Fracture and Fatigue Control in Steel Structures

S. T. ROLFE

C O N S I D E R A B L E effort has been devoted to the prevention in the design process, the probability of brittle fracture
of brittle. fracture* in manufactured structures such as incidence in structures of many types would appear to be
aircraft and pressure vessels, where large numbers of es- increasing. Therefore, the designer should become more
sentially identical structures are fabricated under closely aware of the conditions under which brittle fracture may
controlled conditions. For example, the emphasis on safety occur and the available methods for preventing brittle
and reliability of nuclear pressure vessels and the ensuing fractures, particularly in view of the current AISC Code
extensive research, as w^ell as stringent controls, have led of Standard Practice, which assigns responsibility for the
to a situation where the probability of a brittle fracture in suitability, adequacy, or legality of a design.
a nuclear pressure vessel is virtually zero. For other types Almost all large complex steel structures are designed
of manufactured structures, the causes of field failures using structural steels that have yield strengths ranging
usually can be remedied by changes in design of subsequent from 36 to 100 ksi. These steels have inherent levels of notch
units. toughness that generally are adequate for most structural
In contrast, other types of structures, such as bridges and applications. However, the fracture behavior of these
buildings, are often individually designed for a specific structural steels and weldments can be affected significantly
function and location. The overall service experience of by temperature, loading rate, stress level, and flaw size, as
steels in these structures has been excellent, so that the well as by plate thickness or constraint, joint geometry, and
designer in the past has seldom concerned himself with workmanship. The effect of temperature on notch tough-
notch-toughness as a design parameter. However, the trend ness is generally well known, but the roles of stress (or
in structural design has been such that the following strain), flaw size, loading rate, and thickness are less well
changes have occurred. known. In addition, it is possible for the inherent notch
1. Structural engineers and architects are designing toughness of these steels to vary depending upon manu-
more complex structures than in the past. facturing variables (thermo-mechanical history), even
2. There is increased use of high-strength, thick, welded though the steel may meet an existing chemistry or tensile
steel members, as compared with lower-strength, test specification.
thinner, riveted or bolted steel members. From a fracture control viewpoint, therefore, the basic
3. The choice of construction practices has become in- problems are as follows. Is it necessary to specify notch
creasingly dependent on minimum cost. toughness for the steels and weldments used in a particular
class of structures, based on the specific design, fabrication,
4. The magnitude and number of types of loadings
and service conditions to which the structures will be
considered in design have increased. subjected? Furthermore, if notch toughness requirements
Because of the above noted changes, the increasing are necessary, what notch toughness level should be spec-
number of structures subjected to severe loadings (such as ified to ensure satisfactory performance at reasonable cost.
offshore drilling rigs), the use of more precise methods of Also, what joining techniques and fabrication controls are
analysis, and the explicit recognition of inelastic behavior required, consistent with the overall service conditions and
S. T. Rolfe is Professor of Civil Engineering, University of Kansas, consequences of failure. It should be noted that notch-
Lawrence, Kansas. toughness requirements often are developed to be used in
* Brittle fracture is a type of catastrophic failure that usually conjunction with good design, fabrication, and inspection
occurs without prior plastic deformation and at extremely procedures, without being specific as to how "good" pro-
high speeds {crack speeds as high as 7,000 fps or possibly cedures are defined.
more). The fracture is usully characterized by a flat fracture Because the cost of structural steels generally increases
surface (cleavage) with little or no shear lips and at average with their ability to perform satisfactorily under more se-
stress levels below those of general yielding. Brittle fractures vere operating conditions, the designer should not arbi-
are not so common as fatigue, yielding, or buckling failures, trarily specify more notch toughness than is required. How
but when they do occur they may be more costly in terms
much notch toughness is sufficient for a particular struc-
of human life and/or property damage.
tural application is a difficult question to answer, and es-

ENGINEERING JOURNAL / AMERICAN INSTITUTE OF STEEL CONSTRUCTION

tablishing the fracture-toughness requirements and the materials and allowable stress levels is based on the ap-
concomitant quality control and inspection requirements propriate realization of the fact that crack-like discon-
for various structural applications should be an important tinuities in large complex structures may be present or may
design consideration. As with most other aspects of design, initiate under cyclic loading or stress corrosion, and that
it is as much an economic matter as a technical one. some level of notch toughness is desirable.
Over the years many different tests have been used to "Fracture mechanics" is a term commonly used to de-
evaluate the notch toughness of steels. These include the scribe a method of characterizing fracture toughness, fa-
Charpy V-notch (CVN) impact test, the drop-weight tigue crack growth, or stress-corrosion crack growth be-
nil-ductility transition (NDT) test, the dynamic tear (DT) havior in terms of structural design parameters familiar to
test, the wide-plate test, the Battelle drop-weight tear the engineer, namely, stress and flaw size.^ Fracture me-
(DWTT) test and many others.^"^ Generally, these tests chanics commonly is subdivided into two general categories:
were developed for a specific purpose. The CVN test is linear-elastic and elastic-plastic* fracture mechanics. Al-
widely used as a screening test in alloy steel development though linear-elastic fracture mechanics techniques are
as well as a quality-control test. In addition, because of established reasonably well as compared with elastic-plastic
correlations with service experience, the CVN impact test fracture mechanics, most commonly used structural metals
often is used in specifications for alloy steels for various do not behave elastically to fracture and thus linear-elastic
structural and pressure-vessel applications. The N D T test fracture analysis techniques are not directly applicable to
is used to establish the minimum service temperature for most structural steels. This is good, because obviously the
various Navy and structural applications, whereas the engineer wants his materials to exhibit gross structural
Battelle D W T T test was developed to relate the fracture general-yielding behavior rather than a brittle type (lin-
appearance of line-pipe steels to temperature. ear-elastic) behavior.
All these tests generally have one thing in common, Elastic-plastic fracture mechanics approaches are not
namely, to produce fracture in steels under carefully con- yet well-defined and, in fact, no widely accepted simple
trolled laboratory conditions. Hopefully, the results of the analysis technique for this type of behavior is available to
tests can be correlated with service behavior to establish the engineer. Considerable research on elastic-plastic
levels of performance for various steels being considered fracture mechanics is underway. However, the research
for specific applications. However, even if correlations are based approaches are yet to be simplified to the point where
developed for a class of materials and structures, they do they can be widely used by engineering designers, although
not necessarily hold for other designs, new operating con- Crack-Opening-Displacement (COD) test methods have
ditions, or new materials, because the results, which are been used in some areas of fracture analysis for large
expressed in terms of energy, fracture appearance, or structures, for example the Alaska pipeline.
percentage deformation, cannot be translated into normal Although research has shown that numerous factors can
structural design and inspection parameters, namely, stress contribute to brittle fractures in large welded structures,
and flaw size. Fortunately, recent advances in the fracture the recent development of fracture mechanics has shown
mechanics field have led to techniques and concepts which that there are three primary factors (conceptually) that
permit a more rational approach to fracture as a part of the control the susceptibility of a structure to brittle fracture.
design process than was possible in the past. These three primary factors are:

