HIGHWAY RESEARCH

RECORD
Number 95

Field and Laboratory Testing of Aluminum

2 Reports

Presented at the
44th ANNUAL MEETING
January 11-15, 1965

SUBJECT CLASSIFICATION
32 Cement and Concrete
33 Construction
34 General Materials

HIGHWAY RESEARCH BOARD

of the
Division of Engineering and Industrial Research
National Academy of Sciences-National Research Council
Washington, D. C.
1965
 Department of Materials and Construction
R. L. Peyton, Chairmat1
Assistant State Highway Engineer
State Highway Commission of Kansas, Topeka

GENERAL MATERIALS DIVISION

John L. Beaton, Chairman
Materials and Research Engineer, Materials and Research Department
California Division of Highways, Sacramento

COMMITTEE ON METALS IN HIGHWAY STRUCTURES

(As of December 31, 1964)
LaMotte Grover, Chairman
Welding Engineer, Air Reduction Sales Company, Inc.
New York, New York
W. C. Anderson, Chief Research and Development Engineer, The Union Metal Manu-
facturing Company, Canton, Ohio
John L. Beaton, Materials and Research Engineer, Materials and Research Department,
California Division of Highways, Sacramento
Gordon Cape, Manager of Technical Research, Dominion Bridge Company, Ltd.,
Montreal, Canada
S. K. Coburn, Research Technologist, Applied Research Laboratory, U. S. Steel
Corporation, Monroeville, Pennsylvania
Frank Couch, Welding Engineer, Bethlehem Steel Company, Bethlehem, Pennsylvania
John R. Daesen, Director, The Galvanizing Institute, Protomatic, Inc., Park Ridge,
Illinois
T. D. Dismuke, Product Engineering Group, Homer Research Laboratories, Bethlehem
Steel Company, Bethlehem, Pennsylvania
E. S. Elcock, Bridge Engineer, State Highway Commission of Kansas, Topeka
Eric L. Erickson, Chief, Bridge Division, Office of Engineering, U. S. Bureau of
Public Roads, Washington, D. C.
Edward A. Fenton, Technical Director, American Welding Society, New York, New
York
James H. Havens, Director of Research, Kentucky Department of Highways, Lexington
G. 0. Hoglund, Alcoa Process Development Laboratories, Aluminum Company of
America, New Kensington, Pennsylvania
Ray I. Lindberg, Research Scientist, Metallurgical Research Division, Reynolds Metals
Company, Richmond, Virginia
Thomas A. Lowe, Research Engineer, Department of Metallurgical Research, Kaiser
Aluminum & Chemical Corporation, Spokane, Washington
Robert A. Manson, Metallurgist, Bridge Section, Minnesota Department of Highways,
St. Paul
Arthur W. Moon, Office of Deputy Chief Engineer, Structures, New York State Depart-
ment of Public Works, Albany
Robert A. Norton, Engineer of Hydraulics and Bridge Maintenance, Connecticut State
Highway Department, Wethersfield
J. C. Oliver, Testing Engineer, Alabama State Highway Department, Montgomery
Joseph W. Pitts, Metallurgist, National Bureau of Standards, Washington, D. C.
Melvin Romanoff, Chemist, Metal Reactions Section, Metallurgy Division, National
Bureau of Standards, Washington, D. C.
Clyde F. Silvus, Bridge Engineer, Texas Highway Department, Austin
Oscar Teitel, Product Engineer, International Pipe and Ceramics Corporation, East
Orange, New Jersey
Lewis A. Tomes, Research Associate, Calcium Chloride Institute, Washington, D. C.
L. E. Wood, Department of Civil Engineering, Purdue University, Lafayette, Indiana
 Foreword
"A Preliminary Study of Aluminum as a Culvert Material" by Nordlin and
Stratfull should be of interest and value to highway drainage structure
engineers, as well as those making research studies of the suitability of
aluminum and galvanized steel for culvert materials under various service
conditions and environments. The paper, together with the discussions,
constitutes a good summary of the information that is available to the
designer of drainage structures, as derived from laboratory studies and
experience, for choosing flexible culvert materials. Likewise, they
provide a good review of the literature on this subject for the benefit of
the researcher who is interested in carrying forward the much needed
further research in this field. The information presented provides a
basis for further research to improve the mechanical and chemical prop-
erties of materials and combinations of materials for flexible culverts.
The bibliography of the original paper is strongly supplemented by the
bibliographies of the discussions.
As might be expected in the light of conflicting opinions as well as
commercial interests, this paper and the discussions given at the tech-
nical session where it was presented have been the subject of much con-
troversy.
The original paper has been criticized by some as appearing to draw
some unwarranted conclusions. This criticism is based on the com-
parative severity of service conditions at the test site chosen, the limited
extent of observations of behavior, and some question as to whether the
procedure used in the laboratory was representative of actual conditions.
Those charged with making recommendations and decisions as to pub-
lication sought advice from disinterested parties. Consensus was that
the paper and the discussions provide a good deal of valuable informa-
tion. Even though there was not always agreement that all of the data
presented verified the accompanying contentions, the data were con-
sidered to be useful and in at least some important way relevant to the
general problem of choosing culvert materials.
Taken all together, the information of the paper and the discussions
seem to provide quite a comprehensive statement of the problem, and
to emphasize most, if not all, of its important aspects. These aspects
including the influence of chemistry and mineral content of soil and water
of the environment, as well as quiescence, erosion, abrasionand other
mechanical factors.
Accordingly, the complete paper and all the discussions have been
published. As the authors note, the paper should be considered as only
a progress report. The results of further studies and longer experience
must be available before definite criteria can be established for choosing
the most suitable culvert materials for the various environments that
may be encountered for highway drainage structures.
The investigation described in "Nondestructive Tests for Detecting
Discontinuties in Aluminum Alloy Arc Welds," by Panian, Patsey and
Sager, was intended to evaluate radiographic and ultrasonic procedures
for detecting 14 different types of discontinuities in TIG (gas tungsten-arc
welding) and MIG (gas metal-arc welding) welds in aluminum alloy plating
and to determine the effects of these discontinuities on the static strength
of the welded joints. The test specimens were made of 1/2-in. and 1-in.
thick material.
 Highway engineers engaged in the design, construction, or inspection
of either steel or aluminum structures should find something of value and
interest to them in this paper. Although the work was done on welded
aluminum alloy plates, most of the information developed on the com-
parison of the radiographic and ultrasonic test methods and on the effec-
tiveness of each method under various circumstances would apply equally
well to either steel or aluminum structures with welded joints made by
almost any arc welding process.
Because of the recent rapid growth of the use of welding for the con-
struct.ion of bridges and large overhead sfructures for directional signs,
highwa:~ engineers have become greatly interested in nondestructive tests
for welds and the advantages and disadvantages of the various test methods.
One of the most controversial subjects related to structural welding
is the choosing of acceptance standards for use with nondestructive test-
ing and inspection methods, which will assure adequate strength without
being needlessly severe and costly. Therefore, the information developed
in this investigation on the effect of defects and discontinuities of various
kinds upon the strength of welded aluminum joints should be of interest
to almost anyone engaged in design for welding. However, one might
hesitate to apply to steel structures such information regarding effects
of various defects upon the strength of aluminum welds, without some
verification of the applicability to steel.
The paper points out certain kinds of defects which are difficult or im-
possible to detect radiographically. Some of these can be detected by
ultrasonic testing. Some of them are surface defects which can be de-
tected easily by visual inspection.
Some of those who have reviewed the paper feel that the investigation
should be considered as only a beginning, and the work has not been
carried far enough to warrant conclusions. Probably the greatest value
of the paper is that it focuses attention upon the various facets which need
deeper investigation. For example, in determining the effects of weld
defects upon strength, the investigators in some cases used specimens
containing grossly defective welds. This work should be carried further
to determine critical sizes or extents of weld defects such as porosity
and slag inclusions.
In all of the work, the size and extent of weld defects were determined
by metallographic examination and examination of fracture surfaces of
tensile test specimens as a basis for evaluating the effectiveness of
radiographic and ultrasonic testing.
The authors point out advantages and shortcomings for both of these
nondestructive test methods. They conclude that they are both valuable
and effective procedures for determining the structural integrity of welds.
They feel that radiography provides a more definite picture of the actual
weld condition. Although they consider it advisable to take time to grind
weld surfaces flush when using the ultrasonic method, to avoid false ex-
traneous indications from surface geometry, they think it probable that
ultrasonic examination will gradually replace radiography for the inspec-
tion of long lengths of weld because the ultrasonic method is faster and
more readily automated.
 Contents
A PRELIMINARY STUDY OF ALUMINUM AS A CULVERT
MATERIAL
Eric F. Norcllin and R. F. Stratfull. . . . . . . . . . . . . . . . . . 1
Discussion: Hugh P. Godard . . . . . . . . . . . . . . . . . . . . . . . . 33
Comments: E. F. Nordlin and R. F. Stratfull . . . . . . . . . . . . 39
Discussion: Thomas A. Lowe . . . . . . . . . . . . . . . . . . . . . . . 47
Comments: E. F. Nordlin and R. F. Stratfull . . . . . . . . . . . . 53
Discussion: A. H. Koepf. . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Comments: E. F. Nordlin and R. F. Stratfull . . . . . . . . . . . . 65
Discussion: John R. Daesen . . . . . . . . . . . . . . . . . . . . . . . . 67
Comments: E. F. Nordlin and R. F. Stratfull . . . . . . . . . . . . 67
Discussion: Ernest W. Harvick . . . . . . . . . . . . . . . . . . . . . . 68
Comments: E. F. Nordlin and R. F. Stratfull . . . . . . . . . . 68
Discussion: S. K. Coburn. . . . . . . . . . . . . . . . . . . . . . . . . . 68
Comments: E. F. Nordlin and R. F. Stratfull . . . . . . . . . . . . 69
Discussion: Albert R. Cook. . . . . . . . . . . . . . . . . . . . . . . . . 69
Comments: E. F. Nordlin and R. F. Stratfull . . . . . . . . . . . . 70
Closure: E. F. Nordlin and R. F. Stratfull . . . . . . . . . . . . . . 70

NONDESTRUCTIVE TESTS FOR DETECTING DISCONTINUITIES

IN ALUMINUM ALLOY ARC WELDS
F. C. Panian, J. A. Patsey and G. F. Sager . . . . . . . . . . . 71
Discussion: Simon A. Greenberg . . . . . . . . . . . . . . . . . . . . . 109
A Preliminary Study of Aluminum as a Culvert
Material
ERIC F. NORDLIN and R. F. STRATFULL
Respectively, Assistant Materials and Research Engineer-Structures; and
Corrosion Engineer, Materials and Research Department, California Division of
Highways

This investigation was initiated because there appeared to be a

possibility of an economic or engineering advantage in the use
of aluminum as a culvert material. The project was sponsored
by the Bureau of Public Roads, and the investigation was per-
formed by the Materials and Research Department of the Cali-
fornia Division of Highways starting in 1961.
On the basis of this accelerated investigation, it is estimated
that under favorable conditions, aluminum may have a service
life up to an estimated 25 years. The anticipated favorable con-
ditions for the use of aluminum are described with regard to the
use of protective coatings, limits for the hydrogen-ion and the
resistivity of the soil and water, and the influence of abrasion
on the durability of the metal.
Because this was an accelerated investigation, the durability
of aluminum as a culvert material should be continuously veri-
fied so as to confirm or modify the results with actual field
experience .

•THE POSSIBILITY of an economic or engineering advantage in the use of aluminum

as a culvert material has resulted in this investigation by the California Division of
Highways in cooperation with the Bureau of Public Roads.
The investigation was initiated on March 31, 1961, under Laboratory Project Au-
thorization 71-R- 6244 and more recently, under R- 53097. The cost of the investiga-
tion has been borne by the California Division of Highways and the Bureau of Public
Roads. The actual investigation and associated tests were performed by the Materials
and Research Department of the 'California Division of Highways. This work supple-
ments previous investigations of culvert materials.
This report not only contains information on the field performance of test culverts,
but also includes the results of laboratory testing and presents recommendations for
the use of corrugated aluminum pipe.

SUMMARY AND CONCLUSIONS

Field test sites and laboratory tests were selected or designed to provide as much
information as possible on the probable corrosion and abrasion resistance of aluminum
in the short time available to reach early decision on usage.
Empirical equations for projecting data developed by other investigators demon-
strates the inconsistencies that are possible in predicting corrosion rates (see Figure
29). For this reason, all data obtained under this study were projected on a straight-
line basis. The purpose of this projection is to assist in the selection of culvert ma-
terials in accordance with California practice which allows only those materials that

Pape1· spcmsored by Cammi ttee on Metals in Highway Structures.

l
2

have an anticipated maintenance-free service life of 25 or 50 years, depending on the

highway design criteria. Straight-line projections allow direct comparison of various
materials. It is recognized that the final maintenance-free life may be less or greater
than the straight-line projection would indicate. For these reasons, the projections of
short-time laboratory test results were only given qualitative consideration and were
not used alone in making recommendations or in anticipating service life.
In general, the data obtained during this investigation agree with the published liter-
ature in that aluminum does not seem to be chemically attacked when the pH of the solu-
tion is near neutral (7. O) . In addition, there is agreement that within the limits of pH
6. 0 to 8. 0 aluminum should be chemically stable providing there are no other controlling
factors, such as:
1. Waters containing heavy metals;
2. Concentration-cell corrosion;
3. Stagnant or quiescent water; and
4. Water containing large quantities of dissolved chemicals.
It is a conclusion of this study that these foregoing factors can be successfully con-
trolled by requiring an aluminum culvert protected by means of a bituminous or other
approved organic type of coating.
At the pH ranges of 5. 0 to 6. 0, and 8. 0 to 9. 0, the chemical stability of aluminum
does not appear to be as clearly defined as when the pH range is 6. 0 to 8. 0. There-
fore, whenever aluminum culverts are to be used in the environmental pH ranges of
5.0 to 6.0, and at 8.0 to 9.0, they should also be protectively coated on the basis of pH
alone.
Although this investigation did not determine any direct relationship between the
resistivity of a soil or water and the corrosion rate of aluminum, it did indicate resis-
tivity values below which corrosion is more likely to occur.
Published data indicate that at those locations where the in-place soil resistivities
were less than 1, 500 ohm-cm, the corrosion of an aluminum pipeline was controlled by
the application of cathodic protection. Also, published aluminum culvert test results
based on observations over a maximum of 3. 5 years of exposure indicated that corrosion
from the flow was observed to be almost nil when the in-place soil or the water resis-
tivity had a mean value of approximately 3, 100 ohm-cm. other reports have indicated
that aluminum has been attacked when the water contained more than 181 parts per
million of calcium carbonate.
On the basis of the foregoing, it is apparent that a resistivity limitation is required
because it is a guide to the relative chemical content of the environment.
Because crossdrains are generally located in the more critical locations, when
aluminum is used, it should be protectively coated regardless of pH. In addition, the
minimum resistivity should not be less than 2, 000 ohm-cm, unless the invert is also
paved. This resistivity value implies that the total dissolved solids in the water or soil
is approximately 450 parts per million, which can include a total of approximately 125
parts per million of sulfates as SQ4 and chlorides as Cl ions.
In culvert locations which are not as economically critical as crossdrains, changes
in the pH, resistivity limits, and coating requirements could be made so as to gather
further experience with this material.
The test results of this investigation indicate that aluminum is sensitive to abrasion.
In fact, the corrosion-inhibiting cladding on the aluminum specimens was penetrated
in all of the laboratory corrosion-abrasion tests as would have been the case with zinc
coatings on steel. The specimens in this test had a velocity of 5 fps, and the abrading
material was ottawa sand. The field data agree with the laboratory tests that aluminum
is not as abrasion resistant as a steel culvert. Therefore, at this time, it appears
necessary to restrict aluminum from indiscriminate use in streams of high flow veloc-
ities containing an abrasive bed load.
This investigation also indicates that flow velocity per se may not be a controlling
factor in the abrasion process. It appears that the degree of abrasion suffered by a
culvert will not only be a function of the velocity, but also of the size, quantity, and
shape of the bed material. Severe abrasion was observed in the test culvert where the
 3

bed contained shattered and angular rocks. Conversely, at another culvert site with
similar calculated flow velocities, a minor amount of abrasive destruction was ob-
served where the material consisted of rounded boulders.
On the basis of this accelerated investigation, it is estimated that under favorable
conditions, aluminum may have an anticipated maintenance-free service life of 25
years. However, the durability of the material should be continuously verified so as to
confirm or modify the recommendations since they are partially based upon laboratory
data.

RECOMMENDATIONS
It is recommended that the durability of aluminum culvert material be continuously
monitored so as to confirm or modify, through added field experience, the culvert use
recommendations that are shown in Table 1.
Current practice of the California Division of Highways establishes the following
minimum design service lives for culvert materials:

A. Crossdrains under high-type pavements 50 yr

B. Crossdrains under intermediate and low-type
pavements
1. With less than 10 ft of cover 25 yr
2. With more than 10 ft of cover 50 yr
C. Crossdrains under highways on temporary
alignment 25 yr
D. Side drains on all projects except under street
connections surfaced with high type pavement 25 yr

A high-type pavement is defined as either asphalt concrete of 0 .15 ft or more in

thickness or portland cement concrete pavement. An intermediate or low-type pave-
ment is defined as asphalt concrete less than 0.15 ft thick or other pavement of any
thickness mixed with liquid asphalt.
The recommended use of aluminum as a culvert material is predicated on analysis
of all available data and a judgment to eliminate those environmental factors which
could result in earlier maintenance contrary to the established minimum design service
lives. Furthermore, because of the lack of long-term field data and the acknowledged

TABLE 1
RECOMMENDED USE OF MINIMUM GAGE THICKNESS CORRUGATED ALUMINUM PIPE ANTICIPATED
25- YEAR ,MAINTENANCE - FREE SERVICE

Flow Conditions 2

Protective pH Less than 5 FPS Less than 7 FPS Greater than 7 FPS Resistivity
Location Continuous
Coating 1 Range (ohm-cm),
Flow
Non- Non- Non- min. value
Abrasive Abrasive Abrasive
Abrasive Abrasive Abrasive

Over side None 6-8 x x x x No x x 2, 000

drain Bituminous 5-9 x x x x No x x 1, 500
Under None 6-8 x x x x No x x 2, 000
drain Bituminous 5-9 x x x x No x x 1, 500
Side None 6-8 x x x x No x x 2, 000
drain Bituminous 5-9 x x x x No x x 1, 500
Cross Bituminous 6-8 x x No' x No x No 2, 000
drain Bituminous
plus paved
invert 5-9 x x x x No' x x None
When _plpe. is bitum\nously ca;ilcd, backfill to have pH of not less than 5. 0 and no resistivity limitation.
2' 1x 11 in column deooLOs recammonded use.
3
May be used if metal gage thickness is increased by 2 numbers over minimum loading requirements.
Note: Subject to approval, other thin film type of di-electric coatings may be used in lieu of a thin film bituminous coating.
Aluminum is not to be used as a section or extension of a culvert that contains steel sections. In areas where the flow con-
tains heavy metals, aluminum shall not be used unless the invert is paved, irrespective of the pH and resistivity.
4

uncertainties of the short-term laboratory data and current field experience, no recom-
mendations are made at this time for an anticipated 50-yr maintenance-free service
life for corrugated aluminum pipe.

FACTORS THAT INFLUENCE THE CORROSION

OF ALUMINUM IN SOILS OR WATERS
Hydrogen-Ion Concentration pH
It has been reported that barring an actual test, aluminum alloys are unsatisfactory
for use when the pH of the solution is greater than 10 or less than 3 (1). other reports
have indicated that aluminum is generally inert or inhibited from accelerated corrosion
when the pH range of the environment is: 4 to 9 (2), 6 to 8 (3, 4), 5.5 to 7 .8 (5), 4 to 8
(6), and 4.5 to 9 (4). - - - -
- Based on the standard free energies of the constituents, and the deduced electro-
chemical behavior of aluminum, the oxide of the metal (hydragillite, Ah03·H:-O) is
theoretically chemically stable within a pH range of 4 to 8. 6, providing the solution is
free of substances which can form soluble complexes or insoluble salts of the metal (5).
As indicated by the foregoing, it is apparent that aluminum is chemically stable in
the near-neutral range of pH (7 .0). However, it has been emphasized in the literature
that the pH of a solution or soil is not the primary control, or a completely reliable
basis for predicting the chemical stability of aluminum (2, 3, 7, 8).
From the preceding, it is apparent that the knowledge oCthe pH of a solution or soil
can be a valuable tool in predicting the durability of aluminum, but other factors must
be considered.
Because of the relatively long service of steel culverts and pipe, the relative influ-
ence of the pH of the environment to the rate of corrosion of this metal has been de-
termined (10, _!!, 12, ~).

Chemicals
It has been reported that in sodium carbonate solutions of greater than 0. 001 normal
concentrations (approximately 60 parts per million), aluminum is significantly attacked
(9). When the mineral acid concentration is less than 0. 001 normal, aluminum is re-
Sistant to corrosion (9). In acid solutions containing only one anion, the rate of corro-
sion increases in the following order: (a) acetate, (b) phosphate, (c) sulfate, (d) nitrate,
(e) chloride (9).
The presence of heavy metals, copper, mercury, cobalt and nickel in waters has
been reported as a cause of the corrosion of aluminum (1, 3, 4, 8).
Aluminum which does not have the highly corrosion resistant Cladding has been ob-
served to have accelerated corrosion when a water contains 0. 09 ppm of copper, 0 . 08
ppm cobalt, and 0. 03 ppm nickel (3).
It has been generally observed that aluminum corrodes in "hard" waters. Although
no correlation was determined between the relative hardness of a water and the corro-
sion rate of aluminum, the reported data indicate that a "very hard" water contains ap-
proximately 180 parts per million or more of carbonates that are calculated as calcium
carbonate (8). Of the nine tests of aluminum in different natural waters containing more
than 180 ppm of hardness, seven of these samples were found to have a pit depth of 40
mils in less than 6 months (8). The greatest reported concentration of copper found in
the survey of these seventeen natural waters was 0 .11 ppm (8).
From the preceding data, it appears that either a complete chemical analysis should
be made of the soils or waters to which aluminum would be exposed or an economical
means for testing these environments for mineral content should be considered.

Electrical Resistivity of the Environment

The electrical resistivity has been found to be an indicator of the relative concentra-
tion of chemicals in a soil or water (10, 11). The greater the electrical resistivity, the
less the concentration of soluble cheiTIIcals.
 5

Generally, no correlation has been found between relative values of resistivity and
an associated corrosion rate of aluminum (2).
It was reported that on one undergroundgas pipeline "hot spot" cathodic protection
was applied to those sections of the pipe which were embedded in a soil with a resis-
tivity of less than 1, 500 ohm-cm (14).
Based upon the preceding lack Ofdata, it appears that the electrical resistivity of
an environment is thus far only of academic interest with regard to inferring a possible
corrosion rate of aluminum. The electrical resistivity of an environment may be of
use when considering that it is an indicator of the highly mineralized solutions which
can cause the corrosion of aluminum and steel.
The chemical contents in parts per million of solutions and soils may be estimated
by the following formulas (18 and_!!, respectively):

900,000
Total dissolved solids (1)
R

784, 000
Sum of sulfates and chlorides (S04 + Cl) (2)
R1. 5

where R =resistivity in ohm-cm.

Bimetallic Corrosion
When aluminum is electrically connected to steel, approximately 1. 2 volts can be
initially developed and can result in an accelerated corrosion rate of the aluminum (15,
16). Aluminum has been used as a sacrificial anode for galvanically inhibiting the cor-
rosion of steel ( 17) .
The degree ofgalvanic corrosion of an aluminum culvert would be considered minor
if the steel in contact with the aluminum were limited to just a bolt. Conversely, if the
situation were reversed with an aluminum bolt in a steel culvert, the aluminum could
corrode rapidly .
From this, it is obvious that judgment must be exercised when coupling dissimilar
metals to aluminum. A steel bolt used in a culvert band coupler would not seriously
affect the aluminum culvert. The intermixing of steel and aluminum culvert sections
should not be done as there could be rapid corrosion of the aluminum over an extensive
area. The zinc on a galvanized steel culvert is generally anodic and will generally
corrode when electrically coupled to aluminum in most neutral or acid solutions. Once
the zinc is gone, the steel then can cause the aluminum to corrode.

Concentration Cell and Crevice Corrosion

Concentration cell corrosion is generally defined as an electrolytic corrosion cell
which is caused by a difference in the concentration of the electrolyte, or differences
in the concentration of metal ions in solution (1, 16).
In effect, a concentration cell can be the inTtiaTcause of corrosion, or as a result
of corrosion started by other causes (1), it can be the mechanism by which the corro-
sion process can continue. -
Crevice corrosion is generally considered as a corrosion cell which is the result of
differential aeration of the solution (1). A crevice type of corrosion cell can result in
severe corrosion of the aluminum because the voltage of an active/passive cell can be
superimposed upon the voltage of the differential aeration cell (1). Although structural
steel is greatly affected by differential aeration corrosion cells-(16), it is unlikely that
this metal could be generally susceptible to what is commonly called an active/passive
corrosion cell in the normal soil or water (19).
In general, the aggressive types of corrosion cells may be caused to form on
aluminum by the following factors:
6

1. Bolted or riveted construction (1, 20) ;

2. Pockets or locations of liquid entrapment (1, 20) ;
3. Nonuniform soil compaction (2); - -
4. Differential aeration ( 1) ; -
5. Stagnant pools of water (21); and
6. Electrical connection to Ierrous metals (16, 20).

CURRENT RESULTS OF FIELD TESTS

The test results of the eight field test culvert installations are given in detail in
Tables 2 through 4, and shown in Figures 1 through 23. These test sites were chosen
because some are the most highly corrosive and abrasive conditions to which an actual
highway culvert will be or has been placed. This was a means of getting accelerated
results. An exception was the culvert at I-Hum-35-C in the northwestern part of Cali-
fornia near Bridgeville. This culvert site is exposed to the environmental conditions
typical of the geographic area , and these conditions are considered to be only moderate-
ly aggressive.

TABLE 2
FJELD SITE TEST DATA

I-Hum-35-C D-Sha-3-B m - Bul- 21-B IV-SCl-5-C IV-SCr-5-A X-SJ - 53- C XI-SD-2-Nat. Cty Xl-Imp-187- F
Location s
Bridgeville Redding Oroville Los Gatos Scotts Xing Alo Vista Sweetwater Br. Salton Sea

In s talled 8-20-61 11-16-61 8-21-61 10-19-61 10- 3-62 8-16-61 9-26-61 9-29-61
Las t in s pe c tion 8-21-63 5- 2-63 5- 3-63 3- 4-63 8-16-63 1-30-64 5-21-63 5-22-63
Test time (yr) 2. 0 1.5 I. 7 1. 41 0. 83 2 2. 4 1. 7 I. 7

Aver"<'e pH 6. 6 3. 3 2. 7 7. 7 3. 7 4. 5-6. 3 8. 3 7. 5
Min . r e sistivity 2, 500 650 165 3, 500 330 620-973 39 6. 5
Na + K (as Na), ppm 14 7 65 178 12, 300 99, 740
Ca, ppm 44 266 102 470 65 170 12, 300
Mg, ppm 88 328 19 26 504 2, 170
co3, ppm Nil Nil Nil Nil Nil Nil Nil
HC03, ppm Nil Nil 204 9 170 180
Cl, ppm Nil 50 516 26 144 14, 920 41, 520
so.., ppm 996 13, 600 132 2, 246 356 2, 220 7, 920
1This installation was removed during the last inspec tion . 'Steel CMP was in place approximately 1 year prior to installation of
aluminum test pipe.

TABLE 3
CULVERT SITE TEST RESULTS'

Estimated Years to Perforation Based on Metal Loss at: 2

Minimum
Time in
Re sis- Upstream Surface Downstream Surface
Location Metal Test pH Minimum
tivlty of Corrugation or Valley of
(yr) Cross-Section
(ohm-cm) Corrugation
Loss
Abrasion Pitting (corrosion surface)

steel 6. 1 41 6. 4 18
I-Hum-35-C 2. 0 6. 6 2, 500
Aluminum 3. 6 3. 6 6. 9
steel 2. 3 2. 3
Il-Sha-3-B I. 5 3. 3 650
Aluminum 0 . 33 0.33
m-But-21 - B
steel
I. 7 2. 7 165
0 . 56 o. 56
Aluminum o. 56 0. 56
steel No test culvert
IV-SCr-5-A
Aluminum
o. 83 3. 7 330
0. 83 0. 63
steel 1. 3 1. 3
IV-SCl-5-C 1. 4 7. 7 3, 500
Aluminum 0. 14 0. 14
Steel 4. 5 lo 620 to 49 49
X-S. J-53-C' 2. 4
Aluminum 6. 3 973 12 12
Steel 6. 7 6. 7
Xl-Imp-187-F' 1. 7 7. 5 6. 5
Aluminum 12 17
steel 25 33
X!-SD-2- Nat.
Aluminum
1. 7 8. 3 39 4. 8 6. 6
Cty'
1
A11 t e st results are based upon metallograph.ic analysis of culvert samples.
2
Estimated years to perforation for all samples were calculated on the basis of a 16-gage metal thickness ..
3
Corrosion loss measured on the soil side of the pipes.
 7
TABLE 4
AVERAGES OF ESTIMATED YEARS TO PERFORATION
FOR 16-GAGE METAL

Max. Cross-
Metal Abrasion Corrosion
Section Loss

(a) All Seven Comparative Field Test Sites

Galvanized
steel 13 21 18
Aluminum 4. 8 i. a 8. 6

(b) Esttmated 1 for Five Test Sites with pH Between

4. 5 and 8. 3

Galvanized
steel 18 21 27
Aluminum 6. 5 1. 9 13
'Test site with pH of 4.5 has a pH range of 4.5 to 6.3.

