Page 385 - Forensic Structural Engineering Handbook

11.16 MATERIAL-SPECIFIC FORENSIC ANALYSES

Specification A20 for pressure vessel steels. The reality is that, with the exception of some
bridge member specifications [covered in ASTM Specification A709, AWS (American
15
14
Welding Society) Code D1.5, and AASHTO (American Association of State Highway
and Turnpike Officials)] and pressure vessel toughness requirements, which are covered in
the ASME (American Society for Mechanical Engineers) Boiler and Pressure Vessel
7
Code, many structures are built without any toughness specification being applied to the
steels utilized and, thus, the potential for brittle fracture is always present. The fact that
more brittle fractures do not occur in structures is attributable to a number of factors. First,
the materials codes used in many applications have evolved over time to the point where
steels with really unsatisfactory toughness are no longer employed, even when toughness
is not explicitly measured. Second, in applications where high levels of toughness are truly
required, codes and specifications have been created that make toughness testing manda-
tory. Third, the toughness of structural steels is strain-rate sensitive. Hence, because of their
size and, therefore inherent inertia, large structures have service loading rates that are rarely
as great as the material experiences in the CVN impact test. Thus, there is an inherent, often
unrecognized, safety factor built into these steels which engineers can utilize; that is., the
steels are tougher in service than the CVN test predicts. The AASHTO bridge code recog-
nizes and utilizes this factor explicitly. Fourth, stresses in some structures are sufficiently
low to preclude brittle fractures. Finally, modern steelmaking practice has evolved in the
direction of lower carbon contents and more sophisticated processing for structural steels,
both of which improve fracture toughness performance.
For some, but not many structural applications, fracture toughness characterizations
using linear elastic or elastic-plastic fracture toughness testing are required. These tests are
typically governed by ASTM Specifications, such as E813, E1152, E1290, and E1921.
Such tests can provide substantially more information about fracture potential in structures,
but are significantly more expensive and require a more sophisticated testing capability to
perform and a more sophisticated analysis of the data to ascertain how the results can be
applied to a given structure as compared to CVN testing. When a post-failure fracture eval-
uation is being performed in which the size of the initiating flaw can be determined, a frac-
ture mechanics based analysis is very useful because it enables the investigator to relate the
size of the initiating flaw, a, the applied stress at the time of the failure, the stress intensity
factor, K, and the material fracture toughness, K . These factors are related, in a simplified
C
way, through the following stress intensity factor relationship:
K = K = σ π aF a t) (11.2)
/
(
C
where σ is the applied stress and F(a/t) is a function that depends on the structure’s geometry,
location or the flaw, and its size, a, in the structure. The units of a is inches (millimeters), the
stress units is ksi (MPa), and the K and K units are ksi√in (MPa√m).
C
A more detailed description of fracture mechanics assessment methodology is given
later in this chapter.

Fatigue and Fatigue-Crack Growth Properties
Historically, fatigue analysis has utilized rotating (bending) or axial (tension) “smooth” bar
specimens to establish stress versus number of cycles to failure curves (S/N curves) to pre-
dict the fatigue life of structures. This type of analysis assumes that the structure or component,
like the test specimen, is flaw-free, and failure results from initiation and subsequent growth of
fatigue cracks. The specimens are sometimes modified to incorporate notches or grooves to
simulate the configuration of actual service components. The “smooth” bar tests for steels usu-
ally show a plateau in the S/N curve below which no failure will occur for millions of cycles,