to different propor- edges were coated with paraffin, and
tions of bicarbonate and other anions a scratch was made on both sides of
in corrosion tests with controlled syn- the specimen just prior to immersion
thetic solutions containing known con- in 18 liters of water for a 3-day test
centrations of sodium salts, eliminat-
ing the possible added influence of
bivalent metal ions. Borgmann ( 5 )
has indicated the relative corrosive-
ness of salts of numerous cations and
anions, exclusive of bicarbonate and
carbonate and largely in concentrations
greater than that in natural waters.
The effect of carbonate ions as an
inhibitor of corrosion was previously
demonstrated by Evans (6) in 1927.
Mears and Evans ( 7 ) , in 1935, de-
scribed in detail the inhibiting effect of
potassium carbonate on solutions con-
taining potassium chloride. These
data, however, concerned strips of steel
partially immersed in solutions of
known concentrations, and provided no
information on the pH of the resultant
mixtures of carbonate and chloride
salts. In other words, although the
potassium carbonate concentration Fig. 6. Total-Immersion Test Setup
varied, the pH was not held constant The specimens used were of black plate
and it also varied for the various pro- steel, free from mill scale. The appa-
portions. Therefore, both pH and ratus was of standard type.
carbonate ion variables affected the re- at a flow velocity of 0.085 fps at room
sults. temperature. A low velocity was de-
Apparatus of the standard type (8) liberately chosen in order to simulate
for total-immersion tests of nonferrous the conditions usually existing in as
metals was constructed for the studies much as 25-50 per cent of any munici-
(Fig. 6 ) . The 1½ × 3-in. specimens pal distribution system, including the
of 0.01-in. "black plate" steel (free service lines.
from mill scale) that were used were Various proportions of sodium bi-
reported to have the following com- carbonate and chloride were used at
position (by p e r c e n t a g e ) : C, 0.07; pH 7 and 9. Likewise, various pro-
Mn, 0.30-0.45; P, 0.015 m a x i m u m ; portions of sodium bicarbonate and
6 T. E. LARSON & R. M. KING Jour. AW WA

sulfate were used at these pH values, words, if 15 or 20 ppm dissolved oxy-

and various proportions of sodium bi- gen was present, the corrosion rate
carbonate and nitrate were used at pH in this range may have risen, particu-
7. Carbon dioxide was employed to larly with increasing proportions of
control pH. The results are shown in sodium chloride.
Fig. 7. It was repeatedly found that the cor-
rosion rate was zero when a' particu-
Discussion of Results lar minimum of alkalinity was present
It was noted that, after the propor- in each test. It was also noted that
tion of sodium chloride or sulfate an intermediate range of,' corrosion
reached a given value, the corrosion existed in which the rate was unpre-

Fig. 7. Eesults of Total-Immersion Tests

The experiments were conducted at room temperature and constant flow velocity
(0.085 fps); the dissolved-oxygen content was 8 ppm. The solid lines represent
boundaries of sectors in which the corrosion rates zvere as shozvn.

rate was not influenced by further ad- dictable under the experimental condi-
dition of these chemicals. Also, the tions. For example, of three speci-
rate was of the same order of magni- mens in a solution in this range, "One
tude whether chloride or sulfate was may have corroded at a rate of' 10 mg
used: The corrosion rates with these per square decimeter per day (mdd),
proportions were therefore assumed to while the two others may have cor-
be governed strictly by the dissolved- roded at a rate of 90 mdd in the same
oxygen content of the water. In other solution. In this range of water qual-
Jdn 1954 LOW-VELOCITY. CORROSION .. 7

ity, the corrosion rate may have been Extreme caution is needed in inter-
inhibited or intensified, depending preting these data or applying i the
upon the extent and location of the conclusions to other conditions.- Con-
corroded area of the specimen. sideration must.be given'to the. fact
It was significant that, for any spe- that, at the low velocities employed in
cific chloride concentration,. corrosion these studies, the electrical migration
rates might be considerably greater for of ions under, the corrosion cell poten-
solutions of low mineral content than tials plays a more important part in the
for those of high mineral content, a process than the relatively slow diffu-
finding that is contrary to the usual sion rate of the dissolved oxygen. At
predictions. It may be concluded that a higher velocity, it might be expected
corrosion rates are controlled more by that oxygen diffusion rates would be
the specific mineral quality than by the the more important factor. Also, the
total mineral content. relatively high mineral content mini-
It was also significant that, for some mizes the effect of pH because the hy-
chemical compositions, corrosion rates drogen and hydroxyl ion concentra-
appeared to be greater at pH 9 than at tions are relatively low. , .
pH 7, while, for the others, the rates One severe criticism of these data
were unchanged. This also is contrary is that no attempt was made to dis-
to the normal predictions on the cor- tinguish between general corrosion and
rosive tendency of water. pitting. Where pitting occurs, the
The relatively lower corrosion rates rate of penetration may be quite high,
experienced with nitrates was surpris- although the corrosion per square dec-
ing. Although a water that contains imeter of the total surface may be
only bicarbonate and nitrate is a rarity, no greater than in areas where general
it should be of interest to make a fur- corrosion is experienced. Mears and
ther study of the effect of small concen- Evans (7), however, have shown that
trations of nitrate on corrosion rates pitting is less likely to occur where
in water containing various mixtures little or no anodic inhibitor is present.
of chloride and bicarbonate. The data obtained in the study un-
Several spot tests with a 9-day im- der discussion appear to be particu-
mersion period yielded results no dif- larly significant as a: starting point or
ferent from the 3-day data. basis of comparison for future studies.
These data are specifically limited to Such investigations! might involve
dissolved-solids concentrations between lower quantities of dissolved oxygen,
200 and 1,200 ppm, under the flow higher flow velocities, different tem-
velocity and temperature conditions'in- peratures, polyphosphates, silicates,
dicated. Figure 7, however, shows the free or combined chlorine, and even
corrosion rates experienced in a test calcium at concentrations approaching
series in which the combined sodium or exceeding its solubility as calcium
chloride and bicarbonate concentra- carbonate.
tions ranged from 60 to 210 ppm.
Here again, it was; noted that the bi- Inhibition by Bicarbonate and Car-
carbonate exerted an inhibitive effect. bonate
In one group of tests with University Normally it is contended that, in
of: Illinois tap water at pH 7 (— 0.4 air-saturated solutions which do not
saturation index), no corrosion was corrode, the dissolved oxygen has been
noted until 60 ppm NaCl was present. responsible for the formation of an
8 T. E. LARSON & R. M. KING Jour. AWWA

