Professional Documents
Culture Documents
.RESEARCH FOUNDATION
Chemistry of
Corrosion Inhibitors
in Potable Water
Subject Area:
Distribution Systems
CHEMISTRY OF CORROSION
INHIBITORS IN POTABLE WATER
Prepared by:
Mark M. Benjamin
Steve H. Reiber
John F. Ferguson
Elizabeth A. Vanderwerff
and
Mark W. Miller
Department of Civil Engineering
University of Washington
Seattle, WA 98195
Prepared for:
AWWA Research Foundation
6666 West Quincy Avenue
Denver, CO 80235
February 1990
This study was funded by the American Water Works Association Research
Foundation (AWWARF). AWWARF assumes no responsibility for the content of the
research study reported in this publication, or for the opinions or statements
of fact expressed i.n the report. The mention of tradenames for commercial
products does not represent or imply the approval or endorsement of AWWARF.
This report is presented solely for informational purposes.
Copyright 1990
by
AWWA Research Foundation
American Water Works Association
Printed in U.S.A.
ISBN 0-89867-506-5
TABLE OF CONTENTS
Page
TABLES...................................................... vi i
FIGURES..................................................... ix
ACKNOWLEDGMENTS............................................. xli i
FOREWORD.................................................... xv
EXECUTIVE SUMMARY........................................... xvii
I. INTRODUCTION................................................ 1
II. LITERATURE REVIEW........................................... 5
Introduction............................................. 5
Action of Corrosion Inhibitors........................... 6
Chemical Inhibitors in Potable Water..................... 7
Phosphate, Polyphosphates, and Zinc Phosphate......... 7
Silicate Inhibitors................................... 9
Combinations of Inhibitors............................ 10
Carbonate-Based Inhibitors: CaCO^ Deposition Indices.. 10
Scale Composition and the Siderite Model of Scale
Format ion.............................................. 12
Summary.................................................. 15
III. METHODS AND MATERIALS....................................... 17
Introduction............................................. 17
Water Quali ty i n the Exposure Loops...................... 17
Seattle Tap Water..................................... 17
Water Quali ty Adjustment.............................. 18
Loop-Water Monitoring................................. 22
Corrosion-Rate Measurement............................... 23
Introduction.......................................... 23
Long-Term Corrosion Rates............................. 24
Pipe-Loop Construction and Coupon Materials........... 27
Surface Preparation for Black- and Galvanized-Iron
Coupons............................................. 33
Surface Preparation for Copper Coupons................ 37
Technique Validation.................................. 40
Electrochemical Corrosion-Rate Measures.................. 40
Introduction.......................................... 40
Polarization Cell and Electric Circuit................ 42
General Polarization Measurements..................... 44
Corrosion-Rate Measurements by Monitoring Oxygen
Depletion.............................................. 50
Purpose of Oxygen-Depletion Corrosion-Rate
Measurements........................................ 50
Page
Oxygen-Depletion Piping Loop.......................... 51
Analysis of Oxygen-Depletion Corrosion-Rate
Measurements........................................ 53
Accuracy and Precision of DO Depletion
Measurements........................................ 55
Analysis of Corrosion Scale.............................. 57
Purpose of Corrosion-Scale Analysis................... 57
X-Ray Analysis of Corrosion Scales.................... 58
Chemical Analysis of Corrosion Scales................. 59
Scale Oxidation.................................... 59
Ferrous Iron Analys i s.............................. 59
Total-Metal Analysis of Dried Scales............... 61
Carbonate Analysis................................. 61
Phosphate.......................................... 62
Copper Composition of Scales....................... 62
IV. CORROSION OF MILD-STEEL PIPE................................ 65
Corrosion Rates Based on Weight Loss of Mild-Steel Pipe.. 65
Overall Weight-Loss Comparison........................ 65
Correlation of Water Quality Variable with Corrosion
Rates............................................... 76
Buffer Intensity................................... 77
Alkalinity......................................... 78
The Langlier Saturation Index...................... 79
pH................................................. 79
Phosphate.......................................... 80
Initial Corrosion Rates............................... 81
Comparison of Long-Term Corrosion Rates in pH 6.0
and pH 8.0 Seattle Tap Water........................ 82
Comparison with Previous Corrosion Research
Conducted Using Seattle Tap Water................... 83
Iron Corrosion Scale Analysis............................ 86
Physical Characteristics of Iron Corrosion Scales..... 86
Corrosion-Scale Weights............................... 88
X-Ray Diffraction Analysis of Iron Corrosion Scale.... 88
Chemical Analysis of Iron Corrosion Scale............. 91
Total Iron Content................................. 91
Ferrous Iron....................................... 92
Carbonate Content.................................. 95
The Siderite Model....................................... 97
Qualitative Correlation of This Research With the
Siderite Model of Iron Corrosion.................... 97
Relationship Between Water Quality Parameters
and the Siderite Model........................... 98
A Quantitative Iron Corrosion Model Based on
Siderite Formation.................................. 104
Response of Scale-Covered Mild-Steel Pipe to Changes
in the Alkalinity and pH of the Conditioning Water..... 107
Corrosion-Rate Measurements........................... 107
Scale Analysis........................................ 113
IV
The Effect of High Conductivity on Initial
Corrosion Rates..................................... 115
Summary.................................................. 117
V. CORROSION OF COPPER PIPES................................... 121
Copper Corrosion Rates................................... 121
Scale Accumulation and Composition....................... 128
Short-Term Water Quality Effects on Aged Copper
Corrosion.............................................. 134
Di ssolved Oxygen...................................... 136
Chlorine Residual..................................... 136
pH ................................................... 140
Short-Term Tests Using Fresh Coupons..................... 142
Summary.................................................. 149
VI. CORROSION OF GALVANIZED PIPES............................... 151
Cumulative Weight Loss................................... 151
Scale Characteristics.................................... 155
Scale Accumulation.................................... 155
Scale Mineralogical Identification.................... 159
Electrochemical Measurements on Galvanized Surfaces......... 161
Validation............................................... 161
Reduction Couples........................................ 164
Di ssolved Oxygen...................................... 164
Chlori ne.............................................. 167
pH ................................................... 167
Summary.................................................. 170
VII. SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS................... 173
Mild Steel............................................... 173
Copper................................................... 175
Galvanized Steel......................................... 176
Recommendations.......................................... 177
VIII. REFERENCES.................................................. 179
I.. APPENDIX Al................................................. 185
Validation of Weight-Loss Corrosion-Rate
Measurement Techniques................................. 185
Summary of the Weight-Loss Technique..................... 191
.. APPENDIX A2................................................. 193
Electrochemical Corrosion-Rate Measurements:
Principles and Validation.............................. 193
Surface-Polarization Procedures.......................... 194
The Leroy Model.......................................... 199
Accuracy of Corrosion Measurements....................... 201
Surface Potential and Pitting............................ 207
Pipe Surface Electrochemistry............................ 208
.I. NOMENCLATURE................................................ 209
LIST OF TABLES
Table Page
1 Seattle Tap Water as Received at the University of
Washington During 1986 and 1987......................... 18
2 Characteristics of Conditioning Loop Water................ 19
3 Summary of Procedures for Black-Iron and Galvanized-
Iron Coupon Preparation/Restoration..................... 36
4 Summary of Copper Coupon Preparation/Restoration
Technique............................................... 39
5 Comparison of Corrosion-Rate Measurement Methods.......... 56
6 Cumulative Corrosion Rates Over Six Months Exposure....... 76
7 Water Quality Variables and Corrosion Rates............... 77
8 Initial Corrosion Rates Found on Mild-Steel Pipe.......... 81
9 Long-Term Corrosion Rates in pH 6.0 and pH 8.0 Seattle
Tap Water............................................... 83
10 Seattle Corrosion-Control Pilot-Plant Studies Average
Corrosion Rates Obtained on Mild-Steel Pipe............. 85
11 Iron Content of Dried Iron Corrosion Scales............... 91
12 Ferrous Iron Content of Iron Corrosion Scales............. 93
13 Carbonate Content of Iron Scales.......................... 96
14 Weight-Loss Data From Changing Water Chemistry
Experiments............................................. 112
15 Ferrous Iron and Carbonate Content of Iron Scales
Exposed to Changes in Water Chemistry................... 114
16 Response of Corrosion Rate to High-Conductivity Water..... 117
17 Composition of Copper Corrosion Scales.................... 133
18 Zinc Concentration in Zinc Galvanized Corrosion Scales.... 158
Al-1 Weight Loss Variation as a Function of Exposure
Duration................................................ 185
Al-2 Weight Loss Variation as a Function of Exposure
Duration Corrected for Variation due to
Surface-Preparation Techniques.......................... 190
A2-1 Comparison of Electrochemical Corrosion Rates with
Weight-Loss Measurements................................ 206
Vll
LIST OF FIGURES
Figure Page
1 Cross Section of Insert, Spacer, and Union, and
Assembled ISWS Corrosion Tester......................... 25
2 Cross Section of Coupon/Sleeve Headpiece Assembly
Used for Weight-Loss Test............................... 28
3 Black-Iron, Galvanized-Iron, and Copper Coupon/Inserts.... 30
4 Pi pe Coupon Exposure Loop................................. 32
5 Black-Iron Coupon With Polishing Tool..................... 34
6 Copper Coupon With Polishing Tool......................... 38
7 Polarization Flow Cell.................................... 43
8 Schematic of Closed-Loop Recirculation System for
Polarization Measurements............................... 45
9 Flow Diagram of Tafel Slope Fitting Routine............... 48
10 Evans Diagram of Polarization on a Copper Surface
Showing Theoretical and Observed Values................. 49
11 Schematic of Pipe Loop for Measuring Corrosion Rate by
Oxygen Depletion........................................ 52
12 Weight Loss of Iron Pipes Exposed to Seattle
Tap Water, pH 8.0 ...................................... 66
13 Weight Loss of Mild-Steel Pipe Sections Exposed to
pH 6.0 Water............................................ 67
14 Weight Loss of Iron Pipes Exposed to the Low-
Carbonate Water at pH 8.0............................... 68
15 Weight Loss of Iron Pipe Exposed to the Low-
Carbonate Water at pH 8.4.............................. 69
16 Weight Loss of Iron Pipe Exposed to the pH 7.2,
High-Carbonate Water................................... 70
17 Weight Loss of Mild-Steel Coupons in P04 -Inhibited
and Noninhibited Waters................................. 71
18 Change in Corrosion Rate With Length of Exposure to
pH 8.0 Water............................................ 72
19 Change in Corrosion Rate With Length of Exposure to
pH 6.0 Water ........................................... 73
20 Change in Corrosion Rate With Length of Exposure to
Low-Carbonate Waters.................................... 74
21 Change in Corrosion Rate With Length of Exposure to
High-Carbonate Water.................................... 75
22 Scale Weight Formed on Iron Pipe Exposed to Different
Water Qualities......................................... 89
23 Total Soluble Ferrous Iron in Equilibrium With Siderite... 99
24 Oxidation Rate of Ferrous Iron in Equilibrium With
Siderite................................................ 101
25 Calculated Buffer Intensity as a Function of pH for
Waters with the Indicated Concentration of Inorganic
Carbon and No Other Weak Acids or Bases................. 103
26 Corrosion Rate of High-Carbonate Scale Exposed to
pH 8.0 Tap Water........................................ 109
27 Corrosion Rate of High-Carbonate Scale Exposed to
pH 6.0 Tap Water........................................ 110
ix
jgure Page
XI
ACKNOWLEDGMENTS
The authors wish to express their appreciation for the financial and technical
assistance provided by the AWWA Research Foundation in general, and the
support and patience of Nancy McTigue and Rick Karlin in particular. We have
also benefited significantly from discussions with members of the project
advisory committee: Man/in Gardels of the USEPA, Steven Hubbs of the
Louisville (Ky.) Water Company, and Greg Reed of the University of Tennessee.
Finally, this document could not have been completed without the diligent
efforts and unfailing good humor of Marcia Buck and Pam Hjelm, word processors
par excellence.
Xlll
FOREWORD
This report is part of the on-going research program of the AWWA Research
Foundation. The research described in the following pages was funded by the
Foundation in behalf of its members and subscribers in particular and the
water supply industry in general. Selected for funding by AWWARF's Board of
Trustees, the project was identified as a practical, priority need of the
industry. It is hoped that this publication will receive wide and serious
attention and that its findings, conclusions, and recommendations will be
applied in communities throughout the United States and Canada.
The Research Foundation was created by the water supply industry as its center
for cooperative research and development. The Foundation itself does not
conduct research; it functions as a planning and management agency, awarding
contracts to other institutions, such as water utilities, universities,
engineering firms, and other organizations. The scientific and technical
expertise of the staff is further enhanced by industry volunteers who serve on
Project Advisory Committees and on other standing committees and councils. An
extensive planning process involves marty hundreds of water professionals in
the important task of keeping the Foundation's program responsive to the
practical, operational needs of local utilities and to the general research
and development needs of a progressive industry.
All aspects of water supply are served by AWWARF's research agenda:
resources, treatment and operations, distribution and storage, water quality
and analysis, economics and management. The ultimate purpose of this effort
is to assist local water suppliers to provide the highest possible quality of
water, economically and reliably. The Foundation's Trustees are pleased to
offer this publication as contribution toward that end.
The goal of this project was to improve the scientific and engineering
understanding of the corrosion process in household plumbing, and thereby
develop improved methods to reduce corrosion.
XV
EXECUTIVE SUMMARY
INTRODUCTION
This research project investigated the corrosion rates of copper, mild-steel,
and zinc-galvanized pipe coupons as a function of the chemical
characteristics of the water they were exposed to. The coupons were short
sections of pipe typical of the kind that might be employed in a household
plumbing system. Several coupons of each material were placed in a holder
and exposed to water of controlled composition over a period ranging from a
few weeks to more than two years. After a predetermined exposure period, the
coupons were removed from the assembly and analyzed for weight loss.
Additionally, corrosion scales that had developed were characterized by
visual examination, by x-ray diffraction to identify their mineral forms, and
by chemical analysis.
The weight-loss analysis gave an indication of the corrosion rate over
periods of months to years. Other techniques were used to evaluate short-
term corrosion rates, particularly the rates that reflected short-term
responses to transient changes in the composition of the circulating water.
These other techniques included the rate of depletion of dissolved oxygen in
the water and electrochemical polarization methods.
Overall, more than 1,000 pipe coupons were subjected to corrosion testing.
The number of individual coupons analyzed, the range of analyses, the variety
of water qualities investigated, and the duration of the tests make this one
of the most extensive studies of corrosion in potable water systems ever
undertaken.
XVll
The final two waters were prepared by adding either 1 or 5 mg/L PO*-P to tap
water and adjusting the pH to 8.0. Overall, the six waters studied represent
a range of possible water quality alterations that a utility might consider
to avoid or mitigate a corrosion problem.
CORROSION OF COPPER
As expected, copper pipe sections had much lower corrosion rates than steel
pipe sections. Copper differed from steel in that even low doses of
phosphate inhibitor dramatically lowered the corrosion rate of copper
coupons. Somewhat surprisingly, changing pH from 6.0 to 8.0 did not reduce
the corrosion rate of copper, in contrast with results of previous
investigators.
Copper corrosion rates were fairly similar in all the phosphate-free loops,
especially after the first two months of exposure. Rates were steady
thereafter and were not responsive to carbonate content or pH. The mineral
cuprite was found in the scale from the pH 8.0 tap water exposure, but not in
xviii
the scale from lower-pH loops. It is possible that although the cuprite did
not form a continuous, protective layer during the course of this study, such
a layer might have formed after longer exposure periods.
The chemical composition of the scale in the high-phosphate loop was
consistent with formation of a mineral of the composition Cu^PO^)?.
However, the scale in the low-phosphate loop contained much less phosphate
than would be required to bond with all the oxidized Cu in the scale. Thus,
just as a small amount of siderite can offer corrosion protection to iron
pipe, it appears that copper pipe can be protected by the presence of a
relatively small fraction of the protective mineral in the scale. The
presence of such a scale also protected the pipe against the short-term
effects of exposure to a more corrosive water.
RECOMMENDATIONS
Among the significant conclusions from this research is that no single action
can be expected to limit corrosion in all common types of plumbing, but that
there are useful measures that will restrict corrosion in any given system.
For steel pipes, it is important to recognize that the Langelier Saturation
Index is probably not a good indicator of corrosion protection. This is
certainly the case in soft waters, where buffer intensity is likely to be
much more important than the degree of saturation with respect to calcium
carbonate. Establishing conditions where siderite formation is favored would
be most effective in minimizing iron corrosion. A high dose of phosphate
XIX
will apparently be of some value in protecting iron pipes from corrosion, but
such a high dose is probably impractical in most systems.
On the other hand, copper corrosion in soft waters can be reduced by small
additions of phosphate. While our results did not show a dramatic effect of
pH on copper corrosion, other reputable studies have found such an effect.
Also, we investigated the beneficial effects of phosphate only at pH 8.0, and
so do not know the effectiveness of phosphate addition at lower pH values.
Where copper corrosion in soft waters is perceived to be a problem, addition
of phosphate and maintenance of a pH near 8.0 appears to be a very effective
corrosion control strategy. Copper piping is used primarily in premise
plumbing, and adjustment of water chemistry as the water enters the building
may be useful in these cases.
Protection of zinc galvanized pipe apparently requires addition of alkalinity
to soft waters. Adjusting the pH to 8.0 reduces corrosion dramatically
compared to maintaining it at pH 6.0, and raising the pH to 7.2 with addition
of large amounts of carbonate alkalinity is also effective. The exact
strategy that is appropriate or preferred in a given system will depend on
the characteristics of the particular system.
XX
INTRODUCTION
Corrosion can also be initiated when the pipe surface is chemically attacked
by aggressive molecules in the water. For example, water in "dead spaces"
near pipe joints or cracks may have chemical characteristics somewhat
different from the chemical characteristics of the bulk water solution. The
difference in chemical composition between the two points leads to a
corresponding difference in the electrochemical potential, which can affect
1
the electrical potential on the pipe surface and cause a corrosion reaction.
Such a condition is referred to as a "concentration cell."
Regardless of the force driving the corrosion reaction, the practical ways to
reduce the corrosion rate are to interfere with the process by making it more
difficult for molecules to either approach or vacate the vicinity of the
anode and cathode, or to alter the electrical potential of the surface and
thereby lower the driving force for the reaction. This is, in essence, what
successful corrosion inhibitors do.
Second, the tests were conducted using more than 1,000 individual coupons
that had been exposed to water of well-controlled and wel1-documented quality
over a period of up to 2 1/2 years. To our knowledge, such an extensive data
base has not previously been compiled in a manner that allows relatively
long-term corrosion rates to be compared as a function of water quality
variables.
The relationships that have been established by this work and the data base
that the project makes available to other researchers will be of significant
value in cases where a corrosion problem currently exists and in the
development of the field of corrosion control.