FRACTURE MECHANICS AND DESIGN

1. Material Toughness—Material toughness can be de-
fined as the resistance to unstable crack propagation in
As a general rule the designer must properly proportion the presence of a notch. For linear-elastic behavior the
his structure to prevent failure by tensile overload (yielding
material toughness is measured in terms of a static
or ductile fracture), compressive instability, and by stable
critical stress-intensity factor under conditions of plane
crack growth (for example, arising from fatigue or stress
stress {Kc), of plane strain {Kic), or for dynamic loading
corrosion) or unstable crack growth (brittle fracture).
(Kid). For elastic-plastic fracture behavior, the material
Design to prevent brittle fracture usually refers to using a
toughness may be measured in terms of ductility related
relatively low allowable stress level, as well as to the
parameters such as Jjc, R-curve, COD, and Equivalent
elimination (as much as possible) of those structural details
Energy Approaches as defined below:
that act as stress raisers that can be potential fracture ini-
tiation sites, e.g., certain weld joint details, holes, inter- J'Integral Technique—A path-independent integral
secting plates, arc strikes, etc. Actually, large complex which is an average measure of the elastic-plastic
structures (welded or bolted), cannot be designed or fab- stress/strain field ahead of a crack. For elastic condi-
ricated without some discontinuities, although good design tions, Jic = Kjc^/E{\ — v'^). A test method for this ap-
and fabrication practices can minimize the original size and proach is currently in development.
number of these discontinuities. It is realized that stress
* Sometimes referred to as ''general yielding", particularly in
concentrations or discontinuities will be present, but the the British literature. The term ''elastic-plastic" connotates
designer assumes that his structural materials will yield the situation where a significant yield zone relative to plate
locally and redistribute the load in the vicinity of these stress thickness of inelastic straining occurs near the crack tip such
concentrations or discontinuities. The selection of structural that the linear-elastic analyses are not applicable.

FIRST QUARTER / 1977

 Resistance-Curve {R-Curve) Analysis—A procedure
used to characterize the resistance to fracture of mate-
O = Stress in Structural Configuratio
rials during incremental slow-stable crack extension,
KR. At instability KR = Kc, the plane stress fracture-
toughness which is dependent upon specimen thick-
ness, as well as temperature and loading rate.
Crack-Opening Displacement {COD) Technique—
Toughness evaluation in terms of the pre-fracture de-
formation at the tip of a sharp crack that shows consid-
erable potential as a fracture criterion; a proposed test
method has been developed by the British Standards
Institution.
Equivalent Energy Approach—An energy approach
based on using test results to predict failure, primarily
of thick walled pressure vessels.
2. Flaw Size—Brittle fractures initiate from flaws or
discontinuities of various kinds. These discontinuities
can vary from extremely small cracks, for example, from
within a weld arc strike (as was the case in the brittle
fracture of a T-2 tanker during World War II), to much
larger weld or fatigue cracks. Even though only small
flaws may be present initially, repeated loading (fa-
tigue), or stress corrosion can cause them to enlarge,
possibly to a critical size where brittle fracture can
occur.
3. Stress Level—Tensile stresses (applied, residual, or
both) are necessary for brittle fractures to occur.
Engineers have known the foregoing facts for many years
and have reduced the susceptibility of structures to brittle
fractures by applying these concepts to their structures,
Fig. 1. Elastic-stress-field distribution ahead of a crack
qualitatively. That is, good design (the use of lower stress
levels and the minimizing of discontinuities) and sound
fabrication practice (decreased flaw size through use of
proper welding procedures and control), as well as the use relation between Kj, a, and a for different structural con-
of materials with good notch toughness levels (e.g., as figurations, as shown in Fig. 2.^ Other crack geometries
measured with a Charpy V-notch impact test), have min- have been analyzed for different structural configurations
imized the probability of occurrence of brittle fractures in and are published elsewhere.^'^^
structures. However, the engineer has not had techniques If the critical value of Kj at failure (Kc, Kj^, or Kid) can
available to permit evaluation of the relative performance be determined for a given metal of a particular thickness
and economic trade-offs between design, fabrication, and and at a specific temperature and loading rate, the designer
materials in a quantitative manner prior to the development can determine theoretically the flaw size that can be tol-
of fracture mechanics. erated in structural members for a given design stress level.
The fundamental concept of linear-elastic fracture me- Conversely, he can determine the design stress level that
chanics is that the stress field ahead of a sharp crack can be can be safely used for a flaw size that may be present.
characterized in terms of a single parameter, Kj, the stress This general relationship between material toughness
intensity factor for flat crack propagation (usually referred Kic or Kc, nominal stress a, and flaw size A is shown
to as opening mode), having units of ksiA/in. This single schematically in Fig. 3. If a particular combination of stress
parameter Kj is related to both the stress level, c, and the and flaw size in a structure (AT/) reaches the Kjc or Kc level,
flaw size, a. When the particular combination of a and a fracture can occur. Thus, there are many combinations of
leads to a critical value of AT/, called Kjc or Kc, unstable stress and flaw size (e.g., oy and aj), that may cause fracture
crack growth occurs. The equations that describe the in a structure that is fabricated from a steel having a par-
elastic-stress field in the vicinity of a crack tip in a body ticular value of Kjc or Kc at a particular service tempera-
subjected to tensile stresses normal to the plane of a simple ture, loading rate, and plate thickness. Conversely, there
crack are presented in Fig. 1. These stress-field equations are many combinations of stress and flaw size (e.g., (JQ and
define the distribution of the elastic-stress field in the vi- ao), that will not cause failure of a particular steel, i.e.,
cinity of the crack tip, and can be used to establish the below the Kjc or Kc line.

ENGINEERING JOURNAL / AMERICAN INSTITUTE OF STEEL CONSTRUCTION

 VMM W//A
0 THROUGH THICKNESS CRACK

tin. Kj - o X / i r r

Kj. = critical value of K

~\ SURFACE CRACK

K - l . l V ^ oVa/Q
Increasing Flaw S i z e , 2 a

H_h2c WHERE Q - f(a/2c)

Fig. 3. Schematic relationship between stress, flaw size, and
material toughness
ITTTT'O
p C o l u m n Research C o u n c i l
C o l u m n Strength C u r v e
EDGE CRACK

Kj -1.12 dv-rr a

TTTT,
i^2§-. 2. A"/ values jor various crack geometries

A useful analogy for the designer is the relation between

applied load P , nominal tensile stress c, and yield or limit
stress Oys in an unflawed structural member, and between
applied load P , stress intensity Ki, and critical stress in-
tensity for fracture Kc, Kj^, or Kj^ in a structural member
with a flaw. In an unflawed structural member, as the load
is increased, the nominal stress increases until a limit
loading (yielding) occurs. As the load is increased in a
structural member with a flaw (or as the size of the flaw
grows by fatigue or stress corrosion), the stress intensity Kj Fig. 4. Column instability and crack instability
increases until a limit condition (fracture at Kc, Kjc, or {after Madison, Ref. 12)
Kid) occurs. Thus, the Kj level in a structure should always
be kept below the appropriate Kc value, just as the nominal
design load is always kept below the limit loading.
Another analogy that may be useful in understanding ness will increase the possible combinations of design stress
the fundamental aspects of fracture mechanics is the com- and flaw size a structure can tolerate without fracturing.
parison with the Euler column instability curve, Fig. 4.^^'^^ At this point, it should be emphasized that the Kc levels
The stress level required to cause instability in a column for most common structural steels are so high that they
(buckling) decreases as the L/r ratio increases. Similarly, cannot be measured directly using existing ASTM
the stress level required to cause instability (fracture), in standardized test methods.^^ Thus, although concepts of
a flawed tension member decreases as the flaw size a in- fracture mechanics can be used to develop fracture-control
creases. As the stress level in either case approaches the guidelines and desirable toughness levels, the state-of-
yield strength, both the Euler analysis and the Kc analysis the-art is such that actual Kjc or Kc values cannot be
are invalidated because of yielding. To prevent buckling, measured for most commonly used structural metals at
the actual stress and L/r values must be below the Euler service temperatures. Therefore, traditional notch tough-
curve. To prevent fracture, the actual stress and flaw size ness tests (e.g., CVN, N D T , etc.) are widely used at the
a must be below the Kjc or Kc level shown in Fig. 4. Ob- present time to specify the notch toughness requirements
viously, using a material with a high level of notch tough- for various structural applications. Examples of the use of

