Page 395 - Forensic Structural Engineering Handbook

11.26 MATERIAL-SPECIFIC FORENSIC ANALYSES

fatigue strength, quality control to prevent undesirable defects and discontinuities that are
susceptible to crack extension and premature failure, and in-service inspection when cyclic
loading is a design consideration. Fatigue resistance is governed by the stress range to
which a given structural detail is subjected and the number of loading cycles. The residual
tension stresses that exist at all as-welded details are a primary reason that the dead load
stress or mean stress has a negligible impact on fatigue strength. Figure 11.11 shows the
fatigue crack surface of a welded cover plate detail where the crack has formed at the weld
toe in the heat-affected zone and has propagated into the flange plate as a semielliptical sur-
face crack until it extended through the plate thickness. Beachmark-like features can be
seen on the smooth crack surface. Once the crack became an edge crack, the rate of crack
growth increased; this is apparent in the rough surface features as the crack extended
toward the beam web.

FIGURE 11.11 Fatigue crack forming at the end of a longitudinal weld attaching a cover plate
to a flange.

Fatigue resistance does not depend on the type of steel or on its tensile properties. For a
given type of detail, the fatigue strength is uniformly described by an exponential
model 36–38 that relates cyclic life, N, to the nominal stress range, S r
N = C S 3 (11.13)
f r
where C is a coefficient that defines the various categories of connection details.
f
Coefficient C is defined in the AASHTO and AISC LRFD Specifications. The stress
f
range S is the nominal stress range in the base metal at the connection detail. It does not
r
include the stress concentration from the weld geometry, as that is incorporated in the type
detail classification and its coefficient C . The fatigue limit for each category is an estimate
f
of the stress range below which no cracking develops. The fatigue limit varies with the
6
detail category. For category A it occurs at about 2 × 10 cycles and increases in life for all
7
other categories reaching about 2 × 10 cycles for category E′.
Most structural systems are subjected to random variable loads that result in a skewed
stress range spectrum. Laboratory studies on variable loaded details have demonstrated that
fatigue cracks will develop when as few as 0.05 percent of the variable cycles exceed the
category fatigue limit (CAFL). Consequently, a suitable cumulative damage criteria must
be applied in order to predict fatigue life. Additionally, laboratory fatigue studies have also
demonstrated that the AASHTO-type weldment fatigue curves can be used to predict both
constant amplitude and variable (random) amplitude loading. Variable amplitude loading,