LITERATURE REVIEW
INTRODUCTION
A wide range of inorganic and organic inhibitors has been used in various
process industries and power-generating plants to protect plumbing from
highly corrosive waters. The vast majority of these inhibitors are
inappropriate for potable water supplies and are not discussed here. The
only chemicals commonly added to potable water for corrosion control are
bases to increase pH, carbonate-containing chemicals to encourage growth of
protective calcium carbonate or metal carbonate scales, orthosilicates and
polysilicates, orthophosphates and polyphosphates, and zinc (Zn). These
chemicals may be added individually or in various combinations. Evidence for
their relative effectiveness comes from a combination of short-term bench and
pilot-scale tests conducted by water utilities or engineering consultants,
observations made by these groups when a system-wide corrosion control
program was put into effect, and long-term investigations conducted by
various research groups.
ACTION OF CORROSION INHIBITORS
The most effective and reliable form of corrosion inhibition involving scale
formation is passivation. In this process, an inorganic scale must first
grow on the metal surface. Typically, the scale first forms at the anode,
but grows to cover both anodic and cathodic sites (Ryder and Wagner, 1985).
Thus, it may act as a diffusion barrier to reactants and products of both the
corrosion half-reactions. In the presence of such scales, many surfaces
display a dramatic transition from being highly favorable to further
corrosion to being highly unfavorable to further corrosion as the electrical
potential at the interface between the scale and the underlying metal
increases. This transition occurs even though the thermodynamic driving
force for corrosion increases with increasing potential. The transition may
be related to a change in scale character from being an ionic conductor to an
electronic conductor (Bockris and Reddy, 1970).
Hatch (1973) reports that zinc does not alter the qualitative mode in which
ortho-phosphate acts as an inhibitor, and therefore the optimum conditions
for inhibition (e.g., pH range) are the same when applying phosphate alone or
with zinc. Nevertheless, he and others have found that zinc does allow use
of a lower concentration of phosphate for the same level of protection, and
causes the protection to develop faster for the same phosphate dose (Ryder
7
and Wagner, 1985; Hatch, 1973). However, there are also reports of systems
where zinc phosphate was ineffective in reducing corrosion of iron, copper,
or lead (Karalekas et al., 1983), or where zinc phosphate was effective but
bimetallic zinc phosphate (ZnNaPCfy) was not, even though the doses of Zn and
o-PO^ were quite similar (Muller and Ritter, 1980).
Silicate Inhibitors
Some of the older work on silicate inhibition suggested that it could protect
mild-steel pipes by formation of a continuous amorphous silica protective
scale, and that this scale could form only after other corrosion products had
formed an "anchoring" scale (Ryder and Wagner, 1985; Lehrman and Schuldener,
1952). However, Wagner (1985) claims that the beneficial effects often
ascribed to silicate can be explained solely by the fact that the silicates
raised the pH of the test water. He and his co-workers find that the same
corrosion reduction can be accomplished by raising the water pH with other
bases, and that the silicate itself provides no specific benefit. While this
may be the case at pH values greater than 8.0, which are generally used in
conjunction with silicate treatment, Hatch (1973) claims that there is little
support for using such high pH values, and Wood et al (1957) indicate that
silicates may in fact provide more protection at pH 7.0 than at higher
values. In any case, studies have shown silicates to have little or no
benefit in reducing corrosion of copper, zinc, iron, or lead under the
conditions in which they are commonly applied (Ryder and Wagner, 1985;
Wagner, 1985; Sontheimer et al., 1985).
Combinations of Inhibitors
More than 100 years ago, Heyer (1888) described the reduction of lead
dissolution from plumbing systems by the formation of a thin protective
calcium carbonate scale on the pipe wall, induced by increasing calcium and
carbonate levels to the solubility level. Heyer's work led to the "marble"
test, still in use today, for the measurement of carbonate stability in water
10
supplies. A great deal of corrosion-control practice today is designed to
control CaCOj deposition, and it has become common to describe the
corrosiveness of a water in terms of its ability to form a protective calcium
carbonate scale on distribution pipes.
Recent research has focused more on the physical and chemical properties of
corrosion scales and has led to the development of a new model for corrosion
inhibition by the growth of protective scales in iron-containing pipes. This
research is reviewed briefly below.
Scale formation is common in any system involving contact between metals and
natural water. The scale layer is often the principal protective agent
against internal corrosion in pipes, and its effectiveness determines the
useful life of the system (Hepburne, 1965). The major components of the
scale are corrosion products (metal and hydroxide ions), dissolved ions, and,
in many cases, calcium carbonate deposited from the water. The role that
corrosion products play in the formation of protective scales is not well
understood. The emphasis placed on CaC03 deposition demonstrates that it has
long been taken for granted that CaC03 by itself could form a protective
scale.
Other researchers (Feigenbaum et al., 1978; McCauley and Abdullah, 1958) have
analyzed iron-pipe scales and identified a number of iron precipitates
enmeshed within the calcite, including siderite (FeCC^), goethite (FeOOH),
magnetite ^630^), and lepidocrocite (FeOOH). Kolle and Sontheimer (1985),
and Kolle and Rosch (1980) have examined scale morphology in detail and
determined that steel pipe scales form a readily identifiable "shell-like
layer" consisting of magnetite and goethite that divides the scale into
regions of differing composition. Beneath the shell-like layer (i.e., nearer
the pipe wall) siderite and goethite are found, while above the layer calcite
and goethite predominate. The authors suggest that a strong shell layer
encourages the production of siderite in the underlying layer (it shields the
13
Fe++ from rapid oxidation) and aids in fixing the scale components to the
metal surface.
The primary reactions take place at the metal/scale interface and are usually
controlled by oxygen diffusion through the scale. The rate of diffusion,
coupled with the buffer intensity of the solution, determines the pH near the
pipe wall. This localized pH exerts strong influence on the competing
secondary reactions.
The tertiary siderite reactions involve the slow oxidation of the ferrous
iron in the siderite to the ferric form and the consequent production of
magnetite and a goethite pseudomorph. These two crystalline forms compose
the shell-like layer observed by Kolle and Sontheimer (1985). The shell
layer is formed directly from the siderite and has a dense impermeable
morphology. Although goethite crystal structure is generally less dense than
siderite, the goethite pseudomorph formed maintains the dense structure of
the siderite. This shell layer is important to the siderite model because it
retards the diffusion of oxygen to the metal surface, slowing the corrosion
rate and creating favorable conditions for further siderite formation. The
gradual conversion of the siderite to the ferric oxide form enhances the
shell layer and thus guarantees the integrity of the protective scale.
SUMMARY
15
decisions about which inhibitor to use and ultimately to significant economic
and possibly public health costs.
16
METHODS AND MATERIALS
The majority of the corrosion-rate tests conducted for this research involved
the use of pipe coupons through which water was pumped. The water was stored
in reservoirs where its quality could be controlled. The following sections
describe the apparatus used in the tests and the preparation and analysis of
the pipe coupons.
INTRODUCTION
Two phases of experiments were conducted. In the first phase, pipe sections
were exposed to waters with a specific water quality in long-term
conditioning pipe loops. Pipe sections from the conditioning loops could be
removed after various exposure periods to determine weight-loss and to
analyze the corrosion scale. Corrosion scales were analyzed by powdered x-
ray diffraction and by wet chemical methods. X-ray diffraction was used to
determine the presence and identity of crystalline mineral compounds. Wet
chemical methods were used to determine the chemical composition of these
scales.
All of the waters studied were prepared from Seattle tap water. The source
waters for Seattle tap water are in mountainous regions receiving heavy
17
precipitation. The local geology in the watershed is largely granitic rock.
The source waters flow down steep streambeds, so there is little time for the
water to contact and dissolve minerals. These conditions create a water with
low alkalinity, low hardness, and low total dissolved solids. Seattle's
water treatment consists of chlorination, fluoridation, and the addition of
lime and sodium hydroxide for corrosion control. The aim of Seattle's
corrosion-control program is to raise the pH to 8.0 to counteract the water's
natural acidity and the acidity contributed by chlorination and fluoridation.
Without corrosion control, the pH would be about 6.0. The treatment does not
maintain a positive Langelier Saturation Index (SI); the SI after treatment
is still about -0.9.
The typical composition of the Seattle tap water used in this study is shown
in Table 1.
TABLE 1
Seattle Tap Water as Received at the University of Washington During
1986 and 1987.
The most important water quality parameters studied were pH and carbonate and
phosphate ion concentrations. Calcium ion concentrations were not varied.
Alteration of either the pH or the carbonate ion concentration (Cj) affects
the buffer intensity, saturation index, and conductivity. For example,
18
raising Cj increases the conductivity. Buffer intensity and SI are both
increased by increases in Cj at fixed pH. SI is also increased by increases
in pH. In contrast, when Cj is fixed, buffer intensity peaks at pH 6.3 and
then decreases as the pH is raised, going through a minimum at pH 8.3 and
then increasing again with further increases in pH (see, for example,
Figure 25).
Seven different waters, each with different water qualities, were studied
(Table 2). These water qualities were chosen in an attempt to isolate the
effects of ortho-phosphate and carbonate concentrations and pH variations on
corrosion rates.
TABLE 2
Characteristics of Conditioning Loop Water
The second conditioning loop water (Water 2) was generally unaltered Seattle
tap water with a pH of 8.0 and an alkalinity of 20 mg/L as CaCC^. Since this
water leaves the tap undersaturated with C02 , its pH drifts downward when it
contacts the ambient atmosphere. To compensate, small amounts of NaOH were
added as necessary to keep pH at 8.0 for this study.
The third conditioning loop water (Water 3) resulted from attempting to strip
water of carbonates by bubbling C02 -free air through Seattle tap water. This
raised the pH to about 8.4, while lowering the carbonate content by about 2
percent. Alkalinity was not changed. This high-pH water had a slightly
lower buffer capacity than unaltered Seattle tap water. Hydroxide ions
accounted for about 2.5 percent of the buffer capacity in this water.
This water was prepared so that the effect of dissolved carbonate species on
corrosion rates could be evaluated. The relatively small change in the
concentration of these species that occurred when the water was sparged with
C02-free gas made such an evaluation impossible. For a few months, the gas
sparging procedure was altered in hopes of increasing the amount of C02
stripped without altering other water quality parameters, but these efforts
were unsuccessful.
Both the pH 8.4 water, which had been decarbonated at alkaline pH, and the pH
8.0 water, which had been decarbonated under acid conditions, had
significantly lower concentrations of the fully-protonated ^03 species than
any of the other loops. Coupons had been in the loops for various exposure
periods when the change in procedure occurred. Rather than discard these
coupons, an effort was made to determine whether the change affected their
corrosion rates. For the black-iron coupons, such an effect was noted, and
the corrosion rates for each low-carbonate exposure period are discussed
separately. However, for the other coupon types there was no significant
difference in corrosion rate between the two periods, and the data have been
grouped and discussed collectively. In these cases, the water quality during
both exposure periods is referred to simply as "pH 8.0 low carbonate" water.
The fourth conditioning loop water (Water 5) had a very high alkalinity
produced by adding sodium bicarbonate. In addition, the pH was reduced to
7.2 using carbon dioxide gas. The high alkalinity and lower pH resulted in a
large increase in buffer intensity; the buffer intensity was almost 4,000
times greater than Seattle tap water at pH 8.0. Even with the high
alkalinity, this water still had a slightly negative SI due to the extremely
low calcium content of Seattle tap water.
Larson and Skold (1957) found that iron corrosion rates significantly
increased as this ratio was increased from 0 to 0.7. The waters they studied
had alkalinities between 75 and 260 mg/L as CaC03 , much higher than Seattle
tap water. Higher alkalinities, plus other differences between their study
and the current one, make it impossible to predict the exact effect of
increased chloride ion concentrations on the corrosion rates measured in this
study. The increased chloride concentration of these two conditioning waters
must be considered when analyzing the results.
When acid, base, and sodium bicarbonate were added to the conditioning
waters, the conductivity was increased. The high carbonate water had a much
higher conductivity than all of the other waters; its conductivity was almost
100 times that of Seattle tap water.
Loop-Water Monitoring
Introduction
VVV^^S/x^^v/N/
K)
Ul 1-'/4 in ' PIPE NIPPLE
10 in (254 mm) LENGTH
1.'/4 in* x in" BUSHING
- v s~~
1 , -i
f
t V - <j
J \^__ A
<ISWS CORROSION TESTER
Figure I. Cross Section of Insert, Spacer, and Union, and Assembled ISWS Corrosion Tester
developed a modified version of the ISWS approach in which multiple coupons
are exposed.
The ISWS procedures for surface preparation of the pipe section coupons
involve several steps that must be performed carefully in order to give
reproducible results. Proper restoration of the coupon surface following
exposure is also critical to the weight-loss measurement. The somewhat
cumbersome sequence for preparing pipe section coupons consists of drying for
24 hr in a desiccator, phenolic stripping of the epoxy-resin exterior
coating, scouring of the interior surface using abrasive powders, exposure to
an inhibited acid solution to loosen deposits and etch the surface, immersion
in an ultrasonic bath, water and acetone rinses, another 24 hr of drying, and
final weighing. Handling of each coupon requires several hours of effort
spread over a two-day period.
Given the number and nature of procedures in the sequence, the possibility of
measurement errors or weight loss during sample preparation is significant.
However, the issue of precision in corrosion-rate measurements of this type
is rarely addressed. Even ASTM standards do not estimate the precision of
the resulting measurement, stating instead that this must be considered a
function of the individual system. In the literature, it is rare to find
confidence intervals or other quantitative indicators of measurement
precision. One study reported a standard deviation of 0.64 milligram per
square decimeter per day (mdd) compared to absolute rates of about 1 to 6 mdd
for corrosion of galvanized steel specimens using ASTM D2688 Method C (Neff
et al., 1987).
This project used a modified version of the ASTM machined-nipple test that
simplified the hardware and preparation technique. To improve the precision
of the measurements, the system was designed to hold a relatively large
number of coupons (pipe inserts), yet still offer ease of access and
simplified handling procedures. The following items are specific advantages
of this modified version:
26
1. The system is capable of supporting coupons made from a variety of
common metal plumbing materials (copper, galvanized-iron and black-
iron pipes).
2. It can be easily duplicated in other laboratories, using readily
available materials.
3. Coupon preparation and restoration procedures are both simple and
reliable.
4. The use of multiple coupons increases the precision of the overall
measurement, extending its use to the analysis of short-term weight
loss tests (exposure periods of less than 14 days).
For 1/2-in. black-iron and galvanized-pipe coupons, both with 0.85-in. OD, a
7/8-in. ID acrylic tube makes a satisfactory sleeve. The annular gap of
approximately .01 in. between the coupon exterior and the sleeve wall
provides an expansion area for the gasket material under compression. The
slight gap also makes loading and unloading the sleeve easier, and allows air
circulation to carry away moisture that may condense on the coupon exteriors.
The transparency of the acrylic is especially useful when seating the coupons
end-to-end, or when checking for possible leaks. All major pieces of the
sleeve assembly are machined from readily available acrylic stock. Other
materials (unions and nipples) are standard 1/2-in. PVC fittings.
27
Headpiece x Plunger
Vvlt UIIIUII
Pipe Coupon -\ / s.
j j "-=--1
IwT^I u
Lr^1 ' 1 1KICHr- nHr
. -/ 4 * i j
2.0"
Gasket
r^ir^.,._jQc *
Flexible Pvc
Nipple j L B- \ Acrylic Sleeve
> <>
1.5* 1,0'
Figure 2. Cross Section of Coupon/Sleeve Headpiece Assembly Used for Weight Loss Test
Standard 1/2 or 3/4 inch copper tubing can also be used (types K, L, or M) as
the coupon material; however, the thin wall of standard copper tubing (type K
is the thickest at 0.05 in.) makes it difficult to stack and seal the ten
coupons in a single sleeve. Because of the minimal butt-end exposure,
alignment of a large number of coupons is a problem that often leads to water
leaks at the many gasket seals. This task proved so troublesome that a
thick-wall copper axle stock was substituted for the standard water tubing.
The thicker walls give a broader shoulder against which to seal; this tubing
is also more durable than thin-wall stock and can be reused. The axle stock
selected has an OD of 0.675 inches and a wall thickness of 0.14 in. This
tubing is a minimum 99.9 percent pure copper, with a maximum phosphorus
content of 0.05 percent. This is a composition typical of most copper tubing
and appurtenances, and is nearly identical to ASTM specifications for the
three common water tubing types, which must have a purity of 99.9 percent and
a phosphorus content less than 0.04 percent (Cruse et al., 1985).
Microscopically, the surface of the axle stock appears free of pits and any
obvious inclusions, identical to the degreased surface of copper water
tubing. The axle stock and tubing form similar corrosion scales when exposed
to the same water for equivalent durations. An acrylic sleeve with an ID of
0.75 in. works well for mounting coupons made of this material.
Black-iron and galvanized-iron pipe coupons are cut from standard 1/2-in.
pipe, ASTM A 120 and ASTM A 120-82, respectively. Figure 3 presents a
photograph of the three coupon types. The ends of the coupons are lathe-cut
to ensure a smooth, flat finish that seals tightly against the gaskets. The
standard length of coupon is 4 cm, producing a coupon weight of roughly 50 g,
a size easily handled on most precision balances. Black-iron and galvanized-
iron coupons have an interior surface area of 18.5 cm , while the thick-
walled copper coupons present an area of 12.5 cm2 . After machining, the
coupon exteriors are degreased to remove cutting oil, polished, stamped with
an identification number, and given an exterior coating of clear acrylic to
protect them during handling and to prevent moisture-induced corrosion. The
acrylic coating is applied to the butt ends of the coupon as well.
29
Figure 3. Black Iron, Galvanized Iron, and Copper Coupon/Inserts
30
Figure 4 is a diagram of the closed recirculation loop and manifold assembly
used to determine the effect of specific water quality conditions on
corrosion of different metal types. Three sleeves, each containing a
different coupon type, are held in the manifold. The sleeves are connected
to the manifold by single-union ball valves, making it simple to isolate and
remove individual sleeves/coupons during loop operation. In all cases, a
flow rate giving a fluid velocity of 1 foot per second (fps) was used,
producing a flow pattern well within the turbulent regime.
One test was conducted after the loops had been in use for approximately 10
months to determine the relative flow rates through the three arms of the
apparatus. In this test, water at the normal flow rate of 1.5 gallons per
minute was pumped through the pH 6.0 and pH 8.0 Seattle tap water loops, and
the flow through each arm was measured independently using a graduated
cylinder to collect the exiting fluid.
Visual inspection suggested that the diameter of both the galvanized and
ungalvanized steel coupons had been reduced by about 15 to 20% respectively,
corresponding to about a 40% reduction in cross-sectional area. The flow
distribution in both loops was about the same, with about half the total flow
(0.75 gpm) passing through the arm with the copper coupons, and about one-
quarter the total flow (0.37 gpm) passing through each of the other arms.