FIRST QUARTER / 1977

 Impact Loading
Plastic

C a, 0)
EULJ jJ
Elastic-
Plastic

0-5 j2
Macro Linear
Elastic (Often
Designated as
"Plane Strain"
Behavior Region'

Temperature •

Fig. 5. Schematic showing relationship between notch toughness test results

and levels of structural performance for various loading rates

such test methods in specifications are the recently devel- ticular structure an engineer may be designing. Also, se-
oped AASHTO material toughness requirements for lection of a fracture criterion alone, without considering the
bridge steels and the ASME toughness requirements for other factors involved in fracture control, will not neces-
steels for nuclear vessels. In both of these cases, concepts sarily result in a structure with the desired margin of safety.
of fracture mechanics were used to develop the desired An example of the use of a fracture toughness criterion
toughness requirements, but the actual material toughness developed for one application but also widely used in many
requirements are in terms of CVN or N D T values based other situations is the 15 ft-lb CVN impact criterion at the
on empirical correlations.^ minimum service temperature, which was established on
the basis of the World War II ship failures. This criterion
has been widely used for various types of structures, even
FRACTURE CRITERIA
though the material, service conditions, structural redun-
A fracture criterion is a standard against which the expected dancy, etc., may be quite different from those of the World
fracture behavior of a structure can be judged. In general War II ships for which the criterion was established.
terms, fracture criteria are related to the three levels of Criteria selection should be based on a careful study of
fracture performance as shown in Fig. 5, namely macro the particular performance requirements for a given
linear-elastic (often referred to as ''plane strain" in the structure. The factors involved in the development of cri-
fracture mechanics literature), elastic-plastic, or fully teria commonly include:
plastic. Although it would appear desirable to specify fully
plastic behavior, this is rarely done because it is almost 1. Service conditions (loadings, temperature, controlling
always unnecessary, as well as being economically unde- stress and strain levels, loading rate, cyclic loading,
sirable in many cases. etc.) to which the structure will be subjected.
For most structural applications, some moderate level 2. Desired level of performance and margin of safety of
of elastic-plastic behavior at the service temperature and the structure under both normal service and extreme
loading rate constitutes a satisfactory performance criterion. loading conditions.
While there may be some cases where considerable inelastic 3. Possible modes and consequences of failure.
behavior is necessary (e.g., dynamically loaded military
There is no single fracture criterion that can be applied
protective structures), or where low toughness level be-
to all structures, because optimum design involves economic
havior can be tolerated, (e.g., certain short-life aerospace
considerations as well as technical trade-offs.
applications where the loading and fabrication can be
At the present time it is difficult to establish notch
precisely controlled), for the majority of large complex
toughness criteria for the following reasons:
structures such as bridges, ships, buildings, pipelines, off-
shore drilling rigs, etc., some moderate level of elastic-plastic 1. Establishment of the specific level of required notch
behavior at the service conditions is satisfactory. The toughness (i.e., the required CVN, Kjc, or K^ value
question arises then as to what level of elastic-plastic be- at a particular test temperature), is costly and time
havior is required and how this level of performance can consuming, and is a subject unfamiliar to engineers.
be insured. Furthermore, it depends on many factors such as the
Unfortunately the selection of a fracture criterion is often particular service loadings, design of structural de-
quite arbitrary and based on service experience for other tails, quality of fabrication, inspection, etc., which are
types of structures that may have no relation to the par- difficult to establish.

ENGINEERING JOURNAL / AMERICAN INSTITUTE OF STEEL CONSTRUCTION

 2. There is no well-recognized single "best" approach. for a particular application should be that which most
Therefore, different experts will have different closely models actual structural behavior. However,
opinions as to the "best" approach, although the commonly the selection of the test specimen is based on
science of fracture mechanics is slowly helping to past experience as well as economics of testing.
overcome this difficulty.
2. Specific Notch Toughness Value or Values—The
3. The cost of structural materials increases with in-
second and more difficult part of establishing a fracture
creasing levels of inherent notch toughness. Thus
criterion is the selection of the specific level of perfor-
economic considerations as well as technical ones
mance using a particular test specimen. The specified
must be included when establishing a toughness cri-
values in any criterion should be an optimization of both
terion. However, any design criterion includes eco-
safe structural performance and cost, and depend to a
nomic as well as technical considerations.
large degree on the design, quality of fabrication, in-
A general fracture criterion defined in terms of the lev- spection, and loading for the particular structure.
els of performance (linear-elastic, elastic-plastic, and
plastic), as described in Fig. 5, must be translated into some EFFECTS OF TEMPERATURE, LOADING RATE,
specific fracture test requirement that insures the desired AND THICKNESS
level of performance. For example, a general requirement
that a structural material exhibit elastic-plastic behavior In general, the notch toughness of most structural steels
at service temperatures is a general criterion that is useful increases with increasing temperature and decreasing
to the engineer. However, because of ambiguity and dif- loading rate. The effect of temperature is well known and
ferences in opinion, this general criterion must be made has led to the transition-temperature approach to designing
specific in terms of a fracture test specimen and some to prevent fracture. However, the effect of loading rate may
specified index value. An example of a general toughness be equally as important, not only in designing to prevent
criterion might be that a low level of elastic-plastic behavior fracture, but in understanding the satisfactory behavior of
is required under the most severe expected service condi- many existing structures built from materials that have low
tions. The specific toughness criterion for this example impact toughness values at their service temperatures.
might be that "all structural steels and weldments used in The general effects of temperature and loading rate on
this assemblage must exhibit 21 ft-lb energy absorption as Kjc and Charpy V-notch behavior are shown schematically
measured in a standardized longitudinal Charpy V-notch in Figs. 6 and 7. The toughness of most structural steels
impact test specimen tested at 32''F." Hopefully, this
particular criterion would have been based on sufficient
laboratory results, service experience, and fracture me- Dynamic Kj^ (Kj^)
Slow Bend K
chanics analysis to insure that the desired structural be-
havior is consistent with economic considerations. The Minimum Service
criterion would then be specified for purchase of materials Temperature for
Dynamic Loading
and quality control during fabrication.
As a result of several large structural failures in the pe-
riod 1967-1972,^'^-i^ as well as a growing concern with the
Temperature
overall reliability and safety of structures, many specifi-
cations are now beginning to include specific minimum Fig. 6. Schematic showing effect of temperature and
toughness requirements. This trend is expected to grow as loading rate on Kjc
regulatory governmental agencies become increasingly
active in the development of mandatory fracture prevention
i
criteria. Recent examples are the ASME Nuclear Code,^^
Ft-Lb shift
AASHTO Material Toughness Requirements,^^ and the in upper-
floating nuclear power plant hull structure toughness re- t; shelf level
quirements^ imposed by the United States Coast Guard Z Slow-bend ->^
1 '
and the Nuclear Regulatory Commission. >
c
There are two general parts to a fracture criterion: >s Temperature 1 \ - Impact
ro shift in
c transition-
1. General Test Specimens to Categorize the Material temperature
~o range
Behavior—Throughout the years, various fracture _Q fl
/ I
criteria have been specified using notch toughness tests 8 / 1 ^Energy level at
^ \ which temperature
such as CVN impact, N D T , D T , and, more recently, ( shift is measured
the fracture mechanics test specimens used to measure
Kic and K^. Test specimens currently used as research Test Temperature -
tools and expected to be used more extensively in the
future for metals in the lower yield strength category are Fig. 7. Schematic representation of shift in CVN transition
Jic, COD, and R-curve specimens. The specimen used temperature and upper-shelf level due to loading rate