Figure l. Field test site, I-Hum-35-C,

mile l.l9: (a) inlet of test pipe-alumi-
num section; (b) samples removed from in-
vert after 2-yr exposure.

Figure 2. Field test site~aluminum, I-Hum-35-C, mile l.19: (a) sample from invert of
aluminum culvert; (b) typical loss of cladding at abrasion surface~2 yr; (c) cladding
intact-2 yr.
8

Figure 3. Field test site~steel, I-Hum-35-C, mile 1.19: (a) sample from invert of
galvanized steel culvert; (b) note loss of zinc and minor loss of steel at abrasion sur-
face~2 yr; (c) zinc abraded but intact~2 yr.

.. ·: . : ~ . .' . . .;\~~~.:~- J::".

'. .= . . . . ·:; ..... ,.
I I
I

II 5!•.lfl
~I .J/ .Ir.I J&'.

Figure 4. II-Sha-3-B, right of Sta. 265±:

(a) field test site; (b) typical invert Figure 5. II-Sha-3-B, right of Sta. 265±:
samples removed after approximately 1.5 (a) cross-section of steel after 1.5 yr of
yr of test. test; (b) cross-section of aluminum.
 9

,y·,

·'
·... ·-~ . .!. _.......:.....:4- ~ . . -.

(b) Rt~

Figure 6. III-But-21-B, right of Sta. 594±: (a) field t est sit e ; (b) invert sa~ples re-
moved after approximate ly 1. 7 yr of test (highly corr osive exposure).

. .• . : .. ,.. .-,._-~ ·~- SEE

OCl.IJI.. A 111 -.11Jl.ollll
V4\•"»u1~1,.._..,.(oA;
m1t•I C MP.• t111 d 111ncllt"ltd I•
~~ ol joinin11 mllol pipes

woll' .1111 3M" l•t;1 bol•• ... ~::11~11p

2' 0.C
<', lH"c11uttltr t unk cinch
•nOll!t,,., •H•(llillQ
woler lo unc1111t WO\~ 2' OC

'htll C M, P. 60•' 101)001

z"'1C."wvod1nti.ori1'19bloc'ol

TYPICAL SECTION

N11r1:
I Th11teelchonn1lrun1conrln1U1u1lron both sides
Qf 1h1 metol CM.P. lor lh• l1n111h ol lh• m1lol
' in1tollallon
2 Tht 2•16" wooden bearing block run1 lho len(llh
,., 11111111 lrlmtt.1l1M

BR. NO. 37-165

5'x 6'x 250' R.C.B.

DETAIL A
Deloil of waler light joinlol
lop1 of melol lnl 11ction1

Figure 7, Abrasion test site, IV-SCI-5-C, Trout Creek.

10

Figure 8. Abrasion test site, IV-SCl-5-C, Sta. 250+25, Bridge No. 37-l65: (a) "as built"
concrete test section at inlet section of test culvert; (b) appearance of concrete test
section after l.4 yr of service showing severe abrasion; (c) view showing loss of ap-
proximately ~ in. of concrete in the concrete test section at the outlet (note deposit
of culvert).
 11

·-

(a)

Figure 9. Abrasion test site, IV-SCl-5-C, Sta. 250+25, Bridge No. 37-165. Samples of
the invert from: (a) galvanized ste el section (note wear of rivet heads); (b) A.D.P.I.
section (note loss of rivets at joint); (c) A.B.A.D,P.I. section .

Al.UflllliUM - ALLOY Cl..AO - MG·llJIH34

( ~~t obra•d aampl•)

Direction of flow

ASBESTOS 8Q(llQEP -'Jfl!tALT DIPPED PAVED INVERT

G VAHIZl!b ST££1.
( b)
Figure 11. Results of abrasion tests,
Figure 10. Abrasion test site , IV-SCl-5-C, IV-SCl-5-C, Sta. 250+25, Bridge No. 37- 165 .
Sta. 250+25, Bridge No. 37 -165: (a) severe Typical cross-sections of pipe invert
abrasion of aluminum after 1.4 yr of serv- after test exposure . (Note: All C.M.P.
ice; (b) severe abrasion of galvanized samples were 10 gage ( o.140±). Steel
steel after 1.4 yr of service (note loss samples are typical of the most abraded
of head of rivet). pipe sections.)
Figure 12 . Field test site, IV-SCr-5-A, right of Sta. 530±: (a) aluminum culvert, field
test site (exposed pipe subsequently backfilled); (b) existing galvanized C.M.P., ap-
proximately 2 yr of service (not placed as part of test program).

Figure 13. Field test site, IV-SCr-5-A, right of Sta. 530±: (a) aluminum invert sample
approximately o.8 yr of test; (b) cross-section of aluminum, nonperforated section.

Figure 14. Field test site, X-SJ-53-C, right of Sta. 6±: (a) field test site: (b)
backfill side: (c) inside (invert). Appearance of cleaned galvanized steel samples
after 2.4 yr of test.
 13

Figure 15. Field test site, X-SJ-53-C, right of Sta. 6±: (a) appearance of inside of
aluminum sample after cleaning (invert); (b) appearance of soil side of aluminum sample
after cleaning.

• r
.. •. .•. ' -.,, -
, ·'~it • .
-

~ ' • . -~:!."'':Li.' t. • .,,

- - . .• - . . ... ~~;'l'

Figure 16. Field test site, X-SJ-53-C, right of Sta. 6±: (a) appearance of galvanized
steel joint after cleaning; (b) cross-section of steel (note partial loss of galvanizing
on both sides).

Figure 17. Field test site, X-SJ-53-C, right of Sta. 6±: (a) appearance of aluminum
joint after cleaning~light-colored areas are corroded sections of pipe; (b) cross-
section of aluminum (note loss of cladding on both surfaces).
14

A11!!1/Ni!/1

( b) (c )
Figure l8. Field test site, XI-SD-2-Nat.Cty at Sweetwater Creek: (a) field test site
at high tide; (b) sample removed from culvert inverts after approximately l.6 yr of
test; (c) backfill side of same culvert samples.
 15

Figure 19 . Field test site, XI-SD-2-Nat.Cty at Sweetwater Creek: (a) appearance of

aluminum after cleaning~l.6 yr of test; (b) cross-section of aluminum (note loss of
cladding and penetration into base metal on backfill side of pipe at bottom of photo).

Figure 20. Field test site, XI-SD-2-Nat.Cty at Sweetwater Creek: (a) appearance of
galvanized steel after cleaning~l.6 yr of test; (b) galvanizing penetrated at localized
spots (top surface of photo).

(a)

-;.:. ~- . ~ -·-
-
··- --
-
-
-
. ~ ~ -

Figure 21. Field test site, XI-Imp-187-F, left of Sta. 498±: (a) field test site: (b)
backfill side of culvert samples~approximately 1.7 yr of exposure. (Dark areas on steel
and light areas on aluminum are locations of corrosion.)
16

Figure 22. Field test site, XI-Imp-187-F, left of Sta. 498±: (a) appearance of aluminum
joint after cleaning; (b) section through aluminurn (note loss of cladding and penetra-
tion of base metal on soil side of alurninurn culvert at bottom of photo).

Figure 23. Field test site, XI-Imp-187-F, left of Sta. 498±: (a) appearance of gal-
vanized steel joint after cleaning~dark areas are rust; (b) section through steel (note
loss of galvanizing and penetration at localized areas).

Abrasion Test Results

The details of the results of the comparative field abrasion tests are shown in
Figures 1 through 3 and Figures 7 through 11 (also see Tables 3 and 4) . Specifically,
the culverts located at I-Hum-35-C and IV-SCl-5-C are the only culverts which could
be considered to have an abrasive environment. From past experience, the former
culvert is considered only an average abrasion culvert, and the latter is known to be
highly abrasive.
The rate of metal loss of the aluminum indicates that it will perforate by abrasion
in approximately one-tenth the time as a steel culvert (Tables 3 and 4).
At periods of a high yearly flow, both abrasion test culverts carry a bed load of
rocks. However, the flow velocity at the test culvert at I-Hum-35-C would range from
10 to 14 fps or about one-half the velocity at the other site. Because of the apparent
2: 1 difference in the calculated flow velocities, it would be tempting to assign this
velocity difference as the cause of the approximately 30: 1 difference in severity of
abrasion damage to the two culverts.
Although not a part of this program, an investigation of a culvert condition was made
in the mountainous vicinity of Redding. This particular 48-in. diameter galvanized
steel culvert was observed to have minor abrasion damage after approximately 7 years
of service.
 17

Cobbles of approximately 6-in. diameter were observed lying in the invert at the
outlet end of this pipe. The calculated flow velocity in the pipe is in the range of 20 to
25 fps.
The reader should be aware that the results of erosion are exceedingly difficult to
explain and formulate objectively to a mathematical certainty. For instance, the
severely damaged test pipe located at IV-SCl-5-C may have had a calculated flow ve-
locity in the range of 2 5 to 30 fps with a bed load of shattered rocks. The minor abrasion
damaged culvert near Redding (Il-Tri-20-A, Sta. 582+73) has a calculated flow velocity
in the range of 20 to 25 fps and has a bed load of rounded boulders. Therefore, it is
obvious that even though flow velocities are highly important, the size and shape (round-
ed or shattered) and hardness of the bed material may be of greater consequence in the
subsequent degree of abrasion of a culvert.
For all practical purposes, no commonly used culvert coating or material would
offer a maintenance-free service life at the highly abrasive test site, IV-SCl-5-C.

Corrosion Test Results

The details of the corrosion field test results are given in Tables 2, 3, and 4, and
shown in Figures 1 through 6 and 12 through 23. Even though some of the test sites are
regarded as being highly corrosive to steel, only three sites had a pH of less than 4. 5,
and the remaining five culverts were installed in sites with a pH range of 4. 5 to 8. 3.
In effect, one-half of the culverts were subjected to a flow or soil which had a pH that
ranged between 6. 6 and 8. 3. For all seven comparative corrosion tests culverts, the
field test data indicate that on the average, the aluminum will be perforated by corrosion
in less time than will galvanized steel.
For the five test sites in which the pH of the soil or flow ranged between 4. 5 and
8. 3, the data again indicated that aluminum would be perforated by corrosion in less
time than will galvanized steel.
As shown by the photographs (Figs. 1 through 23), the removed sections of aluminum
are not generally attacked by small areas of random pitting, but at large areas of the
pipe surface. Therefore, the corrosion is not considered to be the result of a minor
and localized imperfection in the protective oxide film on the surface of the aluminum.
Instead, the appearance of the large areas of corrosion on the soil contacting surface
of the pipe, inside the laps, around the rivet holes, and beneath silt, strongly suggests
that the corrosion is the result of a concentration cell. This concentration cell appears
to be the result of the soil causing a partial shielding of the metal from oxygen and in
one case (XI-lmp-187-F) further complicated by the result of a differential concentra-
tion of soil salts in direct contact with the culvert.
With the exception of the culverts carrying the highly acid runoff, the corrosion at-
tack of the aluminum was most severe on the backfill side of the pipes and in the joints.

LABORATORY TESTS
Corrosion-Abrasion Test
In an attempt to compare the relative corrosion-abrasion resistance between galva-
nized steel and aluminum, these metals were separately exposed to solutions of various
pH and resistivity. The testing equipment (dubbed the "wash machine") is shown in
Figure 24. In each test, four each of the 4 x 8-in. similar metal specimens were
clamped so as to rotate with the drum at a speed of approximately 5 fps. These speci-
mens were electrically isolated from direct metallic contact to the drum by means of
rubber spacers attached to the ends of the specimen. In addition, electrical isolation
was further accomplished by the plexiglass multipurpose observation and access win-
dows which were also used to clamp the samples in place during the test.
Prior to testing, all specimens were degreased with benzene, washed, and scrubbed
with soap, and then thoroughly rinsed with Sacramento city tap water.
Some pilot testing of galvanized steel indicated that the corrosion rate of this com-
posite material would change so rapidly with time that each test would probably require
more than two weeks. Therefore, to expedite results, the zinc was prestripped from
18

TABLE 5
LABORATORY CORROSION-ABRASION TEST DATA
Solution Measurements Ottawa'
Test Distilled Chemicals Used in Test
Designated Water Sand
No . Metal pH Max . Range Resistivity (gm) (gm) Formula Grams
of pH (ohm-cm)

Aluminum 9.0 7 . 7-9.8 100 4,000 4, 000 Na, co, 40

NaCl 25
4 Aluminum 8.8 8. 6-9. 6 100 4,000 4, 000 CaC03 4
NaCl 25
Aluminum 8.7 8 . 7-9 . 5 100 4, 000 4, 000 CaC03 20
NaCl 25
6 Aluminum 10.5 10 . 3-10 . 7 100 4, 000 4, 000 Na ,co, 60
NaCl 25
7 Aluminum 8,0 7 . 3-8.2 100 4, 000 4, 000 NaCl 25
8 Aluminum 3.9 2,2-5.6 100 4, 000 4, 000 CH,COOH 845
NaCl 25
9 Aluminum 3.6 3.5-3 . 9 100 4, 000 4, 000 C1sHs204s 32
NaCl 25
10 Aluminum 6.3 6.2-6 . 4 100 4, 000 4,000 Na OH 1. 6
KH2PO• 4.9
NaCl 25
11 Aluminum 5.0 4 . 2-6.7 100 4,000 4, 000 Na OH 7 . 47
K:iC,H,01H20 46 . 04
NaCl 25
HCl 30
12 Steel+ 9.2 7.9-9 .8 100 4,000 4, 000 Na2B•01 · 10 H,O 40
zinc NaCl 25
13 steel + 6. 3 6. 2-6 . 4 100 4, 000 4, 000 Na OH 1. 6
zinc KH2PO• 49
NaCl 25
14 Steel 6.3 6 . 2-6 . 5 100 4, 000 4, 000 Na OH 1. 6
KH2PO, 49
NaCl 25
15 Steel 8.8 8 .6-9.3 100 4, 000 4, 000 ca co, 20
NaCl 25
16 Steel 7.5 7.0-8 . 8 100 4,000 4, 000 NaCl 25
17 Steel 4.5 3 . 4-4. 9 100 4,000 4, 000 C16Hs2046 32
NaCl 25
18 Steel 5.2 5 . 1-5.6 100 4, 000 4, 000 KH2PO, 60
Na OH 0.5
NaCl 25
19 Steel 6.7 5. 5-9 . 9 1, 000 4, 000 4, 000 Na OH 0.042
KH2PO, 5.0
20 Steel 7.5 7 . 2-7 .9 1, 000 4, 000 4, 000 NaCl 2.2
21 Steel 9.1 8 . 9-9 .6 1, 000 4, 000 4, 000 CaC03 20
NaCl 2.1
22 Steel 4.4 4.1-6 . 3 1, 000 10, 000 4, 000 KHCaH.O, 20
23 Aluminum 4. 8 4.1-5 . 5 1, 000 10, 000 4, 000 KHC,H,04 20
24 Aluminum 9. 1 8.8-9 . 4 1, 000 4, 000 4, 000 CaC03 2.0
NaCl 2.0
25 Aluminum 7. 5 7.2-7 . 7 1, 000 4, 000 4,000 NaCl 2.1
26 Steel 7.5 7.2-7.8 5,000 10,000 4, 000 NaCl 1.08
27 Steel 9.1 9. 0-9 .8 5,000 10, 000 4, 000 ca co, 40
NaCl 0.5
28 Steel 7. 4 7. 1-7 . 4 1,000 10,000 4, 000 NaCl 4.4
29 Aluminum 7. 5 7.0-7 . 5 1,000 10,000 4,000 NaCl 4,4
30 Aluminum 7. 5 6. 8-7.9 5,000 10,000 4, 000 NaCl O.4 to
1.0
31 Aluminum 9.0 9 .0-9.7 5,000 10,000 4, 000 NaCl 0 . 33
CaCQ3 40
32 Aluminum 7. 5 6 . 8-8 . 5 1, 000 10, 000 4, 000 NaCl 4. 1
1
Ottawa sand is: Standard Sand 20-30, ASTM designation C-l9Q.

all galvanized specimens with a solution of hydrochloric acid which was chemically in-
hibited from attacking the steel. In this manner, the average testing period for each
sample was reduced to approximately 8 days.
The details of the chemicals, etc., used in this test are shown in Table 5. The pH
of the test solutions varied from the designated values. The designated pH value is that
value at which the solution was maintained for the greatest period of time.
 19
TABLE 6
LABORATORY CORROSION- ABRASION TEST RESULTS OF STEEL

Years to Perforation-16 Gage

Days Res is-
Te st pH of tivity 10()\(. Minimum Corrosion
No. Abrasion
Test (ohm-cm) Weight Cross-
Surface Surface
Loss Section

14 6.3 9 .9 100 4.39 0.41 0 , 41 1.66

15 8.8 9.2 100 0.48 0.07 0 . 08 0.09
16 7.5 7.5 100 0.21 0.06 0 . 08 0 , 12
17 4.5 7.9 100 0.24 0 . 16 0 . 27 0,58
18 5.2 10 . 6 100 1. 76 0.11 0 . 25 0.13
19 6.7 7.8 1, 000 1. 76 0 . 24 0 . 52 0.37
20 7.5 7.7 1, 000 0.18 0.09 0 . 14 0 , 14
21 9.1 10.1 1, 000 0.98 0 . 11 0 . 17 0 . 15
22 4.4 8.0 1, 000 0.22 0.38 0 . 54 0 . 74
26 7.5 7 .8 5,000 3 . 24 0.20 0 . 29 0 . 24
27 9.1 7.8 5,000 1.05 0.44 1. 31 1.31
28 7.4 8.6 1, 000 0 . 53 0 . 10 0.11 0 . 18

Note: No galvanized steel used in this test. Except for perforation by weight
loss, all test results are based upon metallographic analysis of samples .
Abrasion surface is the upstream side of the corrugation. Corrosion is
downstream side or valley of corrugation.

TABLE 7
LABORATORY CORROSION-ABRASION TEST RESULTS OF ALUMINUM

Years to P erforation-16 Gage

Days Resis-
Te st
pH of tivity 10()\(. Minimum Corrosion
No. Abr asion
Test (ohm-cm) Weight Cross-
Surface Surface
Loss Section

3 9.0 15 . 6 100 4.22 0.47 0. 88 0 . 47

4 8.8 14 . 9 100 0 . 53 0.70 0 .81 1.63
5 8.7 6.8 100 3 . 01 0 . 56 0.45 0 . 56
6 10.5 9. 1 100 0.12 0.10 0 .20 0 . 12
7 8.0 9. 8 100 2 . 34 0 . 46 0 . 46 1 . 07
8 3.9 3.6 100 0 . 34 0.09 0. \7 0 . 14
9 3.6 7 .3 100 0 . 75 0.20 0 . 30 0.34
10 6.3 7.9 100 2.22 0.43 0. 52 1.30
11 5.0 'I. 7 100 0.24 0.23 0 .36 0 , 36
23 4.8 7.8 1,000 1.36 0.23 0 .29 1.28
24 9.1 7.8 1, 000 1.14 0.43 0 .26 1.29
25 7.5 10. 0 1, 000 2.48 0.41 0. 41 0 . 82
29 7.5 9. 9 1, 000 1.92 0 . 36 0 . 40 1.08
32 7.5 36.2 1, 000 3 . 24 0.91 1. 32 1.48
30 7.5 8 .3 5,000 1. 62 0.34 0.3<1 0 . 68
31 9.0 7.6 5,000 0 . 94 0.19 0. 19 0 , 84

Note: Except for perforation by weight loss, all test results are based upon
metallographic analysis of samples . Cladding was penetrated on abrasion
surface in all tests. Abrasion surface is the upstream side of the corruga-
tion. Corrosion surface is the downstream side or the valley of the cor-
rugation.

It should be noted when referring to Tables 6 and 8 that initial pilot testing of the
galvanized specimens also indicated that within the allotted short testing period, the
zinc coating could protect the steel from corrosion where abrasion would be less severe
such as on the downstream side of the corrugation. Thus, it is expected the estimated
years to corrosion perforation for steel would be greater than those shown in the fore-
mentioned tables had the specimens been galvanized.
Test Results-Corrosion. -The details of the corrosion-abrasion tests for each
metal are shown in Tables 6, 7, and summarized in Table 8. The extrapolated years
to perforation are presented on the basis of four types of measurements:
20
TABLE 8
SUMMARY OF LABORATORY CORROSION-ABRASION
TESTS, 16-GAGE METAL

Max. Cross- Abrasion Corrosion Weight

Metal
Section Loss Surface Surface Loss

(a) Averages of Estimated Years to Perforation

Plain steel 0.20 0. 35a 0.48 1.3

Aluminum 0.39 0.46 0.84 1. 7

(b) Averages of Estimated Years to Perforation for

pH of 6. 0 to 8 . 0 Only

Plain steel . 0 .18 0.26a 0.45 1. 72

Aluminum 0.40 0.43 0.99 2.12
aGenerally corrosion pits and not metal loss from simple
abrasion.

~emovable
plexiglas

4- by B-in.
test specimen

coated
steel
drum

rim speed
approx . 5 fps

Figure 24. Corrosion-abrasion testing machine (steel drum 24 in. in diameter, 8 in .

deep).

1. Maximum cross-section loss;

2. Just the abrasion surface or the upstream side of the corrugation which had
initial contact with the sand;
 21

4.5
le-i

4.0 - •-100 ohm cm

o - 1000 ohm cm
6-5000 ohm cm

3. 5 - Note:
U)
Weight loss was based Designated pH

~fo >- RangeI of pH

U) on the weight of 16 go.
0
_J
stee I sheet.
..,
I
solution
1- during le st.
:J: 3.0
(.!)
a.I
31:
~
0
Q 2 .5
0
1-
Ul
a:
;-:; 2.0
>-
0
a.I
l-
fe-i I

s
0 1.5
0..
<{
a:
I-
x
a.I
~
1.0 ,..,.......,

1 ......

0.5 ,...... ri

- ~

0
0 2 4 6 B 10 12
HYDROGEN-ION CONCENTRATION pH

Figure 25 . Laboratory corrosion-abrasion test of steel, extrapolated years to 100%

weight loss vs pH.

3. The corrosion surface which is any section of the corrugation except the abrasion
surface; and
4. By means of 100 percent weight loss of the specimen.
In this particular laboratory corrosion-abrasion test with highly aerated solutions,
aluminum generally showed twice the resistance to perforation from corrosion as did
plain or bare steel. However, this procedure did not test the effect of concentration
cell-type corrosion on aluminum or steel, nor did it show the benefit that might be
gained had the steel specimens been galvanized.
Because of the corrosion characteristics of these two metals, it would be expected
that aluminum would not be as adversely affected by an aerated solution as would steel.
Conversely, in quiescent solutions, the corrosion resistance of aluminum is reduced
as was indicated by other tests performed.
22
4.5
'
Designated ' pH

•-100
I
, I
ohm cm
4.0 ~
Range of pH solulfon
0-1000 ohm cm
during test .
.0.-5000 ohm cm
Note:
Weight loss was based
3. 5 ~
on the weight of 16ga.
(/) aluminum shPet
(/)
0
....I
1-
:I: 3.0 T
Cl
w
3::
';/!.
0
Q 2.5 ~

g ~ ..
(/)
a: •
~ 2.0
>- t-o-1
c
w
I-
< H-~
....I
0 1.5
a.. .,.
<
a:
I-
x 1->4
w
1. 0
~

0.5
.......

..i
0
0 2 4 6 B 10 12
HYDROGEN-ION CONCENTRATION - pH

Figure 26. Laboratory corrosion-abrasion test of aluminum, extrapolated years to lOO%

weight loss vs pH.

Disregarding the resistivity of a solution, the data in Figure 25 indicate that steel
could rapidly corrode in aerated solutions where the pH is less than approximately 5. 0
and greater than 7 .0. However, in the case of steel, it is misleading to infer that steel
has its greatest corrosion resistance when it is subjected to an environment with a pH
range between 5. 0 and 7. 0. Further analysis of these data show that for the steel test
series, the pH of the solution is an important factor in the corrosion rate only when the
pH is less than approximately 7. 3. At pH values of less than approximately 7. 3, the
resistivity and the pH of the solution are the controlling factors. At greater pH values
(7. 3 or greater), the resistivity is the primary control of the relative corrosion rate of
steel.
The data in Figure 26 indicate that aluminum is more resistant to corrosion in the
pH range of approximately 5. 5 to 8. 5. An analysis of the data did not indicate any
clear-cut trend in the influence of resistivity on the rate of corrosion. It is suspected
 23
JO

8
/
Cf) ~/
Cf)
v
g
I- 6
v
,, /
I
(.!)
.P
w
5:

~4
~
v /
,,,.,,.-
w ~~
.
v ,/
u ~s~
a:
w
a.. >1 v v-

--- --
_/ 1-10. '?:.8--
2

l_.o"
/"
~
--::;.;.---
\€.SI

.......-
._... ~ i - -
2 3 4 5 6 7 8 9
DAYS OF TESTING

Figure 27 . Laboratory corrosion-abrasion test, r eproduc ibility of plain steel.

,,,,...

v
3.0

/
_..0'!.
l~~O· ~

'I..°'
"'~0·'1..-,
/
z
t-

JV
"'
~ 1.0
"'
Q,

/
A
ct'
v
0
0 10 20 30 40
DAYS OF TESTING

Figure 28. Laboratory corros ion-abrasion test, reproducibility of aluminum .

that the aluminum was more sensitive to the types of chemicals than to the concentra-
tions of the different chemicals used in this test.
Figures 27 and 28 are shown to depict the accuracy in reproducing a single type of
test. From these data, it is obvious that the individual test results probably have a
test accuracy of ± 20 percent.
24

*=Author of equotion - + - - - - l - - - 1 - - - + - - /-,-<q-- - 1 1 - - - - - l

T,t =Time /
D,p,d =Pit Deplh

--- - ---
A,K =Constants

J:
I-
"-
"'
0

I-
"-

Tl ME

Figure 29. Time vs depth of pitting.

..,
I • ,

Figure 30 . Laboratory corrosion-abrasion test, steel. Cross-sections of plain steel

test sample after approximately 8 days of testing. Note minor abras ion loss of metal
(left photo) which was caused by Ottawa sand and a specimen velocity of approximately 5
fps. Note lack of corrosion in this test .

All of the reported test data were extrapolated on a straight-line proportional basis
to the particular end point; i.e., metal perforation or 100 percent weight loss. Such
methods of extrapolation of data are not recommended as being highly accurate but are
a means for comparison of test results. An equation which includes a factor of de-
creasing rate of corrosion with time was not used. Therefore, these data imply an ex-
aggeration of the numerical difference of the corrosion rates which were measured at
the end of each test.
Since equations are available which include a factor describing the decrease in the
corrosion rate with time, Figure 29 shows that there is a choice of three for steel
(24, 25, 26) and one for aluminum (8).
-Figure29 should not be construed to indicate that the corrosion rate of one metal is
clearly less than the other. This is because the required constant for each equation
may be many-fold greater or less than the other. Therefore, when the constants are
included in the equations, the result could be that one metal may perforate in a few
days while the other metal may require years to perforate.
 25
FLOW . . _ .

Figure 31. Laboratory corrosion-abrasion test, aluminum. Cross-sections of aluminum

test sample after approximately 8 days of testing. Note typical loss of cladding (left
photo) which was caused by Ottawa sand and a specimen velocity of approximately 5 fps.
Note lack of corrosion in this test.

TABLE 9
SOLUTIONS USED IN THE CONTINUOUS SUBMERSION TESTS

Re sis-
Test Tap Water Grams of
pH tivity Chemical
No. (gm) Chemical
(ohm-cm)

1 4.3 1, 000_ 10, 000 KHCaH404 22

2 7.5 1, 000 10,000 NaCl 5.2
3 9.0 1, 000 10, 000 CaC03 10
NaCl 5.0

TABLE 10
CHEMICAL ANALYSIS OF SACRAMENTO CITY TAP WATER

Total Hard-
Alk. Cl Ca Mg Na Fe N F
Solids ness

83 36 20 3 11 8 4
Nil 0.1 Nil Nil
to 113 to 76 to 78 to 21 to 19 to 18 to 8

Resistivity = 8,ooo ohm-cm.

pH= 7.2.
Milligrams per liter.
Chemical analysis from California Domestic Water Supplies, Department of
Public Health, 1962.