These flow rates correspond to linear velocities of 1.9 ft/sec through the
copper coupons and 0.45 ft/sec through the other two arms, if we ignore any
reduction in cross-sectional area due to scale growth. If we use the
estimate of 40% reduction in cross-sectional area for the steel coupons, the
linear velocity through them becomes 1.1 ft/sec.
Although there may have been some effect on mass transport of reactants and
corrosion products due to the differences in flow rates among the three
loops, these effects were probably small, since flow through all the loops
was well into the turbulent regime. The effect on the rate of scale erosion,
which is related to the linear velocity of flow through the pipes, may have
been more significant.
31
Pipe Coupon
Sleeves
Rotometer
(Flow Indicator)
Flow Control.
Valve
Pump
32
There were two practical ways to conduct the long-term exposure experiments.
In one, headloss could be equalized in the three arms, by hydraulic linkages
via the headers above and below the pipes. Alternatively, the flow rates
through the three loops could have been equalized using valves and flow
measuring devices such as rotameters. In neither case would the linear
velocities through the loops be constant over the exposure period. The
variation in linear velocity is expected to be less in the system chosen than
in the alternative, since scale growth would decrease the flow rate and
thereby lessen the effect of reduced cross-sectional area. In this work, the
apparatus was set up in such a way that the total flow through all the loops
was constant, but the relative flows through the three arms varied while the
headloss through each arm was equivalent.
34
rotational speed will shorten the life of the brush (which is approximately
25 cleanings) and possibly abrade noncorroded metal from the coupon wall. A
relatively low speed of 2,000 rpm was used with this toolpiece. All but the
thickest corrosion scales were easily removed with a few passes of the brush;
even heavily pitted surfaces could be quickly cleaned.
Galvanized iron coupons have nearly the same ID as black-iron coupons and
were worked with the same toolpiece. However, because zinc is a softer metal
than steel, it was especially important not to abrade the coupon surface by
applying pressure against the coupon wall. Fortunately, the zinc corrosion
scales are less adherent than iron and can be removed in one or two quick
passes of the brush without disturbing the underlying galvanized (zinc)
layer.
Whether the coupon test surfaces were being prepared for the first time or
were being cleaned following exposure, use of the rotating brush required a
thoroughly dry surface. When dry, the scales are friable and can be reduced
to a fine powder that brushing action quickly disperses. If wet, the scale
will form a gritty paste that is difficult to remove and causes excessive
abrasion to the underlying metal. For a thin scale, a few hours of drying in
a desiccator was sufficient. Thicker scales (greater than 1 mm depth)
required up to 24 hr. Drying ovens were not used, since they may soften the
exterior acrylic coating. However, rapid drying could be achieved using a
35
hot-air blower (the type used for paint stripping) if care was taken to avoid
concentrating the heat on the coupon exterior.
Following use of the rotating brush, the interior surfaces of the coupons
were given a final polishing using clean, dry cotton swabs. This removed
residual powdered scale and generally left the metal surface in a condition
close to its original form. Because of the acrylic coating, the coupon
exterior and butt-ends could be quickly cleaned with a lightly moistened
tissue. To avoid soiling the exterior surfaces, cotton gloves were worn when
handling the coupons. After final polishing, the coupons were blown dry,
then stored in a desiccator prior to weighing. Excluding the initial drying
time, a group of 10 exposed black-iron or galvanized-iron coupons could
easily be cleaned, polished, and weighed in 1 hr, compared to ASTM procedures
requiring 3-4 hr of work spread over a period of 48 hr.
TABLE 3
Summary of Procedures for Black-Iron and
Galvanized-iron Coupon Preparation/Restoration
36
Surface Preparation for Copper Coupons
Unlike zinc or iron, corroding copper forms a dense, uniform scale that is
tightly bound to the metal surface. The protection afforded by this scale is
largely responsible for copper's relatively low corrosion rates. The
tightness of the scale makes it more difficult to remove than either zinc or
iron scales, and the removal process is further complicated by the metal's
softness, which precludes the use of steel-bristle polishing tools.
38
corrosion weight loss from copper may be an order of magnitude less than that
of black iron; hence, careful polishing was critical to the precision of the
weight-loss measurement. The procedure is summarized in Table 4.
TABLE 4
Summary of Copper Coupon Preparation/Restoration Technique
Shortly after the start of this research, it became obvious that the copper
coupons would not generate adequate corrosion scale in a reasonable time
period with which to perform x-ray analysis. To solve this problem, flat
copper plates similar in composition to the copper pipe coupons were exposed
to the same water qualities.
The plates were cut to a size that allowed mounting in the specimen holder of
the x-ray diffraction equipment. Prior to being placed in the exposure-loop
reservoir, the surface tarnish was removed using steel wool. Five plates
were installed on a plexiglass plate that was approximately 1 in. wide and 36
in. long. Each plate was held by plastic compression devices at specific
intervals such that the lowest plate was 6 in. below the water surface.
39
While the loop water was being changed during the thrice-weekly water quality
maintenance, the plates were kept submerged in loop water in a separate
basin.
Technique Validation
Introduction
40
presented in Appendix A2. The polarization cell and analytical procedures
are described in the following section.
4!
1. The new polarization cell is easily constructed and can be used with
a variety of potentiostatic devices without having to rely on
expensive proprietary instrumentation.
2. The cell uses actual plumbing materials (black-iron, copper, and
galvanized-iron pipe sections) as the corrosion-measurement surface.
3. Cell geometry duplicates the hydrodynamics of pipe flow and its
impact on the corrosion surface.
4. The technique allows one to determine the values of fundamental
variables controlling the corrosion process (e.g., the anodic and
cathodic Tafel slopes) for the specific surface and water quality
conditions of interest.
Ag-AgCl
Reference
Electrode
Flow Out
Platinum Counter
Electrode
Flow In
To Potentiostat
43
potentiometer capable of set-point polarizations could be substituted.
Analog output from the potentiostat was converted to digital form by a
Keithley Series 500 Data Acquisition System (Keithley Co., Cleveland, OH).
Data was then transferred to an IBM-PC-based microprocessor for statistical
analysis, graphing, and storage.
3 COUPON
SLEEVE
TO
POTENTIOSTAT
STIRRER
HEATER PUMP POLARIZATION
CELL
45
or the cathodic reduction processes. This means that polarizing the surface
by impressing a current density of several milliamps/cm^, as would be done
for a rapidly corroding surface, is not appropriate, since it alters both the
physical and electromechanical character of the electrode surface. The
problem is further complicated by large spatial and temporal variations in
the baseline corrosion potential (Ecor). An individual site may show a
potential shift of + or - 10 mV within a 5-min period. Shifts as great as 50
mV in a day have been observed, especially on iron. This supports Uhlig's
hypothesis that anodic and cathodic regions are constantly in motion,
establishing a "dynamic" surface equilibrium where the favorability of
specific half-cell reactions changes as the corrosion process proceeds
(Uhlig, 1971). Because anodic and cathodic sites may shift, measurement of
polarization resistance at a particular point on the surface must be
completed quickly before the baseline Ecor (from which the polarization
offset is measured) can change substantially.
46
Electrochemical parameters were calculated from the polarization data by
fitting each point to the modified Butler-Volmer equation and adjusting the
estimates of the Tafel slopes to minimize the overall error between observed
and predicted polarization currents. The procedure is summarized in the flow
diagram of Figure 9. In essence, it is a brute force approach that tests
every feasible combination of anodic and cathodic Tafel slope and ultimately
selects the pair that gives the best fit. As such, it lends itself readily
to a computer-based solution; a BASIC language program similar to that of
other researchers (Gerchakov et al., 1981) was developed to speed the
cumbersome calculations. After establishing appropriate search boundaries
and iteration increments, a corrosion current was calculated using the Leroy
Model and an initial selection of Tafel slopes. The Leroy Model (see
Appendix A2) is a statistical approach that predicts the corrosion current
density from a limited set of polarization data and a given set of Tafel
slopes.
i
Calculate Corrosion Rate
(Leroy Model)
i
Calculate Predicted Corn Current for each Offset
(Modified Butler-Volmer Eqn.)
fail
i
Error Limit Test at Each Polarization Offset
Measured vs. Predicted Polarization Current
pass
no
Exceed Iteration Bounds
yes
i
Select Minimum SSE - Report
(Goodness of Fit Test)
48
Enlargement
Observed Polarization
Seattle Tap Water Values
dechlorlnated
pH = 7.2
Cond. = 50 uS
1100 _
Iron corrosion rates were also determined by monitoring the rate of oxygen
consumption in the corrosion reaction. Dissolved oxygen acts as the electron
acceptor in the corrosion reaction; for each mole of iron oxidized to ferrous
iron, one-half mole of 02 is consumed. Therefore, the relationship between
iron corrosion rate and oxygen-depletion rate is more straightforward than
electrochemical measurements, which can be affected by scale development and
changes in water chemistry. Electrochemical corrosion-rate measurements are
especially difficult on heavily scaled iron pipe because of uneven corrosion
rates over the metal surface and because of variation in scale thickness and
characteristics.
Except for the pipe sections to be tested, the mini-loop was composed
entirely of glass and plastic materials. The desired water chemistry was
established in a 4-L glass reservoir. Compressed gases such as C02 and N2
could be sparged into the bottom of the reservoir. After the desired water
chemistry was established, the reservoir was isolated from the circulation
loop. Oxygen-depletion corrosion rate measurements are faster when the ratio
of the pipe surface area to the circulation water volume is high. The
circulation loop contained only about 350 ml of water, which allowed for a
measurable DO decrease in a few hours when one to three pipe sections were
installed in the mini-loop.
51
fi) ) l ) )HD>-
Iron Pipe Sections
xA/W*
Heat Exchanger
l~/ Flow
Meter
4 Liter
Reservoir
Figure 11. Schematic of Pipe Loop for Measuring Corrosion Rate by Oxygen Depletion
Water was circulated using a nylon-lined centrifugal pump. The flow rate was
kept constant at a flow velocity of 0.2 m/s.
Certain precautions were taken to control factors that might interfere with
any of the measurements or interpretations. For example, bacteria present in
the iron scale or growing in the corrosion loop would cause higher DO
depletion rates than the rates for metal oxidation alone. Therefore, the
circulation loop was acid-washed between tests (after removal of pipe test
sections) in order to prevent bacterial growth. If any air leaks were
present, oxygen transfer into loop circulation water would have lowered the
apparent 02 depletion rate. To avoid this, the circulation loop was pressure
tested to detect leaks. Finally, since chlorine can act as an electron
acceptor in the corrosion reaction, water for these tests was dechlorinated.
The ratio of oxygen consumed to the quantity of metal oxidized depends on the
oxidation state of the corrosion products. Therefore, an exact conversion
between the oxygen depletion rate and the corrosion rate requires knowledge
of the iron oxidation state in corrosion scales.
In our experiments iron scales were analyzed for the percentage of iron
present in the ferrous state. The results of this analysis were used in the
conversion between oxygen depletion rates and corrosion rates. A sample
conversion is shown below for a scale in which 40 percent of the iron is in
the ferrous state and the remaining 60 percent is in the ferric state. In
this example it is assumed the three iron pipe sections each with 18.5 cm2 of
surface area are installed in the 0.350 liter mini-loop, and a 1 mg/L DO
decrease occurs over 60 min.
(0.4 x 2 e~) + (0.6 x 3 e") = 2.6 e" released per Fe atom oxidized
1 mg 07 56 mg Fe/mmole x 4 e~/mmole Oo
x 0.35 liters x
liter 32 mg 02/mmole 2.6 e"/mmole Fe
= 0.940 mg Fe oxidized.
0.940 mg Fe 525,600 min
Y _ __________ _ Y
1 cm3 Y
1,000
_
mils/2.54
___________ . _
cm
_ __ __________ _
60 min. year 7800 mg Fe 3 x 18.5 cm^
=7.5 mils/year (mpy).
If water quality conditions change, the oxidation state of iron in the scale
may be altered. If the oxidation state of iron in the scale is unknown or if
it is changing, the oxygen depletion rate cannot be correlated exactly with
corrosion rate. Thus, under transient conditions oxygen-depletion corrosion-
rate measurements must be viewed with reservations. Still, oxygen-depletion
corrosion-rate measurements can provide more information than can weight loss
alone about corrosion-rate changes under changing water chemistry conditions.
54
Accuracy and Precision of DO-Depletion Measurements
55
TABLE 5
Comparison of Corrosion-Rate Measurement Methods
56
The precision of DO-depletion corrosion-rate measurements appeared to depend
on the rate of oxygen depletion. In our experiments the rate of oxygen
depletion was fast when corrosion rates were high and when a greater number
of pipe sections were used during testing. As shown in Table 5, duplicate
measurements at high corrosion rates were fairly precise. The first three
DO-depletion corrosion-rate measurements shown in the Table were made when DO
depletion was rapid (1 mg/L decrease in about one hour). In these tests,
measurements were made with three pipe sections installed in the mini-loop.
Measurements made on the 645-day-old coupon, however, were much less precise.
For this coupon a 1 mg/L DO decrease took several hours, because the
corrosion rate was slow and because only one pipe section with this exposure
history was available for testing. Pipe sections with long periods of pre-
exposure were the least likely to have duplicates with the same exposure
history.
Visual examination of the scale reveals scale uniformity and the extent of
metal surface coverage. The presence of tubercles generally indicates
pitting corrosion as opposed to uniform metal loss. Large tubercles decrease
pipe carrying capacity and increase pumping costs.
The thickness, hardness, and adherency of the scale to the metal surface can
be judged as the scale is physically removed from the pipe surface. These
scale characteristics may indicate the degree of corrosion protection
afforded by the scale. These features can also be used to predict the
likelihood of metal release into the flowing water, since thick loose scales
can be easily sloughed off by changing water chemistry and flow conditions.
58
In this analysis, scale samples were ground to a fine powder and then sieved
through a 200-mesh screen to eliminate inadequately ground particles. The
powder was placed on an aluminum slide and exposed in a Phillips x-ray unit,
model 3005/230, with goniometer model 42202. The x-ray source was a copper
vacuum tube aligned to the K-alpha wavelength.
Corrosion scales were removed from the pipe surface and analyzed for total
metal (iron, zinc, or copper), ortho-phosphate, and carbonate content. The
scales from the iron coupons were also analyzed for ferrous iron content.
Scale Oxidation
Iron scale components generally oxidize to ferric hydroxide when dried in the
atmosphere (Sontheimer et al., 1985). Therefore, as the scale dries its
composition changes. In order to characterize scale composition as it exists
on the pipe surface, scales were analyzed before they were dry. After pipe
sections were removed from the conditioning pipe loops, the wet scale was
immediately scraped off and dissolved in a dilute hydrochloric acid (HC1)
solution. After dissolution, wet scales were analyzed for ferrous iron,
total iron, and carbonate. Some dried scales were also analyzed. The same
analysis methods were used for the dried scale as for the wet scale, except
that the dried scale was weighed before dissolution. The analysis of dried
scale was used to determine which scale components were most resistant to
oxidation.
Ferrous iron was analyzed to evaluate the ratio of ferrous iron to total
iron. Iron scale was allowed to dissolve overnight in a mixture of
hydrochloric acid and deoxygenated water. After dissolution, the iron
solution was diluted using deoxygenated water to obtain an approximate
concentration of 0.25 mg/L ferrous iron. The solution was then analyzed in a
light spectrophotometer using 1,10-phenanthroline as the coloring agent.
Hach reagents and a Milton Roy Company Spectronic 1001 spectrophotometer were
59
used in both the ferrous iron and total iron tests. The 1,10-phenanthroline
will only react with ferrous iron. To analyze for total iron, the 1,10-
phenanthroline is added along with sodium hydrosulfite, a reducing agent
which reduces all the iron in the solution to ferrous iron. Solutions were
first analyzed for ferrous iron and then diluted again (if necessary) and
analyzed for total iron.
This analysis cannot detect small amounts of ferrous iron in the presence of
large amounts of ferric iron. The 1,10-phenanthroline will cause some ferric
iron reduction even if no other reducing agent is added. The analysis of
iron standards prepared entirely with ferric iron indicated the presence of
about 2 percent ferrous iron. Therefore, the analysis of scales containing 2
percent or less ferrous iron were not reliable.
Another problem with the ferrous iron analysis was that some ferrous iron
might have oxidized to ferric iron during dissolution and dilution, even
though the pH was kept low and deoxygenated water was used for dilutions.
Iron solutions were prepared using known amounts of ferric ammonium sulfate
and ferrous ammonium sulfate, and these controls were analyzed using the same
procedures as were used for scale samples. One ferrous/ferric control was
run during each ferrous iron scale analysis. The ferrous/ferric control
solutions lost an average of 12 percent of their ferrous iron during
preparation and analysis. This 12 percent average loss is based on nine
ferrous/ferric controls; ferrous loss ranged between 9 percent and 14
percent. To compensate for this loss, ferrous iron scale sample results were
increased by 12 percent as well. For example, a result of 32 percent ferrous
iron is reported as 36 percent (1.12 x 32 percent) ferrous iron to compensate
for the oxidation suspected to occur during analysis.
Error in the ferrous iron analysis and in all the other iron scale analyses
could be caused by scraping off an unrepresentative scale sample. Soft scale
components were removed much more easily than the hard tenacious scale layers
that formed on some of the iron pipe surfaces. On the other hand, very wet
soft scales often clung stubbornly to the pipe surface and scraping utensil,
and it was difficult to collect the resulting mass for analysis. As reported
by Sontheimer et al. (1985), Kolle and Kuch used dilute acid to remove iron
60
scale for analysis. The dilute acid might have been preferable for
completely removing wet soft scale components from pipe surfaces. However,
the dilute acid removal method was not used in this study in order to be
consistent with the scale removal methods used for weight-loss analysis, and
also because of the potential for dissolving iron metal along with the scale.
Dried scales were analyzed for their metal content. The scale was removed
from the pipe surface, air dried, and then ground into a powder. Next, the
scales were dried in a 105 C oven for 4 hr, cooled in a desiccator, and
finally weighed into a dilute hydrochloric acid solution. These solutions
were analyzed for total metal by atomic absorption spectrophotometry.
Carbonate Analysis
The carbonate analysis used in this research was developed by soil scientists
for determining the carbonate mineral content of soils (Bundy and Bremner,
1972). In this method, a potassium hydroxide (KOH) solution is used to
collect the C02 that evolves when carbonate minerals are dissolved in acid.
Corrosion scale was placed in a flask along with a potassium hydroxide trap.