FIRST QUARTER / 1977

 various yield strengths has been related to the room-tem-
perature yield strength of the steel and can be approximated
Slow-bend Load (€=-10*5 sec. -I) by the following equations:
Intermediate Strain-rate Load (* =• 10-3 sec. "1)
Dynamic Load (^ -10 sec. -1)
Tshift = 2 1 5 - 1 . 5 (jys (for 36 ksi < Uys < 140 ksi)
60 r,/../, = 0.0 (for(T^,>140ksi)

where
Tshift ^ absolute magnitude of the shift in the transition
temperature between slow-bend loading and
rapid dynamic loading, degrees F
(jys = room-temperature yield strength, ksi
g 30 Because of this shift, increasing the loading rate can de-
crease the fracture-toughness value at a particular tem-
perature for steels having yield strengths less than 140 ksi.
The change in fracture toughness values for loading rates
varying from slow-bend to dynamic rates is particularly
important for those structural applications that are loaded
slowly, such as bridges.
J_
As a specific example of the use of the loading-rate shift
-400 -300 -200 -100 0 NOT 100 in the development of fracture criteria, assume that a
Temperature, F structure is loaded at a slow-loading rate of 10~^ in./in./sec
and that the fracture toughness of the material is as shown
Fig. 8. Effect of temperature and loading rate on fracture
toughness of A36 steel
in Fig. 9. If stress-flaw size calculations show that a Kjc
value of about 60 ksiVTiL would insure satisfactory
structural performance, the results presented in Fig. 9 show
tested at a constant loading rate undergoes a significant that this behavior can be obtained at about +40°F dy-
increase with increasing temperature. Thus, the general namically (€ = 10 in./in./sec), at about —90°F at an in-
effect of a slow loading rate, compared with impact loading termediate loading rate, and at about — 150''F for a slow-
rates, is to shift the fracture-toughness curve to lower loading rate.
temperatures, regardless of the test specimen used. Ex- Since it is usually much easier and less expensive to
amples of this shift in behavior with loading rate are pre- conduct impact (dynamic) tests than intermediate-loading
sented in Figs. 8 and 9 for an A36 structural steel and an rate tests, criteria can be established on the basis of one
A572 Grade 50 structural steel, respectively. loading rate, d d the results "shifted" on the basis of a
The magnitude of the temperature shift between slow- laboratory test conducted at a different loading rate. The
bend loading and very rapid dynamic loading in steels of recently developed American Association of State Highway
and Transportation Officials (AASHTO) material
toughness requirements were based on this reasoning.^^ It
should be emphasized that this criterion can only be used
A572 Grade 50 Steel ( a ^ = 50 ksi)
with those materials that exhibit a shift in transition be-
havior with changes in loading rate. The magnitude of this
- • Slow-bend load {k ^ 10"^ sec"!)
• Intermediate strain-rate load (e== 10- •3 sec-1) shift can be considerable and helps to explain why many
^ Dynamic load (^ = 10 sec"') structures operate successfully at service temperatures well
^ below their "dynamic" transition temperature.
B = 0.4 S = 0.4
Qualitatively, the effect of increasing specimen or plate
. 60 - \ • / "^ 6=0.4^^ thickness is to promote a more severe state-of-stress,
• / namely, plane strain. A triaxial state-of-stress occurs at the
/ • tip of a sharp discontinuity in a thick plate and this reduces
AJ the apparent ductility of the material to a lower-bound
• ^- value. Conversely, the apparent fracture toughness of
20 - • materials can increase with decreasing plate thitkness, as
NDT
a result of the relaxation of the lateral constraint in the
vicinity of the notch tip. This apparent increase in tough-
t i l l 1 1 1 1
ness is controlled solely by the thickness of the plate, even
-300 -250 -200 -150 -100 -50
Temperature, F
though the inherent metallurgical properties of the material
remain unchanged. Thus, the minimum toughness of a
Fig. 9. Effect of temperature and loading rate on fracture particular material occurs at specimen thicknesses large
toughness of A572 Grade 50 steel {ays = 50 ksi) enough so that the state-of-stress is plane strain.

ENGINEERING JOURNAL / AMERICAN INSTITUTE OF STEEL CONSTRUCTION

 FRACTURE CONTROL 3. Determination of the relative efficiency and trade-offs
of various design methods to minimize the possibility
The objective in structural design of large complex struc-
of brittle fracture in a member or failure of the
tures such as bridges, ships, pressure vessels, aircraft, etc.,
structural system.
is to optimize the desired performance requirements relative
to cost considerations (i.e., the overall cost of materials, 4. Recommendation of specific design considerations
design, fabrication, and operation), so that the probability to ensure the safety and reliability of the structure
of failure (and its economic consequence) is low. To achieve against brittle fracture. This would include recom-
these objectives, engineers make predictions of service loads mendations for desired levels of material perfor-
and conditions, calculate stresses in various structural mance, as well as material selection, design stress
members resulting from these loads and service conditions, levels, weld performance, design of details, fabrica-
and compare these stresses with the critical stresses in the tion, inspection, and maintenance.
particular modes that may lead to failure of the structure. For those cases where crack growth is a possibility, the
Various criteria are then selected so that failure does not total useful design life of a structural component can be
occur by any of the pertinent failure modes. estimated from the time necessary to initiate a crack plus
Possible failure or limit modes usually considered are: the time to propagate the crack from sub-critical dimensions
to the critical size. The life of the component can be pro-
1. General yielding or excessive plastic deformation
longed by extending the crack-initiation life and/or the
(straining)
sub-critical-crack-propagation life. Consequently, the
2. Buckling or general instability, either elastic or crack-initiation, sub-critical-crack-propagation, and un-
plastic stable-crack-propagation characteristics of structural
3. Sub-critical crack growth (through fatigue, stress materials, as well as their fracture behavior, are primary
corrosion or corrosion fatigue), leading to loss of considerations in the formulation of fracture-control
section or unstable crack growth guidelines for structures. Unstable crack propagation is the
4. Fracture, either ductile or brittle, leading to either final stage in the useful life of a structural component
partial or complete failure of a member subject to failure by the fracture mode. This stage is gov-
Although other failure modes exist, such as general cor- erned by the material toughness, the crack size, and the
rosion or creep, the above mentioned failure modes are the stress level. Consequently unstable crack propagation
ones that usually receive the greatest attention. Further- cannot be attributed only to low material toughness, or only
more, of these four failure or limit modes, structural en- to high stress levels, or only to poor fabrication, but rather
gineers usually concentrate on only the first two and assume to particular combinations of all the above factors. How-
that proper selection of materials will prevent the other two ever, if any of these factors is significantly different than
failure modes from occurring. This reasoning is not always that which is normally found in a particular type of
true and has led to several large structural failures. In a structure, experience has shown that for most structures
complete structural design, all possible failure modes should the possibility of failure is generally increased.
be considered. Figure 10 is a schematic showing the three stages of total
In the case of brittle fracture or fatigue, many of the life behavior (crack-initiation, sub-critical-crack-propa-
fracture-control guidelines that have been followed to gation by fatigue, and unstable-crack-propagation, either
minimize the possibility of brittle fractures in structures
are familiar to structural engineers. These guidelines in-
clude the use of structural materials with good notch
toughnesSj^elimination or minimization of stress raisers,
control of welding procedures, proper inspection, etc. When Flaw Size,
these general guidelines are integrated into specific re-
quirements for a particular structure, they, become part of
a fracture-control plan. A fracture-control plan is therefore
a specific set of recommendations developed for a particular
structure and should not be indiscriminately applied to
other structures.
The four basic elements of a fracture-control plan
are:
1. Identification of the factors that may contribute to the
brittle fracture of a structural member or to the failure
of an entire structure; includes description of service
conditions, loadings, and/or deformations. Numbers of Cycles of Loading