Test Results-Abrasion. - Figures 30 and 31 show the results of abrasion on plain

steel and aluminum when corrosion was practically absent. In all tests there was no
noticeable wear on the abrasion surface of the steel. The abrasion surface is the up-
stream surface of the corrugation. Generally, the steel pitted on the abrasion as well
as on other surfaces of the steel.
26

TABLE 11
RESULTS OF CONTINUOUS SUBMERSION TESTa
(Estimated Years to Perforation for 16-Gage Metal)

Metal Sample pH Years

Galvanized steel 1 4.3 Steel was

2 4.3 unaffected
Aluminum 1 4.3 2.9
2 4.3 2.9
Galvanized steel 1 7.5 Steel was
2 7.5 unaffected
Aluminum 1 7.5 2.9
2 7.5 3.7
Galvanized steel 1 9.0 Steel was
2 9.0 unaffected
Aluminum 1 9.0 2.9
2 9.0 3.3
13.Test solutions had a resistivity of 1,000 ohm-cm and test
period was 70 days.

Figure 32. 70-day laboratory test of continuous submersion of galvanized steel: (a) pH =
4.3, resistivity = l,000 ohm-cm, galvanizing intact, no corrosion of steel; (b) pH =
7.5, resistivity = 1,000 ohm-cm, galvanizing intact, no corrosion of steel; (c) pH =
9.0, resistivity = 1,000 ohm-cm, galvanizing intact, no corrosion of steel.
 27

Figure 33. 70-day laboratory test oi' continuous submersion of aluminum-solution pH

4.3, resistivity = 1,000 ohm-cm. Note corrosion at edges near rivet hole, role marks
and where the two pieces of aluminum overlapped (right photo).

Figure 34. 70-day laboratory test of continuous submersion of aluminum-solution pH =

17.5 resistivity = 1,000 ohm-cm. Note corrosion at edges near rivet hole and where the
two pieces of aluminum overlapped (right photo).

The typical loss of the aluminum cladding on the abrasion surface after an average
of 8 days of testing is shown in Figure 31. At the conclusion of Test No. 32 (36 days) ,
the face of the shear ed leading edge of the aluminum test pa nels p eel ed back for a dis-
tance of approximately 1/i.6 in. as a result of the impact of the spe cimen with the ottawa
sand at a velocity of approximately 5 fps.
After the mounting and polishing of all metallographic specimens, the steel was
etched for 30 seconds with a solution of nitric acid (HN03) and amyl alcohol (CsHuOH).
The aluminum specimens were etched for approximately 10 minutes with concentrated
sodium hydroxide (NaOH) solution.
28

Figure 35. 70-day laboratory test of continuous submersion of aluminum~solution pH =

9.0, resistivity = 1,000 ohm-cm. Note corrosion at edges near rivet holes where the two
pieces of aluminum overlapped (left photo), and corrosion in the long scratch (right
photo).

(a)
Figure 36. Laboratory test in fog room: (a) approximately l yr of exposure of galva-
nized steel and no corrosion of steel; (b) 117 days of exposure of aluminum (note cor-
rosion at edges near rivet holes and at the line where the two pieces of aluminum over-
lapped).

Continuous Submersion
The results of this laboratory test are given in detail in Tables 9, 10 and 11, and
shown in Figures 32 through 35.
The corrosion rate of the metal in this test was determined by micrometer measure-
ments rather than by metallographic analysis. Basically this test consisted of sub-
merging duplicate specimens of either riveted aluminum or riveted galvanized steel
metal in a plastic container containing the described test solutions. There was no in-
termixing of galvanized steel or aluminum in any container. Both metals were culvert
stock and were riveted by a commercial culvert fabricator. The culvert sheet metal
and rivet materials are those which are commercially specified as culvert stock.
 29

The pH and resistivity of the solutions were maintained to the proper level by peri-
odic additions of the chemical additives. After the first 30 days of test, all of the solu-
tions were replaced with a fresh test solution. There was no stirring or attempt to
aerate the test solution.
An effort was made to have the test specimens in a quiescent water which would be
similar to that found in bogs or marsh areas. Also, the resistivity was kept at a con-
stant value of 1, 000 ohm-cm. On the basis of steel corrosion, a solution resistivity
value of 1, 000 ohm-cm is generally not considered as being highly corrosive, but it is
also not disregarded as being non-corrosive.
In all cases the zinc on the galvanized steel is intact and there is no corrosion of the
underlying steel after 70 days of testing (Fig. 32).
In all cases, the aluminum was attacked at the metal laps, edges of the plate, near
the rivet hole, and sometimes at scratches and also sheet rolling marks due to the cor-
rugating process (Figs. 33, 34, and 35).
The overall corrosion of the aluminum was less in the solution of pH 7. 5 than in the
4.3 and 9.0.
The results of this test indicate that among other variables, a concentration cell type
of corrosion attack is a common denominator in the causes of corrosion of aluminum
in quiescent solution. Also, aluminum can aggressively corrode in solution of pH 4. 3
and 9.0.

Laboratory Test in Fog Room

The fog room used for this test is a concrete curing room which is maintained at ap-
proximately 73.4 F and 100 percent relative humidity by means of temperature controls
and water fogging equipment. The fog room can be construed as a misnomer as droplets
of water are continuously being dispersed throughout the chamber and seem more like
rainfall.
The pH of the atomized water is 8. 2 and the resistivity is 6, 300 ohm-cm.
Figure 36 shows the appearance of galvanized steel after approximately one year of
testing and the zinc is intact. Also shown is the typical result of 117 days and also 94
days of exposure of the riveted aluminum samples to the fog environment. In this case,
the aluminum has been attacked near the rivet hole, cut edges where the plates were
in contact, and also at the line where the two pieces overlapped. Apparently this cor-
rosion attack is the result of a concentration cell.
By means of a micrometer, the depth of corrosion was determined and extrapolated
on a straight-line proportional basis to a calculated time to perforation. The results
of these measurements are given in Table 12.

TABLE 12
RESULTS OF FOG ROOM TEST

Est. Years to
Days of
Metal Sample Perforation for
Test
16-Gage Metal

Galvanized steel 1 ±365 Steel was unaffecteda

Aluminum 1 94 3.2
2 94 3.2
3 94 3.2

asample was from previous testing .

30

OTHER FIELD TESTS OF ALUMINUM CULVERTS

An excellent and comprehensive study of the field performance of aluminum culverts
was reported by Lowe and Koepf (2). Although the authors did not report any rates of
corrosion, they did include their observations on the appearance of the culverts. The
reported condition of the pipes visually ranged from an unaffected condition to the ex-
treme where the pipe wall was perforated. In many cases, the resistivity of the in-place
soil or flow and also the pH was tabulated.
As wa s indicated (2), it is obvious that the majority of the reported installations had
no problems involving corrosion be cause approximately 60 percent of their data indicate
that the visual condition of the culvert was unaffected or the metal was stained. It is
assumed that stained aluminum is not evidence of corrosion and indicates a relatively
unaffected condition (3).
The authors (2) did not mathematically present their findings regarding the influence
of soil pH or reSistivity on the corrosion rate of aluminum. However, there appea r to
be some general mathematical relationships which could be of value.
For instance in Table 13, the reported condition of the culverts has been listed in an
assumed rank of corrosion severity that varies from unaffected to perforated. In rank-

TABLE 13
NATIONWIDE FIELD TEST RESULTS OF ALUMINUM CULVERTS (!)
1
Reported Average Average Mean Estimated 2 Average Average Mean
1

Culvert Acid Alkaline Resistivity Rate of Acid Alkaline Resistivity

Condition pH pH (ohm-cm) Corrosion pH pH (ohm-cm)

Unaffected 6,2 7 .9 2, 100

Nil 6.0 7 .8 3, 100
Staining 5.D 7. 7 3, 300
Etching 5.~ 8. 0 600 Light to
5. 6 7. B 2, 000
Pitting 5. 7 7. 7 4, 700 moderate
Cladding
removed 2.B 150
Severe 3 .0 250
Perforated ~.l 300
1
Geornetric Mean.
2
This estimate is speculation. The estimated rate of corrosion is entirely based upon the terminology
that was used in the report for describing the visual appearance of the culverts. No rates of corrosion
were reported.
Maximum years of service of reported culverts were 3. 5.

TABLE 14
CULVERT SITE TEST RESULTS BASED ON INSPECTION OF NOVEMBER 23, 1964'
(Addendum; see also Table 3)

Estimated Years to Perforation Based on Metal Loss at: 2

Time in Minimum
Upstream Surface
Location Metal Test pH Resistivity Minimum Downstream Surface or
(yr) (ohm-cm) of Corrugation Valley of Corrugation
Cross-Section
Loss (corrosion surface)
Abrasion Pitting

Steel 6.1 41 6.4 18

I-Hum-35-C 2.0 6.6 2, 500
Aluminum 3. 6 3. 6 6.9
Steel 2' 3 2, 3
JI-Sha-3-B3 I. 5 3. 3 650
0,33
Aluminum 0. 33
Steel 0. 56 0.56
!Il-But-21-B'
Aluminum
I ,7 2. 7 165
0 . 56
0 . 56
Steel No test culvert
IV -SCr-5-A' 0,83 3.7 330
Aluminum 0.83 0.83
steel 1.3 1. 3
IV-SC1-5-C 3 1.4 7 '7 3, 500
0 , 14
Aluminum 0. 14
Steel 4.5 to 620 to 49 49
X-SJ-53-C' Aluminum 2.4 12 12
6. 3 973
5 Steel 24 24
XI-lmp.J87-F 3.2 7. 5 6, 5
Aluminum 34 37
Steel 8. 0 B. 4
XI-SD-2-Nat. Cty 3. 2 8.3 39 11
Aluminum B.8

'All toat result~ a1·c based u1>1111 111otnllogn111h!c n1u1!yRIS of culvert samples.
1
Estinintod ymrs lo po.rforn.Uon J.oJ' uJl snmples were calculated on the basis of a 16-gage metal thickness .
';\ lumlnum only pl!rforal.Cd within test limo.
'Alu111ln11111 ttnd r,n!Y11nlzed s1ccl per!orate!I wllhln test ilme.
'co1·ro!!llcm loss munsm·t>t.I cm l hC!' soil si de of U1c plpcs.
 31
TABLE 15 ing the relative condition of the culverts,
COMPARATIVE FIELD TEST SITES BASED the more severe condition noted was arbi-
ON INSPECTION OF NOVEMBER 23, 1964
(Addendum; see also Table 4) trarily assigned to represent the rank of
Max. Cross- the culvert. For instance, if the culvert
Metal Abrasion Corrosion
Section Loss was reported as "mottled stain. No at-
(a) All Seven Comparative Field Test Sites tack. Random pitting of clad in invert,"
17
this culvert was assigned to the "pitting"
Galvanized steel 13 21
Aluminum 8 ,5 1. 9 11 classification in Table 13. For each of
(b) Estimated1 for Five Test Sites with pH
these culvert conditions, the acidic pH's
Between 4. 5 and 8. 3 of less than 7. 0 were arithmetically aver-
Galvanized steel 18 21 25
aged. The same was true of pH's that
Aluminum 12 1. 9 17 were greater than 7. 0. In addition, the
1
Test site with pH of 4. 5 has a pH range of 4, 5 to 6. 3 , least resistivity of the in-place soil or
water were averaged on the basis of the
computed geometric mean (27) which is

Geometric mean ( 1)

where n = number of observations, and X = observed value.

The geometric mean of the resistivity values was used because of the extremes in
values that are normally found in resistivity measurements.
Although the validity of this analysis of data in Table 13 has not been verified, it is
interesting to note that there seems to be a reasonably implied correlation of the data.
This is implied by the observation that the severity of corrosion increases with de-
creasing pH and resistivity.
In the subject report (2), it was stated that extensive experience has indicated that
if aluminum is not attacke d by corrosion after periods of a year or more, then the
aluminum metal may be considered to be relatively inert to the environment. Converse-
ly, it should also be true that if significant corrosion of the aluminum occurs at an
early exposure period, then aluminum should sustain some rate of corrosion until dis-
integration.
From Table 13, it appears that the anticipated performance of aluminum could be
satisfactory when the pH ranges between 6. 0 and 7. 8. It is highly probable that when
the pH of the environment exceeds these values, the aluminum could corrode at a rate
that would vary from minor to severe.
The resistivity measurements were determined for the most part on an in-place soil.
Therefore, they may not be accurately reproducible owing to the fact that these values
are highly dependent upon the seasonally variable moisture content of the soil.
Normally, soil resistivity measurements used in culvert corrosion technology are
based on the minimum value. The minimum resistivity is normally less than the in-
place soil resistivity. Therefore, care should be exercised when directly comparing
the in-place field values to the minimum resistivity of a soil (10).

REMARKS
There are few published data concerning the service life of aluminum when used un-
derground or as a culvert. The longest reported service life for this material as a
culvert is 3. 5 years (2).
For underground applications of aluminum pipe, reports of up to 15 years have been
published (22). As reported, the 388 total miles of aluminum pipeline with an estimated
average ofseven years of service, only 8 to 9 miles have had to be replaced because
of corrosion. None of the failed pipe was coated or received cathodic protection. Of
this total reported pipe length of 388 miles, approximately 25 percent of its total length
is protectively coated. In addition, approximately 30 percent of the total length of the
pipelines received cathodic protection. Cathodic protection was not necessarily applied
to coated pipe. The reported wall thickness of these pipelines varied from an equivalent
32

corrugated metal pipe gage of approximately 16 to a reported maximum which would _be
approximately equivalent to 8-gage thickness. Thin-gage pipe wall thickness was in
the minority .
The review of the literature shows that some aluminum facilities have corroded
when placed underground or as a carrier of water. Except for broad generalities, spec-
ific criteria for predicting the service life of aluminum as a culvert are not available.
Past experience with the use of galvanized steel culverts without a means for esti-
mating service life, resulted in 63 percent of all of the culverts (7; 000) in just one of
the eleven California highways districts needing replacement or repair within 30 years
of service (23). From this past experience, it is obvious that caution has to be exercised
before a material should be allowed to be used randomly in large quantities on highway
projects.
Because of the concentration-cell type of corrosion which has been observed in the
laboratory and on the backfill side of the culverts in the field test sites, no aluminum
cross-drains should be placed in critical locations without being bituminously or other-
wise protectively coated.
[Tables 14 and 15 are addenda to original paper.]

ACKNOWLEDGMENTS
This investigation of the corrosion of metal culverts was conducted as one of the
activities of the Materials and Research Department of the California Division of High-
ways, in cooperation with the Bureau of Public Roads.
The authors wish to express their appreciation to J. L. Beaton, Materials and
Research Engineer, for his advice and direction during this study; also to the numerous
personnel of the California Division of Highways and those of the Materials and Research
Department who extended their aid and cooperation during this study.

REFERENCES
1. Pryor, M. J. The Corrosion of Wrought Aluminum Alloys. Richland, Washing-
ton; A lecture in March 1954 Educational Program of the Columbia Basin Chapt .
of the ASM, and a publication of Kaiser Aluminum and Chemical Corp., Dept.
of Met. Res., March 1955.
2 . Lowe, T. A . , and Koepf, A. H. Corrosion Performance of Aluminum Culvert.
Highway Research Record No. 56, pp. 98-115, 1964.
3. Sawyer, D. W., and Brown, R.H. Resistance of Aluminum Alloys to Fresh
Waters. Corrosion, Vol. 3, No. 9, p. 443, 1947.
4. Haygood, A. J., and Minford, J. D. Aluminum Cooling Towers and Their Treat-
ment. Corrosion, Vol. 15, No. 1, p. 36, Jan. 1959.
5. Deltombe, E., and Pourbaix, M. The Electrochemical Behavior of Aluminum.
Corrosion, Vol. 14, No. 11, p. 16, Nov. 1958.
6. Shatalov, A. Y. Effet de pH sur le Comportement Electrochemique des Metaux
et Leur Resistance a la Corrosion. Nauk, U.S.S.R., Doklady Akad, Vol. 86,
p. 775, 1952.
7. Lorking, L. F., and Mayne, J.E. 0. The Corrosion of Aluminum. J our. Appl.
Chem., Vol 11, p. 170, May 1961.
8. Godard, H. P. The Corrosion Behavior of Aluminum in Natural Waters.
Canadian J. Chem. Eng., p. 167, Oct. 1960.
9. McKee, A. B., and Brown, R.H. Resistance of Aluminum to Corrosion in
Solutions Containing Various Anions and Cations. Corrosion, Vol. 3, No. 12,
p. 595, Dec. 1947.
10. Beaton, J. L., and Stratfull, R. F. Field Test for Estimating Service Life of
Corrugated Metal Pipe Culverts. Highway Research Board Proc. , Vol. 41,
pp. 255-272, 1962.
11. Stratfull, R. F. Field Method of Detecting Corrosive Soil Conditions. Univ. of
Calif., L.A., Proc. 15th Calif. Street and Highway Conference, I. T. T.E.,
p. 158, 1963.
 33

12. Stratfull, R. F. A New Test for Estimating Soil Corrosivity Based on Investiga-
tion of Metal Highway Culverts. Corrosion, Vol. 17, No. 10, p. 115, Oct. 1961.
13. Stratfull, R. F. Highway Corrosion Problems. Materials Protection, Vol. 2,
No. 9, p. 8, Sept. 1963.
14. Whiting, J. F., and Wright, T. E. Cathodic Protection for an Uncoated Aluminum
Pipeline. Corrosion, Vol. 17, No. 8, p. 9, Aug. 1961.
15. U. S. Dept. of Commerce, Office of Technical Services. Corrosion Prevention,
Part M. of Maintenance and Operation of Public Works and Public utilities.
NAVDOCKS, TP-Pw-30.
16. H. H. Uhlig (ed.). Corrosion Handbook. New York, John Wiley and Sons, 1948.
17. Verink, E. D., Reid, K. K. , and Diggins, E. R. Current Output of Light Metal
Galvanic Anodes as a Function of Soil Resistivity. Paper published in Cathodic
Protection, by the Nat. Assoc. of Corrosion Engineers, 1949.
18. Betz Handbook of Industrial Water Conditioning. Philadelphia, Penn., W. H. and
L. D. Betz, 1953.
19. Pourbaix, M. Corrosion, Passivity and Passivation from the Thermodynamic
Point of View. Corrosion, Vol. 5, No. 4, p. 121, April 1949.
20. Evans, U. R. The Corrosion and Oxidation of Metals. New York, St. Martins
Press Inc., 1960.
21. Whiting, J. F., and Godard, H. P. The Corrosion Behavior of Aluminum in the
Construction Industry. Canada, The Eng. Jour., June 1958.
22. Aluminum Pipeline Case History Data. NACE, Tech. Unit. Comm. T-2M and
Task Group T-2M-1, Materials Protection, Vol. 2, No. 10, p. 101, Oct. 1963.
23. Beaton, J. L., and Stratfull, R. F. Corrosion of Corrugated Metal Culverts in
California. Highway Research Board Bull. 223, pp. 1-13, 1959.
24. Putnam, J. F. Soil Corrosion. Proc. Am. Petroleum Inst. (IV) 16, 66, 1935.
25. Fetherstonhaugh, E. P. Discussion of Underground Corrosion. Proc. Am. Soc.
Civil Engr. Vol. 101, p. 828, 1936.
26. Logan, K. H., Ewing, S. P., and Denison, I. A. Soil Corrosion Testing.
Philadelphia, Penn. Symposium on Corr. Test Procedures, ASTM, 1937.
27. ASTM Manual on Quality Control of Materials. ASTM, Spec. Tech. Puhl. 15-C,
Jan. 1951.

Discussion
HUGH P. GODARD, . Aluminium Laboratories Limited, Kingston, Ontario, Canada. -
On the basis of an accelerated investigation, Messrs. Nordlin and Stratfull concluded
that, under favorable conditions, aluminum culverts may have a service life of up to
25 years. They decided that to obtain this life, certain criteria for pH, water and soil
resistivity must be adhered to. These authors realized the uncertainty of their con-
clusions, since they recommended periodic examination of existing culverts to confirm
or modify their views .
By contrast, I will endeavor to support the view that aluminum culverts will last far
in excess of 25 years in the great majority of waters and soils, even without paint or
other protection. I suggest also that the criteria selected by Nordlin and Stratfull are
not applicable or necessary for predicting the service life of aluminum culverts.

Introduction
In any problem involving the possible corrosion of a metal, it is necessary first to
select the criterion by which the extent of corrosion should be judged. This is import-
ant, since although too mild a criterion may lead to premature failure, too severe a
criterion will lead to an unreasonably expensive structure. In the case of metal cul-
verts, it is suggested that the sole criterion of corrosion is the continued ability of the
culvert to support the overburden and normal live loads for which it was designed.
34

In natural waters and soils, most aluminum alloys, and certainly all of those which
would be considered for culvert construction, do not suffer uniform or general corro-
sion. That is to say they do not waste away by general thinning. If corrosion attack
does take place, it is localized, and usually in the form of pitting, in a random pattern
over the surface of the metal. Further the pits are of small diameter-usually less
than 1/a in. The effect of a pit on the mechanical strength of a sheet is in proportion to
the cross- sectional area of metal removed, which is negligibly small compared to the
total cross- section. Accordingly, it can be safely predicted that the pitting of an alumi-
num culvert will have no appreciable influence on the load-bearing capacity of the cul-
vert unless the pits become so numerous and so large that an appreciable cross- section
of the metal is consumed-a very improbable condition as the available evidence will
demonstrate.
Most of the literature on the use of aluminum in water and soil pertains to pipelines.
For this use, the primary criterion of corrosion is perforation, since even the first
hole causes a loss of the fluid conveyed, and must be repaired promptly and at some
expense. As just pointed out, this is not the case with culverts. In reviewing the litera-
ture, the authors failed to appreciate this distinction, and as a result have included
criteria such as water composition, water resistivity, and soil resistivity. These give
no information on the loss of strength due to pitting, although they have some value in
predicting the tendency to pitting and the rate of penetration.
Water composition (which determines water resistivity) affects the incidence and rate
of pitting of aluminum and is important when considering pipelines, tanks and water
handling equipment, but is of little consequence in the case of culverts for which small
perforations would be unimportant.
Soil resistivity also affects the incidence and rate of pitting. In addition, it gives
information on soil battery effects which occur in the case of pipelines which traverse
several soil types. By contrast, metal culverts are normally relatively short, and
buried in one ty..Pe of soil. This further reduces the value of soil resistivity readings
in culvert considerations.
The influence of pH on the corrosion of aluminum alloys is dependent on the specific
ions which cause the pH, and hence, pH by itself, is not a reliable criterion in judging
the corrosivity of an environment to aluminum. For example, aluminum is fully re-
sistant to concentrated nitric acid at pH 1, to acetic acid at pH 3, and to ammonium
hydroxide at pH 13. Even in concentrated sodium carbonate solution at pH 11, after an
initial period of activity, a protective surface film forms on aluminum which then be-
comes highly resistant. In studies of aluminum corrosion, pH has not been found to be
a significant variable in either waters or soils.
The authors described the corrosion on 8 aluminum culvert installations in California .
Concern was expressed in that patches of aluminum surface were corroded, as distinct
from point pitting. The authors apparently did not realize that this is the normal be-
havior of an Alclad aluminum surface in soil. An unclad product would have shown only
pin-point pitting attack. The corroding area was protecting the core alloy exposed by
the pits. Experience on the rate of consumption of cladding in seawater suggests that
the rate of cladding consumption drops sharply with time, and that a linear projection
of several years data is unduly pessimistic.
Unfortunately, the data given for the California culverts in Tables 3 and 4 are not
presented in a form that can be appreciated. However, it would appear that the loss of
cross-section of aluminum culverts due to corrosion was appreciably less than that for
galvanized steel. The method of extrapolating to obtain years to perforation was not
given, but in my experience this cannot be calculated from early corrosion data on a
clad aluminum product, since the rate of consumption of cladding is not known, the
minimum area of cladding that must be removed to permit pitting into the exposed core
metal is not known, nor is the rate of penetration of the exposed core metal.
The laboratory erosion and corrosion tests described are of very dubious value in
predicting the field service life of aluminum culvert. It is suggested that examination
of typical installations cited by Lowe and Koepf (28) would be far more rewarding.
 35

Corrosion of Aluminum by Surface Waters

The corrosion of aluminum by natural surface waters has been described by Sawyer
and Mears (29), Godard (30) and Sverepa (31). There is no general thinning of the metal.
If corrosiondoes occur iITakes the form oTsmall diameter pits. The rate of perfora-
tion decreases with time, according to a cube root curve (30). Seligman and Williams
(32), Porter and Hadden (33), Davies (34), and others havemade laboratory studies on
the influence of wat er composition on the pitting of aluminum in water.
However, the pitting action of natural water on aluminum is of limited importance
in culvert considerations in view of the small influence of the pits on load bearing
strength.

Corrosion of Aluminum by Soils

There is rather less information on the corrosion of aluminum by soils, but there is
sufficient to draw some general conclusions that can be applied to culverts.
Logan (35) and later Romanoff (36) reported corrosion data for 2- x 6-in. sheet
coupons ofthree aluminum alloys in five soils after 10 years. The maximum pit depths
are given in Table 16, along with the loss of weight of the 2- x 6-in. specimens, ex-
pressed as a percentage loss of the original weight (based on 0. 062 inch sheet) .

TABLE 16
NATIONAL BUREAU OF STANDARDS 10- YEAR TEST-ALUMINUM
B DRIED IN FIVE SOILS

Alloy 1100 Alloy 3003

Soil
Max. Pit Max. Pit
No. Wt. Loss Wt. Loss
Depth Depth
(mils)
(% of original)
(mils)
(% of original)

13 21 1.2 45+ 5.4

29 62+ 100 62+ 13.8a
42 62+ 5.0 14 2.8
43 6 2.6 13 3.2
45 46+ 6.8 20 4.6

ase cond sample destroyed .

TABLE 17
BRITISH IRON AND STEEL RESEARCH ASSOCIATION-ALUMINUM
BURIED IN FIVE SOILSa

Max. Pit Depth (mils) Wt. Loss (% of original)

Location
Sheet Pipe Sheet Pipe

Benfleet 66 41 1.0 1. 6
Pits ea 39 33 0.2 0.2
Rothamstead 0 0 0.1 0.2
Gotham 86 30 0.1 0.2
Corbyb 125+ 6.4

a1100 Alloy: 10- X 15- 0.125-in. sheet; 1-in. diam., 15 in. long, 0 .062-in .
wall pipe.
bcinder embankment.
36

TABLE 18
BRITISH NON-FERROUS METALS RESEARCH ASSOCIATION
TEN-YEAR TEST-ALUMINUM BURIED IN SIX SOILsa

Max. Pit Wt. Loss

Location Alloy Depth
(% of original)
(mils)

Benfleet 1100 43 4.3

2014 Alclad 20 6,0
Pits ea 1100 32 1. 9
3003 0 1.2
5154 0 1.2
2014 Alclad 24 1.9
Corby 1100 122 64 . 9
2014 Alclad 187+ 90.8
Edinburgh 1100 87 3.4
3003 24 12.5
5154 49 8.2
2014 Alclad 67 3. 5
Woburn 1100 0 2.4
Wye 1100 47 0.9

a4- X 8- X 0.187-in. sheet specimens .

While at first sight the destruction of

TABLE 19 3 of 4 aluminum samples in soil 29 might
ALCOA 7- TO 8- YR TEST IN ONE LOCATION be regarded as serious, it should be real-
(New Kensington, Pa.) ized that this was the most corrosive of
47 soils tested (to steel) and is thus hardly
Maximum Pit Depth (mils)
Alloy an average American soil.
2 Yr 4 Yr 7 to 8 Yr The sheet specimens were too thin to
determine maximum pit depths for both
3003 30 58 50
6061-T6 35 73 78
alloys in 3 of the other 4 soils, but con-
6063-T5 64 98 80 version of the weight losses due to pitting
6063-T5 Alclad 5 6 5a to percent loss of the original specimen
weight indicates that, except for soil 29,
aAlthough almost all of the attack was conf'ined to
the clad layer, 3 or 4 small diameter pits did less than 7 percent of the metal was cor-
occur to a maximum depth of 73 mils where areas of roded in 10 years. If this is applied to
cladding were eaten away.
culverts, which are exposed to soil on
only one side, the figure is then 3. 5 per-
cent in 10 years.