The KOH trap consisted of a beaker containing exactly 5 ml of KOH. The KOH
was freshly prepared using ^-free water and protected from atmospheric
contamination. This beaker was suspended inside the flask, and the top of
the flask was sealed with a rubber stopper. A small-diameter glass tube
inserted into the rubber stopper was sealed with a soft rubber septum. Next,
a hydrochloric acid solution was injected through the rubber septum using a
syringe. The acid solution flowed down the glass tubing and covered the
scale at the bottom of the flask. As the scale dissolved overnight in the
acid solution, the carbonate present in the scale was evolved as C02 gas and
was collected in the KOH trap. Although it was assumed that all the
carbonate scale dissolved, it was not possible to test this assumption
independently. The next day, the KOH solution was removed from the flask,
and its carbonate content was determined by analysis in a total-carbon
analyzer. The carbon concentration in the KOH was proportional to the
61
carbonate in the scale. The acidic iron solution in the flask was then
analyzed for total metal. The results obtained were expressed as the mole
ratio of carbonate ion to metal.
Phosphate
The weight loss due to corrosion of copper coupons could be evaluated with
acceptable accuracy and precision by weighing the coupons before and after
the polishing process described previously. However, it was difficult to
collect and characterize all the powdery scale that was released during
polishing. Also, acid dissolution of the scale directly from the coupon led
to dissolution of unacceptable amounts of metallic copper. Therefore, an
alternative approach was used for evaluating the oxidation state and
composition of this scale.
The approach used followed the method of Budesinsky (1977) for the
differentiation of oxidized and metallic copper in ores. In this method, the
62
oxidized copper is selectively dissolved by exposure to a solution comprising
36 g ammonium chloride plus 280 ml ammonia hydroxide, diluted to 1.0 L with
distilled water. Budesinsky added sodium pyrophosphate to his etching
solution to mask interferences from iron (III). Since the copper coupons in
this study were not expected to have such interferences, and since it was
desirable to analyze some of the scales for their phosphorus content,
pyrophosphate was not included in these solutions.
During the etch procedure, the interior of a copper coupon is filled with the
ammonia etch solution (approximately 3 ml). After a predetermined contact
period, the etch solution is drained into a volumetric flask. The procedure
is then repeated, refilling the coupon with fresh etch solution and, after an
equivalent contact period, adding the solution to the first volume collected.
The collected solutions are then diluted to a known volume and analyzed for
copper and, in some cases, ortho-phosphate.
The etching contact period varied from 10 to 30 min per solution. Two 10-min
etches of a clean coupon caused a 6.0 mg weight loss, of which 5.0 mg was Cu.
Two 30-min etches caused a weight loss of 11.8 mg, of which 10.3 mg was Cu.
It is not clear whether these weight losses represented dissolution of
oxidized Cu which was on the surface of the clean coupons prior to the test,
or whether the etching process led to some oxidation. Weight loss from
coupons that had been exposed to loop water was always at least three times
that of the corresponding clean coupon, except for one coupon from the 5 mg/L
P loop that lost a total of only 2.3 mg in two 10-min etches. It was
concluded that the procedure can provide a reasonable qualitative and
semiquantitative estimate of the copper content of collected scales.
63
CORROSION OF MILD-STEEL PIPE
Iron weight-loss data are shown as a function of exposure duration for the
various water types in Figures 12 through 17. The corresponding cumulative
average corrosion rates are shown in Figures 18 through 21. In all of these
figures, each data point represents one mild-steel pipe section. Multiple
data points on the same day represent analysis of several individual coupons.
Average corrosion rates were calculated by dividing weight loss by days of
exposure and then converting the results into mils per year (mpy) (1 mil =
1/1000 in.). Thus the rates shown were averaged over the total exposure
period. If the corrosion rate of a pipe section changed during the exposure
period, then the actual corrosion rate occurring at the end of the exposure
period would be different from the cumulative corrosion rate shown.
In the pH 6.0 and the pH 8.0 waters with 0 and 1 mg/L o-P04 -P, weight-loss
data were obtained for pipe sections with almost 700 days of exposure. Few
other corrosion studies have obtained such long-term weight-loss data under
controlled water quality conditions. The oldest pH 8.0/5 mg/L P coupons had
560 days exposure; the oldest pH 7.2/high-carbonate pipe section had 460 days
exposure; the oldest pH 8.4/low-carbonate pipe section had 255 days exposure;
and the oldest pH 8.0/low-carbonate pipe section had 180 days exposure.
A six-month test period has frequently been used in previous potable water
corrosion studies. Table 6 shows average corrosion rates for pipe sections
with approximately six months exposure.
65
5000
4000-
3000-
Weight Loss o
(mg)
o
2000-
o
o
1000-
Figure 12. Weight Loss of Iron Pipes Exposed to Seattle Tap Water,
pH 8.0 (Water 2 from Table 2)
5000
4000-
3000-
o
Weight Loss
(mg) 8
2000- o
ON o
8
1000-
o
4000-
o
o
3000-
Weight Loss
(mg)
2000-
1000-
Figure 14. Weight Loss of Iron Pipes Exposed to the Low Carbonate
Water at pH 8.0 (Water 4 from Table 2)
5000
4000-
3000-
Weight Loss
(mg)
2000-1
vO
1000-
Figure 15. Weight Loss of Iron Pipe Exposed to the Low Carbonate
Water at pH 8.4 (Water 3 from Table 2)
5000
4000-
3000-
Weight Loss
(mg)
2000-
1000-
4000-
I
B
1000-
B
Figure 17. Weight Loss of Mild Steel Coupons in PO4 - Inhibited and
Non-Inhibited Waters.
20-
18- o
o
16-
o o
o
14-
12-
Corrosion
Rate (mpy) 10
8- o
8
8 o
6-
4-
2-
0
0 200 400 600 800
Exposure (days)
18-
16-
14-
o
12-
Corrosion 8 ^
Rate (mpy) 10 '
o ft 8
0 O
8- o
o o o 0 o
6- 0 0
o o
o
0
4-
2-
n-
200 400 600 800
Exposure (days)
20- i :
18-
o
16-
14-
0
o 0
12 *
Corrosion
o pH 8.4. Alk. 20 mg/l (Water 3)
Rate (mpy)
10- pH 8.0. Alk. 13 mg/l (Water 4)
8-
6-
4-
2-
A -
10- o
o
Corrosion
Rate (mpy)
5-
8 o
o
o o
o
8
100 200 300 400 500
Exposure (days)
Number of
Water Type Corrosion Rate Data Points
The pH 8.4 low-carbonate water appeared more corrosive than the pH 8.0
Seattle tap water and the pH 6.0 water; however, limited data were collected
for this water type. The pH 8.0 low-carbonate water was the most corrosive,
producing corrosion rates more than double those in Seattle tap water at the
same pH. Corrosion in the pH 8.0 loop slowed considerably after 9 months.
Table 7 shows water quality parameters and the average corrosion rates for
pipe sections with six months of exposure in the loops with no phosphate
inhibitor.
76
TABLE 7
Water Quality Variables and Corrosion Rates
** -1.1
pH 8.4, low Cj 19.2 + 0.2 8.0 13 0.016
pH 8.0, low Cj 13.3 8.4 20 0.024 -0.5
pH 6.0 10.0 + 1.2 6.0 10 0.310 -3.2
pH 8.0 8.6 + 2.2 8.0 20 0.025 -0.9
pH 8.0, high CT 2.0 0.9 7.2 400 96.5 -0.4
Buffer Intensity
Of the several water quality parameters considered, buffer intensity had the
strongest correlation (an inverse one) with long-term corrosion rates. The
high-carbonate water had a much higher buffer intensity and a much lower
corrosion rate than all of the other water types. The pH 8.0 low-carbonate
water had the lowest buffer intensity and the highest corrosion rates. The
pH 8.4 low-carbonate water had a slightly smaller buffer intensity and a
greater corrosion rate than the pH 8.0 Seattle tap water. The higher
77
corrosion rate in the pH 8.4 water compared to the pH 8.0 water is consistent
with the minimum in buffer intensity at pH 8.3. Hedberg and Johansson (1984)
also found that iron corrosion rates increased when the pH was raised above
8.0.
The only exception to the inverse correlation between buffer intensity and
corrosion rates was the pH 6.0 water, which had a higher buffer intensity
than pH 8.0 Seattle tap water but was slightly more corrosive at long
exposure periods. This exception might have been due to the high chloride
concentration in the pH 6.0 water. The high chloride concentration of this
water is reflected in Table 7 by its high Larson Index. Chlorides could also
be partially responsible for the high corrosion rate in the pH 8.0 low-
carbonate water.
The fact that buffer intensity did not predict the relative long-term
corrosivity of pH 6.0 water compared to pH 8.0 water points to the complexity
of the relationship between water chemistry and iron corrosion behavior. It
is difficult to find any one water quality parameter that will always succeed
in predicting relative corrosivity. In general, however, these results
showed a significant inverse correlation between buffer intensity and
corrosivity. This correlation agrees with the research results of Stumm,
Sontheimer, Hedberg, and others.
Alkalinity
78
The Langelier Saturation Index
pH
There was also no apparent correlation between pH and corrosion rates. The
pH 8.0 low-carbonate water had the same pH as Seattle tap water, but was more
corrosive. The pH 6.0 water had the lowest pH, but was definitely not the
most corrosive. The pH 6.0 water also had the highest chloride concentration
and the highest Larson index; however, it still was less corrosive than the
higher pH, low-carbonate waters. Perhaps the higher buffer intensity gave
some protection against the aggressive chloride ion and the low alkalinity in
the pH 6.0 water.
Many previous researchers have found pH 8.0 water to be more corrosive than
pH 6.0 water. Larson and Skold (1957) and Stumm (1960) both found lower
corrosion rates in waters near pH 6.0 compared with waters at pH 8.0 and
above. To lower the pH, those studies used CC^, which does not increase the
chloride concentration or lower the alkalinity. In the research presented
here, the solutions were acidified with HC1, which raises the chloride
concentration and lowers the alkalinity. The corrosive attack of chloride
ions is a plausible explanation for the difference between this study and the
two previous studies, because chlorides have been found to promote iron
corrosion. On the other hand, both of the previous studies lasted only 100
days. At 100 days exposure, corrosion rates were comparable in the pH 6.0
and the pH 8.0 waters in the research presented here (see Figures 12 and 13).
79
If Stumm or Larson and Skold had continued their studies for a longer time,
the results of their studies may have agreed more closely with the research
presented here.
Phosphate
The weight-loss data for the loops containing 1 mg/L P04 -P have quite a bit
of scatter. In general, they suggest that addition of this much ortho-
phosphate had no effect on iron corrosion rates when compared to the PO^-free
control water at pH 8.0. The addition of 5 mg/L P04 -P did reduce the
corrosion rate by about two-thirds compared to the control. While
concentrations as high as 5 mg/L may be added to new pipe systems for a few
weeks to generate a protective scale layer quickly, common practice is to add
80
no more than about 1 mg/L PO^-P as a maintenance dose. The reason for this
is to avoid a dramatic increase in the cost and the P load on sewage
treatment plants receiving wastewater from an affected community. Indeed, in
localities such as Seattle with open water supply reservoirs, even 1 mg/L
PO^-P would cause algal growth problems in the water distribution network.
Because an unrealistic maintenance dose of PO^ seemed to be required to
significantly reduce iron corrosion rates in the systems in this study, the
discussion of iron corrosion and its inhibition will emphasize the results
from the phosphate-free loops.
TABLE 8
81
Table 8 shows that reliance on short-term corrosion tests can lead to errors
when the results are used to predict long-term corrosion behavior. Until 60
days exposure, the pH 7.2 high-carbonate water appeared more corrosive than
the pH 6.0 water; however, after long-term exposures, the low-pH water was
clearly more corrosive. Pipe sections exposed to pH 8.0 Seattle tap water
also had much higher initial corrosion rates than those exposed to pH 6.0
water. Later, corrosion rates in the pH 8.0 water appeared to be the same as
in the pH 6.0 water.
These results agree with those of both Larson (1957) and Stumm (1960), who
found that initial corrosion rates did not predict corrosivity at longer
exposures. In those studies, corrosion rates measured after a week were not
well-correlated with corrosion rates measured after several months of
exposure.
The corrosion rates in the pH 6.0 water were very low initially (less than 60
days exposure), but then seemed to increase. After about 60 days exposure,
corrosion rates stabilized and then slowly declined. A plausible explanation
for this trend is that corrosion initiation sites form more slowly in the pH
6.0 water than in the higher-pH waters, but that once sites are initiated,
corrosion increases rapidly until a protective scale develops, at which time
corrosion rates decline once again.
Seattle tap water at pH 8.0 caused a weight loss similar to pH 6.0 Seattle
tap water until about a year of exposure. Thereafter, almost all of the pH
6.0 pipe sections had higher weight loss than the pH 8.0 pipe sections with
similar exposure periods. Table 9 shows a comparison of corrosion rates in
the pH 6.0 and pH 8.0 waters after a year or more of exposure.
82
Corrosion rates in the pH 8.0 water declined steadily during the entire test
period. No similar consistent decline was apparent for the iron pipe
sections exposed to the pH 6.0 water.
TABLE 9
Long-term Corrosion Rates in pH 6.0 and pH 8.0 Seattle Tap Water
(Calculated by Cumulative Weight-Loss Method)
Comparison with Previous Corrosion Research Conducted Using Seattle Tap Water
83
Hoyt et al. (1982) and Herrera and Hoyt (1984) also evaluated the effect of
corrosion-control water treatment in Seattle source waters. They used
six-month-long pilot-plant studies and slightly different chemical treatment
methods than those used in the Kennedy study. In the 1982 study, Cedar river
water was dosed with lime. In the 1984 study, Tolt river water was treated
by two methods: one test was conducted by adding lime and soda ash, and
another test was conducted with lime and sodium bicarbonate. In the Hoyt et
al. and Herrera and Hoyt studies, the final pH of the treated waters was
about 8.0.
Water quality characteristics and iron corrosion rates from the corrosion-
control pilot-plant studies are shown in Table 10. As shown in the table,
the corrosion-control treatment methods used in these studies produced only
minor changes in iron corrosion rates. In the Kennedy studies, treatment
slightly decreased mild-steel corrosion rates in both the Tolt and Cedar
river waters. In the Hoyt and Herrera studies, treatment increased mild-
steel corrosion rates slightly in the Tolt river water and decreased them
slightly in the Cedar river water.
The reason this type of corrosion treatment had little effect on mild-steel
corrosion rates may have been that the water chemistry changes produced by
treatment had opposing effects on iron corrosion rates. In all of the
studies, alkalinity and hardness were added as a corrosion-control treatment.
The treatment raised the pH and lowered the buffer intensity of the test
waters. According to the siderite model, and in concordance with the results
obtained in this study, the lower buffer intensity could have caused higher
corrosion rates. On the other hand, these treatment methods also resulted in
increases in alkalinity and calcium ion concentrations and decreases in the
Larson Index. Since increases in alkalinity and calcium ion concentrations
and decreases in the Larson Index generally lower iron corrosion rates, these
changes may have partially compensated for the lower buffer intensity. In
this study, corrosion rates in the pH 8.0 Seattle tap water were also not
very different from the corrosion rates of iron pipe exposed to the pH 6.0
water.
84
TABLE 10
Seattle Corrosion-Control Pilot-Plant Studies
Average Corrosion Rates Obtained on Mild-Steel Pipe
85
IRON CORROSION SCALE ANALYSIS
All of the low-alkalinity waters produced iron scales that were similar in
physical appearance and very different from the scale produced in the pH 7.2
high-carbonate water.
The low-alkalinity waters all produced thick, loose scales. The exterior
scale layer was a dark orange-brown crust. This outer layer also contained
flecks of yellow-colored scale. The crust was rigid but could easily be
broken through to reveal a soft porous interior. The interior was composed
of fine granular material that was black or greenish-black. This scale could
be removed by gentle scraping. On pipe sections exposed to pH 6.0 and pH 8.0
Seattle tap water, well-developed scale was approximately 1/8 in. thick. In
the low-carbonate, high-pH waters, scales were slightly thicker. In all of
the low-alkalinity waters, scale thickness was very uneven; scales consisted
of tubercles (mounds of scale) overlying wide shallow pits in the metal
surface.
In all water types, corrosion generally occurred first near metal stress
points such as the machine-cut edges and the butt weld. Machining and
welding stress the metal's crystal lattice and cause these areas to be more
susceptible to corrosive attack (Evans, 1960). Therefore, the extent of
metal deformation found on a particular pipe section could have influenced
initial weight loss. Pipe-to-pipe variations in welding and machining
explain some of the variation in weight loss among pipes with the same
exposure history.
86
high carbonate water, corrosion never completely covered the pipe surface,
even in the oldest pipe section with 460 days of exposure.
87
pipe and in older pipes exposed to low-alkalinity waters resembles the
photographs of siderite from a study by Kolle and Rosch (1980).
Corrosion-Scale Weights
The weights of corrosion scales that formed on iron pipe sections are
shown in Figure 22. The low-carbonate, high-pH waters produced the thickest
and heaviest scales. These waters also produced the highest corrosion rates,
thereby emphasizing the fact that thick scale formation does not necessarily
provide corrosion protection. In fact, although a thick scale developed,
corrosion rates in the pH 8.0 low-carbonate water did not decline at all
during the six months of testing.
On the other hand, scale formation had a very dramatic effect on corrosion in
the pH 7.2 high-carbonate water. The formation of a protective scale during
the first 60 days of exposure correlated with a sharp decline in corrosion
rate. In this water type, the total weight loss at 56 days was not
appreciably different from that at 200-300 days. Similarly, the scale weight
at 60 days is approximately the same as the scale weight at 200 days. These
data suggest that little or no corrosion occurred after formation of this
protective scale layer.
Scale weight in the 1 mg/L PO^P loop was comparable to that in the pH 8.0
loop with no P04 . In the 5 mg/L P04 -P loop, scale weight was about one-third
of that in the loops with 0 or 1 mg/L-P04 -P. These scale weights correlate
reasonably well with the relative corrosion rates in the three loops. There
was no evidence that scale weight or corrosion rate had reached a plateau in
either of the loops with P04 inhibitor.
Air-dried iron scales were analyzed using powder x-ray diffraction. The
results of this analysis are shown below.
7000
6000-
5000-
o pH = 6.0 (Water 1)
pH = 8.0 (Water 2)
4000- ;; PO4 - 1 mg/l (Water 6)
A Low Carbonate (Water 3,4)
Scale D PO4 - 5 mg/l (Water 7)
8
3000 H M High Carbonate (Water 5)
0
a
B a
2000- D
o
o
1000-
Figure 22. Scale Weight Formed on Iron Pipe Exposed to Different Water Qualities.