2. Establishment of the relative contribution of each of

these factors to a possible brittle fracture in a member Fig. 10. Schematic showing three regions in the total life of a
or to the failure of the structure. structure subjected to fatigue loading

FIRST QUARTER / 1977

by rapid fatigue crack growth, ductile tearing, or fracture). Reduce Tensile Design Stress—Large effect on life
The question of when does a crack "initiate" to become a (Region I) because the rate of fatigue crack growth is
"propagating" crack is somewhat philosophical and de- decreased significantly as the applied stress range is
pends on the level of observation of a crack, i.e., crystal decreased (cri curve compared with (J2 curve). Design
imperfection, dislocation, microcrack, macrocrack, etc. In stress range {(Tmax "~ (ymin) is the primary factor to
an engineering sense, the initiation stage is that region in control.
which a very small initial discontinuity or crack grows to
become a measurable propagating crack in fatigue. The Reduce Initial Flaw Size—Large effect on life (Region
sub-critical-crack-growth stage is that region in which a II) because the rate of fatigue crack growth for small
propagating fatigue crack follows one of the existing flaws is very low. Quality of fabrication and inspection
crack-growth laws,^ e.g., da/dN = ^(AAT)^. The unstable is the primary factor to control.
crack-growth stage is that region in which either fatigue
crack growth is very rapid, or a brittle fracture occurs, or
ductile tearing occurs. All three situations in the unstable Increase Material Toughness—
crack-growth stage result in loss of section and failure oc- a) Large effect on life in moving from plane strain be-
curs very quickly, although failure by ductile tearing is havior to elastic-plastic behavior (Region III). The
usually preceded by large deformations. AASHTO Material Toughness Requirements for
Figure 11 illustrates the effect of tensile stress level, flaw bridge steels insure this level of performance under
size, and material toughness (the three primary factors that intermediate rates of loading.
control susceptibility to brittle fracture) on the life of a
structure subject to fatigue loading. Note that these factors b) Small effect on life in moving from elastic-plastic
are related to the three levels of performance, i.e., plane behavior to plastic behavior (Region IV), because the
strain, elastic-plastic behavior, and plastic behavior, dis- rate of fatigue crack growth becomes so large that
cussed earlier (see Fig. 5). The following observations may even if the critical crack size (^^r) is increased sig-
be made concerning the effectiveness of these control factors nificantly, the effect on the remaining fatigue life is
in improving service life: small. Failure mode may change, however.

(Improvement in Life Due

to Large Improvement
n Notch Toughness)

(Improvement in Life Due to

Smaller Initial Flow Size)

Number of Cycles of Fatigue Loading

Fig. 7 /. Schematic showing effect of notch toughness, stress range, and flaw size
on improvement of life of a structure subjected to fatigue loading

10
ENGINEERING JOURNAL / AMERICAN INSTITUTE OF STEEL CONSTRUCTION
 For most structural applications subject to fatigue, The distinguishing feature is whether or not, in the
some moderate level of elastic-plastic behavior at event of fracture of a primary structural member, the
service temperature and loading rate constitutes a design load can be transferred to other members, whether
satisfactory performance criterion. However, un- they are initially classified as primary or secondary. If
usually severe service requirements may require so, the structure is multiple-load-path; if not, it is sin-
special material toughness standards. In extreme gle-load-path. In this sense, multiple-load-path struc-
cases, structural material is required vv^hose notch tures are usually more resistant to failure than single-
toughness is such that the material does not fail by load-path structures. For example, if a single member
brittle fracture even under the most severe operating fails in a single-load-path structure, the entire structure
conditions to w^hich the structure may be subjected. may collapse, as occurred with the Silver Bridge at Point
The use of HY-80 or HY-130 steels for submarine Pleasant, W. Va.^"^ Conversely, if a single member fails
hull structures is an example of this method. How^- in a multiple-load-path structure, the entire structure
ever, as shown in Fig. 11, this method is not very ef- probably will not collapse. This type of behavior was
fective in increasing the life of structures subjected demonstrated in the failure of the Kings Bridge in
to fatigue loading, such as bridges. However, the Australia. ^^ At the instant of failure, the failed span in
increased notch toughness certainly is a desirable the Kings Bridge contained three cracked girders. One
property and does result in the change of failure mode girder had cracked while still in the fabrication shop. A
from brittle to ductile fracture. second failed during the first winter the bridge was
opened to traffic, a full 12 months before the failure of
Although the above three methods are the basic design the bridge. Failure of a third girder led to final failure,
approaches to the control of brittle fracture in most struc- although architectural concrete sidewalls (which added
tural applications, there are other design methods which to the multiple-load-paths of the overall structure)
can minimize susceptibility to, or the consequences of, prevented complete collapse. Similar examples of the
brittle fracture (should the phenomenon occur). These importance of multiple-load-paths can be cited.
other design methods include: Therefore, lower notch toughness can be used in
A) Provide multiple-load paths or structural redundancy, members of multiple-load-path structures than in
so that a single fracture cannot lead to complete failure members of single-load-path structures if a constant
of the structure. From a fracture behavior viewpoint, factor of safety is to be provided for the structure.
multiple-load path structures are different from re- Moreover, fatigue-crack propagation in multiple-
dundant structures. A redundant structure is one in load-path or redundant structures occurs essentially
which the laws of statics are insufficient to solve for the under constant maximum deflection, which corresponds
loads and stresses, and thus the structure is indetermi- to a decreasing stress-field intensity.
nant from an analysis viewpoint. If a single member Thus, cracks propagating in multiple-load-path or
fractures, one degree of redundancy may be removed redundant structures may eventually arrest and, al-
(e.g., an effective hinge may be formed), but the structure though individual structural components will have to be
is still stable. replaced or repaired, complete failure of the structure
A multiple-load-path structure is defined by the is not expected to occur as long as sufficient redistribution
particular geometry of the members used to make up the of load can occur.
structure. For example a simply-supported single-span B) Provide crack arresters so that, in the event that a crack
bridge structure is determinant (non-redundant) because should initiate, it will be arrested before catastrophic
the reactions can be determined by the laws of statics. If failure occurs. Crack arresters or a fail-safe philosophy
the geometry of this single-span determinant structure (i.e., in the event of "failure" of a member, the structure
is a single wide-flange shape such that failure of the is still "safe") have been used extensively in the aircraft
single tension flange leads to collapse of the bridge, then industry, as well as in the shipbuilding industry.
the structure is also a single-load-path structure.
However, if the geometry consists of eight independent C) Insuring that the loading rate is slow is an effective
wide-flange shapes with a concrete deck, then the method of fracture control. Many structures are loaded
structure is a multiple-load-path structure and is much at slow to intermediate loading rates, where their notch
more resistant to complete fracture than the single toughness is quite satisfactory on the basis of the load-
member structure. Lateral bracing of girders and trusses ing-rate shift. This leads to an understanding of why
also provide multiple-load paths in the event of failure there are so few brittle fractures in older structures, even
of a primary structural member. Another example is a though the notch toughness of the steels in these struc-
truss member composed of multiple shapes (e.g., 4 to 10 tures would be considered to be very low on the basis of
eye-bar members parallel to each other), as opposed to impact loading-rate tests. Thus, if the structure can be
one structural shape (e.g., a wide-flange shape in ten- designed such that it is loaded slowly, so that the con-
sion). The former is a multiple-load-path member, while trolling toughness parameter is Kjc rather than Kjd, the
the latter is a single-load-path member. possibility of fracture is reduced considerably.

