106 / 352
102
In order to eliminate grain size effects, it is usually recommended that the specimen
thickness,
B
, be greater than 30 grain diameters /28, 29/. In some cases, such as in large-
grain (~3 mm) lamellar γ-α
2
Ti-Al intermetallic or α-β titanium alloys, the required
specimen sizes would be too expensive, test loads very high, and the component
dimensions probably be less than 30 times the grain size. Then, testing should be perfor-
med on thickness representative of the component. Curvature of the crack front and side-
to-side variation in crack length due to thickness can be a problem in thick specimens.
4.2.3. Specimen pre-cracking
The method by which a notch is machined depends on the specimen material and the
desired notch root radius (
ρ
). Saw cutting is applicable to aluminium alloys, but for a
notch root radius of
ρ
=
0.25 mm milling or broaching is required. In low- and medium-
strength steels notch can be produced by grinding, but for high-strength steels, nickel-
base super alloys, and titanium alloys electrical discharge machining may be necessary.
The specimen might be side polished to allow monitoring of crack growth, using
standard metallography practices, if possible, sometimes followed by etching. For is too
large or small specimen hand grinders, finishing sanders, or handheld drills can be used
with pieces of polishing cloth to apply the abrasive and create a satisfactory viewing
surface. These techniques are quick and easy to apply, and they are often used when
visual measurements are made only during precracking and subsequent measurements are
made by automated techniques such as electric potential or compliance.
The
K
-calibration functions from ASTM E 647 and E 399 are valid for sharp cracks
within the range of crack length specified. Consequently, before testing begins a sharp
fatigue crack that is long enough to avoid the effects of the machined notch must be
present in the specimen. The process that generates this crack is termed precracking.
Loads for precracking should be selected such that the
K
max
at the end of precracking does
not exceed levels expected at the start of a test. For most metals, precracking is a simple
process that can be performed under load or displacement control, with moderate growth
rates (10
-5
m/cycle) using Δ
K
from growth curves in the literature. To decrease the pre-
cracking time, common practice is to initiate crack at a load above that which will be
used in test, and after that reduce it. Crack growth can be arrested above the threshold
stress-intensity factor value due to formation of the increased plastic zone ahead of the tip
of the advancing crack. The loads should be shed no faster than 20% (per increment of
crack extension) from the previous load increment. As the crack approaches the final size,
this percentage can be decreased.
The amount of crack extension between each load decrease must also be controlled. If
the step is too small, the influence of the plastic zone ahead of the crack may still be
present. To avoid transient (load-sequence) effects in the test data the load range in each
step should be applied over a crack-length increment of at least (3
π
) (K'
max
/
σ
YS
)
2
, where
K'
max
is the terminal value of
K
max
from the previous load step. This requirement ensures
that the crack extension between load sheds is at least three plastic zone diameters.
The influence of the machined starter notch must be eliminated so that the crack tip
conditions are stable. For C(T) and M(T) specimens the final precrack should be at least
10% of the thickness of the specimen or equivalent to the height of the starter notch.
Two additional considerations regarding crack shape are the amount of crack variation
from the front and back sides of the specimen and the amount of out-of-plane cracking.
Due to microstructural changes through the specimen thickness, residual stresses