TABLE 20
SOILS INCLUDED IN ALCAN TESTS

Resistivity
No . Location Soil Type
(ohm-cm)

1 Arvida, Que. Stiff clay

2 Kingston, Ont. Clay-silt 2,000
3 Toronto, Ont. Sandy
4 Shawinigan, Que. Sandy
5 Wetaskiwin, Alta. Solodized solonetz 1, 700
6 Wetaskiwin, Alta. Angus Ridge loam 1, 800
7 Wetaskiwin, Alta. Solonetz drag 1, 750
8 Guelph, Ont. Rocky clay 165,000
9 Hespeler, Ont. Sandy loam 210, 000
10 Brampton, Ont. Clay loam 185,000
11 Fredericton, N. B. Sandy and clay loam 165,000
12 Aulac Station, N.B . Salt marsh 600
 TABLE 21
FIRST ALCAN 5-YR TEST-ALUMINUM BURIED IN FOUR SOILS
(1948-1953)

Alloy

1100 3003 3003 Alclad 5052 6063-T5 6061-T6

Soil
No.
Max, Pit Max. Pit Max. Pit Max. Pit Max. Pit Max. Pit
Wt. Loss Wt. Loss Wt. Loss Wt. Loss Wt. Loss Wt. Loss
Depth Depth Depth Depth Depth Depth
(%of orig.) (%of orig.) (%of orig.) (%of orig.) (%of orig.) (%of orig.)
(mils) (mils) (mils) (mils) (mils) (mils)

1 29 0.2 31 0.1 3 0,4 3 0.1 42 0.3 32 0,5

2 27 0.4 22 0.2 4 0.2 4 0.2 31 0.4 20 1.0
3 33 0.1 24 0.1 1 0.1 1 0.1 32 0.1 64+ 0.6
4 2 0.5 2 0.4 1 0.5 1 0.4 2 0.4 36 0,7

TABLE 22
SECOND ALCAN 5-YR TEST-ALUMINUM BURIED IN EIGHT SOILS
(1958-1963)
-
Alloy

Soil 3003 3003 Alclad 5083 Type 6061-T6 6061 Alclad

No.
Max. Pit Max, Pit Max. Pit Max. Pit Max. Pit
Wt. Loss Wt. Loss Wt. Loss Wt. Loss Wt. Loss
Depth Depth Depth Depth Depth
(%oforig.) (%oforig,) (%of orig.) (% of orig.) (%of orig.)
(mils) (mils) (mils) (mils) (mils)

5 0 0.2 0 0.1 5 0.1 22 0.1 0 0.3

6 35 0.7 2 0 .6 0 1.0 25 0.7 0 1.0
7 7 0.1 5 0.2 5 0.1 10 0.1 0 0.2
8 15 0.2 0 0.1 16 0.1 64+ 0.1 12 0.4
9 15 0,1 0 0.1 6 0.1 11 0.1 0 0.1
10 0 0.1 0 0.1 0 0.1 18 0.1 19 0.2
11 24 0.2 0 0.2 43 0.2 64+ 0.3 50+ 0,3
12 64+ 5.1 64+ 20.3 64+ 6.4 64+ 12 . 2 50+ 7.7 c.o
...;i
38
TABLE 23 Gilbert and Porter (37) reported weight
BURIAL PLOT VALUES losses and pit depths for llOO aluminum
alloy tubing and sheet buried five years in
Resistivity (ohm-cm) five soils in Great Britain (Table 17).
Perforation occurred only at Corby
Site No. (soil 5) where the site was located in a
Well Poorly
Drained Drained cinder embankment (between tracks on a
railroad siding) . Here too the weight loss
5 1,700 600 was appreciably higher, though still mod-
8 165, 000 14,000 erate, especially when halved for one side
11 165,000 20,000 corrosion. Cinders are known to be cor-
rosive to metal due to sulfuric acid pro-
duced from residual sulfur.
Only last month Campbell (41) reported
ten-year results from anothers eries of
tests at three of these sites, and three others for AAllOO alloy, plus results on a few
other alloys at two sites (Table 18).
Once again the Corby cinders proved very aggressive to aluminum, while corrosion
elsewhere was small (maximum 6.25 percent-one side, at Edinburgh).
Sprowls and Carlisle (38) have reported 7- to 8-yr burial tests on several aluminum
alloys at one location (NewKensington, Pa.) . The specimens were 5-ft lengths of 11/ 4 -
in. diameter pipe with a wall thickness of 0.280 in. The maximum pit depths are given
in Table 19. The results suggest a decrease in the rate of penetration with time, and
appreciable protection by the cladding alloy. Although some penetration of the cladding
did occur, it was limited to a few areas.
The present writer has obtained pit depth and weight loss results on many aluminum
alloys buried for 5 years in the 12 Canadian soils given in Table 20 (39).
Table 21 gives maximum pit depths and percentage weight losses for the first test
.series. Perforation of the 0.064-in. sheet occurred only in 6061-T6 alloy in one loca-
tion. The prevention of penetration by cladding was also demonstrated. Taken together,
the pit depth and weight losses indicate that while some pitting occurred the loss of
metal was negligible.
In a second series (1958-63) 0. 064-in. sheet specimens were buried at 8 sites across
Canada, selected with the help of the Canadian Department of Agriculture as represent-
ing the main soil types. The maximum pit depths for a number of alloys and the per-
centage weight losses are given in Table 22.
With the exception of site 12, which was land reclaimed from the sea, weight losses
were negligible, as with the four central Canadian soils tested previously. At sites 5,
8 and 11 burials were made in both well and poorly drained soils of the same composi-
tion and structure at nearby locations. To indicate the lack of correlation between
corrosivity and soil resistivity, the values for the burial plots are given in Table 23.
In all cases corrosion was appreciably less (both maximum pit depths and weight losses)
in the poorly drained soil.

Summary
In 26 soils (6 American, 12 Canadian and 8 British), corrosion of aluminum alloys
that might be used for culverts was negligible in 23 soils, and appreciable in only 3.
One of these was salt marshland reclaimed from the sea, another was the most cor-
rosive (to steel) of 47 tested by the Bureau of Standards, and the other was a cinder
embankment. It is significant that there are well over 200 miles of unprotected buried
aluminum pipe in western Canada in a wide variety of soils. There are also over 150
buried thin wall (0. 060 in.) irrigation systems across Canada protected only by a single
coat of bituminous paint. There is also an appreciable mileage of unprotected buried
aluminum pipelines in the U.S.A. (40).
On this evidence, together with the examination of 500 actual aluminum culverts (of
20, 000 installed) presented by Lowe and Koepf (28), the writer contends that the large
majority of aluminum culverts will have a service life far in excess of the 25 years
tentatively suggested without conviction by Nordlin and Stratfull.
 39

References
28. Lowe, T. A., and Koepf, A. H. Corrosion Performance of Aluminum Culvert.
Highway Research Record 56, pp. 98-115, 1964.
29. sawyer, D. W., and Brown, 'R.H. Resistance of Aluminum Alloys to Fresh
Waters. Corrosion, Vol. 3, pp. 443-456, 1947.
30. Godard, H. P. The Corrosion Behaviour of Aluminum in Natural Waters. Can.
Jour. Chem. Eng., Vol. 38, pp. 167-173, 1960.
31. Sverepa, 0. Corrosion of Aluminum and Its Alloys in Waters of Various Com-
position. Werk. u. Korr., Vol. 9, pp. 533-535, 1958.
32. Seligman, R., and Williams, P. The Action on Aluminum of Hard Industrial
Waters. Jour. Inst. Met., Vol. 23, pp. 159-192, 1920.
3 3. Porter, F. C. , and Hadden, S. E. Corrosion of Aluminum in Supply Waters.
Jour. Appl. Chem., Vol. 3, pp. 385-409, 1953.
34. Davies, D. E. Pitting of Aluminum in Synthetic Waters. Jour. Appl. Chem.,
Vol. 9, pp. 651-660, 1959.
35. Logan, K. H. Underground Corrosion. U.S. Nat. Bur. Stds. Circ. C-450,
p. 127, 1945.
36. Romanoff, M. Underground Corrosion. U. S. Nat. Bur. of Stds. Circ. 579,
April 1957.
37. Gilbert, P. T., and Porter, F. C. Tests on the Corrosion of Buried Aluminum,
Copper, and Lead. British Iron and Steel Res. Assoc., Spec. Rept. No. 45,
pp. 55-74, July 1952.
38. Sprowls, D. 0., and Carlisle, M. Resistance of Aluminum Alloys to Underground
Corrosion. Corrosion, Vol. 17, pp. 125-132, 1961.
39. Godard, H. P. Unpublished Results, Series 1-1948-1953, 4 locations; Series
2-1958-1963, 8 locations.
40. Dalrymple, R. S. Aluminum Pipeline Case History Data. Materials Protection,
pp. 101-104, Oct. 1963.
41. Campbell, H. S. Corrosion and Protection of Aluminum Alloys Underground.
Jour. Inst. Met. (U.K.). Vol. 93, pp. 97-105, 1963.

E. F. NORDLIN and R. F. STRATFULL, Comments.-Hugh P. Godard's discussion

contains numerous comments and supporting data. We believe it is necessary to make
specific reference to the major points involved. For this reason we have used an un-
usual format for our closure wherein we comment on each major point.

Statement 1 (Godard)
"In the case of metal culverts, it is suggested that the sole criterion of corro-
sion is the continued ability of the culvert to support the overburden and normal live
loads for which it was designed."

Comments
It is agreed that one criterion of the results of corrosion is the continued adequacy
of culvert as a structure. However, there are two other criteria which should be in-
cluded if a culvert is perforated by corrosion. These criteria are (a) the penetration
of water through corrosion perforations can increase the moisture content of the over-
burden (embankment) and thus could reduce its structural strength in supporting loads,
and (b) when significant areas of the culvert are perforated by corrosion, the exposed
backfill material may be removed by scour. Whether or not these latter criteria are
applicable or could be critical will depend on the particular culvert site. As an example,
we are now designing a freeway in which a critical problem is to prevent the contact of
drainage waters with the soils on which this highway will be built. After inundation of
40

the foundation soil, a test section at this location settled approximately 14 ft below its
original elevation. In this general area, a county road was temporarily closed because
runoff water contacted the embankment causing foundation settlement which disrupted
the roadbed. A temporary closure of a freeway must be prevented. A corrosion-caused
hole in a culvert on this freeway would be considered to be serious even though the pipe
were still structurally adequate.

Statement 2 (Godard)
"However, the pitting action of natural water on aluminum is of limited importance
in culvert considerations in view of the small influence of the pits on load bearing
strength."

Comments
In our report, Figures 4, 6, and 13 show the complete loss of sections of the alumi-
num culvert invert. At the present time, the aluminum culvert shown in Figure 4, II-
Sha-3-B, has complete losses of metal in the bottom of the pipe which range up to 1 ft
in length. The waters which flow through these three culverts are natural waters.
They do not emanate from a mine or a commercial source. These particular test cul-
verts were placed in the same drainage channels in which metal culverts were previous-
ly used when corrosion testing was not practiced in the normal course of highway con-
struction.

Statement 3 (Godard)
"They decided that to obtain this life, certain criteria for pH, water and soil resis-
tivity must be adhered to." . . . "I suggest also that the criteria selected by Nordlin
and Stratfull are not applicable or necessary for predicting the service life of aluminum
culverts." . . . "In studies of aluminum corrosion, pH has not been found to be a
significant variable in either waters or soils."

Comments
No. 1. Lowe and Koepf, in their report on aluminum culverts (2), state: "The pH
range of 4. 0 to 9. 0 removes the prospect of chemical attack on the- oxide film."
No. 2. Godard states (48), "The oxide film on aluminum generally is stable in the
pH range 5-8." -
No. 3. A conclusion of Technical Committee T-4E of the National Association of
Corrosion Engineers is: "Conclusion 4. Aluminum is not resistant to waters contain-
ing copper salts or with a pH less than about 6- 6. 5." (42)
No. 4. A handbook published by an aluminum company states: "Aluminum is readily
attacked by strong alkaline or strong acid solutions with the exception of nitric acid and
some concentrated organic acids." (43)
The definition of a strong acid haSbeen implied to be a number greater than 4. 5 by
Lowe and Koepf ~) in Table 4 of their paper as indicated by the following:

Very strongly acid to slightly acid (pH 4 . 0-6. 5)

Very strong acid to neutral (pH 4 . 5-7. 3)
Extremely acid to neutral (pH 3 . 0-7 . 3)
Extremely acid to neutral (pH 4.0-7.0)

No. 5. If the corrosion testing criteria are not applicable or employed, then with
the resultant lack of specific information an aluminum company advises: "Aluminum
pipe offers excellent resistance to corrosion by many soils. However, some soils may
cause corrosion; and in the absence of specific information, it is advisable to protect
aluminum pipe by suitable coatings and wrappings similar to those used for steel pipe."
(44)
- No. 6. "This laboratory's experience is that fairly hard, mildly alkaline water such
as in the Great Lakes System tends to pit aluminum, while soft, mildly alkaline or
mildly acid waters do not." (47)
 41

Statement 4 (Godard)
"Accordingly, it can be safely predicted that the pitting of an aluminum culvert will
have no appreciable influence on the load-bearing capacity of the culvert unless the pits
become so numerous and so large that an appreciable cross-section of the metal is
consumed. "

Comment
No criteria are given by which to predict safely the pitting density of aluminum when
it is used as a culvert.

Statement 5 (Godard)
"Soil resistivity also affects the incidence and rate of pitting. In addition, it gives
information on soil battery effects which occur in the case of pipelines which traverse
several soil types. By contrast, metal culverts are normally relatively short and are
buried in one type of soil. This further reduces the value of soil resistivity readings
in culvert considerations."

Comments
Lowe and Koepf (2) state in their report on aluminum culverts: Tentative Conclusion
3. "The corrosion attack observed on the soilside surface of some aluminum culvert
is believed to be the result of nonuniform soil compaction rather than of borderline pH
or resistivity conditions. Such lack of uniformity causes concentration cells whose
activity is influenced by soil resistivity. Good compaction at the time of installation
can reduce attack from such cells." . . . "That resistivity influences the processes
of corrosion of buried metals is seldom disputed. There are many cases, however,
indicating that other factors play an equal or perhaps more significant role in corrosion
of buried culverts . "

Statement 6 (Godard)
The following are H. P. Godard's comments regarding the N.B.S. underground
tests of aluminum. (45)
"The sheet specimens were too thin to determine maximum pit depths for both alloys
in 3 of the other 4 soils, but conversion of the weight losses due to pitting to percent
loss of the original specimen weight indicates that, except for soil 29, less than 7 per-
cent of the metal was corroded in 10 years. If this is applied to culverts, which are
exposed to soil on only one side, the figure is then 3. 5 percent in 10 years."

Comments
No. 1. Soil can be deposited on the inside of a culvert by the flow as shown by
Figure 8 of our paper.
Comment on these data by the National Bureau of Standards (45): "The aluminum
alloys were susceptible to intergranular corrosion. In the advanced stages, this type
of attack caused ridges and blisters to occur on the surface, beneath which was a white
powder on some of the specimens. The unalloyed specimens were the best of the group.
Table 56 shows the loss of weight and maximum penetration of the thin aluminum speci-
mens, exposed approximately 10 years, and similar data for the same soils on zinc and
iron for comparison. None of the thin materials was satisfactory for use unprotected
in the corrosive soils to which they were exposed. Great strides have been made during
recent years in the development of aluminum alloys which might be more corrosion
resistant than the specimens buried at the Bureau's test sites."
[Authors' Note: The thickness of the aluminum test specimens at the N.B.S. site was
0. 062 in. which is thicker than the equivalent 16-gage aluminum culvert sheet (O. 060
in.).]
42

No. 2. C. J. Walton, in his discussion ( 50), has stated, "Loss in weight measures
the total amount of corrosion; but such data often do not measure reliably the effect of
the corrosion on other properties, such as actual depth of attack or actual loss in ten-
sile properties. In case of the aluminum alloys, depths of attack calculated from weight
losses were always much less than that measured by microscopic examination of cross-
sections; and the losses in strength calculated from weight loss data were much less
than the actual losses in strength as determined by tension tests. Thus, weight loss
data provide a basis for comparing the total amount of corrosion of different kinds of
metals, but are not adequate for evaluating the relative effects of the corrosion on other
properties."
No. 3. Included in Table 24 are some of the results of the ASTM atmospheric tests
(50) where the changes in the physical properties of several aluminum alloys are com-
pared to the results of corrosion. It is of interest for the reader to compare the amount
of corrosion (Table 24) and those amounts of corrosion of aluminum when placed under-
ground as submitted by Godard in his discussion . In culvert applications, a significant
loss of elongation within one year conceivably could contribute to structural failure of
a culvert.

Statement 7 (Godard)
"There are also over 150 buried thin wall (0.060 in.) irrigation systems across
Canada protected only by a single coat of bituminous paint."

Comments
The majority of culverts which are used in highway construction are 16-gage or in
the case of aluminum would be 0.060-in. thick. With the exception of invert paving,
the protective coating recommendations in the paper only include a single coat of bitu-
minous or other suitable material.

Statement 8 (Godard)
"On this evidence, together with the examination of 500 actual aluminum culverts
(of 20, 000 installed) presented by Lowe and Koepf (28) , the writer contends that the
large majority of aluminum culverts will have a service life far in excess of the 25
years tentatively suggested without conviction by Nordlin and Stratfull."

Comments
No. 1. None of the tabulated data submitted as evidence by Godard has a reported
testing time of greater than 10 years.
No. 2. None of these data submitted by Godard for up to 10 years of reported test-
ing were correlated to demonstrate that a large majority of the specimens will have a
service life far in excess of 25 years.
No. 3. From the corrosion data submitted by Godard in Tables 16 through 19 and
21 through 22, it is obvious that the majority of specimens were not destructively cor-
roded within 10 years . However , it is also of prime importance to know the magnitude
of the minority which were seriously corroded. In Figures 37 through 40 we have
plotted on probability paper the data submitted by Godard. The plotting positions of the
data were calculated in a recommended manner. (46, 48)
In all cases, the tabulated weight loss data were corrected to a culvert wall thick-
ness of 0. 060 in. by direct proportion. This was done because in some cases the re-
ported percentage of weight loss in Godard's tables was for sheets as thick as 0. 280 in.
Therefore, these weight losses were corrected to a thickness of 0. 060 in. be cause if a
reported weight loss of an 0.280-in. thick material was approximately 21 percent, then
 43
TABLE 24
RESULTS OF ATMOSPHERIC CORROSION TESTS OF SEVERAL ALUMINUM ALLOYS AND RELATED
CHANGES IN STRUCTURAL PROPERTIES (50)

Actual Loss Max. Pit Depth Loss in Elongation (%)e

Calculated Loss
in Tens ile in Tensile .Mlis, Micro-
Strength (,;)a' b Strength (%)h, c s cop!c Exam.ct Pre-Machined Panels
Alloy (20 yr) (20 yr) (20 yr) Tension Spec. (20 yr)
(1 yr)
New La New La New La New La
York Jolla York Jolla York Jolla New La York Jolla
York Jolla

Alclad 2017-T3 0 0 3.5 2.6 1.4 2.9 1. 5 4. l 4

3003-H14 8.3 7.0 3.4 2.6 6.4 10.2 31 42 +2 34
1100-H14 6.8 8.2 4.4 2.8 8.4 14.0 19 . 7 69 , 9 14 54
6051-T4 11.6 19.6 4.1 3.5 6. 7 12.1 38.9 76. 7 30 64
2017-T3 6.9 19 , 9 5. 9 10.4 7.1 20.3 46 , 3 91 , 7 15 58
Avg. 6.7 10. 9 4.3 4.4
3
Actual loss in tensile strength was measured on tension specimens machined from panels.
bFrom Table IV ( 50) .
ccalculated fromWeight loss data, assuming corrosion to be perfectly uniform.
dFrora Table D ( 50) .
eFrom Table C (~) ,

80
__
,
. ·tttt 1-l+ltl+f:t+
rttnm111T1

' -- OF 5 YEAR UNDERGROUND

DISTRIBUTION
- CORROSION TEST RESULTS OF ALUMINUM --1 ~
00.
- ,,
,_
-- -
70
,_ -~ - -
-- - ·- ,_ -
~
. Thickness of Test :
,_
:= I=
_ .... .
.. Sheet (0.064")

-
: 1
'.3 60 ,_ -- .. --
ll l l

- . -
::!:
J:
r

h: !IO
-- I
:
-- ~-

w
Cl
- ,_ ,_
-~

.
~-

.
I::
a.. •10
,_
- :
--
--
·- ·- ,,_
,_
·-_,_ -
,_
l -
~
NOTES
-
~
~
,_ =t::: - I - Alloys were 3003 8 6061 ~
:~
20
,_ - f Ef placed in the same soils. =
..
··-- I
.. .
=
:: ~
10
I

r. ·
Data from tables 7 and 9,

.
H. P. Godard.

~~
,_
·- ·- ,_
- r
· fl-
~
0
Plain (n=20)
Alclad (n=20)
,.-··· ==
0 .01 0050.1
-
02
-
. ·- - r
- -~-u•H·H-

99.e 99.9
-
99.9
05 I 2
' 20 '50
FREQUENCY - PERCENT
<lj) 50 60 70 80 90 95 98 99

Figure 37 .
44

_.__
->--

80 90 95 98

FREQUENCY-PERCENT
Figure 38.
 45
ao ,~

so DISTRIBUTION OF UP TO 10 YEARS UNDERGROUND

CORROSION TEST RESULTS OF ALUMINUM .. _.!_ .
- !!!

·- --
70

·- ~

'.
. ~ -
en
::::! 60 -- -- : ,_ -
::!:
I
J:
I-
- -
~ 60 ,__
0 - ---....
I-
a: - - .....
40
-
::!:
:::>
::!:
x __
,_,_
,
:___ : L_

!Iii-I
~ 30
,_
- Data from tables 1,3,
w 4,5,7,9,NOTES
H.P. Godard
.
- • Plain (n=82)
,_ . ~ o Alclad (n=27) _ ,_
10
·- -
·-
0
0 .01 0 .05 01 0.2 0.5 I 10 :io &0 405oso10 eo 90 95 ae sg 998 99.9 99.99
FREQUENCY - PERCENT

Figure 39 .

this would be the approximate equivalent of 100 percent weight loss for an 0. 060-in.
thick specimen. Other than the preceding thickness correction and the plotting, no
other mathematical conversions of the reported data were performed.
In the foregoing analysis, the following assumptions were made:
1. The data submitted by Godard is a representative and random sample of the ex-
pected performance of aluminum in soils.
2. The random data demonstrate that the large majority of aluminum culverts will
have a service life far in excess of 25 years.
3. The data for all listed alloys and soils verify that corrosion criteria are not
applicable or necessary for predicting the service life of aluminum culverts.
On the basis of the preceding assumptions, the random placement of aluminum cul-
verts without corrosion testing or culvert coatings could result in the following:
1. The random use of aluminum culverts of 0. 060 in. indicate the possibility that.
up to approximately 20 percent of the culverts will be perforated by pitting within ap-
proximately 5 years of service (Fig. 37).
2. The random use of aluminum culvert of 0. 060-in. thickness indicates the possi-
bility that approximately 10 percent of the culverts can have a weight loss of greater
than 5 percent within 5 years of service (Fig. 38).
3. The random use of aluminum culverts with a wall thickness of 0.060 in. indicate
the possibility that up to approximately 25 percent of the culverts will be perforated by
pitting within approximately 10 years of service (Fig. 39).
4. A small percentage of culverts could have a 100 pe:rcent weight loss of metal
within 10 years. If the weight loss of 7 percent (see Godard's Statement 7 with Com-
46

I I I I
2 6 IO 20 30 40 ~ 60 70 DO 90 95 98

FREQUENCY-PERCENT

Figure 4o .
 47

ment No. 1) is used, then it seems to be possible that up to approximately 20 percent

of the culverts with a wall thickness of 0. 060 in. will probably be in an unsatisfactory
condition for use within approximately 10 years of service (Fig. 40).
The evidence of the corrosion test results of aluminum indicates that corrosion con-
trol measures are necessary if one does not wish to accept a significant percentage of
culverts with perforations and significant weight loss in less than 10 years.

References
42. Service Life of Pipe Exposed to Domestic Waters. N. A. C. E. Tech. Comm. Rept.,
Publ. 60-11. Corrosion, Vol. 16, p. 453t, Sept. 1960.
43. Sheet and Plate Product Information. Kaiser Aluminum and Chemical Sales, Inc. ,
Copyright, 1953.
44. Process Industries Applications of Alcoa Aluminum. Copyright, 1955.
45. Romanoff, M. Underground Corrosion. Circular 579, Nat. Bur. of Stds., p. 92,
1957.
46. U. S. Bur. of Pub. Roads, Office of Research and Development. Contract CPR
11-8718, M-W Tech. Rept. No. 201. Raleigh, N. C., Miller-Warden Assoc.,
Aug. 1963.
47. Wright, T. E., and Godard, Hugh P. Laboratory Studies on the Pitting of Alumi-
num in Aggressive Waters. Corrosion, Vol. 10, No. 6, p. 195, June 1954.
48. Godard, Hugh P. The Corrosion Behavior of Aluminum. Corrosion, Vol. 11,
No. 12, p. 542t, Dec. 1955.
49. Aziz, P. M. Application of the Statistical Theory of Extreme Values to the
Analysis of Maximum Pit Depth Data for Aluminum. Corrosion, Vol. 12, No.
10, p. 495t, Oct. 1956.
50. Walton, C. J., and King, William. Resistance of Aluminum Base Alloys to 20-
Year Atmospheric Exposure. ASTM Sy mp. on Atomospheric Corrosion of Non-
Ferrous Metals, ASTM Spec. Tech. Publ. No. 175, p. 21-46, 1955.
51. Putilova, I. N., Balezin, S. A., and Barannik, V. P. Metallic Corrosion
Inhibitors. New York, Pergamon Press, p. 122, 1960.
52. Stratfull, R. F. A New Test for Estimating Soil Corrosivity Based Upon an In-
vestigation of Metal Highway Culverts. Corrosion, Vol. 17, No. 10, p. 115-
118, Oct. 1961.
53. Walton, C. J., McGeary, F. L., and Englehart, E. T. Compatibility of Alumi-
num with Alkaline Building Products. Corrosion, Vol. 13, No. 12, p. 807t,
Dec. 1957.

THOMAS A. LOWE, Department of Metallurgical Research, Kaiser Aluminum &

Chemical Corporation. -This discussion has been prepared with the intent of construc-
tively commenting on a paper entitled "A Preliminary Study of Aluminum as a Culvert
Material, " by Eric F. Nordlin and R. F. Stratfull. These comments are not intended
to stand by themselves, but to complement discussions offered on this paper by Dr.
Hugh Godard and by A. H. Koepf. As a corrosion research engineer for a major alumi-
num-producing company, I have actively and directly participated in the aluminum cul-
vert program. My work started at the inception of the product's development six years
ago and has since involved an extensive, thorough, and continuing evaluation of corro-
sion performance in a large number of culvert installations which encompass many
types of soils and service widely distributed throughout the United States.
The results of field tests of aluminum culvert are valuable, since they provide a
broader background of experience to compare with recommendations of producers and
to compare with results of culvert tests being conducted by other agencies. We are
seriously concerned, however, about the conclusions of Nordlin and Stratfull respecting
aluminum culvert. If these conclusions are accepted by the highway authorities, the
48

effect will be to discriminate unfairly against aluminum culvert and to discourage its
use. On the basis of our knowledge of aluminum and our experience with the metal,
both in general and in the form of culvert, we do not believe the authors' conclusions
are justified. We have carefully reviewed the paper, and we disagree with: (1) the in-
terpretation of information and data taken from the literature; (2) the experimental ap-
proach; and (3) the analysis of the reported results.

Summary
The authors' paper should be read in its entirety, and, if possible, the major refer-
ences, some of which we discuss here, should also be reviewed. Neither the work of
Nordlin and Stratfull, nor the papers which they reference, justify the conclusions which
the authors have reached.
The narrow limits imposed on the use of aluminum culvert by the State of California
are not supported by the data which they have accumulated. Even those data are in
question since they involved conditions that are not representative of those normally
encountered in culvert installations.
Our inspections of hundreds of bare aluminum crossdrains and sidedrains throughout
the United States, in soil conditions varying from purposefully aggressive to the more
normal, show no evidence to support these restrictive limits, or the assumedly aggres-
sive conditions which the authors conceive. These many installations, and, in fact,
those installed by the State of California, have performed in a manner consistent with
what we have come to expect. On the basis of this broader experience, aluminum cul-
vert would be expected to:
1. Provide corrosion performance superior to that of galvanized steel in soils within
the pH range of 4.0 to 9.0 and having a minimum resistivity above 1, 500 ohm-cm.
(Field experience indicates this value can be lowered considerably, but further exposure
is needed to confirm it.)
2. Provide better corrosion performance than galvanized steel in installations ex-
posed to flow of brackish or sea water.
3. Suffer attack, as does galvanized steel, in runoff from pyrite areas whose pH at
any time drops below 4. 0 .
4. De more resistant than galvanized steel to the normal erosion-corrosion cycles
encountered by drainage structures in areas of erosive runoff. Our experience in such
installations has been reported more fully in the discussion of this paper by A. H. Koepf.

Literature Reference
For those unfamiliar with the corrosion characteristics of aluminum, it is natural
to assume that those characteristics will be similar to other metals commonly used in
construction. Such an assumption is not true. Reference to the literature, as attempted
by the authors, is an excellent means for familiarizing oneself with the subject of alumi-
num corrosion. It is important, however, to fully digest the intent of any reference,
along with the significance of all data presented.
It would be desirable to discuss each of the references given by Nordlin and Stratfull
which concerns aluminum, but space will not allow. Instead, certain references will
be selected for comment and are listed again at the end of this discussion. The serious
investigator is encouraged to read some of these references so that he might appreciate
the danger of misapplying or misinterpreting statements or data from those references.