11
FIRST QUARTER / 1977
AASHTO MATERIAL TOUGHNESS REQUIREMENTS
The recently developed AASHTO material toughness A572 G r a d e 50 Steel I t 44
requirements for bridge steels are based on the observation • Impact
75 75
that the maximum loading rates in bridges^^'^^*^"^ are closer 50 h _ • Intermed iate-stra in-rate
behavior predicted by
to slov^-bend loading rates than to impact loading rates. shifting the impact data %
A Intermediate-strain- •
• • w
In fact, the loading times in bridges are greater than 1 rate behavior predicted % • 7
40 — from K^^ data tested ^ ^
second, which corresponds to a strain rate of less than 10~^ ••
sec~^ on the elastic-plastic boundary in the vicinity of a
at e = 1 0 - 2 sec"^ ^ f ^
• • mV . •
crack tip. Thus, a strain rate of 10~^ sec~^ (intermediate •1 • • / •
9- 30 h • •J • ••I •
loading rate) is used as a conservative measure of the /•• • #•
maximum strain rate for bridges. :/•• • /
20 h A/
The AASHTO toughness requirements for bridge steels,
which are based on testing at temperatures above minimum / • / •
service temperatures and at impact CVN strain rates, are 10 •/•• •/*•
such that the transition from plane strain to elastic-plastic . I • (•
behavior under intermediate loading rates will occur below ^ J^ NOT

the minimum service temperature.

1
-200
1
-150
1
-'100
_J
-50
__L
0 _±^ 50 100
L _J
150
The Kic data obtained by testing A36 and A572 Grade Temperature, F
50 steels at a strain rate of 10~^ sec~^ (intermediate loading
rate) indicate that fracture does not occur under plane- Fig. 12. Charpy V-notch energy absorption behavior for
strain conditions when the test temperature is greater than impact loading and intermediate rate loading of standard
about —80°F (see Figs. 8 and 9).25'26 Because Kjc tests are CVN test specimens
expensive and difficult to conduct, and because of the ap-
parent correspondence between Kic test results and CVN
test results,^^'^^'^^ the CVN test was selected as the refer- The 15-ft-lb CVN impact toughness requirement at
ence test for the AASHTO fracture-toughness require- +40°F for steels of 50 ksi yield strength or less was based
ments. The fracture-toughness transition temperature is on a —30° F minimum operating temperature. The pre-
the temperature at which the fracture toughness of the steel ceding procedure can be used to develop toughness re-
begins to increase rapidly from plane-strain behavior to quirements for any minimum operating temperature. The
fully ductile behavior. The CVN test results, Fig. 12, show resulting toughness requirement for 50 ksi yield strength
that this transition behavior at 15 ft-lbs under intermediate steels is 15 ft-lbs in an impact test at a test temperature that
rates of loading at —80°F should ensure a non-plane-strain is 70° F higher than the specified minimum operating
fracture behavior at a minimum operating temperature of temperature. Thus, the CVN test temperatures and the
—30°F for a 50 ksi yield strength steel. minimum operating temperatures are linearly related. To
Although the intermediate-loading-rate test is the test minimize the proliferation of a variety of testing temper-
that more properly describes the expected service perfor- atures, and the resulting problems in the design and fab-
mance of bridge steels, the standard impact-loading-rate rication of steel bridges, the variable testing temperatures
CVN test is much easier to conduct and analyze and is less were comprehended by establishing three zones of service
expensive than an intermediate-loading-rate CVN test. temperatures and providing temperatures and CVN impact
Consequently, the difference in fracture-toughness behavior values for each zone. The three zones of service tempera-
at the two strain rates was used to develop the toughness tures and the corresponding test temperatures and mini-
values in terms of the impact test rather than an interme- mum toughness values for bridge steels are presented in
diate-loading-rate test. The temperature shift between the Table 1.
CVN (and Kjc) curves of a 50 ksi yield strength steel tested The general relationships between service temperatures
at a strain rate of 10"-^ sec~^ and at an impact strain rate and test temperatures for A36 steel satisfying the re-
of 10 sec~^ (10,000 times greater) was on the order of quirements of each of the three service-temperature zones
120°F (see Fig. 9). Consequently, the requirement of a 15 are shown in Fig. 13. These results show that, because of
ft-lb CVN impact value at +40° F corresponds to a 15 ft-lb the loading-rate shift, CVN-toughness levels greater than
CVN value under an intermediate strain-rate at --80° F, 15 ft-lbs are expected at intermediate loading rates ap-
which in turn corresponds to a non-plane-strain fracture proximately 70° F below the impact testing temperature.
behavior at an assumed minimum operating temperature In terms of the N D T temperature measured using drop
of — 30°F. Thus, a CVN fracture-toughness requirement weight test specimens subjected to impact loading, the
of 15 ft-lbs at +40° F was imposed on all primary member minimum service temperature is approximately 70°F
components in tension and of 50 ksi yield strength steels for below NDT.
bridge applications. This same requirement was also im- The specifications of the American Society for Testing
posed on all primary member components in tension for and Materials (ASTM) for A572 Grade 50 and A588
bridge steels having yield strengths less than 50 ksi, which steels require a minimum yield strength value of 50 ksi.
is a conservative requirement for these steels. Consequently, these steels as actually produced may have
12
ENGINEERING JOURNAL / AMERICAN INSTITUTE OF STEEL CONSTRUCTION
 Class I significantly greater than 50 ksi, the CVN test temperature
(>0°F)
30 r
Specification
Impact
was decreased incrementally as the yield strength increased.
Slow Intermediate
The CVN requirements of 15 ft-lbs at + 7 0 ^ 4-40° and
120°F + 10°F for Zones 1, 2 and 3, respectively, were restricted
15
to A572 Grade 50 and A588 steels having yield points be-
|Servi<^ -NDT
I Temp. tween 50 and 65 ksi, inclusive. When the yield strength of
-160 -120 -80 -40 80 these steels exceeds 65 ksi, the temperature for the CVN
Class II value for acceptability was reduced by 15°F for each in-
|(-30°F to -1°F)
30 r Impact crement of 10 ksi above 65 ksi.
Slow Intermediate
The above philosophy, which is based on fracture-
15 L ^/^^/ 120°F mechanics concepts, was used to develop toughness re-
quirements for bridge steels of 100 ksi yield strength
I jTemp .|
_L_
~ NDT
_i_ (ASTM A514 and A517). These steels show a temperature
z
> shift of 60° F between static and impact loading rates (Figs.
14 and 15). Moreover, increasing the design stress (which
generally requires a higher yield strength steel) results in
more stored elastic energy in a structure. Thus, the fracture
toughness of the steel should also be increased to ensure the
same degree of safety against fracture as the structure with
the lower design stress. The resulting fracture-toughness
requirements for high-strength bridge steels also are pre-
sented in Table 1.
In summary, the basis for the AASHTO material-
Fig. 13. AASHTO material toughness specifications toughness specification, which is in the recently developed
for A36 steel
ASTM A709 Standard Specification for Structural Steel
for Bridges, is fracture mechanics. However, the desired
yield strengths that are higher than 50 ksi. The data in Fig. level of performance is outside the range of linear-elastic
14 show that the magnitude of the temperature shift be- fracture-mechanics behavior at the service temperatures
tween static and impact loading rates decreased with in- and loading rates for bridges. Thus, Kjc values cannot be
creased yield strength. The magnitude of the decrease in measured directly by existing fracture-mechanics tests, and
the temperature shift is about 15°F for every 10 ksi increase correlations between Kjc and CVN test results were used
in yield strength. To ensure the same fracture behavior for to establish the material-toughness requirements shown
A572 Grade 50 and A588 steels having yield strengths in Table 1. These toughness requirements depend on the