(Water 3 and 4 Data are not separated)
Water Type Iron Compound Detected
Goethite was identified in all of the iron scales in the P04 -free loops,
while siderite was found only in the high-carbonate scale. These results
suggest that siderite was a constituent of the hard tenacious scale layer
that composed much of the high-carbonate loop scale. Scale x-ray diffraction
patterns were checked for other iron compounds such as hematite, magnetite,
other crystalline forms of ferric hydroxide, green rust, and for calcium
solids such as calcite and aragonite; however, no other compounds were
identified.
90
Chemical Analysis of Iron Corrosion Scale
Dried, powdered iron scales were analyzed for their total iron content, and
the results are shown in Table 11. The iron content was higher in scales
that were stored a long time before analysis than in scales that were more
recently sampled. Oven-drying the scales at 105 C also appeared to increase
the iron content, probably by driving off bound water.
TABLE 11
Iron Content of Dried Iron Corrosion Scales
The compositional changes that occurred during storage and air-drying show
that analyses of dried scale cannot be used to draw definitive conclusions
91
about the scale as it exists on the pipe surface. After oven-drying and
storage, the iron content of most scale samples approached the 63 percent
iron content of goethite, and goethite was found in the x-ray analysis of all
the dried scales.
Ferrous Iron
The results of ferrous iron analyses are shown in Table 12. Unless otherwise
noted, the results.in this table are for wet scale samples. Some iron
corrosion scales appear to oxidize quickly when air dried. For instance,
less than 1 percent ferrous iron was found in a Seattle tap water scale
removed from a pipe surface and then oven dried for 4 hr. Appreciably more
ferrous iron was found when this scale was dissolved in dilute hydrochloric
acid and analyzed immediately after the pipe section was removed from the
conditioning pipe loop.
For several measurements, scale samples were removed from two different pipe
sections with the same exposure history. These duplicate analyses are shown
as two entries in Table 12. With the exception of one high-carbonate scale
sample, the duplicate ferrous iron results were quite close. The differences
found in these duplicate samples may have been due to either analytical error
or to actual pipe-to-pipe variability.
In Table 12, iron scales which formed in pH 8.0 and pH 6.0 Seattle tap water
are grouped together. In general, the ferrous iron content of these two
scale types was similar and tended to increase with the length of the
conditioning period. One exception to this trend occurred in a pipe section
92
TABLE 12
WET SCALES
% Fe+2
Water Type Exposure Period (% of Total Iron)
(days)
Seattle Tap Water
pH 8.0 172 16
pH 8.0 203 40
pH 6.0 351 27
pH 8.0 473 30 and 32
pH 6.0 528 32 and 36
Low-Carbonate Water
pH 8.0 and 8.4 139 41 and 46
256 48 and 48
285 36
394 36
High-Carbonate Water
pH 7.2 198 days 6 and 24
203 days 7
205 days 5 and 7
284 days 7
DRIED SCALESt
% Fe+2
Water Type Exposure Period (% of Total Iron)
(days)
*Analysis done immediately after scale removal from the pipe surface;
scales were not allowed to air dry.
^Analysis done on dried iron corrosion scales.
93
exposed for 203 days. Since scale thickness also increased with exposure,
increased ferrous iron correlated with thicker scales. Likewise, scales
formed in the low-carbonate waters were the thickest and contained the
highest percentage of ferrous iron. In contrast to the pH 8.0 and pH 6.0
scale samples, the ferrous iron content of the low-carbonate scales appeared
to decrease slightly after about nine months exposure.
The most dramatic difference between the high-carbonate scales and the other
scales is that the small amount of ferrous iron in the high-carbonate scales
was apparently resistant to oxidation. There was no detectable loss of
ferrous iron when high-carbonate scale samples were air-dried. This
oxidation-resistant ferrous iron may have been present as the siderite
mineral detected by x-ray analysis. The resistance of siderite to oxidation
has been previously reported by Ghosh et al. (1967), who found no measurable
oxidation of siderite solids that were submerged in oxygenated water for two
weeks. The amount of ferrous iron detected in the high-carbonate scale
samples suggests that 5-10 percent of the total iron is in the form of
siderite in these scales.
In the research presented here, the ferrous iron content of scales was
inversely related to the scale's protectiveness. The pH 7.2 high-carbonate
water produced a scale that contained the least ferrous iron and provided the
most corrosion protection, whereas the low-carbonate waters produced a scale
that had the most ferrous iron and provided the least corrosion protection.
This relationship is in direct contrast to the results reported by Sontheimer
94
et al. (1981). In that study, the scales containing the highest percentage
of iron in the ferrous state were the most protective ones. In protective
scales, ferrous iron was about 90 percent of the total iron, while other less
protective scales were only about 40 percent ferrous. The difference between
the results presented here and Sontheimer's results may have been caused by
the different age of the scales that were analyzed. In the Sontheimer study,
scales were analyzed after only a week or two of exposure. In the study
presented here, scales were several months old. Since the ferrous iron
content of nonprotective scales seemed to increase with the length of
exposure, the initial fraction of ferrous iron in these scales might have
been very low.
At any rate, the results presented here support the thesis that oxidation-
resistant ferrous iron in the form of siderite is an important component of
protective scale. The easily oxidizable ferrous iron formed in low-
alkalinity waters was not as protective. Even though the oxidation-resistant
ferrous iron (presumably as siderite) composed only 5-10 percent of the high-
carbonate scale, this scale was highly protective.
Carbonate Content
Both the freshly sampled iron scales and the dried, powdered iron scales were
analyzed for their carbonate content. The results are shown in Table 13.
Results are given as the molar ratio of carbonate to iron. If all the
carbonate found in the scale was present as FeC03 , then the molar ratio of
carbonate to total iron represents the fraction of total iron present as
FeC03 .
The carbonate results exhibited a similar pattern to those for ferrous iron.
In scales formed in pH 8.0 or pH 6.0 Seattle tap water, or in the low-
carbonate water, the carbonate disappeared when the scales were dried. In
the scale formed in pH 7.2 high-carbonate water, the carbonate was retained,
and dried scales had approximately the same carbonate content as wet scales.
In the scales formed in the low-alkalinity waters, the mineral form of the
carbonate present was not determined, since no carbonate minerals remained in
95
TABLE 13
WET SCALE
C03 - 2/FeT C03 2 YFe+2
Water Type Exposure Mole Ratio Mole Ratio
(days)
DRY SCALE
C03 - 2/FeT C03 2 7Fe+2
Water Type Exposure Mote Ratio Mole Ratio
(days)
96
scales that were dried for x-ray diffraction analysis. The transient nature
of the carbonate may indicate that it was not in a crystalline mineral form.
Similar to the ferrous iron content, the carbonate content of wet scales
seemed to increase with length of the scale-formation period.
Table 13 also shows the mole ratio of carbonate to ferrous iron. As shown by
this ratio, scales formed in the high-carbonate water contained approximately
the same molar amount of ferrous iron and carbonate. This may mean that all
the ferrous iron present in the high-carbonate scales was present as FeC03
(siderite).
To summarize the above results, siderite was identified only in the most-
protective iron scales. The ferrous iron and carbonate detected in these
scales were not lost by oxidation or decomposition when the scale was removed
from the pipe surface and dried. The scale was thin, dark, hard, and
adherent. Scale analysis suggests that siderite composed 5-10 percent of the
scale formed in the high-carbonate water. The siderite formed in a highly
buffered, high-alkalinity water at near neutral pH. Siderite was not
detected in the less-protective scales formed in low-alkalinity waters.
Most of these results are consistent with the siderite model proposed by
Sontheimer et al. (1981). In that model, high buffer intensity and
alkalinity are thought to promote siderite formation, and the presence of
siderite is thought to promote the formation of protective iron scales. Only
the detection of moderate amounts of ferrous iron and carbonates in the
less-protective scales fails to correlate with this model.
Both carbonate and ferrous iron are necessary for siderite formation and are
therefore important elements of the siderite model. The thermodynamic
equation governing siderite solubility is: Ks = io~ 10 - 68 = (Fe+2 )(C03 ~ 2 ) at
o o
25"C, where (Fe ) and (C03 ~^) are the molar activities of ferrous iron and
carbonate in solution.
Dissolved ferrous iron can form aqueous complexes with both hydroxide and
bicarbonate anions. The primary complexes that form under normal drinking
water conditions are FeOH+ , FeHC03+ , and Fe(OH) 2 (aq). Since only
uncomplexed ferrous iron affects siderite solubility, these complexes
increase the concentration of dissolved ferrous iron necessary to form
siderite. Soluble ferrous hydroxide and carbonate complexes were included in
the total soluble ferrous iron shown in Figure 23. The contribution from
98
-2
Log Ferrous
Iron Cone. 4
(M/L)
-6-
pH
Figure 23. Total Soluble Ferrous Iron in Equilibrium with Siderite. Data Points
Represent Water 2 and Water 5 on Upper and Lower Lines, Respectively
ferrous iron complexes was slight; uncomplexed ferrous iron is the dominant
ferrous iron species throughout most of the pH region shown. The ferrous
hydroxide complex, FeOH+ , begins to predominate above pH 9.3.
0.15-
0.12-
0.061
0.03-
Figure 24. Oxidation Rate of Ferrous Iron in Equilibrium with Siderite. Data Points
Represent Water 2 and Water 5 on Upper and Lower Lines, Respectively
with siderite is 1.1 x 10"^, and the oxidation rate is 1.8 x 10" 4 mol/L-min.
In the high-carbonate water at pH 7.2, the ferrous iron concentration in
equilibrium with siderite is 3.3 x 10~ 6 , and the oxidation rate is 1.4 x 10" 6
mol/L-min, almost 100 times less than in pH 8.0 Seattle tap water. In the pH
7.2 high-carbonate water, both the higher carbonate concentration and the
lower pH contribute to a lower iron oxidation rate and therefore promote
siderite formation.
102
5- Ct = 9.1 E-3 M/L
4-
Buffer Intensity 3-
mM/L-pH unit
2H
Figure 25. Calculated Buffer Intensity as a Function of pH for Waters With the
Indicated Concentration of Inorganic Carbon and No Other Weak Acids or Bases.
Data Points Represent Water 2 and Water 5 on Lower and Upper Lines, Respectively
A Quantitative Iron Corrosion Model Based on Siderite Formation
A complete model of iron corrosion and scale formation would be very complex.
Such a model would include the oxidation reactions of iron metal and ferrous
iron, the reduction reaction of oxygen, the precipitation reactions of all
potential iron solids, and the changes occurring in already-formed iron
solids such as the dehydration and oxidation of precipitated ferrous solids
to form goethite. Such a model would also have to account for transport of
reactants to the metal surface and of products away from the surface. This
transport is influenced not only by the flow of water through the pipe but
also by the properties of the developing scale and by the nature of the
anodic and cathodic reactions.
On some metals, equilibrium modeling can be used to predict the water quality
conditions that will cause protective scales to form. Unfortunately, the
equilibrium modeling of iron corrosion scale formation is severely limited by
incomplete knowledge of the identity and crystalline structure of solids that
exist in the scale (Snoeyink and Kuch, 1985). Even with x-ray diffraction
analysis and wet chemical analysis, it was not possible to determine the
complete composition of the corrosion scales. The siderite model suggests a
general explanation for how water quality affects the formation of protective
scale, but it does not account for all of the reactions that may occur in the
iron corrosion process.
104
Using these assumptions, the production rate of ferrous iron from the
corrosion reaction was compared with the oxidation rate of ferrous iron.
From this comparison, it was determined whether it was possible for the
ferrous iron concentration to build up enough for siderite to precipitate.
Because reliable data was not available for the chemical conditions at the
metal surface, the chemical conditions were assumed to be the same as those
in the bulk solution. The pH and total carbonate concentration of the bulk
solution were used in calculations. The oxygen concentration was assumed to
be the saturation value in equilibrium with an atmospheric oxygen
concentration of 0.21 atmospheres partial pressure.
In the above calculations, the ferrous iron production rate was about an
order of magnitude different than its oxidation rate for water qualities
similar to both Seattle tap water and the high-carbonate water. Since the
difference between the two rates was large, the conclusions drawn from the
model are fairly insensitive to small variations in the assumed values of the
105
corrosion rate~and scale thickness. The assumptions of a 2-mm-thick scale
layer and a corrosion rate of 6 mpy are reasonable approximations based on
the results of this research. However, the conclusions from the model would
be the same even if slightly different values had been used.
These rough calculations illustrate how oxidation rates could limit the
potential for siderite formation by not allowing the ferrous iron to build up
to concentrations high enough for siderite precipitation. Although this
crude model cannot represent the real iron corrosion and scale formation
mechanisms because important details such as diffusion and concentration
gradients are ignored, its correspondence with observed trends is satisfying.
The scale formed in Seattle tap water contained ferrous iron; however, this
ferrous iron oxidized as the scale dried. It is possible that this ferrous
iron was incorporated into a mixed ferrous/ferric solid like FegC^ or
amorphous green rust. Much lower ferrous iron concentrations are necessary
for formation of these mixed ferrous/ferric solids than are required for
siderite formation. The ferrous iron produced by corrosion in Seattle tap
water is therefore either oxidized or precipitated as a mixed ferrous/ferric
solid before it can build up to concentrations high enough for siderite
precipitation. The formation of these mixed ferrous/ferric solids might
explain why the ferrous content of scales in our study did not correlate with
scale protectiveness as they did in the experiment reported by Sontheimer et
al. (1981). Although ferrous iron was present in the scales, none was
present in the siderite form.
Corrosion scales may hinder the diffusion of oxygen to the metal surface and
cause reducing conditions to develop beneath scale layers. If siderite was
present in the small amounts of hard, tenacious scale formed in the low-
alkalinity waters after long exposure periods, it may have been formed by
this mechanism. When less oxygen is present, ferrous iron oxidation rates
decrease and higher concentrations of ferrous iron can result. This may
cause the formation of siderite even when conditions in the bulk solution are
unfavorable. Even though siderite only composed 10 percent or less of the
high-carbonate scale, this scale was still protective, presumably because the
other components of the scale retained the dense structure of siderite. In
106
the siderite model, scales do not have to be entirely composed of siderite to
be protective. Siderite is thought to form first near the metal surface and
then promote the formation of protective scale layers, but the protective
layers may be composed of other minerals such as ferric hydroxides that
retain the siderite structure. In the absence of siderite, precipitated
ferrous and ferric minerals form loose, amorphous, and nonprotective scales.
In this study, 90 percent to 95 percent of the iron in the protective scales
which formed in the high-carbonate water was in the ferric form. X-ray
diffraction analysis suggests that the ferric iron was present as goethite.
This goethite may have retained the dense structure of siderite.
Corrosion-Rate Measurements
107
third experiment, two pipe sections initially exposed to the pH 8.0 Seattle
tap water were monitored after exposure to pH 7.2 high-carbonate water.
The results of these experiments are shown in Figures 26 through 28. The
initial oxygen-depletion corrosion rate is shown on the left-hand edge of
each graph. The initial rate shown on the graph is the average of three
measurements taken on the pipe sections shortly before their exposure to the
new water type.
When Seattle tap water pipe sections were placed in the pH 7.2 high-carbonate
water, the corrosion rate initially increased. After several weeks, however,
the corrosion rate decreased below the rate measured before exposure to the
high-carbonate water.
2.5-
2.0-
Corrosion
Rate, mpy
o
vO
20 40 60 80 100 120
Days exposed to pH 8.0 Water
10-
Corrosion g-
Rate, mpy
40 60 80
14-
12-
10 H
8
Corrosion
Rate in 8-
Seattle
Tap Water
61
4-
2- Q
0
0 10 20 30 40 50 60 70 80 90 100
Number of Days in High Carbonate Water
Figure 28. Seattle Tap Water Scale (from Water 2) Exposed to High Carbonate Water (Water 5)
TABLE 14
Weight-Loss Data From Changing Water Chemistry Experiments
High-carbonate pipe sections exposed to pH 8.0 Seattle tap water had a higher
weight .loss than pipe sections exposed only to high-carbonate water for the
same total exposure period. In contrast, corrosion-rate measurements by the
dissolved-oxygen-depletion method showed no increase in corrosion rates after
the high-carbonate pipe sections were exposed to pH 8.0 Seattle tap water.
These conflicting results may be explained by referring to Figure 16, the
cumulative weight loss graph for pipe sections exposed to pH 7.2 high-
carbonate water. From this graph, weight loss in the high-carbonate water
does not appear to increase with exposure period. This may be because very
little corrosion occurred on high-carbonate pipe sections after the formation
of a protective scale layer. Thus, if these pipe sections had been left in
the high-carbonate water instead of being transferred into Seattle tap water,
their dissolved-oxygen-depletion corrosion rates would probably have
112
decreased to near zero. Therefore, even though the corrosion rates of the
high-carbonate pipe sections did not increase in pH 8.0 Seattle tap water,
they also did not decrease significantly, and so they still experienced more
weight loss. Unfortunately, two factors increase the difficulty of comparing
the two types of corrosion-rate measurements in this experiment: the
significant pipe-to-pipe weight-loss variations in the high-carbonate water,
and the large variations that occur when measuring very low corrosion rates
by the dissolved-oxygen-depletion method.
Scale Analysis
After the experiments were complete, the pipe sections were examined
visually, and in some cases the scale was removed for analysis. Over the
course of the experiment, the appearance of the scales did not change
appreciably. However, the high-carbonate pipe sections exposed to the pH 6.0
water and the pH 8.0 Seattle tap water did seem to thicken somewhat. They
also appeared to gain a higher percentage of soft dark material than that
which was present in regular high-carbonate scales. The pH 8.0 Seattle tap
water scale exposed to the high-carbonate water still had the lumpy, thicker
structure indicative of tubercle formation. However, the tubercles were less
pronounced than those on regular pH 8.0 Seattle tap water pipe sections of
the same exposure period.
The removed corrosion scales were analyzed twice for their ferrous iron and
carbonate contents; immediately upon removal (while still wet) and after oven
drying. The results of these analyses are shown in Table 15. For
comparison, Table 15 also shows representative values found in regular pH 7.2
high-carbonate and pH 8.0 Seattle tap water scales.
113
TABLE 15
Ferrous Iron Content: Results given as the percentage of iron in the ferrous
state.
Seattle tap water scales; 203 days of exposure.
Wet Scale: 40
Dry Scale: < 1
Seattle tap water for 4 months; transferred to the high-carbonate water for 4
months; 230 days of total exposure.
Wet Scale: 18
Dry Scale: 9 and 10
High-carbonate water for 4 months; transferred to Seattle tap water for 4
months; 233 days of total exposure.
Wet Scale: 19
Dry Scale: 11
High-carbonate water scale; 203 and 205 days of exposure.
Wet Scale: 5-7
Dry Scale: 6
Carbonate Content: Results given as the mole ratio of carbonate to iron metal.
Seattle tap water scales; 203 days of exposure.
Wet Scale: 0.072
Dry Scale: <0.01
Seattle tap water for 4 months; transferred to the high-carbonate water for 4
months; 230 days of total exposure.