Influence of pH
Deltombe and Pourbaix (54) are listed by the authors as setting forth a pH range of
5. 5 to 7. 8 over which aluminum is inert or "inhibited from accelerated corrosion."
This reference reports a chemical thermodynamic treatment through which a potential-
pH equilibrium diagram of the system aluminum-water was developed from standard
free energies of certain constituents. The general electrochemical behavior of alumi-
num was deduced from the diagram.
 49

Deltombe and Pourbaix have assumed for their model that hydragillite, Ai203 · 3 H20,
(usually called gibbsite in the U.S.) is the oxide on the metal surface. With this model,
they predict that the gibbsite-covered aluminum surface is passive, or is corrosion re-
sistant, over a pH range of 4.0 to 8.6. The point is that Deltombe and Pourbaix have
interpreted the behavior of aluminum in terms of the soluble A1+++ and Al02 species
and in terms of the solid Al 0 and Ah03 · 3 H20. They cannot, nor do they attempt, to
equate the surface oxide films that normally occur on aluminum with their reference
models. These films are too complex, and they vary in composition with the medium
in which they are in contact. Therefore, not only have the authors misinterpreted the
information contained in this reference, they have incorrectly quoted the pH range of
5.5to7.8.
The pH range of 5. 5 to 7. 8, which Nordlin and Stratfull mention, is actually the range
in which Deltombe and Pourbaix found Al(OH)J to have minimum solubility. This range
is of no real significance since, as Deltombe and Pourbaix mention, "The aluminum
hydroxide gel is not stable. It crystallized eventually to give the monohydrate of
boehmite, crystallizing in the rhombohedral system. It then gives the trihydrate or
bayerite, crystallizing in the monoclinic system, and finally another trihydrate, hydra-
gillite, crystallizing in the same system. This evolution of the hydroxide of aluminum
is known as "aging." The diverse hydrates formed in the course of aging are charac-
terized by greater and greater stabilities, and concomitant variations in all their pro-
perties, particularly in their solubilities in acids, bases and pure water."
The paper by Deltombe and Pourbaix and that by Nordlin and Stratfull reference the
work of Shatalov (55). Deltombe and Pourbaix reproduce the graphs of Shatalov in their
paper which indicate the influence of pH on the rate of corrosion aluminum. These
graphs show essentially zero corrosion rates over a pH range wider than the 4. 0 to 8. 6
suggested by Deltombe and Pourbaix, than the 4. 0 to 9. 0 recommended by Kaiser
Aluminum, and certainly wider than the 6. O to 8. 0 specified by Nordlin and Stratfull.
Confirmation for the influence of pH determined by Shatalov is found elsewhere in the
literature ( 5 6) .
Similar detail is in order with respect to Nordlin and Stratfull' s interpretation of the
pH ranges suggested in other references. Rather than attempt such a detailed discus-
sion, it can be stated that several of the references have been misinterpreted. All but
one of them represent the coverage of aluminum performance in applications very dif-
ferent from culvert, nevertheless they support a stability of the aluminum oxide film
over a pH range of 4.0 to 9.0.
To summarize our position on pH, we believe that the range suggested by Nordlin
and Stratfull has been arrived at arbitrarily. It has no basis of experience, either in
their work or in the literature. From a practical standpoint, we know that there are
few soils which fall outside the pH range of 4 to 9. Therefore, we must conclude that
other factors influence corrosion performance, if we are to explain the few cases of
corrosion which have been noted.

Chemical Compatibility
The reader might gain misleading conclusions from statements under the chemicals
section of the paper. A reference is made to attack by sodium carbonate solutions (57).
One is warned against accepting data without learning the conditions under which the-
data were obtained. In the McKee and Brown study (57), also referenced by Nordlin
and Stratfull, the one sentence discussing tests with sodium carbonate reads, "Although
alkalinity produced by the presence of sodium hydroxide or resulting from hydrolysis
of a sodium salt such as the carbonate causes appreciable corrosion, aluminum may or
may not be resistant to such solutions, depending upon the nature of other ions present."
The important point is that the tests in this study were of only 48 hours' duration. A
glance at Figure 41 (Fig. 1, 58) shows that, with time, there is a striking decrease in
the corrosion rate of aluminum in a sodium carbonate solution much more concentrated
than those used by McKee and Brown (57). A study of corrosion which does not properly
assess the influence of time on corrosion rate can give erroneous results, as indicated
by the data shown in Figure 41.
50
AVERAGE CORROS ION RATE OF ALUMINUM
IN SODIUM CARBONATE SO LUTIO NS

16 0 . .. .. . 606 1

15
><. .. .. .. 5083
(Taken from graph in "Corrosion Preve ntion
and Control '' magazine, November, 1963.)

6 months
1 day l week
0
10 100 1,000
Exposu re Time ( Da ys )

Figure 41. Rapid decrease in corrosion rate with increasing exposure, of aluminum alloys
6061 and 5083 to 10 percent (by wt) and to saturated sodium carbonate solutions at roo~
temperature.

As mentioned by the authors, the presence of heavy metals in waters can cause at-
tack of aluminum. Incidence of attack caused by heavy metals, however, is rare. The
case referred to in the paper involved the water supply of Altoona, Pa. An unusually
high content of heavy metals was found in this water , which, in combination with other
characteristics, caused the water to be particularly aggressive. Since such waters are
infrequently encountered, they should receive little consideration. The success of
aluminum irrigation tubing, and the absence of detrimental attack during the five years
since since aluminum culvert was introduced is evidence of aluminum's compatibility
with nearly all "natural" waters. (A distinction is made between "natural" and "pro-
duced" waters , the former being those in equilibrium with air, such as runoff, lake, or
river water, as opposed to the latter which are not , such as spring or well water. Pro-
duced waters are not normally encountered in culvert applications. )
The claddingonaluminum culvert is intended to mitigate any attack that mi ght be
caused by an unusual water. Such cladding has helped prolong the useful life of alumi-
num in such applications as cooking utensils ( 58), hot water heaters, heat exchanger
tubing, and irrigation pipe. Attack spreads laterally along the layer of more anodic
cladding, rather than into the core alloy. Should subsequent pitting of the exposed core
alloy penetrate the metal, it will have an insignificant effect on the strength of the cul-
vert.
While on the subject of cladding, it might be well to comment on the authors' termi-
nology "corrosion inhibiting cladding." The cladding does not, nor is it intended to,
inhibit attack. As mentioned above, its function is to control the manner of attack, if
any should occur .

Electrical Resistivity
Nordlin and Stratfull have made a serious misinterpretation of resistivity readings
provided in the report by Lowe and Koepf (59). Those readings were made in the field
using a Model 263 A Vibroground equipped with a wiring harness for obtaining average
resistivities at depths of 2. 5, 5. 0 and 10 ft. The procedure is given in the original
paper.
 51

The Vibroground is the most widely used instrument for making soil resistivity
surveys before designing cathodic protection systems for buried oil and gas pipelines.
This instrument gives an average reading of a hemisphere of soil whose radius is de-
termined by pin settings. This average will include surface soil as well as soil at the
designated depth.
ff resistivity values from 2. 5 ft to 10 ft are increasing with depth, the more shallow
soils have a lower resistivity than indicated by the average. For example, if a soil has
a resistivity of 1, 000 ohm-cm at 2. 5 ft and 5, 000 ohm-cm at 10 ft, the surface layers,
if isolated, would have a resistivity lower than 1,000 ohm-cm. The soil at the 2.5-ft
depth would have a resistivity somewhere between 1, 000 and 5, 000 ohm-cm, such as
2, 500 ohm-cm.
We believe that depths of 2. 5, 5. 0 and 10. 0 ft cover the great majority of culvert
installations in existence. We therefore get a reasonable indication of soil resistivity
at the culvert depth while getting a "feel" for the complete range of material which may
have been used as backfill.
One cannot arbitrarily average the values reported and have a meaningful value. Further-
more, most of the installations which we reported fall below the mean value listed by Nordlin
and Stratfull, and these installations, as a whole, are showing excellent performance.
For longer term data, it is interesting to look at results reported by the National
Bureau of Standards in their 10-yr test. Four of the five soils included in that program
would be classified as aggressive soils by corrosion engineers. Nevertheless, bare,
not clad, aluminum is withstanding the rigors of those four environments better than
zinc or steel. Based on our knowledge of the corrosion characteristics of aluminum,
it would continue to be superior for years to come. Clad aluminum would show even
better corrosion performance.
The 1, 500 ohm-cm value quoted from another reference (61) concerned a pipeline
which was cathodically protected at "hot spots." Long-line currents which gather on
pipelines make low resistivity soils a potential hazard. Culverts are not subject to such
currents; consequently, there is no need to consider resistivity from that aspect.
Furthermore, we have cladding to protect against particularly aggressive soils.

Bimetallic Corrosion
The concepts which Nordlin and Stratfull present on this subject are generally cor-
rect. However, they neglect the influence of surface films on the activity of galvanic
cells. Only aluminum alloys specifically designed to provide cathodic protection can
be used to protect steel. These alloys will corrode more freely and will not be greatly
affected by surface films. Most aluminum alloys will not provide such protection, as
evidenced by the literature (62) on aluminized steel. For aluminized steel, aluminum
provides some protection only in the presence of significant chloride concentrations,
such as in marine environments.

Concentration Cell and Crevice Corrosion

In the many culverts which we have examined, we have found no problem of prefer-
ential corrosion at laps, either circumferential or longitudinal. All common metals of
construction, including steel and galvanized steel, are subject to concentration cell and
crevice corrosion if conditions exist which promote such attack. As for the possibility
of active:passive cells supplementing crevice attack, the reader should understand that
any oxygen-passivated metal, such as chromium, stainless steel, or aluminum is sub-
ject to such attack. Again we repeat that, even in the culvert exposed to the muck at
Gramercy, Louisiana, we have not seen evidence of preferential attack of lapped sur-
faces in the many field installations examined.

Laboratory Tests
We do not feel that the compatibility of aluminum, or any other metal, can be real-
istically evaluated by exposing that metal to chemical solutions in the laboratory. One
cannot reproduce the soil electrolyte chemically.
52

There is the added problem of length of exposure for the tests reported by Nordlin
and Stratfull. Short-term tests, particularly with no provision for determining time/
rate data, are meaningless. The data in Figure 41 illustrate this point, as does the
reaction of concrete on aluminum. After a general etching of 0. 001 in. to 0. 002 in.
during the setting period, aluminum is unaffected when embedded in concrete.
As for the abrasion tests, field performance of a metal exposed to erosive flow is
not solely determined by its resistance to abrasion. The ability of the metal to with-
stand countless erosion-corrosion cycles is the true criterion. Such a criterion re-
quires a time factor not easily included in laboratory evaluations.
Actual installations, observed periodically over a few years, give a true picture of
the comparison of galvanized steel and aluminum in such erosive flows. We have had
a number of such installations under surveillance, one of which we reported in some
detail in a previous paper (59). Similar field tests by another state highway department
have provided identical results to those which we have observed. Pictures of the in-
place culvert, as well as photomicrographs of cross-sections taken from that installa-
tion at the end of 2. 0 and 4. 3 years were shown during our discussion of this paper at
the 44th Annual Meeting of the Highway Research Board.

Field Tests
It is unfortunate that the authors chose such extreme conditions for their field tests
rather than exposures more representative of California soils. Even in the eight sites
reported , no indication of the general soil- side performance is provided for the three
sulfuric acid sites or for the two abrasion sites. The highly acid runoff would be ex-
pected to affect the soil-side of the invert, where it leaks through joints or perforations
of the invert. The remainder of the soil- side surfaces would not be so affected. Surely,
the soils representative of those three sites will be used as backfill for culverts not
exposed to acid runoff, just as the soil in contact with culvert I-HUM-35-C will be in
contact with other culverts not exposed to erosive runoff. In effect, soil corrosion is
being evaluated at only three of the eight sites, a fact confirmed in the authors' paper,
Table 3. Thus, the use of aluminum culvert by the State of California is based on the
performance at only three sites.
A further weakness of the California tests is that five or six of the eight test culverts
were not installed under normal conditions. Most of these installations were made in
ditches adjoining the highway where the culvert was merely covered with a mound of
dirt. There was no opportunity for compaction of the "backfill" which a normal installa-
tion experiences. The importance of such compaction is pointed out in several refer-
ences (59, 63).
The authors give no data or description of the comparative performance of the cul-
vert materials from one inspection time to another. Progress of attack, if any, cannot
be determined.
It is hoped that the authors will provide more detail concerning inspection results in
their next report on the subject. In the meantime, the reader is asked to consider
carefully the procedure that he might follow before embarking upon a field test program
of any type of material. Test conditions should duplicate those to be experienced in
service.

References
54. Deltombe and Pourbaix. The Electrochemical Behavior of Aluminum. Corrosion,
Vol. 14, No. 11, pp. 496t-500t, 1958.
5 5. Shatalov, A. Y. Effet du pH sur le Comportement Electrochemique des Metaux
et Leur Resistance a la Corrosion. U.S.S.R., Doklady Akad. Nauk., Vol. 86,
pp. 775-777, 1952.
56. Aluminum in Chemical Engineering. Corrosion Prevention & Control, pp. 41-43,
May 1963.
57. McKee, A. B., and Brown, R. H. Resistance of Aluminum to Corrosion in Solu-
tions Containing Various Anions and Cations. Corrosion, Vol. 3, No. 12,
pp. 595-612, 1947.
 53

58. Wei, M. W. The Corrosion Rates of Aluminum. Corrosion Prevention & Con-
trol, pp. 34-35, Nov. 1963.
59. Lowe, T. A., and Koepf, A. H. Corrosion Performance of Aluminum Culvert.
Highway Research Record No. 56, pp. 98-115, 1964.
60. Romanoff, Melvin. Underground Corrosion. Nat. Bur. of Stds. Cir. 579, pp. 92,
19-20; 1957.
61. Whiting, J. F., and Wright, T. E. Cathodic Protection for an Uncoated Aluminum
Pipeline. Corrosion, Vol. 17, No. 8, p. 9, Aug. 1961.
62. Evans, Ulick R. The Corrosion and Oxidation of Metals. London, Edward
Arnold, pp. 640- 641, 1960.
63. Romanoff, Melvin. Corrosion of Steel Pilings in Soils. J our. of Res. of the Nat.
Bur. of Stds., Vol. 66C, No. 3, July-Sept. 1962.

E. F. NORD LIN and R. F. STRA TFULL, Comments. - For the most part, Lowe dis-
agrees with almost everything in our report. Generally, he has detailed the reasons
for his difference of opinion. Because of the numerous points of disagreement, we
are commenting on each major point. However, some of the major points brought out
by Lowe were also discussed by Koepf and are included in our comments on the latter
discussion.

Statement 1 (Lowe)
"On the basis of this broader experience, aluminum culvert would be expected to:
1. Provide corrosion performance superior to that of galvanized steel in soils within
the pH range of 4.0 to 9.0 and havingaminimum resistivity above 1, 500 ohm-cm."

Comments
No. 1. No data have been submitted by Lowe which demonstrate that the corrosion
performance of galvanized steel culverts has been studied in all of these soils and that
the superior corrosion performance of aluminum culverts has been comparatively de-
termined.
No. 2. No data have been submitted by Lowe which demonstrate that laboratory test-
ing or field data have been mathematically correlated to demonstrate that aluminum
culverts would provide corrosion performance superior to that of galvanized steel in
soils within a pH range of 4. 0 to 9. 0 or having a minimum resistivity above 1, 500 ohm-
cm.
No. 3. In his discussion, Lowe made reference to the published paper (2) he co-
authored with Koepf. A paper by Stratfull (52), which was also used as a reference in
the paper by Lowe and Koepf (2), defines arid describes minimum resistivity to be the
result of a laboratory type of test. To our knowledge, no method has been established
for correlating an in-place field resistivity obtained by the method employed by Lowe
and Koepf (2) with the minimum soil resistivity. In their paper or in his discussion,
Lowe has not indicated that they have actually determined the minimum resistivity of
a soil.
In their paper (2), Lowe and Koepf, in apparent support of their minimum resistivity
recommendations,- refer to one paper by Whiting and Wright (14) in stating, "A minimum
soil resistivity of 1, 500 ohm-cm has been suggested as a threshold value below which
corrosion of aluminum may occur (4)." Whiting and Wright (14) make no reference to
the term "minimum soil resistivity-:-" There is no other mention of a minimum soil
resistivity of 1, 500 ohm-cm in the text to the paper (2) by Lowe and Koepf. Therefore
we are not aware of the basis for the minimum soil resistivity recommendation of 1, 500
ohm-cm.
54

It should be noted that the application of the type of cathodic protection such as used
by Whiting and Wright (14) is not necessarily limited to soils of a particular resistivity.
This type of protection is also applied to pipelines where a significant soil resistivity
differential exists and the lower limit of resistivity can be much higher than or even
less than 1, 500 ohm-cm.

Statement 2 (Lowe)
Deltombe and Pourbaix (54) are listed by the authors as setting forth a pH range of
5. 5 to 7. 8 over which aluminum is inert or 'inhibited from accelerated corrosion' .
. . . "Therefore, not only have the authors misinterpreted the in.formation contained
in this reference, they have incorrectly quoted the pH range of 5. 5 to 7. 8." . . . "This
range is of no real significance since, as Deltombe and Pourbaix mention: 'The alumi-
num hydroxide gel is not stable. . . '." . . . "Similar detail is in order with respect
to Nordlin and Stratfull' s interpretation of the pH ranges suggested in other references."

Comments
No. 1. We do not believe we have misquoted or misinterpreted the information pre-
sented in the paper by Deltombe and Pourbaix (5). On Page 499t, they state: "Accord-
ing to laboratory tests the minimum solubility Of Al( OH) 3 lies between pH 5. 5 and 7. 8."
This paper also states, "When alkali is added to a solution of an aluminum salt, or
acid to a solution of an aluminate, one obtains a precipitate, hydroxide gel, correspond-
ing essentially to the composition Al(OH)J and amphoteric in nature." . . . "The last
stage of aging of the aluminum hydroxide gel in caustic soda corresponds, according to
Fricke and Mey ring, to the formation of hyd1·argillite Ah03 · 3 H20, of which Fricke and
Jucaitis have calculated the solubility product (Al02)(H+) '° 2. 5 x 10- 15 or 10- 14 • 60 . "
In relating the work of Shikkor, Messrs. Putilova, Balezin and Barannik (51) state:
"Shikkor also established that the solution rate of aluminum in alkalies is almost in-
dependent of the purity of the metal, and on the basis of his experiments concluded that
the solution rate is determined not by the formation of micro-galvanic cells but, most
probably, by the production of a film of amorphous aluminum hydroxide on the metal
surface and its subsequent slow dissolution."
No. 2. McKee and Brown (9) state: "In direct contrast to sodium hydroxide solu-
tions, low rates of attack were obtained with ammonium hydroxide solutions. This
wide difference in corrosion rates in two different alkaline solutions can be explained
by the great difference in the solubility of the corrosion product in the two solutions."
Figure 17 in the McKee and Brown paper (9) shows the effect of potassium nitrate
and ammonium nitrate on the solubility of aluminum hydroxide Al(OH)J in ammonium
hydroxide.
From the preceding it appears that other authors attach importance to aluminum
hydroxide.
No. 3. Lowe's statements are based upon the following sentence in our paper, and
we quote: "Other reports have indicated that aluminum is generally inert or inhibited
from accelerated corrosion when the pH range of the environment is: 4 to 9 (2), 6 to
8 (3, 4), 5.5 to 7.8 (5), 4 to 8 (6), and 4.5 to 9 (4)." -
-This sentence as written does not contain an incorrect quotation because we have not
directly quoted any references as specifically stating "generally inert or inhibited from
accelerated corrosion." The listed pH ranges will be found in the cited references.
The authors believe they are justified in using the nonspecific terms, "generally
inert or inhibited from accelerated corrosion," because reference was made to five
publications that varied not only in scope but also in the terminology which was used
in reporting their observations.
The use of the word "indicated" in a sentence without quotation marks does not neces-
sarily imply a direct quotation. In fact, the use of the word "indicated" may imply that
a further analysis of data is being reported.
For example, the following two statements are contained in two papers (2, 53) that
comment on the results of the same investigation (9). The published paper;- itself, by
McKee and Brown (~), does not show data or contain text that describe the corrosion
 55

rate test results or the effect on the oxide film in terms of ranges of pH values per se.
In this subject paper, the criterion of good corrosion resistance was stated to be below
5 mils/year.
(a) Statement by T. A. Lowe and A. H. Koepf (2): "Aluminum oxide is generally
inert to chemical attack within the range of pH 4 to-9 (io)."
(b) Statement by C. J. Walton, F. L. McGeary, and E. T. Englehart (53): "It has
4
been indicated by McKee and B.1;own that in exposures to neutral or nearlyneutral solu-
tions, pH 4. 5 to 8. 5, the film is fortified by the formation of additional hydrated alumina
to increase its resistance to the new environment."
It will be noted that statement (b) included the word "indicated," and neither state-
ment (a) or (b) had contained quotation marks.

Statement 3 (Lowe)
"Deltombe and Pourbaix reproduce the graphs of Shatalov in their paper which in-
dicate the influence of pH on the rate of corrosion of aluminum. These graphs show
essentially zero corrosion rates over a pH range wider than the 4. 0 to 8. 6 suggested
by Deltombe and Pourbaix, than the 4. 0 to 9. 0 recommended by Kaiser Aluminum, and
certainly wider than the 6. 0 to 8. 0 specified by Nordlin and Stratfull."

Comment
With reference to the reproduced graphs of Shatalov, Deltombe and Pourbaix (5)
conversely state: "In Figure 3b, these same results have been transferred to a graph
with linear co-ordinates, which emphasizes the slow rate of corrosion between pH 4
and pH 8, and the rapid increase outside these limits."

Statement 4 (Lowe)
"One is warned against accepting data without learning the conditions under which
the data were obtained. In the McKee and Brown study ( 57), also referenced by Nordlin
and Stratfull, the one sentence discussing tests with sodium carbonate reads, 'Although
alkalinity produced by the presence of sodium hydroxide or resulting from hydrolysis
of a sodium salt such as the carbonate causes appreciable corrosion, aluminum may or
may not be resistant to such solutions, depending upon the nature of other ions present.' "

Comment
Under Conclusions in the McKee and Brown study (9 or 57) to which Lowe refers,
the following sentence related to sodium carbonate may alsobe found: "Conclusion 9.
Aluminum is resistant to sodium carbonate solutions up to 0. 001 normal concentrations,
either in the presence or in the absence of sodium chloride, but in higher concentrations,
the behavior is similar to that in sodium hydroxide solutions. 11
In our paper, the sentence regarding the corrosion of aluminum in sodium carbonate
reads: "It has been reported that in sodium carbonate solutions of greater than 0. 001
normal concentrations (approximately 60 parts per million), aluminum is significantly
attacked (Q) . "

Statement 5 (Lowe)
"Produced waters are not normally encountered in culvert applications."

Comment
Lowe in his discussion defines a spring and well water as produced water. In Cali-
fornia it is not unusual for culverts to convey spring water or for underdrains to inter-
cept subterranean water .

Statement 6 (Lowe)
"While on the subject of cladding, it might be well to comment on the authors' termi-
56

nology 'corrosion inhibiting clading.' The cladding does not, nor is it intended to, in-
hibit attack."

Comments
No. 1. In H. P. Godard's discussion of our paper, he states: "In natural waters
and soils, most aluminum alloys, and certainly all of those which would be considered
for culvert construction, do not suffer uniform or general corrosion. That is to say
they do not waste away by general thinning. If corrosion attack does take place, it is
localized, and usually in the form of pitting, in a random pattern over the surface of
the metal."
Based upon the pitting criterion stated by Godard, Figures 37 and 39 indicate that
the percentage of clad aluminum samples which had zero mils of pitting was far greater
than those samples which were not clad. With zero mils of pitting, the cladding cannot
be acting galvanically to the base material. Therefore, as corrosion did not occur on
the cladding in more cases than it did on unclad aluminum, it appears that the cladding
may be correctly termed "corrosion inhibiting."
No. 2. Lowe states in his discussion: "The cladding on aluminum culvert is intend-
ed to mitigate any attack that might be caused by an unusual water." From this state-
ment it also seems reasonable to assume that the cladding is "corrosion inhibiting."

Statement 7 (Lowe)
"One cannot arbitrarily average the values reported and have a meaningful value.
Furthermore, most of the installations which we reported fall below the mean value
listed by Nordlin and Stratfull and these installations, as a whole, are showing excellent
performance."

Comment
Table 25 indicates that we find most of the readings published by Lowe and Koepf (2)
do not fall below the reported mean value which is shown on Table 12 in our report. -
The mean resistivity values listed in Table 13 of our paper were based upon the one
which was the least in-place soil resistivity value reported by Lowe and Koepf (2) for
each culvert site. In addition, these values were segregated according to what appeared
to be the corrosion condition of the pipe. Approximately 40 percent of the least in-place
resistivity values resulted in a mean of 2, 000 ohm-cm or less. In 33 out of 39 cases,
Lowe and Koepf show three in-place soil resistivity values for each culvert site.
In Table 25, every resistivity value was used without regard to the condition of the
culvert or the pH of the environment.

Statement 8 (Lowe)
''We believe that depths of 2 . 5, 5 . 0, and 10 . 0 ft cover the great majority of culvert
installations in existence. We therefore get a reasonable indication of soil resistivity
at the culvert depth while getting a 'feel' for the complete range of material which may
have been used as backfill."

Comment
In California we have had difficulty in duplicating field resistivity measurements
that are obtained during different seasons of the year.

Statement 9 (Lowe)
"Nordlin and Stratfull have made a serious misinterpretation of resistivity readings
in the report by Lowe and Koepf (59). Those readings were made in the field using a
Model 263 A Vibroground equipped with a wiring harness for obtaining average resistivi-
ties at depths of 2. 5, 5. 0 and 10 ft."
 57

TABLE 25 Comments
TOTAL RESISTIVITY READINGS (2) We do not agree that we have made a
ABOVE AND BELOW MEAN VALUE serious misinterpretation of Lowe and
Koepf's published (2, or 59) resistivity
Mean readings . - -
Number Number
Value Reference (12) in our paper relates the
Above Below
(ohm-cm) work of Stratfull wherein he empirically
correlates the culvert corrosion test
3, 100 54 53 method (10) of pH and minimum soil re-
2,000 60 47 sistivity to test methods which utilize the
250 103 4 average soil resistivity. The latter is the
method employed by Lowe and Koepf (2).
Total 217 104 The test method published by StratfulC( 12)
was found to be a more accurate test for
estimating soil corrosivity.
In our paper under discussion, we have
shown by the following statements that
average soil resistivites as obtained by Lowe and Koepf (2) can be significantly differ-
ent than the "minimum soil resistivity." "Although the validity of this analysis of data
in Table 13 has not been verified, it is interesting to note that there seems to be a rea-
sonably implied correlation of data." . . . "The resistivity measurements were de-
termined for the most part on an in-place soil. Therefore, they may not be accurately
reproducible owing to the fact that these values are highly dependent upon the seasonally
variable moisture content of the soil. Normally, soil resistivity measurements used
in culvert corrosion technology are based on the minimum value. The minimum resis-
tivity is normally less than the in-place soil resistivity. Therefore, care should be
exercised when directly comparing the in-place field values to the minimum resistivity
of a soil (10)."

Statement 10 (Lowe)
"Only aluminum alloys specifically designed to provide cathodic protection can be
used to protect steel." . . . "Most aluminum alloys will not provide such protection,
as evidenced by the literature (62) on aluminized steel. For aluminized steel, alumi-
num provides some protection only in the presence of significant chloride concentrations,
such as in marine environment."

Comments
No. 1. H. P. Godard states (48): "It is well known that aluminum stands high in
most galvanic series and hence provision must be made to avoid galvanic corrosion
when using aluminum in contact with other metals. This is one of the most common
practical corrosion problems with aluminum and one that can be eliminated if attention
is given to joint design and care of construction."
No. 2. It is our understanding that the cladding used on aluminum culvert sheets is
specifically designed to provide cathodic protection.

Statement 11 (Lowe)
"In effect, soil corrosion is being evaluated at only three of the eight sites, a fact
confirmed in the authors' paper, Table 3. Thus the use of aluminum culvert by the
State of California is based on the performance at only three sites."

Comment
Of the eight aluminum culvert sites, four were shown (see Figs. 4, 6, 10, and 13)
to be perforated by corrosion or destroyed by abrasion. Because of this destruction,
we assumed that the observations of corrosion on the backfill side of the culverts had
been misleadingly influenced by the flow leaking through the perforations.
58

Statement 12 (Lowe)
"The authors give no data or description of the comparative performance of the cul-
vert materials from one inspection time to another."

Comment
Tables 2, 14, and 15 in our paper report the results of two inspections on some
culverts.

Statement 13 (Lowe)
"These comments are not intended to stand by themselves, but to complement dis-
cussions offered on this paper by Dr. Hugh Godard and by A. H. Koepf."