Table 1. AASHTO Fracture-Toughness Specifications for Bridge Steels*

Energy Absorbed (ft-lbs)

ASTM
Designation Thickness Zone 1** Zone 2** Zone 3**

A36 15@70°F 15@40°F 15 @10°F

A572t Up to 4 in. mechanically fastened 15@70°F 15@40°F 15 @10°F
Up to 2 in. welded 15@70°F 15@40°F 15 @ 10°F
A440 15@70°F 15@40°F 15 @ 10°F
A441 15@70°F 15@40°F 15@10°F
A242 15@70°F 15@40°F 15 @ 10°F
A588t Up to 4 in. mechanically fastened 15@70°F 15@40°F 15 @10°F
Up to 2 in. welded 15@70°F 15@40°F 15 @10°F
Over 2 in. welded 20 @ 70°F 20@40°F 20 @ 10°F
A514 Up to 4 in. mechanically fastened 25 @ 30°F 25@0°F 25@-30°F
Up to 2-1/2 in. welded 25@30°F 25@0°F 25 @ - 3 0 ° F
Over 2-1/2 in. to 4 in. welded 35@30°F 35@0°F 35@-30°F

*See also ASTM A709-75 Standard Specification for Structural Steel for Bridges.
**Zone 1: Minimum service temperature 0°F and above. Zone 2: Minimum service temperature from - -1 to —30°F. Zone 3: Minimum service
temperature from —31 to — 60°F.
tif the yield point of the material exceeds 65 ksi, the temperature for the CVN value, for acceptability shall be reduced by 15°F for each incre-
ment of 10 ksi above 65 ksi.
13

FIRST QUARTER / 1977

 -160 1 1 1 1 1 1 •1 I

GO 1
-140 h
4> ,'
80

I
Predicted ^Dynamic K.
U.-120 h Slow-Bend K T J \
60
mperature Shift
-100 40 = -60F J

20 0 Actual Slow--Bend K j ^ H
J -80 Test Results
0 1 1 1 J 1 L_ 1 1

-350 -300 -250 -200 -150 -100 -50 50

-60
Test Temperature, F
*
5 -40 Fig. 15. Effect of temperature and loading rate on fracture
toughness of A517 steel
-20