Wet Scale: 0.097
Dry Scale: 0.072 and 0.087
High-carbonate water for 4 months; transferred to Seattle tap water for 4
months; 233 days of total exposure.
Wet Scale: 0.095
Dry Scale: 0.092
High-carbonate water scale; 203 and 205 days of exposure.
Wet Scale: 0.054, 0.086, and 0.092
Dry Scale: 0.011
114
The most striking result of this analysis is that when the coupons from the
pH 8.0 Seattle tap water loop were exposed to the pH 7.2 high-carbonate
water, new scale containing ferrous iron and carbonate formed, and these
components remained after the scales were dried. Oxidation-resistant ferrous
iron and carbonate had previously been found only in high-carbonate scale.
This may indicate that siderite was formed in these scales after exposure to
the high-alkalinity water. The formation of siderite correlates with the
decreased corrosion rates measured toward the end of the exposure period
involving the high-alkalinity water.
The high-carbonate scale exposed to pH 8.0 Seattle tap water retained its
oxidation-resistant ferrous iron and carbonate content. The wet scale,
however, had more ferrous iron than the dry scale and had a higher ferrous
iron content than most of the regular high-carbonate wet-scale samples. This
high-carbonate scale appeared to contain some of the easily oxidized ferrous
iron normally found only in the Seattle tap water scales. This may mean that
some of the ferrous iron in these scales was incorporated into less-
protective, mixed ferric/ferrous solids instead of into the more-protective
siderite layer.
The corrosion rates of pipe sections exposed to pH 8.0 Seattle tap water
increased immediately after the coupons were exposed to the pH 7.4 high-
carbonate water, and later decreased. The initial increase may have been
caused by either the higher conductivity of the water or its lower pH. To
investigate this issue, a few short-term experiments were conducted where the
conductivity of the exposure water was increased while the pH was held
constant.
In these experiments the corrosion rate of prescaled pipe sections from the
Seattle tap water loop was first measured by the dissolved-oxygen-depletion
method. The coupons were then exposed to one of two higher-conductivity
waters, and the corrosion rate was measured again. In one solution,
115
conductivity had been increased by addition of sodium chloride, and in the
other solution the same increase in conductivity had been induced by addition
of a mixture of hydrochloric acid and sodium bicarbonate. The final pH (8.0)
and conductivity (350 /imhos/cm) of both solutions was the same, but the
former had a higher chloride content and the latter a correspondingly higher
bicarbonate content.
As shown in Table 16, the corrosion rate increased in both cases, and the
magnitude of the increase was comparable. The absence of a significant
difference between the corrosion rates in the two high-conductivity solutions
suggests that the chloride ion was not acting primarily as a specific
promoter of corrosion; rather, the conductivity seemed to act as a
nonspecific corrosion promoter, with chloride and bicarbonate contributing
equally to the effects.
These experiments provide an excellent example of the type of data that can
be collected using the DO depletion method for evaluating corrosion rate.
For these experiments, weight loss would have been too small to detect and,
as noted earlier, electrochemical measurements on iron surfaces are very
imprecise due to variations in electrical potential along the surface.
Understanding the strengths and limitations of each approach allows the
experimenter to choose the best method for the particular experiment and
thereby obtain the most accurate results.
116
TABLE 16
Response of Corrosion Rate to High-Conductivity Water
Trial #1: Pipe sections were exposed to pH 8.0 Seattle tap water for 62
days before testing.
'Trial #2: Pipe sections were exposed to pH 8.0 Seattle tap water for 47
.days before testing.
tTrial #3: Pipe sections were exposed to pH 8.0 Seattle tap water for 62
days before testing.
Stumm (1960) also showed that a higher solution conductivity can cause higher
initial corrosion rates. After an initial conditioning period, Stumm found
that the effect of high solution conductivity was masked by the formation of
protective scales. In the study presented here, corrosion rates decreased
after the Seattle tap water pipe sections were exposed for several weeks to
the high-carbonate water, presumably because a more protective scale
developed. Like Stumm's experiments, the protective scale counteracted the
corrosion-rate increase caused by the increased conductivity.
SUMMARY
To summarize this part of the work, the corrosion rates of mild-steel pipes
and the composition and structure of corrosion scales were related to water
quality variables.
117
The most protective iron corrosion scale was formed in a high-alkalinity,
moderate-pH water (alkalinity 400 mg/L as CaC03 , pH 7.2). Pipe sections
coated with this scale had very low corrosion rates (about 2 mpy or less).
The protective scale was thin, hard, dark, and very adherent to the metal
surface. Siderite was found in this scale by x-ray diffraction analysis.
This was the only scale where siderite was detected. Wet chemical analysis
of this corrosion scale before and after air-drying showed that 5-10 percent
of the iron was in the ferrous state. Small amounts of carbonate were also
found in both wet- and dried-scale samples; the ratio of carbonate to total
iron was approximately the same as the ratio of ferrous iron to total iron.
This suggests that most of the ferrous iron in this scale was present as
FeC03 , i.e., siderite.
Scales formed in the low-alkalinity waters were thick and loose, consisting
of large tubercles covering shallow pits. After nine months to one year of
exposure to low-alkalinity water, small amounts of hard, tenacious scale were
formed under the soft, loose scale layers. Goethite (a-FeOOH) was the only
mineral detected in the x-ray diffraction analysis of dried low-alkalinity
scales. However, goethite was not the only compound present in these scales,
since ferrous iron and carbonate were detected in scale samples before they
were air dried.
118
In general, increases in corrosion rates correlated with decreases in buffer
capacity. In two waters, the buffer capacity of Seattle tap water was
decreased by bubbling the water with CC^-free air. Both of these waters were
more corrosive than regular pH 8.0 Seattle tap water. The pH 7.2 high-
carbonate water had by far the highest buffer capacity and was much less
corrosive than any of the low-alkalinity waters. The pH 6.0 Seattle tap
water was the only exception to the correlation between buffer capacity and
corrosivity; it had a higher buffer capacity than Seattle tap water, but was
also more corrosive. At first, the pH 6.0 water produced similar corrosion
rates to the pH 8.0 Seattle tap water, but after a year of exposure the pH
6.0 water was more corrosive.
119
CORROSION OF COPPER PIPES
Figures 29-32 show the results for weight loss in the four loops where ortho-
phosphate was not present, and Figure 33 presents the data for all these
loops on a single graph. Considering that the weight loss represents about
0.02 percent to 0.4 percent of the total weight of the coupon, and that every
data point reflects an independent experiment, the results for each of the
four data sets are remarkable for their internal consistency. In fact, in
each loop the cumulative weight loss not only increases quite steadily with
duration of exposure, but the data are reasonably linear. It appears that
the corrosion rates in the pH 6.0 Seattle tap water and the pH 7.2 high-
carbonate water may be somewhat greater during the first few weeks of
exposure than thereafter, but the corrosion rate in each of the loops is
nearly constant from about 50 days of exposure until the end of the test,
which lasted between 400 and 700 days.
150-
Weight Loss
(mg)
o
o e
50- o 8
100-
Weight Loss
(mg)
N) o
OJ
o
50-
o
8
o
2(XH
o
o
8o
150-
Weight Loss
o
(ing)
0 8
100-
50-
08
o
200-
150-
Weight Loss
(rag) 8
o
100-
(O
00
8
o
50-
Figure 32. Weight Loss of Copper Pipe Exposed to Seattle Tap Water
Adjusted to pH= 6.0 (Water 1)
250
200-
g
D
150-
Dn e o pH= 6.0 (Water 1)
Weight Loss pH= 8.0 (Water 2)
(mg) 8
o pH= 8.0 Low CO3 (Water 4)
100-
n pH= 7.2 High COS (Water 5)
NJ
50-
cflo
i _
*
ol
220 -
i
180- Q Cathodic Tafel Slope
* Anodic Tafel Slope
Ba,Bc
(mV/dec)
100 -I
60
20
2.0
1.5:
lo i.o:
(uA/cm2)
o.s:
1
o.o
.01 i 10 100 1000
Surface Exposure (days)
Figure 34. Variation in Corrosion Current and Tafel Slopes on an Aging Surface Exposed to
pH 6.0 Tap Water (Water 1). Error Bars Represent One Standard Deviation of
Replicate Measurements on Different Surfaces
virtually no further change during the next 200 days. The figure also shows
the changes in Tafel slopes over the 200-day period. The increase in both
slopes indicates that both the oxidation and reduction reaction are being
inhibited over time, which is consistent with the formation of a thin scale
gradually covering the pipe surface, blocking both anodic and cathodic sites.
The corrosion rates during the majority of the test period can be estimated
from the slopes of the graphs. Within the limits of precision attainable
from such an analysis, the corrosion rates in the loops with pH 6.0 or 8.0
tap water and with pH 7.2 high-carbonate water were identical from day 50
forward. Almost all the data points for the pH 8.0 tap water loop indicated
that less corrosion had occurred in this loop than in the other two loops
after equivalent exposure periods, but these differences seemed to reflect
different rates during the first 50 days of testing, rather than
substantially different rates thereafter. The corrosion rate in the pH 8.0
low-carbonate loop was about one-third of that in the other three throughout
the test period.
Figure 37 displays the results for the accumulated scale on the coupons in
the phosphate-free systems after various exposure histories. Whereas the
weight loss is a measure of the overall corrosion rate, scale weight reflects
combined effects of corrosion rate, chemical composition of the scale, and
128
250
200-
1501
o pH= 6.0 (Water 1)
Weight Loss
(mg) 8 pH= 8.0 (Water 2)
o
100- pH= 8.0 PO4= 1 mg/L (Water 6)
50-
o
<p >
ol
D C
70-
60-
50-
o pH= 6.0 (Water 1)
20- o
o
t
10-
a
a
70-
60-
50-
o pH= 6.0 (Water 1)
Scale
Weight 40 ~ pH= 8.0 (Water 2)
(mg)
pH=8.0LowCO3(Water4)
30-
B a pH= 7.2 High CO3 (Water 5)
o DO
20-
- S-'
10-
Figure 37. Scale Weight on Copper Pipes Exposed to Different Water Qualities
the tendency of the scale to adhere to the pipe walls and resist erosion.
Although the pH 6.0 and 8.0 tap water loops and the high-carbonate loop had
similar corrosion rates, coupons exposed to pH 8.0 Seattle tap water had much
more adherent scale than any of the other coupons.
The composition of the scale in the copper coupons was analyzed by dissolving
the scale in an ammonia solution. In the four loops under discussion, copper
represented between 69 percent and 84 percent of the scale that dissolved
(Table 17). The quantity of scale which could be collected from the usual
polishing procedures was insufficient to use for x-ray diffraction analyses.
Rather, flat copper plates were exposed to the water in the pH 6.0 and 8.0
tap water loops to generate enough scale for this purpose. The scales formed
in this way in both loops contained a cuprous oxide mineral known as cuprite,
Cu20. This mineral is 89 percent Cu, 11 percent 0 by weight, which is
somewhat more copper-rich than the analyses of the scale indicated. The wet
chemical analysis is more consistent with the scale being either a cupric
oxide, CuO, or a hydrated cuprous oxide, Ci^O't^O, which are 80 percent and
79 percent copper by weight, respectively. It'is possible that the scale
contained solids such as these that were amorphous and therefore were not
detectable by x-ray diffractometry. No plates were exposed to water in the
low- and high-carbonate loops.
The one time that copper was analyzed in the recirculating water, essentially
100 percent of it was present as particulate matter. This may have been
either cuprite or a cupric oxide (CuO), since oxidation of Cu(I) to the
cupric form would have been thermodynamically favorable in the oxygenated
132
TABLE 17
Composition of Copper Corrosion Scales
Loop 1
Coupon #103 500 21.4 16.6 78
Loop 2
Coupon #118 500 57.0 39.1 69
Loop 3
Coupon #124 445 24.1 17.6 73
Loop 4
Coupon #212 71.0 55.0 78
Coupon #211 53.0 41.0 81
LoopB
Coupon #207 330 2.3 1.1 48
Loop 6
Coupon #328 190 29.6 20.9 71
Clean Coupon
Coupon #557 6.0 5.0 83
133
water. Efforts to quantify the fraction of copper in each oxidation state in
either the attached or suspended scale were unsuccessful.
There was a good deal of imprecision in the scale weight data for the
phosphate loops because the total amount of scale present was so small.
Overall, the order of increasing scale accumulation was the same as the order
for increasing corrosion rate, with apparently a somewhat larger fraction of
the scale remaining attached to the coupon in the phosphate loops than in
those that contained no ortho-phosphate. For instance, after 400 days
exposure, the cumulative weight losses in the control, low-phosphate, and
high-phosphate loops were in the approximate ratio 14:2.4:1. The
corresponding ratio for scale weights was 5.8:1.8:1.
Chemical analysis of the scale after several hundred days exposure indicated
that the scale in the low phosphate loop was about 73 percent copper, which
is in the same range as the copper content of the scales in the phosphate-
free systems. Only 3 percent of the scale weight could be accounted for as
o-P04 , the remainder apparently being oxide ions or bound water. The scales
from the high phosphate loop contained only 48 percent copper by weight.
Cupric phosphate, ^(PO^, is 50 percent copper, so this appeared to be a
likely candidate for the major compound forming the scale. However, the
scale was only 4 percent P04 by weight, compared to a theoretical value of 50
percent for ^(PO^. It is not clear what compound other than bound water
could make up the remainder of the scale, if these analyses are correct.
o pH 6.0 (Water 1)
pH 8.0 (Water 2)
0.8- a pH 8.0, PO4 1 mg/l (Water 6)
pH 8.0, PO4 5 mg/l (Water 7)
0.6-
Instant.
Corrosion
Rate (MPY)
0.4-
U)
0.2-
0.0
50 100 150 200 250
Exposure Period (Days)
Figure 38. Comparison of Instantaneous Corrosion Rates on Copper Coupon Surfaces Under
Various Water Quality and Inhibitor Conditions
Dissolved Oxygen
Chlorine Residual
Similar experiments were carried out where the free chlorine residual was
varied over the range of values that may be encountered in a water supply
system. The pH was maintained at 7.0 in all these experiments. The results,
displayed in Figures 40 and 41, indicate that even small chlorine residuals
can dramatically increase copper corrosion currents on both new and aged
surfaces. However, if 1 mg/L o-P04 -P is present during aging, the effect of
free Cl is dramatically diminished. These tests also showed that, as
expected, the corrosion potential is sensitive to chlorine residual, becoming
more oxidizing with increasing chlorine. For both fresh and aged surfaces,
the effect of as little as 0.1 to 0.2 mg/L free chlorine was noticeable, and
the corrosion current continued to increase approximately linearly with
increasing chlorine throughout the range of residuals typically applied in
water supply systems (up to about 1 mg/L).
136
260
m Cathodic Tafel Slope
Ba.Bc
(mV/dec.)
20
0.5
0.4.
lo 0-3
((iA/cm2)
0.2
0.1.
0.0
0 4 6 10
Dissolved Oxygen (mg/L)
Figure 39. Variation in Corrosion Current and Tafel Slopes as a Function of Dissolved Oxygen
(pH 7.5, Temp 23 C, Conductivity 80 |iS, Flow Velocity 1 fps, Surface Exposure 5 Days)
220.
180-
Re
(niV/dcc.) 140 -"
100-
60
160
120-
Ra 80 <
(mV/dec.)
40-
00
0.
4-
3-
lo
2.
Figure 40. Variation in Corrosion Current and Tafel Slopes on Copper Surfaces as a Function of Free
Chlorine Residual (Conductivity 100 ^S, DO 10 mg/l)
220.
180.
Be
(mV/dec.)
140.,
100
140,
100.
Ba
(mV/dec.)
60.
u>
vO
20
3.0
2.5-
Aged Surface, Water 2
260 Day Exposure
2.0 :
Aged Surface, Water 6
lo 1.5-i Phosphate Inhibited
(u.A/(cm2) 220 Day Exposure
1.0-
0.5
0.0-9
Because the readings were made after only a few hours of equilibration to the
new solution conditions, it is not surprising that any protective scale that
had developed on the aged surface retained its protectiveness during the
test. The slight increase in corrosion current that was observed in this
system may have been due to partial loss of the scale or to the fact that the
driving force for the oxygen-reduction reaction is increased as pH is
decreased. By contrast, the response of the relatively young, unprotected
surface, which had been exposed to Seattle tap water for approximately three
140
220
B Fresh Cu Surface
ISO- Aged Surface, Waier 2
Be 140- a a + a
(mV/dec.)
100-
60
160.
120-
80- * * * *
Ba Q D D
(mV/dec.)
40-
3-
lo 2-
(\i\lcm2)
1-
0.
5 6 7 pH 8 9 10
Figure 42. Variation in Corrosion Current and Tafel Slopes on Copper Surfaces as a Function of pH
(Conductivity 100 nS, Dissolved Oxygen 10 mg/1)
days, approached that of a fresh surface. This suggests that if low pH
conditions persist in a system any protective scale that formed previously
might dissolve, leading to dramatically increased corrosion currents.
Several short-term experiments were conducted using fresh coupons. All these
experiments started with a short exposure to pH 7.5 water containing no
inhibitor, after which either pH or o-P04 concentration was changed.
2-
Corrosion
Current
(uA/cm2)
1-
.
0
0 50 100 150 200 250
Exposure (hours)
1.0 -
Corrosion
Current
(uA/cm2)
0.5 -
Figure 44. Corrosion Current for Coupon Exposed to 1 mg/l PO4 at Indicated Time (pH 7.5, Initial PO4 0 mg/l)
260
220 -
System Flush to Remove
Orthophosphate Inhibitor
180 -
CD
El
Ba, Be
(mV/dec.)
Addition of 1 mg/L of
Orthophosphate
100 -
0.6
0.5-
0.4-
lo Oo "
(HA/cm2)
0.2-
0.1 -
0.0 i 1
0 10 15 20 25 30
Experimental Progression (exposure days)
Figure 45. Effect of Orthophosphate Inhibitor Addition on Corrosion Current and Tafel Slopes for a
Relatively Fresh Copper Surface (pH 7.5, Temperature 23 C, Conductivity 80 [iS,
DO 8.2 mg/I, Flow Velocity 1 fps)
corrosion current began to increase once again, but not as rapidly as it had
decreased. One week after the inhibitor was eliminated, the corrosion
current had increased to about half of the initial, pretest value.
146
I 7.5 5.5 5.8 6.6 7.0
2.0
1.5-
Corrosion
Current 1.0-
-~j (uA/cm2)
0.5-
0.0
50 100 150 200 250
Exposure (hours)
180-
Be 140-
(mV/dec.)