Comments
It appears that the comments by T. A. Lowe also contradict the discussion by H. P.
Godard with regard to pH and resistivity limitations for aluminum culverts. For ex-
ample, Lowe states: "On the basis of this broader experience, aluminum culvert would
be expected to: 1. Provide corrosion performance superior to that of galvanized steel
in soils within the pH range of 4. 0 to 9. 0 and having a minimum resistivity above 1, 500
ohm-cm."
Conversely, H. P. Godard states: "In studies of aluminum corrosion, pH has not
been found to be a significant variable in either waters or soils." . . . "To indicate
the lack of correlation between corrosivity and soil resistivity, the values for the burial
plots are given in Table 23." . . . "I suggest also that the criteria selected by Nordlin
and Stratfull are not applicable or necessary for predicting the service life of aluminum
culverts."

A. H. KOEPF , Kaiser Aluminum.-The paper, "A Preliminary Study of Aluminum as

a Culvert Material, " has been prepared to describe data from a first inspection of eight
culvert sites where aluminum alloy and galvanized steel culverts were exposed to ag-
gressive environments. Several laboratory tests relating aluminum alloy to steel were
also conducted. From the data obtained, the authors have estimated service life ex-
pectancy of aluminum alloy culvert.
Discussion of the report is required in two stages. First is a general review of the
background fundamentals upon which this report appears to be based, for in this area,
the authors' method of analysis is open to question. Errors in understanding and extra-
polating results noted negate most of the value which may be attached to the conclusions
of the authors. Second in discussion is a review of the report details. In this area,
there is much to be learned from the data, particularly when stripped of opinion so the
data may be judged from the merits of their compilation. In spite of differences of
opinion, however , the authors are to be complimented for the presentation of field data
in a subject where more knowledge is necessary.

General Discussion
Several major points need to be established. (1) The report is based solely upon
performance of only eight sites, all of which were so placed to develop data for extreme
exposures ; yet the pattern of the paper is to rely heavily on statistical averaging of
these few extremes for predictions of life. These two conditions, one of exposure and
a second of analysis cannot be reconciled, and certainly do not represent a normal ap-
proach to research. (2) The progression of corrosion behavior of alclad aluminum
alloys follow a distinct pattern which can be quite different from the progression of
galvanized steel. The authors did not recognize this in their analysis and opinions ,
creating impressions which vary widely from the behavior expectations of aluminum
 59

TABLE 26
ESTIMATED YEARS-TO-PERFORATION OF GALVANIZED
STEEL FIELD TEST CULVERTS
(Based upon a method of predicting years-to-perforation
adopted by the California Division of Highways)

Estimated Years-to-
Culvert Installation Conditions Perforation by
Location
Site for Corrosion Evaluation (16 gage) Corrosion
(10)

A I-HUM-35-C Average coastal soil, mild acid

Bridgeville moderate abrasion site 22
B II-SHA-3-B Strong sulfuric acid flow from
Redding sulfide soil leaching 0
c III-BUT-21-B Very strong sulfuric acid flow
Oroville from sulfide soil leaching 0
D IV-SCL-5-C Abrasion site- not considered
Los Gatos in corrosion summary as sand
backing used 50
E IV-SCR-5-A Sulfuric acid flow from leaching
Scotts Crossing silty peat backfill elsewhere 0
F X-SJ-53-C Silt much in invert area only
this area reported 2-14
G XI-SD-2-NAT CTY Clay muck in invert area and
National City granular select backfill else-
where with both Urban runoff
and salt water tidal flow.
Only invert area and stream
on soil side reported 8
H XI-IMP-187- F Salt saturated alkaline
Salton Sea 3
Average years-to-perforation of seven corrosion sites
(Site D not included) 5. 0 to 6. 8 years
Average years-to-perforation of four corrosion sites
with pH in 4. 5 to 8. 3 range (Sites B, C, D, E, not
included) = 8. 7 to 11. 7 years

alloy. (3) The cumulative mechanisms of wastage of galvanized steel and aluminum
alloys in erosive, abrasive, and abusive flows are not the same. The authors indicate
they have no way to rate these conditions, yet categorically conclude culvert perform-
ance life across the same full range of bedloads. (4) The data obtained on galvanized
steel, supported by previous work by the authors, showed short life expectancy at these
sites; while the same work acknowledges areas of long life exposure on steel under less
severe exposures. This disparity may also be applied to aluminum alloy culvert, but
was not done in this instance.

The Eight Field Sites

A description of the location of the eight test sites cited in the paper is contained in
Table 26. The soil or water properties existing at each of these sites is charted in
Figure 42. The recommended lower limits for galvanized steel for 25- and 50-yr time
for perforation established by the Division of Highways (10) are shown along with the
60

Figure 42.

lower limit recommendations for long life for aluminum alloy culverts as developed by
the aluminum industry. Figure 42 and column 4 of Table 26 indicate clearly that all
test sites represent extremes, thus offering possibilities for rapid acquisition of know-
ledge if each site is evaluated comparatively but separately from other non-similar sites.
A general description of each site was not included in the paper. Therefore, column 3
of Table 26 is included for reference.
The preparation of the paper bases service predictions on application of statistical
averaging. Earlier work was based upon as much as 7, 000 sites (10) from which
statistical averaging can be expected to produce well-supported results. From the pre-
vious data and restated in this paper, there is some trend of linearity of wastage in
corrosion of galvanized steel which, of course, improves the accuracy of predictions
of years-to-perforation of steel.
The same statistical averaging approach was used to analyze the corrosion perform-
ance of aluminum alloy culverts. However, in this case, this was done with but eight
specimens with all but one (A) substantially outside of the recommended application
range for either steel or aluminum, as shown in Figure 42.
Application of statistical analysis is only as good as its base data. It is readily ap-
parent that when all sites are beyond the normal product working range and limited to
 61

eight specimens general averaging, adopted by the authors, produces results of little
value. As an example, consider the use of this approach on galvanized steel life. The
prediction formula of the Division (10) shows the seven corrosion sites for steel to have
an average years-to-perforation of5.0 to 6.8 years. Discounting Sites B, C, and E in
sulfuric acid with rated times of zero years, the remaining four corrosion sites aver-
age 8. 7 to 11. 7 years-to-perforation. On the basis of this data alone, using statistical
averaging, steel culverts which are expected to resist perforation for more than 10
years would need bituminous coating. A blanket conclusion such as this is obviously
invalid. It is well established that in many exposures steel will perform well for many
times that period. However, this is the exact analogy and statistical base upon which
the authors' conclusions on aluminum service life were derived and coating require-
ments established.

Progression of Corrosion
The progression of corrosion of alclad aluminum alloys has been established by a
number of investigators. The unique characteristics which resist or arrest corrosion
are the basis upon which the wide use of aluminum in corrosive exposures may be con-
sidered. The need for enlightened understanding of the stages of corrosion of aluminum
is mandatory if proper credence is to be placed in the uses of the material which have
been proven by time. Where the exposure is noncorrosive or of a mildly corrosive
nature, the surface may be observed to perform in several manners. It can appear
stained, a result of differential light diffraction from oxide buildup of varying thick-
nesses. It may show a random nonprogressive pit with hard corrosion product buildup.
Neither case represents corrosion with proceeds at a linear rate; in fact, the surface
performance might be improved as a result of oxide buildup. In this first phase of
exposure, the aluminum is structurally unaffected.
When the exposure becomes more corrosive to aluminum, the cladding proceeds to
provide anodic protection of the base metal. Electrochemically, the protection may be
likened to that of zinc on steel only to a limited extent. Cladding is anodic to the alumi-
num alloy core by a small potential difference and thus appears to be more active than
zinc on steel in early stages. It cannot be effective unless it does suffer corrosion.
However, in so acting, deposits of corrosion products inhibit further electrochemical
current and the corrosion cell action becomes self-arresting. Zinc on steel reacts
somewhat differently, more like a coating. Because of the high potential difference be-
tween zinc and iron, the removal of the zinc coating, once penetrated by corrosion,
may well proceed at a higher unit rate than on alclad aluminum. Neither the zinc nor
the iron have self-healing oxides and progressive corrosion may be expected. Thus in
the second phase, the aluminum will show evidence of corrosion relatively quickly and
just as quickly show evidence of the self-arrestment of surface corrosion. Galvanized
steel will suffer uniform attack on the surface, sometimes becoming arrested if the
surface buildup is completely contained by the environment, such as soil. A second
type corrosion has been observed to occur in culvert most frequently in heavy or salt-
laden soils which may be likened to concentration cell activity. Such soils are usually
fine grained and dense with very poor oxygen circulation and are characterized as silts
or mucks or soils of very low resistivity.
A third type of aluminum corrosion occurs when the chemical level is so high that
the surface aluminum oxide can be chemically attacked and corrosion will proceed ag-
gressively as fast as reaction rates allow. In the same exposures, zinc and iron are
also readily attacked. The example of such progressive corrosion occurs in strong
sulfuric acid exposures.
Each of the eight test sites falls within a category broadly defined by the types of
corrosion above. Sites A and Dare in the noncorrosive soils; Sites F, G, and Hare in
soils causing some, but not progressive-type, corrosion; and Sites B, C, and E are in
aggressive environments.
The authors did not consider the step-wise characteristic behavior of aluminum alloys
which does not lend itself to life predictions by linear projections except in aggressive
corrosion exposures. The possibility of progressive corrosion behavior may be rea-
sonably confirmed by making two or more inspections over a period of time. This paper
62

P~/\1:- ,lo NEnc. E l'olEit.<;i"f 100·

Vs. Ve.Loerr'{ or- W/ITf.~
f.tl'il f<.=1<,.. Sn..e-
lllcu>P1tJ<:r l1 M•\ \)C:SJ5W to
FoR. C.Ou.IEflT flA\J/.l<os
So
40

ao
CA~ll'o2JJoA
'SITE- D -t-r---~'"<·

\l>

r--
e
~ ~
I-
u.
'--'
4-
:;> r.:'\
0
-l 3 ,C) f<AT1rJer-
1L -- t;l!,061<:lN
~
'L
t-:3
~Q
\!) ..t
~~<!J (~' RA'fl\\/C:r
<.)~ i ~ l'Je>N ~i<os111r=
~ U)
I!]
'2 < o.e
.:;i.u..
0 o. .t,,
:J
<:t.
~
0.. ., /;\~i RAT1 We,,

"'"' . .. No El"f'f<..T

o.1.

\'l:~K.. VE-LOCJ'N OF 'v./"'"~

( 'f~/GE"c..)
\0 )5 '1.0 2S

Figure 43 .

does not check this very important point, even though over a year elapsed after the re-
ported inspections and the completion of the paper.
When the function of the cladding is not fully considered and service life is reported
by linear projection from corrosion depth measurements using a single observation the
results are obvious. The standard thickness of clading on aluminum culvert sheet is 5
percent of the total on each side and when performing its function may be corroded to
this depth. Using the linear means of projection, based upon measured depth of corro-
sion by the years-to-perforation extension method, alclad aluminum alloys become
rated at 10 or 20 years, such as Sites F, G, and H. Such an extrapolation is meaning-
less as it considers nothing other than that the cladding is functioning. A prediction of
life can only be established from several inspections. However, these unrealistic ex-
trapolations were made by the authors and have become prime base data for the pre-
viously described statistical averages. Thus, the years-to-perforation for Sites A, F,
 63

G, and H have been reported as generally the same, 6.6 to 17 years, further reducing
the questionable process of averaging.

Erosion and Abrasion

The wastage of aluminum alloy or galvanized steel by combined action of water velo-
city and size of bed-load cannot be conveniently considered in the single category of
"abrasion." The authors acknowledge this in comparing the performance of Sites A and
D. In spite of this, however, the conclusions on service life are made on the basis of
such a single broad category .
Contrary to the opinions stated by the authors, work has been done in studying the
effects of abrasion on culvert. The method of establishing such a study has been to
develop an expression of energy level exerted by a combination of rock size and velocity
of flow and relate this level to the effect on culverts. Energy level of a single rock
may be developed from equations of dynamics using approximations of surface friction.
Using this approach as a comparison basis, Figure 43 was developed. In order to re-
late the effect of the energy level on culverts, a series of rating conditions have been
established for aluminum culvert. The five rating conditions are as follows:
A - No effect on the aluminum surface
B - Slight roughening of crown of invert corrugation, but no significant metal removal
as a result of flow
C - Erosion; slight abrading of the corrugation surface at an estimated rate of 0. 002
in. per year (25-year life)
D - Abrasion; abrading of the corrugation surface at an estimated rate of 0. 002 in.
to 0.005 in. per year
E - Abuse; rapid abrasion of the corrugation surface at an estimated rate exceeding
0.005 in. per year
A number of investigations have been made and the rating condition lines have been
superimposed on the previously calculated energy level curves. The completed curves
and rating lines for the first time allow a method of determining cause and effect so that
the spectrum of abrasion may be properly described.
An energy rating system for culvert must, of necessity, be an approximation. Wide-
ly varying flow rates and velocities, maximum sizes, gradation of bedload, and shape
of rocks during each period of exposure require that any system be used with judgment
based upon experience. Nonetheless, it is felt that these tend to strike an average in
their effect on the surface. In order to lend uniformity in ratings, flow velocities for
this analysis are based upon the condition of a projection inlet culvert flowing two-thirds
full at the entrance. This condition represents a flow which could occur frequently in
the service period and if the culvert flows full or half full, the velocity in the culvert
will not vary widely. Rock size is established by approximating a mean- peak size which
can be expected to pass through the pipe during the higher flows, based upon observation
of the site or prior knowledge of stream bed behavior. The rating system was established
to consider fractured rocks rather than rounded rocks. The lesser area of contact
for sharp rocks would, of course, concentrate more stress at the impact point and
would be expected to abrade more. Limited observations appear to call for reduction
of ratings for flows of rounded rocks one or two levels .
Using this rating system, results of the report's abrasion Sites A and D have been
plotted on Figure 43 and may now be properly assessed as abrasion and abuse. The
conclusions of the paper may then be amended to state that these conditions could be
expected. The laboratory coupon tests with sand flow represent another condition of
either rating B or C .
Wastage of metal in culvert inverts is a result of combined corrosion and erosion.
When the mechanics of corrosion are superimposed upon an energy spectrum, it is
then possible to evaluate the cause of early failure of metal culverts due to "abrasion."
This process on galvanized steel usually takes the form of combined corrosion-erosion
over a period of several years. On aluminum alloy culvert, the failure must normally
be due to erosion or abrasion alone, as rarely does corrosion of aluminum appear in
64

abrasive flows. Once again, arbitrary statistical comparisons are meaningless without
the exercising of judgment in analysis of the data which have been presented.
The laboratory corrosion-abrasion tests, when considered, offer ample evidence of
the considerably different means to approach equivalent end results. The aluminum
results showed generally low-corrosion wastage but some displacement of the cladding.
The steel results showed the relatively rapid wastage of metal as a result of corrosion
assisted by the scrubbing action of the sand to expose new metal. While the authors did
not mention it, the zinc layer on the galvanized steel specimens in this test were rapidly
abraded to expose bare steel so that, except for a starting lag, galvanized steel may be
expected to perform similarly to the bare steel.
The principal conclusion one can draw from these tests is confirmation that, in nor-
mal culvert service, the corrosion-erosion wastage cycles must be considered as ap-
plicable with steel on the corrosion side and aluminum on the erosion side of the cycles.
It is interesting to note that the calculations and observations for energy ratings indicate
that, generally, bedloads with rocks up to 2 in. in size do not appear to be deleterious
to aluminum as erosion culvert service but that smaller particles can propagate the
corrosion-erosion cycle on galvanized steel. In this range, it does appear aluminum
alloy is equal to, or superior to, galvanized steel as the laboratory tests confirm.
The Division of Highway's method for definition of abrasion flow design limits at
Q10" of 5 or 7 ft/sec is an excellent one, particularly as it limits the probable rock
sizes at the entrance of the culvert which may be carried into the pipe.
As the water velocity and rock sizes in the bedload increase, the impact effect begins
to increase markecUy, following a form approximated as up to V6 • At higher energy
levels, the erosion in the erosion-corrosion cycle increases markedly and the aluminum
may be expected to begin to be wasted by abrasion more rapidly than steel. Meanwhile,
the corrosion rate on steel would remain relatively constant across most of the range.
At some point, the cumulative wastage of aluminum and galvanized steel would be com-
parable and above this in highly abrasive or abusive flows, the galvanized steel will
perform better, though the design life of both will be shortened.
The report's culvert test Sites A and D are performing within the projections of the
abrasion hypothesis. Site A, rated as abrasive to aluminum, is confirmed. The prin-
cipal wastage on steel is due to corrosion assisted by bedload scrubbing to remove zinc,
and later, iron oxide, while aluminum was subjected to localized abrasion and no cor-
rosion. Site D, rated as abusive to aluminum clearly demonstrated it. Galvanized
steel was abraded considerably, but as the time was so short, the corrosion part of the
cycle was virtually nonexistent. Site D contains a great deal of valuable information of
use in design of inverts and invert protection, and it is unfortunate the authors did not
discuss this.
Concluding the discussion on abrasion, it is important to indicate that the corrosion-
erosion cycle must be considered in evaluation; that mathematical approximations based
upon impact energy do exist from which levels of abrasive flow may be derived; that
aluminum performs well in normal erosion flows but poorly in abusive flows. The con-
clusions of the authors upon which the report recommendations on abrasion for all con-
ditions of flows are based were arrived at without placing the data into some evaluation
form on severity of exposure.

Conclusions
The report in question is just that. Data were obtained from eight culvert test sites
subjected to aggressive exposures, each representing an extreme. This was super-
ficially supported by limited laboratory abrasion, water immersion and fog tests. The
data developed were noted in some detail, but did not include observations and measure-
ment details which are necessary to complete understanding of performance. The con-
clusion and opinions of the authors were arrived at through a combination of statistical

'cQ 10 is defined as the 10-yr storm flow which will develop.Entrance headwater depth
equal to the height of the culvert, or the condition of flowing full at the entrance.
The velocity at the entrance is obtained by dividing Q10 by the area at the entrance.
 65

averaging with an inadequate data base and by not thoroughly understanding the observed
performance.
Contrary to the opinions stated in the report, aluminum alloys are performing as
well as, or better than, might be predicted from knowledge of the exposures-indicating
it likewise will perform well in milder normal exposures.

E. F. NORD LIN and R. F. STRA TFULL, Comments. -

Statement 1 (Koepf)
"The paper, 'A Preliminary Study of Aluminum as a Culvert Material, ' has been
prepared to describe data from a first inspection of eight culvert sites where aluminum
alloy and galvanized steel culverts were exposed to aggressive environments."

Comments
Tables 2, 14, and 15 in the subject paper report the results of two inspections on
some culverts.

Statement 2 (Koepf)
"The need for enlightened understanding of the stages of corrosion of aluminum is
mandatory if proper credence is to be placed in the uses of the material which have
been proven by time . "

Comments
The maximum previously reported amount of time in which aluminum had been used
as a culvert material was 3. 6 years (2). This is believed to be an insufficient amount
of time to definitely establish a long-term corrosion pattern of aluminum as a culvert
material.
The results of other tests of the underground corrosion resistance of aluminum for
up to 10 years are shown in Figures 37, 38, 39, and 40, which are included in our
comments on Godard's discussion, These tests were all performed on comparatively
small specimens that do not necessarily encompass corrosion variables that can occur
as the result of larger dimensions and methods of fabrication such as found in culverts.

Statement 3 (Koepf)
"The recommended lower limits for galvanized steel for 25- and 50-yr time for
perforation established by the Division of Highways (10) are shown along with the lower
limit recommendations for long life for aluminum alloy culverts as developed by the
aluminum industry . "

Comments
No. 1. Reference is made to aluminum industry (2) recommendations, as shown in
Figure 42 of Koepf's discussion, for a lower resistivTI:y limit of 1, 500 ohm-cm and an
indicated pH range of 4 to 9. The aluminum industry (2) has not demonstrated by
mathematical verification that these criteria are applicable to aluminum alloy culverts.
In Figure 42, "Resistivity (ohm-cm)" is not defined. The 25- and 50-yr curves for
galvanized steel (10) represent the minimum resistivity of a soil sample removed from
the culvert channelor the resistivity of the culvert flow and are not an average in-place
soil resistivity measurement.
No. 2. The aluminum industry (2) did not define how many years is considered to
be "long life" for aluminum alloy cUiverts.
66

Statement 4 (Koepf)
"The authors did not consider the step-wise characteristic behavior of aluminum
alloys which does not lend itself to life prediction by linear projections except in ag-
gressive corrosion exposures."

Comments
No. 1. Reference is made to our Figures 37 through 40. These figures indicate
two general behavior patterns for aluminum alloys. For the reported periods of testing,
the indicated patterns of underground corrosion behavior as shown in Figures 38 and 40
are: (1) aluminum has a weight loss of 0 .1 percent and 0. 2 percent, or (2) the results
of testing indicate a log-normal distribution of greater amounts of weight losses. In
Figures 37 and 39, the indicated underground corrosion behavior patterns are: (1) zero
mils of pitting, or (2) a normal distribution of increasing amounts of pit depths.
None of these Figures (37 through 40) demonstrate that when aluminum is randomly
placed in all types of soils, there is a corrosion characteristic which will prevent the
significant deterioration of all culverts of 0. 060-in. wall thickness by the defined cor-
rosion criteria of weight loss or pit depth.
No. 2. "Aluminum alloys are quite resistant to sea water. A corrosion rate of
about 0. 4 mpy or one-tenth the rate for steel was found for Ale an 578 (US 50 52) at
Harbor Island, N. C." (48).
This corrosion rate isa linear 0.0004 in. per year and thus one might assume that
sea water is not an aggressive environment to aluminum.
No. 3. McKee and Brown (9) state: "In solutions such as acetic acid and sodium
hydroxide (see Figures 1 and 2) the weight loss of aluminum varies linearly with time."
. . . "Therefore, weight losses in ammonium hydroxide solutions were determined
after two days and after seven days, and the calculated penetration rate was based on
the difference between the two-day and the seven-day weight losses." Since they con-
clude in their report that aluminum is resistant to corrosion in ammonium hydroxide
and acetic acid solutions one can assume that these are environments which are not
aggressive to aluminum.
McKee and Brown (9) reported most of their test results in the linear units of
mg/sq cm/day and penetration-mils/year. These are linear descriptions of the rates
of corrosion of aluminum.
Some of the data shown in our paper may be converted to the terms of mg/sq cm/day
by dividing the weight per unit area (sq cm) of 16-gage aluminum or steel by the report-
ed years (times 365 days) to 100 percent weight loss.

Statement 5 (Koepf)
"The possibility of progressive corrosion behavior may be reasonably confirmed by
making two or more inspections over a period of time."

Comments
Sometimes it is extremely difficult to establish and reasonably confirm the pattern
of progressive corrosion on the basis of two or more samples randomly selected at dif-
ferent times. For example, in the paper by Lowe and Koepf (2), the microsections of
aluminum shown in their Figures 3 and 4 after 1. 0 years of exposure appear to be more
corroded than those shown in Figures 6 and 7 after 3. 1 years at Royal City, Washington.
This is indicated by comparing Figure 3 to Figure 6 and Figure 4 to Figure 7. Thus,
one might question the validity of any one sample as reasonably confirming a minimum,
average or maximum amount of corrosion of the culvert. This factor is also demon-
strated on Table 19, which was submitted as a part of this discussion by H. P. Godard,
wherein 2 out of 4 samples had less depths of pitting after 7 to 8 years than at 4 years.

Statement 6 (Koepf)
"Contrary to the opinions stated by the authors, work has been done in studying the
effects of abrasion on culverts."
 67

Comments
It is not clear as to what is meant by Koepf' s statement as this was not stated or
implied in our paper. For example , abrasion tests were performed as a part of this
inve stigation under discussion.
We wish to compliment Koepf for his approach to the effects of abrasion on culverts.
His chart , "Peak Kinetic Energy vs. Velocity of Water and Rock Size Including Limit
Design of Culvert Ratings," s eems to be a reasonable approach to the evaluation of the
effects of abrasion. However, we do not have information or a reference where this
hypothesis with regard to culverts ha s been documented or confirmed by a previous
publication in a technical journal.
Koepf' s abrasion hypothesis would be enhanced if the 22 categories of abrasion and
rock size were verified by more than the two observations which he shows on the chart.

JOHN R . DAESEN, Director, The Galvanizing Institute.-A test sponsored by Bureau

of Public Roads, which results in recommendations for use of aluminum culverts for an
estimated life of 25 years, should state clearly and briefly in its summary, conclusions
and r ecommendations the limitations discovered, namely:
Uncoated aluminum culverts are not recommended for use in
soils of average pH below 6 or over 8, or with soil resistivities
below 2, 000 ohms-cm, or with abrasive flow over 7 fps; and
should not be used, bare or coated, when the flow contains
heavy metals unless the invert is paved.

The following comparative relationships, developed by the test, between galvanized

steel and aluminum should be clearly indicated.
From Tables 3 and 4, in 7 field test sites, most of them highly aggressive, with pH
ranging from 2. 7 to 8. 3, the life of aluminum was estimated to average 9 or 48 percent
that of galvanized steel, based on abrasion and corrosion, respectively. In 5 test sites
with pH between 4. 5 and 8. 3 the ratio of life was as above, but the average life for both
materials was 38 percent longer for both materials than in the broader pH range.
From Table 8, in laboratory corrosion-abrasion tests , bare steel was estimated to
have a life (to perforation in 16 gage) of 51 to 76 percent of that of aluminum. Gal-
vanized steel, in pilot tests, showed far greater resistance than either bare steel or
aluminum ("each test would probably require more than two weeks"). The test results
of bare steel against aluminum in this exposure are therefore without value in predict-
ing compara ti ve life.
From Table 11, in continuous submersion tests at pH 4.3 to 9.0, aluminum had an
estimated life of 2. 9 to 3. 7 years (to perforation in 16 gage) while galvanized steel was
unaffected (70-day test).
From Table 12, in laboratory fog room tests, aluminum had an estimated life of 3. 2
years to perforation (16 gage) while galvanized steel was unaffected (one year test of
galvanized steel).
As the connection between the reported results and the estimate of a 25-yr life for
aluminum culverts where recommended is not indicated, it can not be presumed that
this test supports a recommended use for a life of 25 years.

E. F. NORDLIN and R. F. STRATFULL, Comments.-J. R . Daesen is entirely cor-

rect in that we were unable to directly correlate our test results with a 25-yr service
life of aluminum culverts. However, as we pointed out, we exercised judgment in re-
lating our test results to a numerical service life. This judgment was based upon pH
and resistivity levels which indicated a minimum corrosion rate for aluminum.
68

ERNEST W. HORVICK, Director of Technical Services, American Zinc Institute. -

The tests carried out by the California Highway Division were certainly technically
conducted, neutral, objective and unbiased. We agree with the discussion of findings
under the heading "Remarks."
It was emphasized that since the paper related to accelerated investigations, actual
service experience would be carefully noted to ascertain the verity of the accelerated
tests.
A culvert in performance represents a dynamic situation in which the material is
exposed to soil, running water and that which is entrained in it. This represents true
performance.
The aluminum soil test data offered in rebuttal only related to test pieces embedded
in soil and represent a static condition in which all of the variables encountered in
culvert performance are not met.

E. F. NORDLIN and R. F. STRATFULL, Comments.-We agree with E.W. Harvick

that test results of small samples placed in the soil can only be indicative of a particu-
lar parameter of the corrosion phenomenon on the soil side or beneath silt in a culvert.
It is reasonable to assume that even these underground test results would have been
different if the dimensions of the samples were drastically altered.

S. K. COBURN, Applied Research Laboratory, U.S. Steel Corporation, Monroeville,

Pennsylvania. -The following comments are offered to supplement the references
given in the paper with respect to the effect of heavy metals, principally copper, on
the pitting tendency of quiescent natural waters in contact with aluminum. Porter and
Hadden, The British Nonferrous Metals Research Association, reported on their studies
concerning the performance of several aluminum alloys in the waters of eight cities (64).
They found that copper concentrations of 0. 02 ppm and greater in stagnant water would
seriously influence the behavior of aluminum alloys. They also found that the solution
in the pits was strongly acid and contained chlorides concentrated some tenfold over
those found in the flowing water.
Sawyer and Brown (65), Aluminum Company of America, indicated that very small
amounts of heavy metals may stimulate corrosion of aluminumbase alloys. The attack
is usually of the pitting type and is accelerated by the presence of chlorides. They de-
scribed the pitting that occurred in aluminum utensils used in Altoona, Pennsylvania,
where the water was found to contain, among other elements, 0. 09 ppm of copper and
O. 08 ppm of cobalt, together with chlorides, sulfates, silicates, and btcarbonates.
Rowe and Walker (66), General Motors Corporation, commented on the harmful
effects on aluminum Ofcopper found in tap water in various parts of the United States.
Presumably they were concerned with the possibility of corrosion of aluminum engine
blocks and/or radiators. They believe that the pickup of copper in domestic water sys-
tems employing copper tubing, together with the bicarbonates and chlorides that are
present, would require the use of corrosion inhibitors to reduce the pitting tendency of
the circulating cooling water .
One investigator (67), commenting on the Rowe and Walker paper, indicated that in
the analysis of 100 natural waters, a range of copper contents was found from less than
0. 001 ppm to 0. 30 ppm with most containing less than 0. 010 ppm (67).
These reports make it clear that the composition of natural waters can have a pro-
nounced effect on the pitting of aluminum culverts. The most important factor in the
occurrence of pitting is the presence of heavy metals in natural waters under stagnant
conditions .
 69

References
64. Porter, F. C., and Hadden, S. E. Corrosion of Aluminum in Supply Waters.
Jour. of Appl. Chem., Vol. 3, pp. 385-409, 1953.
65. Sawyer, D. W., and Brown, R. H. Resistance of Aluminum Alloys to Fresh
Waters. Corrosion, Vol. 3, pp. 443-457, 1947.
66. Rowe, L. C., and Walker, M. S. Effect of Mineral Impurities in Water on the
Corrosion of Aluminum and Steel. Corrosion, Vol. 17, pp. 353t-356t, 1961.
67. Comments on paper by Rowe and Walker. Corrosion, Vol. 17, pp. 597t, 1961.