invisible oxide film. Evans ( 9 ) cites are present. Although the ferrous hy-
many studies with electron diffraction droxide theory may be perfectly cor-
and X-ray techniques which indicate rect when dealing with corrosion prod-
films of ferric oxide. In fact, steel ucts derived from solutions of sodium
treated with pure oxygen for a suffi- chloride, the actual initial corrosion
ciently long period under proper con- product' is soluble ferrous chloride,

Fig. 8. Solubility of Ferrous Ion

The solubility of FeCO3 (solubility product, 2.11 × 10-11) is considerably less than
that of ferrous hydroxide in natural waters.

ditions remains corrosion resistant un- hydrolyzed insoluble ferrous hydroxide

til the film is broken or attacked. being secondary.
It has often been stated that the cor- It will be noted in Fig. 8, however,
rosion product adjacent to the metal is that the solubility of ferrous carbonate
ferrous hydroxide. It is extremely is considerably less than that of ferrous
doubtful that this assertion holds true hydroxide, although it is obviously
when carbonate or bicarbonate ions greater than that of ferric hydroxide.
Jan. 1954 LOW-VELOCITY CORROSION 9

This appears to indicate that any in- lem. Natural waters contain a corro-
consistency in an oxide coating would sion inhibitor, varying in concentration
be protected immediately by ferrous or proportion from one supply to an-
carbonate rather than ferrous hydrox- other. Without basic data on the pri-
ide. mary, partial, or total inhibition of this
Reconsidering the data that showed natural ingredient, there is no hope of
greater corrosion rates at pH 9 than at correlating the observations made on
pH 7 for the same mineral quality—and the effectiveness of other inhibitors or
accepting the assumption that ferrous methods of treatment against corrosion.
carbonate provides an inhibitory film— 4. A possible explanation has been
the increased corrosion rates at pH 9 provided for the inhibitory effect of
can be explained by the fact that fer- bicarbonate and carbonate alkalinity.
rous hydroxide is more readily formed
at the higher pH and is also more References
readily oxidized. Such localized pre-
1. LANGELIER, W. F. The Analytical Con-
cipitates adhering to the metal provide trol of Anticorrosion Water Treatment.
a physical barrier to oxygen diffusion Jour. AWWA, 28:1500 (Oct. 1936).
and permit the metal surface underly- 2. LARSON, T. E. The Ideal Lime-softened
ing them to become anodic to the ex- Water. Jour. AWWA, 43:649 (Aug.
posed surface. W h e n sufficient alka- 1951).
linity (not O H " ) is present, how- 3. WHITMAN, G. E.; RUSSELL, R. P.; &
ALTIERI, V. J. Effect of Hydrogen Ion
ever, the flaws in an oxide coating- Concentration on Submerged Corrosion
are protected by ferrous carbonate be- of Steel. Ind. Eng. Chem., 16:665
fore the ferrous ion concentration can (1924).
become large enough to form ferrous 4. BAYLIS, J. R. Factors Other Than Dis-
hydroxide. The relative structure of solved Oxygen Influencing the Corro-
ferrous carbonate and ferrous hydrox- sion of Iron Pipes. Ind. Eng. Chem.,
18:370 (1926).
ide and their reactive properties with 5. BORGMAN, C. S. Initial Corrosion Rate
dissolved oxygen are beyond the scope of Mild Steel. Ind. Eng. Chem,, 29 :815
of this paper. (1937).
6. EVANS, U. R. Practical Problems of
Summary Corrosion. J. Soc. Chem. Ind., 46:347
(1927).
1. There is need for fresh-water cor- 7. MEARS, R. B. & EVANS, U. R. The Prob-
rosion research. ability of Corrosion. Trans. Faraday
2. T h e experiments described dem- Soc, 31:529 (1935).
onstrate the behavior of solutions at 8. FRASER, O. B. J.; ACKERMAN, D. E.; &
SANDS, J. M. Controllable Variables in
corrosion cell electrodes. the Quantitative Study of Submerged
3. Basic data have been obtained to Corrosion of Metals. Ind. Eng. Chem.,
which other data can be related in 19:332 (1927).
order to provide an organized ap- 9. EVANS, U. R. Metallic Corrosion Pas-
proach to the water corrosion prob- sivity and Protection. Edward Arnold
& Co., London (1937). p. 53.