100-
60
160
120-
Ba 80-
(mV/dec.) D
*
40-
Preformed Scale
PO4 soln. (5 mg/L)
Preformed Scale
lo No Inhibitor
(|iA/cm2)
Figure 47. Effects of pH Shift on Phosphate Inhibitor Scale (pH Maintained at 7.0 until Time 0 then
Shifted to 5.9, (Conductivity 100 nS, Dissolved Oxygen 10 mg/l)
Taken together, these results suggest that a significant acid transient will
cause a corresponding increase in copper corrosion in any system. This may
be related to partial or complete dissolution of any protective scale layer
that had previously built up. However, the extent and duration of the
corrosion response can be mitigated substantially by the presence of ortho-
phosphate. On fresh coupons, the pH change caused a rapid initial increase
in corrosion rate regardless of the presence or absence of phosphate.
However, in the presence of of ortho-phosphate, this initial increase was
rapidly reversed; when phosphate was absent, this rapid increase persisted.
On coupons with pre-existing protective scales, lowering pH in the presence
of ortho-phosphate led to a gradual increase in corrosion, while lowering it
in the absence of phosphate caused a much larger short-term increase in
corrosion and a comparable long-term change.
SUMMARY
One or five mg/L o-PO^-P lowered copper corrosion rates very dramatically.
The effect was apparent from the start of the test, and after a few months of
exposure, corrosion of copper in these loops nearly ceased altogether.
The only mineral that could be positively identified in the scales in any of
the phosphate-free loops was cuprite, Cu0. A small fraction of the corroded
copper remained attached to the coupon walls in these loops, with the
fraction being somewhat greater at higher than at lower pH. Scale in the
5 mg/L P loop had a Cu concentration consistent with (^(PO^, but the scale
149
contained much less P than would be required for that mineral to represent a
significant fraction of the scale. Apparently, only a small fraction of the
scale need be 113 (PO^ or some other P-containing mineral to confer good
corrosion-protection properties on the scale.
150
CORROSION OF GALVANIZED PIPES
While corrosion was faster in the low-pH loop than in the other loops from
the inception of the test, the main factor that distinguished this loop was
that its corrosion rate remained constant, while corrosion rate declined
significantly in all the other loops after a weight loss of about 250-300 mg.
Each coupon contained about 500 mg Zn in the galvanized layer, and therefore
it is not possible to relate the change in corrosion rate to complete loss of
this layer. However, the consistency of the weight loss at the time when the
corrosion rate changed suggests that some significant factor changed when
about half of the Zn had corroded. Perhaps at this point portions of the
underlying Fe surface are being exposed, and the lower corrosion rate
reflects the loss of Zn surface area in conjunction with a change in the
electrochemical characteristics of a Zn/Fe surface versus a pure Zn surface.
In any case, regardless of the mechanism involved, in all the loops other
than the one at pH 6.0 the average corrosion rate slowed substantially after
about 250 mg had been lost from each coupon. Because of this, the total
weight loss in the pH 6.0 loop was on the order of twice that in the other
loops after six months but about 7 to 10 times greater after 18 months.
151
5000
4000-
3000-
o pH= 6.0 (Water 1)
Weight Loss
(rag) pH= 8.0 (Water 2)
1000-
800-
600-
Weight
Loss o pH= 8.0 (Water 2)
(mg) pH= 8.0 Low CO3 (Water 4)
a o o
400- Op
o D n pH= 7.2 High CO3 (Water 5)
o
200-
& an 0 *
800-
600-
o pH- 8.0 (Water 2)
Weight
Loss pH- 8.0 PO4- 1 mg/L (Water 6)
(mg)
O D pH- 8.0 PO4- 5 mg/L (Water 7)
400-
o
o B
a
200- D B
cP
o
0 D
Scale Accumulation
The data characterizing the amount of scale that accumulated on the coupons
was more scattered than the weight-loss data, but in most cases trends within
data sets and distinctions between data sets were nevertheless discernible.
In the loop containing the most corrosive water (pH 6.0), scale accumulation
was unremarkable for 1 to 1 1/2 years and then increased tremendously (Figure
51). As shown previously, there is no corresponding change in the corrosion
rate. An obvious hypothesis is that the change in scale-accumulation rate
correlated with disappearance of the Zn layer, and that the accumulating
scale is primarily a hydrous iron oxide. The second part of this hypothesis
is consistent with the few analyses conducted during the course of this
research on the chemical composition of these scales (Table 18). None of the
scales contained less than 43 percent Zn except the scales from the pH 6.0
loop after 473 days exposure, which contained only 5 percent Zn.
In the two pH 8.0 loops with no added PO^, scale appeared to accumulate for
four to six months, after which a substantial portion of the scale was
released from the pipe surface (Figure 52). The change from a positive to a
negative accumulation rate corresponded reasonably well with the decrease in
corrosion rate (as determined by weight loss) described above. This change,
therefore, may be related to a switch from primarily Zn corrosion to
primarily Fe corrosion. However, there were substantial differences between
the long-term corrosion of the coupons from the pH 8.0 loops with no added
PO^ and the long-term corrosion of ungalvanized coupons. For instance, the
approximate steady-state scale weight for the last 200 days of the test in
the pH 8.0 loops is about an order of magnitude less than the scale weight
after 200 days of exposure of ungalvanized iron coupons to the same waters.
Furthermore, the Zn content of the scale in one coupon in the pH 8.0 low-
carbonate loop was 40 percent after 218 days, i.e., after the apparent
release of much of the previously accumulated scale. Finally, the rate of
weight loss from the galvanized coupons in the pH 8.0 loop was much lower
than in the pH 6.0 loop, while the weight-loss rates for nongalvanized iron
155
5000
o o
4000-
1000-
D
a D D
Q fl
" oflo
300-
o
100-
o o
0 o
Figure 52. Scale Weight Formed on Galvanized Pipe Exposed to Various Water
Qualities in the Absence of Phosphate
TABLE 18
Zinc Concentration in Zinc Galvanized Corrosion Scales
August, 1987
Zinc Concentration in
Dried Scale (Percent)
Loop 1 Zinc
473 days Zn 106 (zinc coating depleted) 5
140 days Zn 153 44
39 days previous analysis Zn 173 & 180 57
Loop 2 Zinc
140 days Zn 158 & 115 51
70 days previous analysis Zn 190 & 193 57
Loop 3 Zinc
141 days Zn 216-1 - 47
141 days Zn 216-2 47
175 days previous analysis Zn 172,175,162 49
Loop 4 Zinc
218 days Zn 165-168 48
135 days Zn 315 47
84 days previous analysis Zn 201 & 207 57
Loop 5 Zinc
316 days Zn 123-1 44
316 days Zn 123-2 44
178 days previous analysis Zn 160,161,162 43
Loop 2 Iron
Fe 301-2 1
158
coupons were about the same at the two pH values. Taken together, these
observations indicate that even if most of the Zn did oxidize early in the
test period, Zn continued to play an important role in controlling the
corrosion rate and restricting the accumulation of scale in these waters.
In the pH 7.2 high-carbonate loop, scale accumulation was always quite low,
and there was no evidence of significant scale release at any time. If
anything, it appeared that scale accumulation reached a steady-state value
rather quickly, and then underwent a small but consistent increase to a new
steady-state value around the time that the corrosion rate decreased.
The scales collected from the coupons generally had weak x-ray diffraction
patterns, if they had any at all. The high-alkalinity loop scales had x-ray
peaks suggestive of a mixed zinc hydroxide/carbonate solid known as
hydrozincite, but no positive identification could be made. The only other
scales that had any x-ray peaks at all were those from the pH 8.0, 5 mg/L o-
P04 -P loop. These scales were positively identified as a hydrated zinc
phosphate known as hopeite. Furthermore, this scale contained 14 percent P,
which matches the expected value for P in hopeite quite well. The coupons
exposed to water containing 1 mg/L o-P04 -P also contained a significant
amount of P (around 8 percent) but less than that expected for pure hopeite.
159
1000
D
D
D
800- n
ca
D
a
600- a
o pH= 8.0 (Water 2)
Scale Weight
Accumulation pH= 8.0 PO4= 1 mg/L (Water 6)
(mg)
a pH= 8.0 PO4= 5 mg/L (Water 7)
400-
200-
o o
o o
o o o o
oo
Validation
161
200
ISO-
160-
Be
(mV/dec.)
140-
120-
100
200
180.
160-
Ba D
(mV/dec.) m a
140-
Ov
10
120-
100.
80
70.
Electrochemical 60-
Measure as a 50.
Percent of
Gravimetric 40^
30-
20
0 50 100 150 200
Exposure Period (days)
Figure 54. Electrochemical Measurement Variation as a Function of Exposure Time. Error Bars Represent
a Single Standard Deviation Around the Mean (pH 7.0, Conductivity 60 ^S, DO 10 mg/l)
Loop#1 Conditions
Loop #2 Conditions
(8 day exp.)
Loop #5 Conditions
Figure 55. Comparison of Weight Loss and Electrochemical Corrosion Rate Measurements
Under Different Water Quality Conditions
Reduction Couples
Dissolved Oxygen
-800 -
-900 -
Surface
Polenlial -1000-
Ecorr (mV)
-1100-
-1200
180
Anodic TafeJ Slope
a Cathodic Tafel Slope
160-
a a o
Ba.Ik
(mV/dcc.)
140-
120
2.0.
1.5.
lo
(H/V/cm2)
0.5.
0.0
02468 10
Dissolved Oxygen (mg/L)
Figure 56. Electrochemical Measurement Variation on a Galvanized Surface as a Function of Dissolved
Oxygen (pH 7.0, Conductivity 100 ^S, Surface Exposure 20 Days in Water 2)
800 _
l/SDe HzD + 2e- = 20HT
Zn Polarization Diagram
600 pH 7.0
conductivity = 100 uS
dissolved oxygen = D.O ng/l
400 _
wo _
U
Surface
-400 Potential
Deaerated
Water
-00 _
-MO _
-1000
-1MO _
L
0.00001 0.0001
Current Density
Chlorine
pH
167
-700
-800-
Surface -900 ^
Potentia l
Ecorr(mV) 1000-
-1100-
-1200
180
160 J
Ba, Be
(mV/dec.)
140-
00 Anodic Tafel Slope
* Cathodic Tafel Slope
120
2.0
1.5.
lo 1.0-
(uA/cm2)
0.5-
0.0
0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
0.2 0.4
Free Chlorine Residual (mg/L)
Figure 58. Electrochemical Measurement Variation on a Galvanized Surface as a Function of Free Chlorine
Residual (pH 7.0, Conductivity 100 ^S, DO at Saturat ion, Surface Exposure 60 Days Water 2)
-700
-800
Surface
-900
Potential
Ecorr(mV)
-1100
-1200
180
Anodic Tafel Slope
Cathodic Tafel Slope
160-
Ba/Bc
(mV/dec.)
ON 140-
120
2.0
1.5 J
lo
(uA/cm2) i.o:
0.5-
0.0
10
SUMMARY
170
increase in scale accumulation rate after 1 1/2 years may have been due to
corrosion of the underlying iron.
171
SUMMARY. CONCLUSIONS. AND RECOMMENDATIONS
This research project investigated the corrosion rates of copper, mild steel,
and zinc-galvanized pipe coupons as a function of the chemical
characteristics of the water to which they were exposed. The coupons were
short sections of pipe typical of the kind that might be employed in a
household plumbing system. Several coupons of each material were placed in a
holder and exposed to water of controlled composition over a period ranging
from a few weeks to more than two years. After a predetermined exposure
period, the coupons were removed from the holding assembly and analyzed for
weight loss. Additionally, corrosion scales that had developed were
characterized by visual examination, by x-ray diffraction to identify their
mineral forms, and by chemical analysis. Overall, more than 1,000 pipe
coupons were subjected to corrosion testing.
MILD STEEL
The corrosion rates of mild-steel pipes and the composition and structure of
corrosion scales were related to water quality variables. The most
protective iron corrosion scale was formed in a high-alkalinity, moderate-pH
water (alkalinity 400 mg/L as CaC03 , pH 7.2). Pipe sections coated with this
scale had very low corrosion rates (about 2 mpy or less). The protective
scale was thin, hard, dark, and very adherent to the metal surface. Siderite
was found in this scale by x-ray diffraction analysis. This was the only
scale where siderite was detected. Wet chemical analysis of this corrosion
scale before and after air-drying showed that 5-10 percent of the iron was in
the ferrous state. Small amounts of carbonate were also found in both wet-
and dried-scale samples, and it appeared that most of the ferrous iron in
this scale was present as siderite.
173
Scales formed in the low-alkalinity waters were thick and loose, consisting
of large tubercles covering shallow pits. After nine months to one year of
exposure to low-alkalinity water, small amounts of hard, tenacious scale were
formed under the soft, loose scale layers. Goethite (a-FeOOH) was the only
mineral detected in the x-ray diffraction analysis of dried low-alkalinity
scales. However, goethite was not the only compound present in these scales,
since ferrous iron and carbonate were detected in scale samples before they
were air dried.
174
COPPER
Copper corrosion rates in these tests were remarkably steady. The only loop
in which they changed substantially during the course of the experiments was
the pH 6.0 loop, in which corrosion rate decreased significantly after about
50 days exposure. Corrosion rates in the pH 7.2 high-carbonate water and the
pH 8.0 tap water were about equal to each other, and to the post-50-day rate
in the pH 6.0 water. The absence of a strong dependence of copper corrosion
rate on pH in the range 6.0 to 8.0 is a surprising result and is generally at
odds with prior literature results. The pH 8.0, low-carbonate water had a
significantly lower rate than any of the other three phosphate-free waters.
Both 1 and 5 mg/L o-P04 -P lowered copper corrosion rates dramatically. The
effect of the phosphate was apparent from the start of the test, and after a
few months of exposure, corrosion of copper in these loops nearly ceased
altogether.
The only mineral that could be positively identified in the scales in any of
the phosphate-free loops was cuprite, Cu20. A small fraction of the corroded
copper remained attached to the coupon walls in these loops, with the
fraction being somewhat greater at higher than at lower pH. Scale in the
5 mg/L P loop had a Cu concentration consistent with Cu3 (P04 ) 2 , but the scale
contained much less P than would be required for that mineral to represent a
significant fraction of the scale. Apparently, only a small fraction of the
scale need be ^(PC^^ or some other P-containing mineral to confer good
corrosion protection properties on the scale.
GALVANIZED STEEL
By contrast, scale accumulation was slight in the pH 7.2 loop with high
carbonate. In two loops at pH 8.0, scale accumulated and was then released
from the surface. Scale accumulation in the loops containing ortho-phosphate
was moderate and attained a steady state within about six months.
176
RECOMMENDATIONS
The intent of this project was to develop a data base which might lead to a
more fundamental and complete understanding of corrosion of common plumbing
materials. This has been accomplished, and both the data and the methodology
will be of value to utilities and corrosion-control practitioners. However,
no single study can investigate the corrosion reaction over the full range of
water qualities typical of drinking water sources or potable water. For this
reason, it is inappropriate to make specific recommendations regarding
corrosion control practices in a given system; it is still the case that
control strategies must be developed on a case-by-case basis.
Similar studies can be, and to some extent are being, undertaken to
investigate the effects of corrosion-control strategies on corrosion rates in
systems with significantly different source water quality. These studies
should be supported so that, over time, an understanding of corrosion
mechanisms can be applied properly, thus reducing the effort and uncertainty
that are now required to develop strategies to minimize corrosion. In the
past, the Langelier Saturation Index and other corrosion indices have been
used inappropriately, and hence unsuccessfully, to predict corrosivities of
water supplies. This study has helped to confirm and extend the siderite
model for corrosion protection in steel pipes. It is reasonable to expect
that more studies of the type described will allow scientists and engineers
to develop protocols for assessing corrosivity for water with different
177
characteristics and plumbing materials, and that these protocols will be
considerably more reliable and useful than those previously available.
178
REFERENCES
179
Feigenbaum, C., Gaylor, L. and Yahalom, J. "Microstructure and chemical
composition of natural scale layers." Corrosion. 34(2) 65-69 (1978).
Freneir, W.W. and Growcock, F.B. "Mechanism of iron oxide dissolution - A
review of recent literature." Corrosion, 40, 663 (1984).
Gerchakov, S.M., Udey, L.R. and Mansfield, F. "An improved method for
analysis of polarization resistance data." Corrosion. 37, 696 (1981).
Ghosh, M.M., O'Conner, J.T. and Engelbrecht, R. "Bathophenanthroline method
for the determination of ferrous iron." Jour. AWWA. 59(7), 897-905
(1967).
Grauer, R., Moreland, P.J. and Pini, G. A Literature Review of Polarization
Resistance Constant Values. National Association of Corrosion
Engineers, Houston (1982).
Hatch, G.B. Corrosion Inhibitors (C. Nathan, editor). National Association
of Corrosion Engineers, Houston, TX, pp. 114-125 (1973).
Hatch, G.B. "Protective film formation with phosphate glasses." Ind. Enqr.
Chem.. 44, 1775-1786 (1952).
Hausler, R.H. "Practical experiences with linear polarization rate
measurements." Corrosion. 33, 117 (1977).
Hedberg T. and Johansson, E. Protection of Pipes against Corrosion. Special
Subject 20, IWSA Congress (1984).
Hepburne, M.L. Corrosion in Piping Systems. Piping Handbook. McGraw Hill
(5th ed., 1965).
Herrera, C.E. and Hoyt, B.P. Seattle Distribution System Corrosion Control
Study, Vol. 2, Tolt River Water Pilot Plant Study. U.S.E.P.A Report,
No. EPA-600/S2-84-065 (1984).
Heyer, C. Ursache und beseitigung des bleiangriffs durch leitungawasser.
Verlag Paul Baumann, Dessau (1888).
Hoyt, B.P., Herrera, C.E. and Kirmeyer, G.J. Seattle Distribution System
Corrosion Control Study, Vol. 1, Cedar River Water Pilot Plant Study.
U.S.E.P.A Report, No. EPA-600/S2-82-026 (1982).
Hudson, H.E. and Gilcreas, F.W. "Health and economic aspects of water
hardness and corrosiveness." Jour. AWWA. 68, 201 (1976).
Ijsseling, F.P. "Application of electrochemical methods of corrosion rate
determination to systems involving corrosion product layers." British
Corrosion Journal. 21, 95 (1986).
Jankowski, J. and Juchniewicz, R. "A four-point method for corrosion rate
determination." Corrosion Science. 20, 841 (1980).
180
Jobin, R. and Ghosh, M.M. "Effect of buffer intensity and organic matter on
the oxygenation of ferrous iron." Jour. AWWA. 64(9), 590-595 (197";).
Karalekas, P.C., Ryan, C.R. and Taylor, F.B. "Control of lead, copper and
iron pipe corrosion in Boston." Jour. AWWA. 75, 92-95 (1983).
Kennedy Engineers. Seattle Internal Corrosion Study, Phases I, II, and III.
Tacoma, WA (1978).
Kolle, W. and Rosch, H. "Untershugen an rohrnetzinkrusteirungen unter
minerologischen gesichtspunkten." Vom Wasser. 55, 159 (1980).