E. F. NORDLIN and R. F. STRATFULL, Comments.-As S. K. Coburn points out,

the influence of relatively small trace amounts of copper in a natural water can have a
significant influence on the corrosion of aluminum.
This point is of further concern because these fractional quantities of impurities
can only be determined by a costly laboratory analysis. The cost of this analysis
would be far and above that currently used by the California Division of Highways when
considering the use of other culvert materials. Furthermore, even with the results of
a laboratory analysis, we are not cognizant of any information which would enable us
to definitely predict a corrosion rate of aluminum.

ALBERT R. COOK, International Lead Zinc Research Organization. -The authors and
the California Division of Highways are to be congratl!lated on a very objective study.
They must necessarily deal with the difficult program of deciding what long-term ex-
perience can be predicted on the basis of short-term tests. In corrosion work, this is
always a difficult and hazardous undertaking. Clearly the authors are justified in their
concern about pitting attack. Since the time to perforation is of vital importance,
general weight loss measurements as opposed to pit depth measurements have very
little relevance to a true evaluation of aluminum as a culvert material.
The mechanism and extent of corrosion in flowing water will be quite different from
that experienced under static water conditions. The pitting characteristics of aluminum
under static water conditions are well known; they can be catastrophic in the presence
of copper ions. Under flowing water conditions perhaps one should be more concerned
about abrasion resistance; abrasion is a hazard with most culvert materials but perhaps
this paper shows it to be a serious hazard for aluminum culverts.
Since aluminum is used for sacrificial anodes one must clearly be concerned about
bimetallic corrosion and possibly stray current corrosion. It would be well for users
to bear in mind that where you have a small area of anode (e.g., aluminum) and large
area of cathode (e.g., steel) such corrosion due to the bimetallic couple can be catas-
trophic in its intensity. I agree with the authors that where the steel is galvanized, the
zinc coating on the steel will give some measure of protection to the aluminum and may
still confer some continued protection to the steel since zinc corrosion products are
often inhibiting when kept in contact with steel.
Where corrosion on the soil side may be a hazard, attention should be drawn to the
National Bureau of Standatds, Circular 579, Underground Corrosion-M. Romanoff.
Here the data show that after being buried for 10 years in a number of corrosive soils
none of the aluminum alloys tested including commercial aluminum and aluminum
manganese alloys were satisfactory for use unprotected. One could speculate that some
newer aluminum alloys may show improvement here, and there would be soils where
satisfactory experience might be expected.
In one test reported on galvanized steel specimens with 3. 08-oz zinc coatings, in 8
out of 10 inorganic soils the zinc coating remained virtually intact after 13 years, while
 70

- in two highly reducing inorganic soils the zinc coating was almost completely removed
during the first few years yet the subsequent attack on the steel was relatively slow as
compared with the controls. A careful study of this excellent report is recommended.
On the basis of relatively long-term evaluations, the value of galvanizing for buried
steel is clearly brought out. It also indicates the need to consider attack on the basis
steel as a criterion of performance rather than corrosion of the zinc coating. As the
authors have pointed out care is necessary in specifying uncoated galvanized steel for
service in specific aggressive environments.
In general, one always feels safer when a proposed application can be related to a
similar one where long experience has given assurance of good results. Fortunately,
there are excellent case histories to support the use of galvanized steel under a wide
variety of conditions and for periods of the order of 30 years.

E. F. NORDLIN and R. F. STRA TFULL, Comments. -The authors agree that in cor-
rosion work any predictions of a rate of corrosion is a difficult and hazardous under-
taking.
The highway engineer must be concerned with the durability of a material in its
anticipated environments because disregard of this factor can lead to abnormal costs
for maintenance. He can no longer accept a material on the basis of recommendations
given in the terms of "maybe," "better," "looks good," etc. As a result, in the Cali-
fornia Division of Highways, culverts are now judged and economically evaluated on the
basis of expected years of service, which can always lead to a difference of opinion.

E. F. NORDLIN and R. F. STRATFULL, Closure.-We thank all of the contributors

to this discussion. Their comments give the readers a broader picture of the use of
aluminum than that given in the paper, alone.
We especially wish to thank Dr. Hugh P. Godard for submitting the previously un-
published data. This information will be of value to many engineers.
We are pleased that there was open discussion and hope that it will result in a diligent
effort to accumulate further engineering data which could clarify and resolve the use of
aluminum as a culvert.
The authors wish to take this opportunity to correct an oversight in their paper
wherein the thickness of metal for the steel and aluminum field test culverts was not
mentioned.
With reference to Tables 3 and 14 of our paper, culverts placed at IV-SCl-5-C were
10 gage, at I-Hum-35-C were 12 gage, while the remainder of the culverts were 14 gage .
All laboratory test samples were 16 gage.
Nondestructive Tests for Detecting Discontinuities
In Aluminum Alloy Arc Welds
F. C. PANIAN, J. A. PATSEY and G. F. SAGER, Aluminum Company of America,
Alcoa Research Laboratories, New Kensington, Pa.

This paper describes an investigation conducted to evaluate radio-

graphic and ultrasonic procedures for detecting 14 types of dis-
continuity in TIG or MIG arc welds in 2219-T87 aluminum alloy
plate, and to determine the effects of these discontinuities on the
static strength of the welded joints. The discontinuities studied
were microporosity, linear porosity, scattered porosity, oxide
inclusions, tungsten inclusions, lack of interpass fusion, lack of
root fusion, lack of side fusion, incomplete root penetration, crater
cracks, longitudinal cracks, craters, underbeadfolds and weld bead
overlaps. The welds were examined metallographically to aid in
establishing or confirming the types of discontinuity present.
Although the above work was done on 2219-T87 aluminum alloy
plate, the nondestructive tests employed could also be applied to
other aluminum alloys .
This investigation represents a portion of the work done on NASA
Research Contract No. NAS 8-5132 on the Arc Welding of 2219
Alloy.

•WELDS of high structural integrity have been made in aluminum alloys for many
years, but as in any other metal, the consistent production of such welds depends upon
the use of suitable equipment and the skill and care of the welder. For these reasons,
appropriate nondestructive testing procedures are necessary for determining the quality
of welds in structures .
While there has been wide experience in welding and inspecting aluminum alloy weld-
ments, there have been few data on the size and distribution of discontinuities in welds
coupled with the effect on the static strength. Another area where there has been little
reported work concerns the relative accuracy of radiography and ultrasonic techniques
in examining aluminum alloy weld structures.
Although some data indicating relative capabilities of these nondestructive tests were
obtained, no attempts were made to establish inspection standards or to develop speci-
fications. These data should be helpful in this connection, but additional research and
development is necessary before realistic inspection standards and specifications can
be established.
Aluminum alloy 2219 is used extensively in welded missile structures because it
exhibits good welding characteristics, uniformity of weld strength, resistance to stress
corrosion cracking, and high strength at ordinary, elevated, and cryogenic tempera-
tures. This investigation was undertaken in conjunction with a NASA research contract
on the arc welding of 2219 alloy to obtain data that would be of use in establishing non-
destructive testing procedures .for welded joints in 2219 alloy plate. Although the radio-
graphic and ultrasonic testing procedures discussed were employed to evaluate welds
in 2219 alloy plate, the same procedures also can be applied to other aluminum alloys
that might be used in highway applications.
Fourteen different types of discontinuities were deliberately introduced into a group
of welded plate samples, which also included relatively sound welds (for controls).

Paper sponsored by Committee on Metals in Highway Structures .

71
72

TABLE I The weld in each sample was subjected to

WELD CONDITIONS CONSIDERED IN INVESTIGATION
radiographic examinations and ultrasonic
tests employing conventional and some
General Type Specific Condition experimental procedures. The two series
Relatively Sound Relatively Sound
of tests were performed to determine
Porosity Microporosity
which of the test methods is the more suit-
Linear Porosity able in each instance. Reduced section
Scattered Porosity tensile and guided bend tests on specimens
Inclusions Oxide Inclusions containing portions of the various welds
Tungsten Inclusions finally were conducted to determine the
Lack of Fusion Lack of Interpass Fusion effect of the weld conditions on joint strength
or Penetration Lack of Root Fusion
Lack of Side Fusion and ductility. Auxiliary phases of the work
Incomplete Root Penetration included metallographic studies to verify
Cracks Crater Cracks the types of dis continuity or to explain their
Transverse Cracks* effects on the properties of the joints, and
Longitudinal Crackst
fracture studies for the same purposes.
Miscellaneous Craters
The weld conditions considered in the
Underbead Fold
Weld Bead Overlap investigation are given in Table 1.
* Transverse cracks were eliminated from the investi-
gation after a number of attempts to produce such PREPARATION OF WELDED PANELS
cracks failed.
Plate of 2219 alloy and filler wire of
Investigated to only a very limited extent because of
difficulty in producing such cracks. 2319 alloy were used for the welded panels
prepared for this investigation. The tensile
properties of the unwelded plate are given
in Table 2. One series of panels was
TABLE 2 fabricated from %-in. plate and a second
TENSILE PROPERTIES OF 2219-T87
from 1-in. plate. The panels were 18 x
PLATE USED FOR WELDED PANELS*
24 in. and were made by joining two 12 x
18 in. pieces of plate by a weld running
Plate T.S. Y.S. Red. of Elong. along an 18-in. edge. In general, single- V
Designation Thickness psi psi Area-% 3
- - - - - butt joints were used in the %-in. panels
Lot A 1/2" 71100 59400 20 lOt and double-V butt joints in the 1-in. panels.
Lot B 1/2" 67750 55200 22 lOt
However, additional %- x 18- x 24-in.
panels were prepared with square butt
Lot C l" 68850 56800 20 9**
joints welded in two passes (one on each
* Properties are averages for two tests. Specimens from side) with various degrees of weld pene-
1/2" plate had nominal diameter of 1/4"; those from l" tration.
plate had nominal diameter of 1/2", All specimens
were taken in the transverse direction. All welds were made by tungsten or
Elongation in l" gage length, consumable electrode inert-gas shielded
**Elongation in 2 gage length.
11
arc welding (TIG or MIG) . These proce-
dures eliminate the need for flux and are
used extensively for the welding of alu-
minum alloys. Joint preparation and weld-
ing procedures were varied to achieve the desired weld conditions. For example, the
degree of penetration was controlled by varying such factors as root spacing, welding
current and welding speed. Lack of fusion was achieved by the use of relatively cold
passes at appropriate stages in the welding operation. Oxide inclusions were introduced
by reducing the flow of inert gas to the point where shielding was no longer sufficient
to prevent oxidation. An oiled liner in the electrode hose was used in some instances
to produce porous welds. Tungsten inclusions were produced with the TIG procedure
by jogging the arc on and off. Craters and crater cracks were produced by a back-
stepping procedure that interrupted the continuity of the welding operation.
In several instances it was necessary to produce "synthetic" defects by procedures
that would not be encountered in ordinary welding operations. For example, longitudinal
porosity was simulated in one instance by drilling small holes in the root pass and then
covering them with a cold pass. These simulated defects were useful for checking the
capabilities of the nondestructive testing procedures but such defects were not included
in the mechanical property tests because the results would be misleading.
 73

TABLE 3

IDEIHIFICATION OF PANELS EXEMPLIFYING

VARIOUS WELD CONDITIONS
Plate Type of Identification of Panel
Nominal Weld Condition Thickness Type of Joint Weld ARL S No. APDL No .
---.-
Relatively Sound 1/2" Single-V Butt TIG 278454 1
Relatively Sound l" Double-V Butt MIG 278515 9B2
Microporosity 1/2" Single-V Butt MIG 278491 9Al
Microporosity 1/2" Single-V Butt MIG 278490 9A2
Microporosity with Some
Scattered Porosity 1/2" Single-V Butt TIG 278462 6A
Linear Porosity 1/2" Single-V Butt MIG 278492 lOAl
Linear Porosity (Artificial) l" Double-V Butt MIG 278501 10B2
Scattered Porosity 1/2" Single-V Butt MIG 278496 11A2
Light Randomly Scattered Porosity l" Double-V Butt MIG 278509 1182
Scattered Porosity l" Double-V Butt MIG 278596 1183
Oxide Inclusions 1/2" Single-V Butt MIG 278598 12A5
Oxide Film l" Double-V Butt TIG 278549 A5
Tungsten Inclusions 1/2" Single-V Butt TIG 278456 53
Tungsten Inclusions 1/2" Single-V Butt TIG 278486 53A
Lack of Interpass Fusion 1/2" Single-V Butt MIG 278493 5Al
Lack of lnterpass Fusion l" Double-V Butt MIG 278516 5Bl
Lack of Root Fusion 1/2" Single-V Butt TIG 278459 26B
Lack of Root Fusion l" Double-V Butt MIG 278511 6B2
Lack of Side Fusion 1/2" Single-V Butt MIG 278597 7A4
Lack of Side Fusion l" Double-V Butt MIG 278518 7Bl
Incomplete Root Penetration 1/2" Single-V Butt MIG 278488 8Al
Incomplete Root Penetration l" Double-V Butt MIG 278504 8B2
Internal Longitudinal Crack l" Double-V Butt MIG 278495 3B2
Craters (Face Pass) 1/2" Single-V Butt MIG 278522 2Al
Craters (Face Pass) l" Double-V Butt MIG 278514 2B2
Craters (Root Pass) 1/2" Single-V Butt TIG 278461 46
Underbead Fold 1/2" Single-V Butt MIG 278502 13Al
Underbead Fold l" Single-V Butt MIG 278505 13Bl
Weld Bead Overlap 1/2" Single-V Butt MIG 278519 14Al
Weld Bead Overlap l" Single-V Butt MIG 278512 14Bl
Reasonably Sound (Used for
Reheat Treating Tests) l" Double-V Butt TIG 278548 A4
Complete Penetration with
Roots of Weld Beads Inter-
penetrating about 1/811 1/2• Square Butt TIG 285185
Same as Preceding 1/2" Square Butt TIG 285321
Penetration Barely Complete
with Roots of Weld Beads
Just Touching 1/2" Square Butt TIG 285322
Incomplete Penetration with
Separation of about 1/64"
Between Roots of Weld Beads 1/2" Square Butt TIG 285323
Incomplete Penetration with
Separation of about 3/64"
Between Roots of Weld Beads 1/2" Square Butt TIG 285324
Incomplete Penetration with
Separation of about 1/16"
Between Roots of Weld Beads 1/2" Square Butt TIG 285187
Same as Preceding 1/2" Square Butt TIG 285326
Incomplete Penetration with
Separation of about 1/8"
Between Roots of Weld Beads 1/2" Square Butt TIG 285221
Penetration Varying from
Complete to Incomplete
(Sample from NASA) I" Square Butt TIG 278550

After a number of unsuccessful attempts to produce welds containing transverse

cracks, it was concluded that the occurrence of this type of defect was extremely un-
likely in inert gas shielded welds made in 2219 alloy plate with 2319 alloy filler, so
further attempts were abandoned. It was likewise not possible to produce longitudinal
cracks in any actual welding operation although a number of restraining schemes were
used. This experience indicates that the occurrence of longitudinal cracks also is un-
likely with the above welding procedure and alloys. A synthetic internal longitudinal
 74

- crack was finally made by bending the panel after one or two passes had been laid down
and then applying a cold pass over the resulting crack.
Table 3 gives the identification of the panels exemplifying the various weld conditions,
the plate thickness, the type of joint and the welding procedure.
If the subsequent radiographic examination indicated that the desired weld condition
had not been achieved, the panel was discarded, and an additional panel was welded with
appropriate alterations in joint preparation and welding practice. In some instances
sections were cut from the joints and examined to confirm the weld conditions.
A further check on the weld conditions was obtained from the ultrasonic, metallo-
graphic and tensile tests and from the examination of fractures in the tensile specimens.
Some additional panels were made during the final stages of the test program when it
became apparent that certain of the welds did not represent the desired weld conditions
or the desired degrees of severity of those conditions.
The welded panels prepared by Alcoa for this investigation were supplemented by
five panels with square butt welded joints received from the George C. Marshall Space
Flight Center at Huntsville, Ala. (ARL S Nos. 290963 to S290967, inclusive). These
panels were submitted primarily for a further evaluation of ultrasonic tests for detecting
slight amounts of incomplete penetration which are discussed in a subsequent portion of
this paper.

RADIOGRAPHIC EXAMINATION
In the initial examination of each panel, practically the entire length of the weld was
radiographed. These radiographs are subsequently referred to as the "full-length"
radiographs. When the ultrasonic tests were completed, a 24- x 5-in. section contain-
ing a 5-in. length of weld was cut from each panel and retained as a reference sample.
The welds in these radiographic panels were re-radiographed after the panels had been
cut from the original weldments. These sections, which are subsequently referred to
as the "radiographic panels," were reserved for the production of additional radiographs
if required.
Two General Electric Co. OX-140 radiographic units were used for the radiographic
examinations. Table 4 gives the exposure conditions and types of film used.
The full-length radiographs provided a valuable basis for screening the panels to
separate weld conditions suitable for use in the investigation from those that were not.
In most instances, the radiographs of the panels selected for further study indicated that
the desired weld conditions persisted over a significant portion of the weld length.
Table 5 lists the weld conditions that could be detected radiographically with reason-
able assurance, those that could not be detected in this way, and those for which detec-
tion was questionable. The radiographic observations in Table 5 are based on three
groups of radiographs: the full-length radiographs, those of the 5" radiographic sections
and those of the tensile and bend specimens .
The weld conditions detected with reasonable assurance were linear porosity, scatter-
ed porosity, tungsten inclusions, lack of interpass fusion, lack of root fusion, lack of
side fusion, incomplete root penetration, craters and crater cracks. Detection of in-
coinplete root penetration in squa1·e butt welds became uncertain when the separation
between the roots of the weld beads was about 1/16 in. or less.
The conditions that were not detected radiographically were microporosity and an
internal longitudinal crack. In the questionable detection category were oxide inclusions,
underbead folds and weld bead overlaps. The oxide inclusions were quite apparent in
the radiograph of a rather extreme example (S278596) produced by welding with a de-
ficiency of shielding gas. However, an oxide film condition (S278549) was not observed
radiographically. This undetected condition prevented proper bonding of weld metal to
plate metal and seriously weakened a portion of the joint. Underbead folds and weld
bead overlaps show up in radiographs as would unusually thick weld crowns or weld
crowns with an unsymmetrical distribution of metal. It is difficult to determine radio-
graphically whether the metal is actually fused into the surface of the plate or merely
folded over mechanically. Ordinarily, underbead folds and weld bead overlaps can be
more readily identified by a visual inspection of the weld.
 75
TABLE 4
RADIOGRAPHIC EXPOSURE CONDITIONS AND TYPE OF FILM*

Position Thickness
in Which Penetrated Exposure
Type of Specimen Radiographed by X-Ray KVP Time Film

24" x 18" panel o f normalt 1/ 2''*. 70-80 135-150 sec . Kodak AA

1/ 2" plate (single· V
groove)

24" x 18" pan e l of normal 1/ 2" 66 5 min . An s c o Supe ray ''B''

1/ 2" plate (squ are
burr joint)

24" x 18" panel of normal l" 105 3 min. Kodak AA

l" pl a te

24 11 x 511 radiographic normal 1/ 2" 66 5 min. An s co Superay ''B"

panel of 1 /2" p lace

24 11 x 511 radio graphic normal l" 94 5 min. An s co Supe ca y "B' 1

from 1/ 2" panel

T e nsile specimen normal 0.4 5-0.47" 66 5 min. An s co Superay "B 1

'

from l /2" panel

Same as preceding transvers e t 1-1/ 2" 94 5 min. An s co Superay ''B''

1
Ten s ile specimen normal 0.95-0.98" 90 5 m in. An s c o Superay ' B''
from 111 panel
Same as preceding transverse l" 94 5 min. Ans co Super ay ' 1 B''

Guid e d bend (face and normal 3/ 8" 63 5 min. An sco Supera y "B "
root from 1/2 11 and 111
panels)

Same a s preceding transver s e 1-1/ 2" 94 5 min . An s co Supe ray "B u

*All exposures made on General Electric Company OX-140 Radiographic Units with a tube current of 5 ma
and a source-to-film distance of 36".
t "Normal " indicates that X-ray beam was normal to plate surface; "transverse" indicate s that specim e n
was radiogrophed in an edgewise position with the X-ray beam parallel to the axis of the weld.

**All specimens except the machfoed tensile and bend specimens were radlographed with the crown on the
welds so the maximum thick ness penetrated by the X-ray beam is somewhat gre crter than the plate thick-
ness.

Radiographs of a number of the weld conditions under investigation are shown in

Figures 1, 2 and 3. Unfortunately, it has not been possible to retain in the illustrations
the degree of detail discernible in the radiographs themselves.
Figure 1 illustrates radiographs of a relatively sound weld, incomplete root penetra-
tion, linear and scattered porosity and lack of interpass fusion. The relatively sound
weld contains some microporosity not discernible in the radiograph. Incomplete root
penetration appears as a sharp line extending horizontally along the midportion of the
weld bead. The linear porosity appears as an irregular scattering of faint spots along
the centerline of the weld bead. The majority of the spots are in the right-hand third
of the weld, but there are several just to the left of the center. Lack of interpass fusion
appears as a discontinuous line of varying width and density extending along the center-
line of the weld bead.
Figure 2 illustrates radiographs of welds containing tungsten inclusions, oxide in-
clusions and craters . Light patches associated with the high density tungsten inclusions
stand out very sharply. The oxide inclusions require closer scrutiny and appear as
small spots of porosity near the right-hand end of the weld bead. The craters appear
as rounded zones with relatively dark centers.
76

S278454 - RELATIVELY SOUND TIG WELD

S278488 - MIG WELD WITH INCOMPLETE ROOT PENETRATION

S278492 - MIG WELD WITH LINEAR POROSITY

S278496 - MIG WELD WITH SCATTERED POROSITY

5278493 - MIG WELD WITH LACK OF INTERPASS FUSION

Figure l. Rad iographs of we lds i n ~ - in. 22l9-T87 plate .
 77

S278456 - TIG WELD WITH TUNGSTEN INCLUSIONS

• ' t '· ..
I .'

TENSILE BEND
S278486-T2 S278486-B3

TENSILE AND BEND SPECIMENS WITH TUNGSTEN INCLUSIONS

S278598 - MIG WELD WITH OXIDE INCLUSIONS AND POROSITY

S278522-Tl S278522-2

CRATERS IN TENSILE BLANKS

(MIG WELD - FACE PASS)

S278461-Tl S278461-T2
CRATERS IN TENSILE BLANKS
(TIG WELD - ROOT PASS)

Figure 2 . Radiographs of weld s in t -in . 2219-T87 plate .

78

S278515 - RELATIVELY SOUND MIG WELD

S278504 - MIG WELD WITH INCOMPLETE ROOT PENETRATION

S278501 - MIG WELD WITH LINEAR POROSITY

5278516 - MIG WELD WITH LACK OF INTERPASS FUSION

5278511 - MIG WELD WITH LACK OF ROOT FUSION

Figure 3. Radiographs of welds in l-in. 22l9-T87 plate .

 79

Figure 4. Radiographs of square butt welds in ~-in. 2219-T87 plate representing various
degrees of penetration: 8285185-R, complete with interpenetration of weld bead roots;
8285187-R, moderately incomplete; and 8285221-R, markedly incomplete.
BO

278518-T2t 278597-T2 278596-Tlt

LACK OF SIDE FUSION SCATTERED POROSITY

278511-T2t 278459-T2 278459-62

LACK OF ROOT FUSION

tFrom l " panels; all others from 1/ 2" panels.

278493-T2 278493-62

LACK OF INTERPASS FUSION

278488-T2 285187-T2* 2B5221-T2*

278488-64 285187-61* 285221-64*

INCOMPLETE PENETRATION
* Square butt joints; all others ¥-groove.

Figure 5. Radiographs of welds in some t ensile specimens ( - T) and bend specimens ( - B) .

 dis1ance in inches from reference edge
290963
2 3 4 5 6 7 8 9 I0 II I2 I3 I4 I5 I6 17 I8 19 20 21 22 23
0
----
- 290964
Cl)Cl).------
Q)~b I 2 3 4 5 6 7 B 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
cQ)- - - - - - - - •- o
o.~
-0 o
VJ
-...
- 290965
.E r--....,1--""2,......---=-3--...,q--...,s,......---=s- - " " '1,......--s= ----=9- - . . ,1""0--.,..,11--..,.,.,,.2- -1'"'3,.......-.,.14.,..---,1"="s- - " "16,......-..,.11= --...,,,.,e:---..,,19: :--=,-=o:----:2:-:1- 22 i3

~Cl) - - - - - - - - - - - 0 0
Q)-
"Co

290966
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

--- 0 0

290967
2 3 4 5 6 7 8 9 10 ti 12 13 IG IS- 16 17 IB 19 20 21 22 23

------'--. 0

continuous ultrasonic indication in weld presumably incomplete penetration

------ radiographic evidence of incomplete penetration
0 location of metallographic sect ion showing complete penetration

• location of meta llogra phic sect ion showing incomplete penetration

Figure 6. Results obtained from the ultrasonic, radiographic and metallographic tests for incomplete penetration .

00
......
 •I

CX>
t-.:i

Figure 7. Radiograph of weld in panel 8290966 after scarfing (about two-thirds original size). Weld starts about l~ in. from
"reference" end of plate (upper right); incomplete penetration is evident for a distance of about three-fourths in. at beginning of
weld; circular indentations (dark spots) may be associated with weld setup or attachment of e~uipment.
 83

The radiographs in Figure 3 include a relatively sound weld containing some micro-
porosity not discernible in the radiograph and welds with incomplete root penetration,
linear porosity, lack of interpass fusion and lack of root fusion. The line of incomplete
root penetration is broader and more diffuse than that in Figure 1. The lack of interpass
fusion is evident as a few dark spots along a very faint, diffuse and discontinuous hori-
zontal line through the mid- portion of the weld. The conspicuous dark spots in the bot-
tom radiograph indicate voids of s ignificant size along unfused portions of the weld root.
Figure 4 shows radiographs of s quare butt welds in %-in, plate with several degrees
of penetration and fusion. The marked incomplete penetration in panel S285221-R(roots
of two weld beads about %in. apart) is plainly discernible a nd is associated with linear
porosity. Panel S285187-R in which the roots of the weld beads are about 1/ 1 6 in. apart,
shows radiographic evidence of the incomplete penetration along only a portion of the
weld length.
Radiographs of the welded joints in a number of machined tensile and bend specimens
are reproduced in Figure 5. The weld conditions represented are lack of side, root and
interpass fusion, incomplete penetration and scattered porosity.
Figure 6 shows the extent of incomplete penetration detected radiographically in the
square butt welds from the George C. Marshall Space Flight Center. Figure 7 is a
radiograph of one of these welds (S290966).