• t I _!•_
40 80 120 160 200 240 280
Yield Strength, ksi specimens had been subjected to cyclic loading, some were
cooled to —SO'^F and then loaded under an impulse load
Fig. 74. Effect of yield strength on shift in transition to a maximum bending stress of 20 ksi for A36 steel and 27
temperature between impact and slow-bend CVI^ test results ksi for A572 Grade 50 steel. The specimens were then
subjected to impulses of either 36 ksi or 50 ksi, the specified
minimum yield strengths of the two steels, respectively. The
total time for the 20 ksi impulse was approximately 1 sec-
ond — about the same time as for the impulses observed
existence of a strain-rate shift^ to obtain the desired in field measurements of truck loadings on short span
toughness levels at a service loading rate (intermediate) that bridges.^^ Since the stress levels in the field measurements
is 70°F below the actual specification loading rate (impact). were generally below 6 ksi, the strain rates in the test im-
In addition, these toughness requirements are also based pulse were well above the strain rates observed in the
on the assumption that the design, fabrication, erection, and field.
inspection procedures for bridges are satisfactory, although
In summary, the test temperature was below the mini-
an exact definition of satisfactory is difficult to obtain. If
mum temperature expected to occur in actual highway
this is the case, as it is for most bridges, then the AASHTO
bridges in the continental United States, the loading rates
material toughness requirements presented in Table 1
were well above the loading rates observed in field mea-
generally are quite adequate. If this is not the case, then
surements, and the beams were subjected to the highest
even making the material toughness requirements more
allowable cyclic stresses and the maximum number of cycles
stringent may not achieve the desired result of a fracture-
specified by AASHTO for cover plate ends at that stress
resistant structure because of fatigue, as was illustrated in
level. Thus, the fracture stress was determined for non-
Fig. 11.
redundant (single-load-path) bridge members containing
the most severe common structural detail and for the most
FRACTURE TESTS OF WELDED BEAMS severe combination of temperature, strain rate, and prior
To verify the adequacy of the toughness requirements fatigue loading that could reasonably be expected to occur
presented in Table 1, beam specimens of A36 steel and in actual highway bridges in the continental United States,
A572 Grade 50 steel were used to study the fracture be- and they still did not fracture until stress levels about 70
havior of simulated bridge members under extreme service percent higher than the maximum design stress permitted
conditions.^^ The specimens were designed to include two by AASHTO were reached.
common structural details that adversely affect fatigue and Even though the principles involved in the development
fracture strength: (1) a cover plate end and (2) a transverse of the AASHTO material toughness requirements for
stiffener. Fatigue tests have shown that the cover plate end bridge steels can be used to develop toughness requirements
is one of the most severe common structural details with for other types of structures, these specific AASHTO
respect to fatigue^^ and probably also fracture.^^ toughness requirements are not recommended for direct
The beam specimens of A36 and A572 Grade 50 steels use in other types of structures. As in the development of
were subjected to a cyclic stress range of 21 ksi for 100,000 fracture criteria for any type of structure, the particular
cycles or more. This loading corresponds to the maximum criterion is dependent on the overall service behavior and
allowable fatigue loading specified by AASHTO for cover experience, loadings, strain rate, design, and details, re-
plate ends in either steel, but is much more severe than the dundancy, consequences of failure, etc., and not just the
cyclic loadings measured in actual bridges.^^ After the fracture characteristics of the materials.
14
ENGINEERING JOURNAL / AMERICAN INSTITUTE OF STEEL CONSTRUCTION
 ACKNOWLEDGMENTS 18. Rolfe, S. T. Fracture Mechanics and Fracture Criteria For
Ship Hull Structures ASTM Workshop on Low Temper-
Most of the material in this paper is taken from the recent ature Properties of Ship Plate, St. Louis, Mo., May 1976.
textbook Fracture and Fatigue Control in Structures— 19. ASME Boiler and Pressure Vessel Code Section III, Section
Applications of Fracture Mechanics by S. T . Rolfe and J . G, American Society of Mechanical Engineers, New York,
M . Barsom, Prentice-Hall, 1977. NY, 1974.
In addition, much of the information also has been 20. AASHTO Materials Part 1, Specifications, American
published in Refs. 2 1 , 32, and 33, and the assistance of W. Association of State Highway and Transportation Officials,
J. Hall and J. M. Barsom is gratefully acknowledged. Washington, D.C, 1974.
21. Barsom, / . M. Development of the AASHTO Fracture-
Toughness Requirements for Bridge Steels Engineering
Fracture Mechanics, 1975, Vol. 7, pp. 605-618.
REFERENCES 22. Cudney, G. R. Stress Histories of Highway Bridges
1. Parker, E. R. Brittle Behaviour of Engineering Structures fournal of the Structural Division, American Society of Civil
John Wiley & Sons, New York, N.Y., 1957. Engineers, Vol. 94, No. ST12, Dec, 1968.
2. Shank, 'M. E., Ed. Control of Steel construction to Avoid 23. Madison, R. B. Application of Fracture Mechanics to
Britde Failure Welding Research Council, New York, N. Y., Bridges Lehigh University Institute of Research, Fritz
1957. Engineering Laboratory Report No. 335.2, Bethlehem, Pa.,
3. Hall, W. J., H. Kihara, W. Soete and A. A. Wells Brittle fune 1969.
Fracture of Welded Plate Prentice-Hall, Inc., Englewood 24. The AASHTO Road Test Highway Research Board of the
Cliffs, N.J., 1967. NAS-NRC Division of Engineering and Industrial Research,
4. Boyd, G. M., Ed. Brittle Fracture in Steel Structures The Rep. 4, Bridge Research, Special Report CID, Publication
Royal Institution of Naval Architects, Published by Butter- No. 953, National Academy of Science, National Research
worth, , London, England, 1970. Council, Washington, D.C, 1962.
5. Tipper, C F. The Brittle Fracture Story Cambridge 25. Barsom, f. M., f. F. Sovak and S. R. Novak AISI Project
University Press, Cambridge, England, 1962. 168—Toughness Criteria for Structural Steels: Fracture-
6. Libowitz,H.,Ed. Fracture, and Advanced Treatise Vol. Toughness of A36 Steels U.S. Steel Research Laboratory
I-VII, Academic Press, New York, N.Y., 1968. Report 97-021-001 (1), Pittsburgh, Pa., May, 1972.
26. Barsom, f. M., f. F. Sovak and S. R. Novak AISI Project
1. Rolfe, S. T, and J. M. Barsom Fracture and Fatigue
168—Toughness Criteria for Structural Steels: Fracture
Control in Structures—Applications of Fracture Mechanics
Toughness of A572 Steels U.S. Steel Research Laboratory
Prentice-Hall, Inc., Englewood Cliffs, N.J., 1977.
Report 97-021-001 (2), Pittsburgh, Pa. Dec, 1972.
8. Fracture Toughness Testing and Its Applications ASTM 27. Barsom, f. M., and S. T. Rolfe Correlations between Kjc
Special Technical Publication No. 381, 1964. and Charpy V-Notch Test Results in the Transition-Tem-
9. Tada, H., P. C. Paris, and G. R. Irwin The Stress Analysis perature Range, Impact Testing of Metals ASTM STP
of Cracks Handbook Del Research Corporation, 1973. 466, American Society for Testing and Materials, p. 281,
10. Sih,G.C. Handbook of Stress Intensity Factors Lehigh 1970.
University, Bethlehem, Pa., 1973. 28. Barsom, f. M. Relationship between Plane Strain Ductility
11. Safety and Reliability of Metal Structures American Society and Kic for Various Steels Trans, of the ASME, f. Engi-
of Civil Engineers Speciality Conference, Pittsburgh, Pa., neering Indust. Series B, 93, (4), 1209, Nov., 1971.
Nov. 1972. 29. Sailors, R. H., and H. T Corten Relationship between
12. Madison, R. B. Application of Fracture Mechanics to Material Fracture Toughness using Fracture Mechanics and
Bridges Ph.D. Thesis, Lehigh University, 1969, and Fritz Transition Temperature Tests, Stress Analysis and Growth
Engineering Laboratory Report No. 335.2, Bethlehem, Pa., of Cracks ASTM STP514, American Society for Testing
fune 1969. and Materials, 1972.
13. Standard Method of Test for Plane-Strain Fracture 30. Schilling, C G., K. H. Klippstein, G. T. Blake, S. R. Novak
Toughness of Metallic Materials ASTM Standard E399, andf. M. Barsom AISI Project 168—Toughness Criteria
1976. for Structural Steels: Low-Temperature Tests of Bridge
Members U.S. Steel Research Laboratory Report 97-
14. Collapse of U. S. 35 Highw^ay Bridge, Point Pleasant, West
021-001 {3), Pittsburgh, Pa. Dec, 1972.
Virginia, December 15, 1967 NTSB Report No. NTSB-
31. Fisher, f. W., K. H. Frank, M. A. Hirt and B. M. McNamee
HAR-71-1, 1971.
Effect of Weldments on the Fatigue Strength of Steel Beams
15. Bennett, f. A., and H. Mindlin Metallurgical Aspects of NCHRP Report 102, 1970.
the Failure of the Pt. Pleasant Bridge fournal of Testing 32. Rolfe, S. T., and W. f. Hall Fracture Mechanics, Fracture
and Evaluation, March 1973, pp. 152-161. Criteria, and Fracture Control in Structures Structural and
16. State Cites Defective Steel in Bryte Bend Failure Engi- Geotechnical Mechanics, edited by W. f. Hall, Prentice-Hall,
neering News-Record, Vol. 185, No. 8, August 20, 1970. Inc., Englewood Cliffs, N.f., 1977.
17. Joint Redesign on Cracked Box Girder Cuts into Record 33. Rolfe, S. T. Fracture Mechanics and the AASHTO Ma-
Tied Arch's Beauty Engineering News-Record, Vol. 188, terial Toughness Requirements for Bridge Steels Canadian
No. 13, March 20, 1972. Structural Engineering Conference, Feb. 23 and 24, 1976.

15
FIRST QUARTER / 1977