Langelier, W.F. "Chemical equilibria in water treatment." Jour. AWWA. 8,
169 (1946).
Langelier, W.F. "The analytical control of anti-corrosion water treatment."
Jour. AWWA, 28, 1500-1521 (1936).
Larson, I.E. "Corrosion by domestic waters." Bull. 59, ISWS, Urbana (1975).
Larson, I.E. and Buswell, A.M. "Calcium carbonate saturation index and
alkalinity interpretations." Jour. AWWA. 34, 1667 (1942).
Larson, I.E. and Skold, R.V. "Corrosion and tuberculation of cast iron."
Jour. AWWA. 49(10), 1294-1302 (1957).
Lehrman, L., Schuldener, H.L. "Action of sodium silicate as a corrosion
inhibitor in water piping." Ind. Enq. Chem.. 44, 1765-1769 (1952).
Leroy, R.L. "Evaluation of corrosion rates from polarization measurements."
Corrosion. 31, 173 (1975).
Lowenthal, R.E. and Marais, C.V.R. Carbonate Chemistry of Aquatic Systems:
Theory and Application. Ann Arbor Science Publishers, Ann Arbor, MI
(1976).
Lumsden, J.F. and Szklarska-Smialowska, Z. "The properties of films formed
on iron exposed to inhibitive solutions." Corrosion. 34, 169-176
(1978).
McCauley, R.F. and Abdullah, M.O. "Carbonate deposits for pipe protection."
Jour. AWWA. 50, 1419-1428 (1958).
Merrill, D.T. and Sanns, R.L. "Corrosion control by the deposition of CaCOo
films." Jour. AWWA. 69, 634 (1977).
Millete, J.R. "Aggressive water: Assessing the extent of the water." Jour.
AWWA. 72, 276 (1980).
J.M. Montgomery Consulting Engineers. Internal Corrosion Mitigation Study.
Prepared for the Bureau of Water Works, Portland, OR (1982).
181
Muller, E.D. and Ritter, J.A. "Monitoring and controlling corrosion by
potable water." Jour. AWWA. 72, 286 (1980).
Neff, C.H., Schock, M.R. and Marden, J.I. "Relationships Between Water
Quality and Corrosion of Plumbing Materials in Buildings--Vol. 1.
Galvanized Steel and Copper Plumbing Systems." EPA/600/2-87/036A (1987).
Pourbaix, M. Lectures on Electrochemical Corrosion. Plenum Press, New York
(1973).
Reiber, S.H., Ferguson, J.F. and Benjamin, M.M. "Corrosion Control in the
Pacific Northwest." Jour. AWWA. 79, 71-74 (1987).
Richards, W.N. and Moore M.R. "Lead hazard controlled in Scottish water
systems." Jour. AWWA. 76, 60 (1984).
Rossum, J.R. and Merrill, D.T. "An evaluation of the calcium carbonate
saturation indices." Jour. AWWA. 75, 95 (1983).
Ryder, R. and Wagner, I. Corrosion Inhibitors in Internal Corrosion of Water
Distribution Systems, AWWARF, Denver, CO, 617-655 (1985).
Ryznar, J.W. "A new index for determining amount of calcium carbonate scale
formed by a water." Jour. AWWA. 36, 472 (1944).
Schock, M.R.and Buelow, R.W. "The behavior of asbestos-cement pipe under
various water quality conditions: Part 2, Theoretical considerations."
Jour. AWWA. 73, 636-751 (1981).
Shull, K.E. "An experimental approach to corrosion control." Jour. AWWA.
72, 280-285 (1980).
Singley, J.E. and Lee, T. "Pipe loop system augments corrosion studies."
Jour. AWWA. 76, 77 (1984).
Singley, J.E. et al. Corrosion and calcium carbonate saturation index in
water distribution systems. Municipal Environmental Research
Laboratory, USEPA (1984).
Snoeyink, V.L. and Kuch, A. Principles of metallic corrosion in water
distribution systems in Internal Corrosion of Water Distribution
Systems. AWWARF, Denver, CO, pp. 1-32 (1985).
Sontheimer, H., Kolle, W. and Kuch, A. Uniform Corrosion and Scale Formation
in Internal Corrosion of Water Distribution Systems, AWWARF, Denver, CO,
pp. 62-88 (1985).
Sontheimer, H., Kolle, W. and Snoeyink, V.L. "The siderite model of the
formation of corrosion-resistant scales." Jour. AWWA. 73, 572-578
(1981).
182
Stern, M. and Geary, A.L. "Electrochemical polarization: Theoretical
analysis of the shape of polarization curves." J. Electrochem. Soc.,
104, 56 (1957).
Strickland, J.D.H. and Parsons, T.R. Fisheries Research Board of Canada,
Ottawa, 49 (1972).
Stumm, W. "Calcium carbonate deposition at iron surfaces." Jour. AWWA. 48,
300-310 (1956).
Stumm, W. "Investigation of the corrosive behavior of waters." ASCE-San.
Engr. Div., 86, 27 (1960).
Stumm, W. and Lee, G.F. "Oxygenation of ferrous iron." Indus. Eng. Chem..
53, 143 (1961).
Swayze, J. "Corrosion study at Carbondale, Illinois." Jour. AWMA. 75, 101-
102 (1983).
Trewick, G.P., Glicker, J., Chow, B. and Sprinker, M. "Pilot plant
simulation of corrosion in domestic pipe materials." Jour. AWWA.
77(10), 74-82 (1985).
Uhlig, H.H. Corrosion and Corrosion Control. John Wiley and Sons, New York
(1971).
Wagner, C. and Traud, Z. Electrochem., 44, 391 (1938).
Whitman, G.W., Russell, R.P. and Altier, V.J. "Effect of hydrogen ion
concentration on the submerged corrosion of steel." Industrial and
Engineering Chemistry. 16(7), 665-670 (1924).
Wood, J.W., Beecher, J.S. and Lawrence, F.S. "Some experience with sodium
silicate as a corrosion inhibitor in industrial cooling waters."
Corrosion, 13, 719t-724t (1957).
183
APPENDIX Al
TABLE Al-1
Weight Loss Variation due to Surface-Preparation Techniques
In the mechanical cleaning procedure, the most serious metal loss occurs on
the copper coupons. Although it is only slightly higher than that observed
on black and galvanized iron, it is of greater significance because of the
substantially lower rate at which copper generally corrodes. The average
restoration loss of 1.5 mg represents the weight loss expected of an
equivalent pipe section corroding at a moderate rate over an exposure period
of from two to four days. For short-term weight-loss measurements, it is
critical to adjust for this restoration loss, and since the standard
deviation of the restoration-induced weight loss is substantial, a larger
number of coupons is necessary to provide a statistically meaningful result.
The number of coupons depends on the desired precision and the length of
exposure.
Galvanized Iron
30- Copper
Coefficient of
Dispersion (%)
00
-J 20-
10-
10 100 1000
188
50
Modified
Coefficient of
Dispersion (%)
20-
oo
sO
D
10-
0
10 100 1000
190
SUMMARY OF THE WEIGHT-LOSS TECHNIQUE
This report reviews new procedures for conducting corrosion-rate tests using
weight loss of metal coupons as the primary measure. A new assembly for
mounting the coupons and a new procedure for preparing and restoring the
metal surfaces are presented. The new procedure offers the advantages of
pipe flow hydrodynamics and multiple coupon mounting. The assembly is easier
to construct than the ASTM standard, and up to 10 coupons can be mounted in a
single sleeve, which makes it simpler to develop a statistically accurate
picture of the corrosion process.
191
APPENDIX A2
Linear polarization has been widely used to measure corrosion rates since
1939. The basic relationship derives from the modified Butler-Volmer
equation and assumes that electrochemical kinetics are solely a function of
the difference between the imposed potential and the potential of the freely
corroding metal surface (Bockris and Reddy, 1973). The modified Butler-
Volmer equation is:
I - {E - Ecor)/Ba . -(E-Ecor)/BC
xmeas
Where:
Imeas = the measured polarization current that flows between the
metal and the counter electrode (equal to the difference
between the partial anodic and cathodic currents, i 0 m -
V,z)
Ecor = the open circuit corrosion potential (freely corroding
potential)
E = the potential imposed upon the metal
I cor = the corrosion current that exists under freely corroding
conditions; it is also the value needed to determine the
corrosion rate.
193
sufficiently displaced from the reversible potentials of the respective
anodic and cathodic reactions (usually differences of 200 mV or greater are
sufficient).
where Rp , the polarization resistance, has units of ohms and is equal to the
slope of the potential versus current plot at [E-Ecor=0]. The factor K
becomes an explicit function of the Tafel slopes.
Equation A2-3 offers a simple and direct means of solving for I cor if an
accurate estimate of Tafel slopes is available. Commercial corrosion-rate
meters make use of this equation by employing an internal "hard wired" value
for K and measuring the polarization resistance at a single offset from Ecor ,
usually less than 20 mv. While these meters offer simplicity and speed,
absolute accuracy suffers because the electrochemical parameters cannot be
adjusted to fit different cell conditions, or the changing conditions of a
The given cell over time (Bandy and Jones, 1976; Hausler, 1977).
Of the three materials studied, the variation of surface potential was the
most pronounced on black-iron coupons. Variation was enhanced because of the
tendency of these surfaces to undergo pitting-type corrosion under the water
quality conditions tested. Conversely, with few exceptions, copper surface
corrosion is uniform under the distribution system conditions studied. The
zinc coating on galvanized pipe is also more uniform in its initial corrosion
patterns than black iron; however, once the galvanized layer has been
penetrated, it is subject to extreme pitting.
The distribution of anodic and cathodic sites on metal surfaces and resulting
variations in corrosion potential has been explored by several authors
(Dexter et al., 1985; Freneir and Growcock, 1984). Experience with this
technique has shown that the more nonuniform the corrosion pattern, the
larger the surface variations in Ecor . With the flexible reference electrode
and the rotating cap of the polarization cell, it is possible to scan the
surface of the test electrode and map the corrosion potential variation.
Figures A2-1, A2-2, and A2-3 present, respectively, a surface scan over a 1-
cm-wide circumferential band of an iron, galvanized-iron, and copper pipe
coupon. Each coupon had been exposed for approximately 96 hr to a 1 ft/sec
velocity of Seattle tap water (temperature = 15'C, conductivity = 55
umhos/cm, dissolved oxygen = 10 mg/L, pH = 7.8).
Figure A2-1 shows the dramatic variation in Ecor associated with the
nonuniform iron corrosion. Although this coupon had been exposed for only 96
hr, distinct pitting sites had been initiated, accompanied by an overlaying
iron oxide scale. The deep valleys of the scan correspond to these areas,
while the flatter higher-potential areas represent the bulk of the coupon
195
-1601
so Surfoca
Potential -200
faV)
-240 -i
Depth
(c)
Depth
(cm)
-640
-680-
-720-
-760
VO Surface
Of
Potential -800
-8401
-880
-9201
-960 7
....
7
/
7
/...../
/ DcPth
(rm)
The potential scans on copper and galvanized iron (Figures A2-2 and A2-3)
show a more stable surface Ecor than iron. The uniform nature of their
corrosion process and scale deposition explains this difference.
Polarization resistance measurements on these surfaces show only slight
spatial variation, and since both copper and zinc scales are generally
thinner than iron scales, polarization measurements can be made anywhere on
the surface regardless of coupon exposure time or scale growth.
2. =1 [I.(lQ.(EJ-Ecor)/Ba . 10 -(Ej-Ecor)/Bc)
Icor = s~ ri 0 (Ej-Ecor)/Ba - iQ-Ctt-EcorJ/Bc^] (A2 " 4)
J *
where j=l...n represent the n data points. Using the entire data set (n
values of E and I), I cor is completed from Eq A2-4. This value of I cor is
then used with Eq 1 to compute the value of Imeas for each value of E. If
any computed value of Imeas differs from the experimental value of more than
some arbitrary cutoff (usually 5 percent to 10 percent of the experimental
value), the set of assumed Tafel slopes is rejected. If all n computed
values of Imeas fall within the criteria, the values of the Tafel slopes are
stored along with the sum of the squares error
^experimental) '* Tafel slopes are then incremented and the calculation
proceeds until all possible combinations established within the original
limits have been tested. Selection of the "best" Tafel slopes uses the
minimum sum of the square of the errors as its goodness-of-fit criteria.
The procedures described appear cumbersome and time consuming, even for a
computer-based solution. However, the task is simplified by the
insensitivity of Eq A2-1 to minor changes in the Tafel slope values, and the
well -documented range of feasible Tafel slope values for the metal of
interest. Because of insensitivity, the incremental adjustments to the Tafel
slopes need not be less than 10 mV. Choosing a smaller value may find a
marginally better fit, but such efforts are not justified considering the
precision of the actual polarization measurements. A large body of
experimentally determined Tafel slopes for each metal has been catalogued by
the National Association of Corrosion Engineers (Grauer et al., 1982). These
data will generally bracket a feasible solution. An iteration boundary of
200
plus-or-minus 100 mV around the anticipated value is usually sufficient to
locate a satisfactory fit.
Figures A2-4 and A2-5 are Evans diagrams constructed from polarization data
collected on black-iron coupons exposed to the Seattle distribution system
for 24 hr and 7 days, respectively. The lines of best fit (polarization
curves) resulting from the parameter estimation technique described above are
overlaid on the six measured polarization points. In this case, the maximum
single error between measured and predicted values is less than 5 percent.
It is generally possible to find a fit that falls within an error range of 5-
10 percent. The dashed lines intersecting at Ecor are extrapolations of the
predicted Tafel slopes from the linear region of the polarization curve. The
current density at the point of intersection represents the current density
of the freely corroding surface.
Figures A2-6 and A2-7 present Evans diagrams of measured and predicted
polarization values derived from copper and galvanized-iron surfaces exposed
for seven days to the Seattle distribution system. In both cases, the
analytical approach was the same as for the black-iron examples. Table A2-1
summarizes the parameters derived from the four polarization examples and
compares electrochemically measured corrosion rates to weight-loss estimates
of corrosion rates made on five coupon sets exposed to identical conditions
for similar durations. Corrosion current densities are converted directly to
a penetration rate by applying Faraday's law:
where s.g. is the specific gravity of the corroding metal and the equivalent
weight (eg. wt.) represents the gram weight of metal oxidized by the transfer
of one equivalent of electrons.
150
100
50
/ solid lines predicted values
/ - Masured values
Offset 0 F I,
(V)
10.2 uA/ca
NJ
o -50
K)
150
BQ - 220 iV/decode
100
SO
solid lines predicted values
* Measured values
Offset OF -I
(V)
K)
O
U)
-50
-100
Saottlo Top Water
pH- 7.8
Cond. * 65uMhos/o
-150 B c - 180 V/decade Te*p 15 deqrees-C
00 10 g/l
:low Vel. - 1 ft/ecc
-200
-1 -0.5 0 0.5 1 1.5 2.5
Log Currant Density CuA/ca2)
150
B0 - 80 V/dacoda
100
50
solid linos predicted values
* Measured values
Offset
(V) ICOT l.B uA/ai
\
N
-50
150
8Q * 150 V/dacoda
100
50
olid Una* predicted values
* Matured values
Offset n r
<V) t
-50
-100
Seattle Tap Water
pH-7.8
Cond. 85uho/c
-150 Bc 120 V/decode Tep "15 degrees-C
uO - 10 TO/1
:lot Vol. - Ift/eed
-200
-1 -0.5 0 0.5 1 1.5 2.5
Log Currant Density (uA/c2)
Galva
Black Black nized
Iron Iron Copper Iron
Exposure Period 24 hr 7 days 7 days 7 days
Exposure Medium Seattle Seattle Seattle Seattle
tap tap tap tap
Fitted Anodic Tafel Slope (mV) 250 220 80 150
Fitted Cathodic Tafel Slope (mV) 250 180 180 120
Corrosion Current Density 10.2 7.2 1.8 1.37
(mA/cmz )
Corrosion Rate - Electrochemical 114 81 41 18
(lM/yr.)
Weight-Loss Measurement 5.5 2.8 1.02 0.92
(5 coupon set) (mg/day)
Corrosion Rate - Weight Loss 134 68 33 24
(/im/yr)
Difference in Weight Loss vs. -15 +19 +24 -25
Electrochemical Corrosion Rate (%)
206
Association of Corrosion Engineers speaks of predicted values being within
one-to-two fold of actual corrosion rates (Grauer et al., 1982). Even ASTM
standards are reluctant to state an acceptable level of precision because of
the large variability encountered between systems. By contrast, the examples
of Table A2-1 come within 25 percent of weight-loss measurements and yield
additional information about the electrochemical processes. The hardware and
analytical techniques described here offer improved accuracy for corrosion
measurements on plumbing materials exposed to tap water. The accuracy
approaches that necessary to distinguish between the more subtle effects of
various chemical corrosion inhibitors or to access the effects of minor water
quality changes.
207
scale growth appear to impair the accuracy of polarization measurements and
make estimation of even a uniform penetration rate imprecise.
208
NOMENCLATURE
a = Standard Deviation
ASTM = American Society for Testing and Materials
Ba = Anodic Tafel Slope (mV/decade of polarization current)
Bc = Cathodic Tafel Slope (mV/decade of polarization current)
C = Degrees Centigrade
COg = Carbonate
Cj = Total carbonate-concentration
Cu = Copper
DO = Dissolved oxygen, mg/L
E = Measured Surface Potential relative to the Ag/AgCl Reference
Electrode (mV)
Ecor = Freely Corroding Surface Potential Relative to the Ag/AgCl
Reference Electrode (mV)
Fe = iron
fps = Feet per second
H2C03 = Carbonic Acid
ty
I cor = Corrosion Current Density (/jAmps/cnr)
Imeas = Measured Current Density Impressed upon the Test Electrode
Surface (/jAmps/cnr)
I 0 = Equilibrium exchange current density, /tA/cnr
nU * IIIm = Partial
ara 4- Anodic Current Generated from the Oxidation of Metal
m to nr
i r z = Partial Cathodic Current Generated from by the Reduction of
' Substance z
ISWS = Illinois State Water Survey
K = Constant, a Function of Tafel Slopes [BaBc/2.3(Ba+Bc )]
Ks = Solubility constant for siderite
L = Liter
209
mA = Milliamperes
mg = Milligrams
mpy = Mils per year
mv/dec. = Millivolts per decade
mV SHE = Electrical Potential Relative to a Standard Hydrogen Electrode
P = phosphorus
P04 = phosphate
Rp = Polarization Resistance, (dV/dl at E-Ecor=0)
rpm = Revolutions per Minute
s.g. = Specific Gravity of the Corroding Metal
S.I. = Langlier Saturation Index
uS = MicroSiemens
Zn = zinc
210
American Water Works Association
TM