ULTRASONIC TESTS
All welds were inspected by an angle-beam shear-wave ultrasonic procedure in which
the ultrasonic beam was directed through the plate and into the weld as shown in Figure
8. Four traverses of the search unit were made along lines roughly parallel to the weld
and extending over its full length. These scans were made along each side of the weld
and on each surface of the plate. Maximum response from discontinuities was obtained
by slightly rotating the search unit about an axis normal to the plate surface and varying
its distance from the weld.
A 2.25-mc lithium sulfate angle-beam shear-wave contact search unit and a Sperry
Type UR Reflectoscope were used for most of the ultrasonic tests. The angle-beam
search unit produced a shear-wave beam refracted to approximately 48 deg in the plate
(angles of incidence and refraction are given with respect to a normal to the plate sur-
face) . Standardization was attained by adj usting the equipment for a 2. 5-in. peak-to-
pe~ indication from a /s in . diameter hole drilled through the % in. dimension and
1

normal to the surface of a% x 21/2 x 14 in. aluminum alloy angle beam reference
plate.
A variable angle 2. 25-mc lithium sulfate contact search unit was also employed for a
limited number of tests. Other tests were conducted with experimental immersion test-
ing apparatus employing higher test frequencies. A special lucite holder in which search
units of the immersion type could be mounted was constructed for the latter tests. This
holder contained a water column to provide coupling between the search unit and the part

search unit

Figure 8. Shear wave search unit in position for an ultrasonic examination of 22l9-T87
weldments.
84

'

Figure 9. Angle-beam search units used for

ultrasonic weld inspection; unit in upper
right is contact unit transmitting e, shear
wave into alwninum at a refracted angle of
48°, other unit is a liquid filled, variable
angle unit.

being inspected. It was designed so that

a shear-wave beam could be directed into
the part at an optimum refraction angle
of 45 deg. Tests were made with this unit
at frequencies of 5 and 10 megacycles
using lithium sulfate transducers of both
the focusing and nonfocusing types. The
three ultrasonic search units are shown Figure 10. Special holder for immersion
in Figures 9 and 10 . tY}le s ear ch units.
After the initial ultrasonic inspections
were concluded, weld beads on a number
of samples were machined or ground flush
with the plate surface and re-examined to determine whether extraneous ultrasonic in-
dications were obtained from bead geometry or discontinuities in the crown of the welds.
Samples containing weld crowns that appeared to be causing extraneous indications with
the fixed-angle contact search unit were re-examined with the variable-angle search
unit. These comparison tests were performed to establish a possible refraction angle
other than 48 deg that might eliminate or minimize the extraneous indications.
The ultrasonic tests were not limited to the inspection of the as-welded panels. The
tensile and bend specimens machined from the panels were also examined ultrasonically.
The examination of the latter provided a further means of determining the effect of the
weld crowns on ultrasonic response. The square butt welds previously referred to were
examined with particular regard for incomplete root penetration.
The results of the ultrasonic tests on joints of the V-groove type are summarized in
Table 5. The conditions readily detected ultrasonically include oxide inclusions, lack
of interpass fusion, lack of root fusion, lack of side fusion, incomplete root penetration
and internal longitudinal cracking.
 85
TABLE 5

EFFECTIVENESS OF RADIOGRAPHY AND ULTRASONIC

INSPECTION FOR DETECTING VARIOUS WELD CONDITIONS

Condition Detected by Condition Detected by

Radiographic Examination Ultrasonic Examination *
Weld Condition Yes Questionable t No Yes Questionablet No

Microporosity x x
Lin ear Porosity x x
Scattered Porosity x x
Oxide Inclusions x x
Tung s ten Inclusions x x
Lack of lnterpass Fusion x x
Lack of Root Fusion x x
Lack of Side Fusion x x
Incomplete Root Penetration** X' X''
Crater Cracks x x
Internal Longitudinal Crack x x
Craters x x
Underbead Fold Xtt x
Weld Bead Overlap Xtt x
*With a few exceptions, ultrasonic tests were made with a 2.25 me lithium sulfate contact angle-beam search
unit producing a shear wave beam refracted at an angle of 48° to the normal in the aluminum. In most cases
it was necessary to grind or machine the weld flush with the plate s urface to avoid extraneous indications
from the weld crown. A few tests made during the latter part of the investigation with a liquid-filled vari-
able-angle search unit indicated that the extraneous indications might also be eliminated or minimized by
introducing the sound beam at a suitable angle.

t This category includes instances where the condition was not detected with certainty or where it was de-
tected in one sample but not in another.
**This condition was investigated in square butt joints as well as in the single-V and double-V groove joints
used for all other phases of the investigation .
0
Detection became uncertain when separation between roots of weld beads was about 1116" or less.
00
Ultrasonic method appears superior to radiographic method for detecting slight incomplete penetration in
square butt welds rut exact limit of detection has not yet been determined.

ttWeJd crown malformations are usually apparent in a radiograph but it might be difficult to relate them to
a specific defect.

The reflectograms reproduced in Figures 11 through 15 show the screen presentations

obtained with relatively sound welds and welds containing a number of the discontinuities.
The weld conditions reliably detected by ultrasonic testing usually gave well-defined
indications. An ultrasonic noise level (hash) like that in Figure 11 was noted in testing
most welds.
The weld conditions that were not detected consistently or with assurance in the
ultrasonic tests include scattered porosity, linear porosity, craters and tungsten in-
clusions. In general, weld bead overlaps, underbead folds and microporosity were
not detected by the ultrasonic tests .
Table 6 gives the ultrasonic response obtained from discontinuities in the square butt
welds with various degrees of penetration. Figure 16 contains reflectograms corre-
sponding to 3 of the square butt welds (8285221, 8285107 and 8285105) representing
various degrees of incomplete penetration. Detection of incomplete penetration in welds
of this type be.comes uncertain when the separation of the two weld bead roots is less
than about '110 inch.
The five square butt welds obtained from the George C. Marshall Space Flight Center
were examined ultrasonically with the conventional 2. 25-mc angle-beam shear-wave
86

1/2 THICK PLATE t" THICK PLATE

278454-TI 278515-TI
relatively sound and some microporosity

278522-T2 278514-TI
craters
Figure ll. Reflectograms showing ultrasonic indications corresponding to various weld
conditions.
 87

11
1/2 THICK PLATE

278493-TI 278516-T2
lack of interpass fusion

278459-T2 278511-T2
lack of root fusion
Figure 12. Reflectograms showing ultrasonic indications corresponding to various weld
conditions .
88

112" THICK PLATE I 11 THICK PLATE

278597-R 278518-TI
lack of side fusion

278488-T2 278504-TI
incomplete root penetration
Figure 13 . Reflectograms showing ultrasonic indications corresponding to various weld
conditions.
 89

112" THICK PLATE 1" THICK PLATE

278496-TI 278509-T2
scattered porosity

278598-TI 278549-T2
oxide inclusions
Figure 14. Reflectograms showing ultrasonic indications corresponding to various weld
conditions.
90

11 11
l/2 THICK PLATE 1 THICK PLATE

278492-TI 278501-R
linear porosity

278486-T2
tungsten inclusions
Figure 15 . Reflectograms showing ultrasonic indications corresponding to various weld
conditions.
 91

A- indication from incomplete penetration

S-285221

B- indication from incomplete penetration C- screen presentation from complete

s- 285187 penetration S-285185

Figure l6. Reflectograms showing ultrasonic indications from varying degrees of

penetra,tion.
 <D
!>:>

complete penetrotron complete penetration

incomplete penetration incomplete penetration

Figure 17. Reflectograms and macrographs from corresponding locat ions in weldment numbered 29c966 ,
 93

TABLE 6

ULTRASONIC INDICATIONS FROM VARIOUS DEGREES

OF PENETRATION IN SQUARE BUTT WELDS.

Measured Bead Ultrasonic

Separation or Indication
S No. Penetration Sought Interpenetration Height

285185 Complete with some Int. Approx.

interpenetration 3/3 2° 0.3" p-pt

285187 Incomplete with a Sep. Approx.

small root separation 1/16" 2.4" p-p

285221 Incomplete with substantial Sep. Approx .

root separation 1/8" 3.2" p-p

285321 Complete with roots

interpenetrating
about 1/8" Int. 5/32"* 0.2"p-pt

285322 Barely complete with

roots just touching Sep. 0.046"* 0.7" p-p

285323 Incomplete with root

separation of aOOut
1/64" Int. 1/32"* 0.4" p-p

285324 Incomplete with root

separation of about
3/64" Sep. 0.066"* 1.0" p-p

285326 Incomplete with root

separation of about
1/16" Sep. 0.060"* 2.2" p-p

* Average of two measurements on metallographic sections: other values are for one section.
t Normal hash level; no isolated indications.

test and with the special immersion test using 5- or 10-mc lithium sulfate search units
in a lucite holder. The results of the conventional tests were shown graphically in
Figure 6, which also shows the results of the radiographic examinations previously
described. Figure 17 shows reflectograms and corresponding macrosections represent-
ing several locations along one of these welds (8290966). In one instance, incomplete
penetration that escaped detection in the ordinary ultrasonic test was detected in the
special test using higher frequencies.

METALLOGRAPHIC EXAMINATIONS
Following the ultrasonic examination of the original 18- x 24-in. panels, transverse
sections through the welds were removed for metallographic examination. In the ma-
jority of the welds, this examination was limited to a macroscopic study at low magni-
fication but in some instances, it was necessary to make examinations at higher magni-
fication to check on specific conditions.
Closely related to the metallographic studies was examination of the fractures in the
specimens subjected to the tensile and bend tests. These examinations were made to
obtain further information on weld structure and soundness.
The macroscopic examinations and fracture studies helped to confirm the weld con-
ditions sought, most of which had previously been identified by radiographic examina-
tion. Metallographic examinations also showed the presence and extent of conditions
such as microporosity, oxide inclusions and incomplete penetration. (In this paper
S278454 - RELATIVELY SOUND WELD S278488 - INCOMPLETE ROOT PENE·
(Etch 103 NaOH) TRATION (Etch 103 NaOH)

S278519 - WELD BEAD OVERLAP 5278502 - UN DE RB EAD FOLD

(Etch 103 NaOH) (Etch 103 NaOH)

S278597 - LACK OF SIDE FUSION

(Etch Keller's)

Figure 18. Macrographs of sections through butt welds int-in. 2219-T87 plate (mag. 2X).
278515 - RELATIVELY SOUND WELD S278495 - INTERNAL LONGITUDINAL CRACK

S278516 - LACK OF INTERPASS FUSION S27851 l - LACK OF ROOT FUSION

Figure l9. Macrographs of sections through butt welds in l-in. 22l9-T87 plate (mag. 2X,
etch lO"/o NaOH).
S278518 - LACK OF SIDE FUSION S278504 - INCOMPLETE ROOT PENE-
(Etch 103 NaOH) TRATION (Etch 103 NaOH)

S278596 - SCATTERED POROSITY

(Etch Keller's)

Figure 20 . Macrographs of sections through butt welds in 1-in. 2219-T87 plate (mag. 2X) .
 5278512 - WELD BEAD OVERLAP 5278505 - UNDERBEAD FOLD

Figure 2l. Macrographs of sections through butt welds in l-in. 22l9-T87 plate (mag . 2X,
etch lo% NaOH).

Figure 22. Crater on root pass of TIG butt weld in 22l9-T87 plate (S-278457, mag. 2X) .
 . .· ....· . ... .I...
.. ·. • I

.... .
···. ..
.·: "'.·'....:
•. .. ' •" .'"'
·.· .·•
..
,.·.. .... ~-
~

# • ••

•. •
...

Figure 23. Macrograph of section through weld and crater (max. 5x, as-polished) .

Figure 24. Macrograph of section through TIG weld in i-in. 22l9-T87 plate; arrow in-
dicates lack of root fusion (s-2781+59, mag. 5x, etch Keller's).
 ·.
. ..
.
.
-: -' ·
,• .
.,
.. ·'·

Figure 25. Macrograph showing microporosity in an otherwise sound TIG weld in ~-in.
2219-T87 plate (S-278454, mag. SX, as-polished),

, • f •

I•

·.....

Figure 26, Macrograph showing microporosity in a MIG weld in ~-in. 2219-T87 plate
(s-278490, mag. 5X, as-polished).
100

.. . .. •:-> •I
.....·',
. 1

,.. . .
' ,,I
~
,;

...
=··

Figure 27. Macrograph showing porosity in a TIG weld int- in . 2219-T87 plate (S-278462,
mag. 5X, as-polished) .

·,·

Figure 28. Micrographs showing oxide inclusions in MIG weld in t-in. 2219-T87 plate
(s-2785 98 , mag. 5oox, etch Keller's).
 101

S285321-Ml S285323-Ml S2B53??-Ml ~;285324--Ml

Figure 29. Macrographs of sections through square butt TIG welds in 2219-T87 plate show-
ing various degrees of penetrat ion (mag. 2X, etch: top row 10% NaOH, bottom row
Keller's).

microporosity is arbitrarily defined as porosity in which the voids have diameters less
than about 0. 01 in. which is approximately the smallest void discernible to the unaided
eye.) Various weld conditions are shown in the macrosections included in Figures 18,
19, 20 and 21.
Figure 22 is a 2x photograph of a crater on the root pass of a TIG butt weld in 1/2-in.
plate. A macrosection through the weld and crater appears in Figure 23. Fine porosity
is rather generally distributed through the weld bead on the face side but the root side
bead is comparatively sound. A more detailed metallographic examination of the section .
failed to reveal any significant effect of the crater on the structure of the surrounding
weld metal.
Figure 24 shows a macrograph of a section through a TIG weld in %-in. plate. An
arrow indicates lack of root fusion. Figures 25 and 26 illustrate microporosity in other-
wise sound welds in %-in. plate. Figure 27 shows rather generally distributed porosity
in a TIG weld in %-in. plate. The larger pores are in the macro range and the finer ones
in the micro range. Figure 28 shows oxide i nclusions in a MIG weld in %-in . plate.
The weld was made with a reduced flow of shielding gas to favor oxide formation. The
macrographs in Figure 29 illustrate various degrees of penetration in the square butt
welds previously referred to .
Metallographic measurements of the extent of interpenetration or separation of the
weld bead roots in the square butt welds were summarized in Table 6, previously re-
ferred to in conjunction with the ultrasonic tests.
DISCUSSION OF NONDESTRUCTIVE TESTS
As previously pointed out (see Table 5) a substantial fraction of the weld discontin-
uities can be detected both radiographically and ultrasonically. Radiographic examina-
102

tion provides more information on the exact nature of the weld conditions but the ultra-
sonic procedure would be faster for large-scale commercial inspection and more ame-
nable to automation.
One shortcoming of the ultrasonic method that probably can be overcome is associated
with extraneous indications originating from the crowns on the welds. In some instances
it is necessary to scarf the weld to determine whether an ultrasonic indication is as-
sociated with an actual defect. A few preliminary tests with the variable-angle liquid-
fill ed search unit suggested that a search unit angulating technique could be developed
for minimizing the extraneous indications originating from the weld bead. Extraneous
ultrasonic indications were also consistently reduced by merely smoothing the weld
crowns before testing.
Substantial incomplete penetration in the square butt welds (root separations in ex-
cess of about 1/16 in.) could be detected by both the radiographic and the ultrasonic pro-
cedures. However, the ultrasonic procedure was much more effective for detecting this
condition when the weld bead roots were separated by the smaller distances associated
with only slight amounts of incomplete penetration.

(Face Surface)
Both Ends

1... 1

_ _ _ _ _ _ Machine this area

Width of weld + Y2 in,

_ _ _ _ _ _L
= -- - - - - - -----··---
L/t..

._________.f f_ _.__
;<C< ~b....___l ~-____.~UT ENDS

ROO~ WELD METAL MACHINED FLUSH

WITH BASE METAL AS SHOWN

SPECIMEN DIMENSIONS
NOTE : Specimen thickness Nominal Total Width at Minimum
greater than decimal dimen-
Thicknes s 1 in. Width, in. Wel~) ;n. Length , in.
(T) (W;) (L)
sion (last No. 1st Col.)
shall be machined in accord- 1/4 to 0. 356 2 lY, 18
ance with the dimension
shown for the next higher
3/8 to 0.475 2 1Y, 20
specimen thickness range. 1/2 to 0.720 2 lY, 22

NOTE: For specimen layout

3/4 to 0.963 2 I Y, 24
allow 3/ 16 in. for sawcut l" to 1.455 lY, l" 28
and finish of edges.
Grea ter t han 1-1/2 lY, l" 30

Figur e 30 . Reduced -section t ensile specimen,

 103

TABLE 7

TENSILE STRENGTHS OF ARC WELDED

2219-T87 PLATE WITH SINGLE-V AND DOUBLE-V BUTT JOINTS*
Tensile Test
APDL Type of Plate Spec . T.S.+
S No. Ho. Weld Thickness Nominal Weld Condition No. psi

278454 TIG 1/2" Relatively sound but some Tl 36200(a)

microporoslty T2 39000(a)
Av . 37600

278515 9B2 MIG ••• Relatively sound buc some Tl 41900(c)

microporosity T2 41900(c)
Av . 41900

278491 9Al MIG 1/ 2" Microporosity Tl 38400(a)

T2 38600(a)
Av . 38500

278490 9A2 MIG 1/2" Microporosity Tl 38900(a)

T2 39100(a)
Av . 39000

278462 6A TIG 1/2" Microporosity with some Tl 32500(c)

scattered m acroporosity T2 357,00(c)
Av . 34100

278492 JOA! MIG 1/2" Linear porosity Tl 34300(a)

T2 34900(a)
Av . 34600

278496 11A2 MIG 1/2" Scattered porosity Tl 34800(c)

T2 33600(c)
Av. 34200

278509 1182 MIG l" Light randomly scattered Tl 29300(c)

porosity T2 29800(a)
Av. 29550

278596 llB3 MIG l" Scattered porosity Tl 32800(a)

T2 37700(a)
Av . 35250

278598 12A5 MIG 1/2" Oxide inclusions Tl 16500(a)

T2 18000(a)
Av . 17250

278549 A5 TIG 1'' Oxide film Tl 25200(c)

T2 19600(c)
Av. 22400

278486 53A TIG 1/2" Tungsten inclusions Tl 34700(b)

T2 29600(a)
Av_ 32150

(Continued)
104

TABLE 7 (Continued)

TENSILE STRENGTHS OF ARC WELDED

2219-T87 PLATE WITH SINGLE-¥ AND DOUBLE-¥ BUTT JOINTS*
Tensile Test
APDL Type of Plate Spec . T.S.+
S No. No. Weld Thickness Nominal Weld Condition No. psi

278493 5Al MIG 1/2" Lack of interpass fusion Tl 32300(a)

T2 31800(a)
Av . 32050
278516 5Bl MIG l" Lack of interpass fusion Tl 33300(a)
T2 32500(a)
Av. 32900

278459 26B TIG 1/2" Lack of root fusion Tl 16500(a)

T2 l 7900(a)
Av. 17200
278511 6B2 MIG ]" Lack of root fusion Tl 27300(a)
T2 27300(b)
Av. 27300
278597 7A4 MIG 1/2" Lack of side fusion Tl 27700(a)
T2 27200(a)
Av. 27450
278518 7Bl MIG l" Lack of side fusion Tl 33100(b)
T2 32700(a)
Av. 32900
278488 BAI MIG 1/2" Incomplete root penetration Tl 9800(a)
T2 IOIOO(a)
Av . 9950
278504 8B2 MIG l" Incomplete root penetration Tl 21500(a)
T2 21100(a)
Av. 21300
278522 2Al MIG 1/2" Craters (face pass) Tl 33700(a)
(Specs. Tl and T2 include T2 30300(a)
craters; T3 and T4 do not) Av. 32000
T3 37700(a)
T4 38500(a)
Av . 38100
278514 2B2 MIG I" Craters (face pass) Tl 32500(a)
(Specs. Tl and T2 include T2 34500(a)
craters; T3 and T4 do not) Av , 33500
T3 40000(c)
T4 42500(b)
Av. 41250
278461 46 TIG 1/2'' Craters (root pass) Tl 24700(c)
(Specs. Tl and T2 include T2 24800(c)
craters; T3 and T4 do not) Av . 24750
T3 34100(c)
T4 32800(c)
Av . 33450
(continued)
 105

TABLE 7 (Continued)

TEN~LE STRENGTHS OF ARC WELDED

2219-T87 PLATE WITH SINGLE-V AND DOUBLE-V BUTT JOINTS*
Tensile Test
APDL Type of Plate Spec. T.S.+
S No. No. Weld Thickness Nominal Weld Condition No. psi

278548 A4 TIG l" Reasonably sound--(Specs. Tl 29900(a) (As welded)

Tl and T3 tested in as-welded T3 25500(a) (As welded)

condition; T2 and T4 reheat- Av. 27700 (As welded)

treated and aged to -T62 T2 46400(a) (Reheat treated)

T4 47100(a) (Reheat treated)
temper before testing.)
Av. 46750 (Reheat treated)

*Reduced-section specimens used for all tensile tests. Single-V butt joints used for all 1/2" plates and double-V butt joints
for all l" plates. All specimens tested in as welded condition.
+Letters in parentheses after tensile values indicate path of fracture: {a) through weld, {b) at edge of weld and {c) partly
through and partly at edge of weld.

TENSILE TESTS
In order to determine the effect of various weld conditions on joint strength and ductil-
ity, reduced-section tensile tests and face and root guided bend tests were made on
specimens representing all but a few of the weld conditions included in the investigation.
These tests were omitted in the case of a few welds containing the synthetic defects
previously mentioned. Because the significance of the bend tests was questionable, their
results also have been omitted.
With one exception, the joints were tested without reheat treatment although it was
recognized that joint properties could be influenced by the heat of welding as well as by
the weld condition. To obtain some indication of this heat effect, four reduced section

TABLE 8

EFFECT OF DEGREE OF PENETRATION

ON TENSILE STRENGTH OF SQUARE BUTT
WELDED SPECIMENS OF 2219-T87 PLATE*

Tensilet
Strength
S No. Penetration psi

285185 Complete. Roots of two weld beads interpenetrate 42,600

one another to depth of about 1/8".

285187 Moderately incomplete. About 1/16" separation 30,000

between roots· of two weld beads.

285221 Markedly incomplete. About 1/8" separation 26,400

between roots of two weld beads ,

*Welds made in 1/2" plate by TIG-DCSP procedure using 2319 filler wire and one pass on each side
of joint. Test specimens conform with Section IX of ASME Boiler and Pressure Vessel Code, 1962
Edition. Values are averages for two tests.

t Reduced-section specimens.
106

tensile specimens were machined from a 1-in. TIG welded plate of 2219-T87 alloy. Two
of these specimens were reheat treated and aged to the -T62 temper (heat-treated 1%hr
at 1, 000° F, quenched in cold water and aged 36 hours at 375° F) before testing while
the remaining two were tested in the as-welded condition. It was not feasible to restore the
original -T87 temper as this requires a strain-hardening step between solution heat
treatment and aging.
In the case of several panels containing weld craters, some of the tensile blanks were
cut to include craters, and others were cut from portions of the same welds that were
free from craters. The type of reduced-section specimen employed for the tensile tests
is shown in Figure 30.
The results of the tensile tests on specimens from the panels with single- V and
double-V joints are given in Table 7. With two exceptions (S278548-T2 and -T4) these
specimens were tested in the as-welded condition.
It appears unlikely that microporosity has had any significant adverse effect on weld
strengthalthough a completely sound weld was not available for comparison. However,
in certain instances the coarser forms of porosity appeared to affect the strength of the
weld adversely. Weld strength also appears to have been adversely affected by oxide
inclusions and film, tungsten inclusions (particularly in the case of specimen S278486-
T2), lack of interpass, root and side fusion, and incomplete root penetration.
Craters appeared to exhibit an adverse effect on tensile strength in spite of the fact
that the actual crater cavities were removed in machining the specimens. The average
reduced section tensile strengths for specimens with and without craters from 2 welded
panels are as follows:

Tensile Strength, psi

Panel S No .
With Craters Without Craters

278461 24750 33450

278514 14250 33500
278522 32000 38100

The data for the last group (S278548) of specimens listed in Table 7 give a rough in-
dication of the extent to which joint strengths have been influenced by the heat of welding.
The average tensile strength of 27, 700 psi for the as-welded specimens was increased
to 46, 750 psi by reheat treatment and aging to the -T62 temper.
Table 8 gives reduced section tensile test data for some of the TIG welded square butt
j oi.nts in 1/2- in. plate . It is evident that even moderately i ncomplete penetration with a.root
,,;eparation of about 1/ie-in. has had a marked effect in reducing the tensile properties of
this type of joint.

EXAl\ilNATION OF FRACTURES
Examination of the fractures in the tensile specimens helped to confirm the following
weld conditions: incomplete root penetration; lack of root, interpass and side fusion;
linear and scattered porosity and oxide inclusions and film. Photographs of tensile
fractures associated with these conditions are shown in Figures 31 and 32. The frac-
tures shown in Figure 31 are from 1/2-in. panels and those in Figure 32 from 1-in.
panels.
In fracture 8278488 (incomplete root penetration) the sawed surface of the unpene-
trated land is plainly evident. Fracture S278459 (lack of root fusion) shows the wavy
lower edge of weld metal that has not fused into the plate. This metal has taken the
imprint of saw marks on the groove. A fissure extends in from the left side for about
half of the specimen width.
The lack of interpass fusion in S278493 appears as a dark line or shallow fissure
extending almost completely across the fracture about 1/a in. below the top surface.
 107

S278488 S278459
INCOMPLETE ROOT PENETRATION LACK OF ROOT FUSION

S278493 S278597
LACK OF INTERPASS FUSION LACK OF SIDE FUSION

S278492 S278598
LINEAR POROSITY OXIDE

Figure 31. Tensile fractures in specimens from welded t-in, 22l9-T87 plates (natural
size or slightly enlarged).

Lack of side fusion in 8278597 appears as a shallow tapering cavity extending from the
left edge about two-thirds across the fracture. There are two isolated cavities near
the right side. A few small cavities associated with the linear porosity appear in the
lower part of fracture 8278492. The dark surface that covers over one-third of the
fracture in 8278598 had a carbon-like appearance which is probably associated with
finely dispersed oxide.
The first fracture in Figure 32 is associated with incom.plete root penetr ation. The
surface of the unpenetrated land in 8278504 forms a conspicuous band about %-in . wide
extending completely across the fracture. The dark areas just above the horizontal
centerline in fracture 8278511 are cavities associated with the lack of root fusion that
this specimen exemplifies.
In the fracture of 8278516 (lack of interpass fusion) a narrow band that is generally
dark but which contains some small white spots extends horizontally across the fracture
about 5/is-in. above the lower edge . This band represents an unsound region associated
with the lack of interpass fusion. Fracture 8278518 (lack of side fusion) exhibits a nar-
r ow band that is generally dark but which contains small lifht spots and patches. This
band extends across the greater part of the fracture about /1s in. above the lower edge.
It represents an unsound portion of the joint associated with the lack of side fusion. The
scattered porosity in fracture 8278596 is conspicuously evident in a zone about 3/is-in.
extending horizontally across the fracture a short distance below the top edge.
108

S278504 S278511
INCOMPLETE
LACK OF ROOT FUSION
ROOT PENETRATION

S278516
S278518
LACK OF
LACK OF SIDE FUSION
INTERPASS FUSION

5278596 5278549
SCATTERED POROSITY OXIDE FILM
Figure 32. Tensile fractures in specimens from welded l-in. 22l9-T87 plates (natural
size or slightly enlarged).

The white areas just above the horizontal centerline in fracture 8278549 (oxide film)
represent portions of the fracture where an oxide film is believed to have prevented
bonding between the weld metal and the metal of the plate. The weld metal has actually
taken the imprint of saw marks on the groove but did not fuse to it.
 109

Fracture studies and a metallographic examination indicated that the lower strength
of the specimens with craters might be related to a concentric ring pattern of coarse
and fine grains apparent in the fracture and thought to be connected with the crater for-
mation, but further work would be necessary to adequately explain the effect of the craters
on weld strength.

CONCLUSION
Radiographic examinations and ultrasonic tests are valuable and effective procedures
for determining the structural integrity of welds in aluminum alloy structures. The
former procedure gives a more definitive picture of the actual weld condition. Never-
theless, it is probable that ultrasonic examination will gradually replace radiography
for the inspection of long lengths of weld because this method is faster and more readily
automated.
Certain of the discontinuities deliberately introduced into the welds for this investiga-
tion significantly impaired strength. It is important to note, however, that a number
of the specimens were extreme examples of the conditions they represented. Welds
containing limited amounts of certain of the discontinuities might be entirely suitable
for many applications. In setting up rejection limits for radiographic or ultrasonic in-
spection of welds, consideration must be given to the requirements for the application
in which the welds are to be used. Otherwise, the unwarranted rejection of usable as-
semblies may result in a substantial financial loss.

ACKNOWLEDGMENT
The authors express their appreciation to the George C. Marshall Space Flight Center
of the National Aeronautics and Space Administration for permission to publish this
paper which is based on work done on Task Order M-ME-TLA-AL-4 of Contract No.
NAS 8-5132.

Discussion
SIMON A. GREENBERG, Industrial Consultant, Flushing, N.Y.-This paper presents
an interesting initial investigation of the defects which might be encountered in inert gas
welding of aluminum. It also offers a first attempt at the evaluation of the feasibility
of defect detection by nondestructive methods and an evaluation of the effects of such
defects on structural integrity.
Being a beginning, the work reported in this paper does not cover any aspect it treats
in depth. The greatest benefit to be derived from this paper will be from the prompting
it gives to further work in exploring each of the subjects covered separately and in
greater depth.
A separate study of the different techniques of radiography and ultrasonics as applied
to aluminum welds would be most beneficial.
As the authors point out, their study of the effects of weld defects was done with welds
containing unusually large amounts of defects. If a study were made to establish the
margin of acceptance of each type of defect, it should lead to reasonable, valid specifi-
cation acceptance standards. This would permit fullest use of aluminum, yet allow for
screening out those welds which are structurally unsound.
Although the authors surely did not intend it, this paper would serve the structural
field very well indeed if it prompted a study of the effects of weld defects in steel on
their structural behavior.
Our present requirements are probably safe, but they are born of opinion and sub-
jective experience. In some instances they are vague for lack of valid data on which
to base more specific requirements.
 110

- A study of weld defects in steel might permit relaxation of some of the requirements
and answer the clamor of many that this be done-which we cannot do at present for
lack of data on which to base such relaxations. It might even be found that acceptance
standards for some defects should be more stringent for some combinations of service
conditions .
Most important, such studies, for whatever materials they are made, will permit
more efficient use of the material and a closer approach to its full utilization in struc-
tural design and fabrication. Thus, we can closer approach the economic design of
structures, something that is too often overlooked by our designers today.