Fatigue crack growth rate data that support standard damage integration packages of the type described in Sections 5.1 and 5.2 are based on constant amplitude testing of cracked specimens.  Typically, multiple specimen tests are conducted at a number of fixed stress ratio (R) conditions so that the complete range of crack growth rate is covered for the mechanical and environmental variations of interest.  For the most part, all tests of this type are covered by ASTM E647 on fatigue crack growth rate testing.

Test conditions that deal with the conditions essential for obtaining near threshold growth rates are further described by ASTM E647.  Substantial care is necessary for correctly controlling the precracking operation and the stress-intensity-factor control conditions in the near threshold region of the fatigue crack growth rate curve (da/dN vs DK) [Yoder, et al., 1981; Wei & Novak, 1982].  Also, ASTM E647 must be supplemented with information relative to control of environmental conditions when these conditions affect behavior.

The ASTM E647 describes the test, as well as the data collection, reduction and reporting requirements.  The test itself requires standard fatigue test capability and utilizes precracked specimens which have widely accepted stress-intensity factor solutions.  The standard currently recommends three specimen configurations, the middle-cracked tension [M(T)], the compact tension [C(T)], and the eccentrically-loaded single edge tension [ESE(T)] specimen geometries, which are shown in Figure 7.2.1.  While the M(T) specimen is generally recommended for all stress ratio conditions, it should be noted that the C(T) and the ESE(T) specimens can only be used for positive stress ratio conditions. 

The primary control exercised during a test is the control of the fatigue forces that are being applied to the test sample.  Most modern servocontrolled, electrohydraulic test machines that are periodically recalibrated using force cells traceable to the National Institute of Standards and Technology (NIST) will result in force control well within the ASTM E647 requirements.  Force cells, of course, should be selected such that fatigue crack growth rate tests are being conducted using forces that are at the higher end of the load cell range to maximize force accuracy.  Specific care should be taken to minimize force errors.  Such errors can cause major errors in reported crack growth rate data since stress-intensity factor (K) is a linear function of force.

Fatigue crack growth rate data are derived from the crack length data (discrete pairs of crack length and cycle count data) and test load data.  Significant errors in crack growth rate behavior can also result if systematic errors in crack length measurement occur since such errors directly affect the calculated stress-intensity factor parameters. ASTM E647 places strict requirements on the measurement of crack size and recommends a frequency of crack length measurement based on the gradient (rate of change) of the stress-intensity factor through the crack length interval in the given test specimen.

Figure 7.2.5 shows a schematic that illustrates the data reduction of a single test’s crack length data to the fatigue crack growth rate format.  The procedures that one uses to differentiate the crack length data have some effect on the individual da/dN vs DK discrete data points.  To ensure some uniformity in this part of the data reduction process, ASTM E647 recommends that either the secant or the 7-point incremental polynomial methods be utilized.  In fact, the standard includes a listing of a FORTRAN computer program that can be utilized to reduce the crack length data according to the 7-point incremental polynomial method.  Other differentiation methods leading to the same data trends for a given test include 5, 7, 9 point incremental, linear, quadratic, and power law least squares fitting schemes and the three-point average incremental slope method utilized by MIL-HDBK-5.  The specific differences that result from differentiating a set of crack length data using different methods are primarily associated with point-to-point data scatter in the a vs N data.  Discussion of the impact of this scatter on design was covered in Section 5.1. 

Figure 7.2.5.  Fatigue Crack Growth Rate Data Reduction Procedure

ASTM E647 recommends that duplicate tests be conducted to establish the crack growth rate behavior for a given set of test conditions (constant and environment).  However, if a complete definition of the growth rate behavior between threshold and fracture is required for a given set of test conditions, six constant load type tests with three different load levels might be required to cover the range.  For determining general trends under a given set of test conditions, shortcut methods are available.  These methods include:

·        methods of periodically increasing the constant amplitude load (by less than 10 percent) as the crack grows, and

·        methods of periodically modifying either the stress ratio or cyclic load frequency during a test. 

These shortcut methods are designed so that only selected intervals of the fatigue crack growth rate data are generated, although a description of the complete da/dN vs DK curve is possible since the entire range of behavior is covered.

When shortcut methods are utilized to obtain a design database, it is recommended that a preliminary test program be conducted to verify the accuracy of these shortcut methods.  The preliminary test program would be based on a sufficient number of both constant load amplitude and shortcut tests to justify the shortcut test methods, since changing test loads, stress ratio levels, cyclic load frequencies and environmental conditions can introduce crack growth transients.  The crack growth transients of most concern are those that modify the interpretation of the mean trend behavior exhibited by the material under the test variations considered.  The preliminary test program should determine the magnitude of the transient and the crack growth increment required to establish steady-state behavior after a new condition is introduced.  The approving agency should review the results of the preliminary test program relative to the impact of transient behavior and to the development of data reduction methods that exclude those intervals of crack length where transient behavior might be exhibited.