PSI - Issue 2_B

P.B.S. Bailey / Procedia Structural Integrity 2 (2016) 3758–3763

3761

4

Author name / Structural Integrity Procedia 00 (2016) 000–000

yet a simple, subtle difference in fitting algorithm causes a 2.4% shift in the result, despite using an idealized dataset without any anomalies from test equipment or physical effects outside the model behavior.

Figure 3: Calculated crack length vs J i for model dataset

Figure 4: J-R curve construction for model dataset

4. Calculations with real data Calculations are run in the same manner as for the model dataset, but the unloading compliance is calculated only on the section after the short dwell and consequent relaxation. In Figure 5 it is clear that the choice of when to measure compliance has a noticeable effect on the result, despite the appearance of Figure 1 that unloading and reloading lines coincide well for this specimen and test system. Furthermore, within the first two unloading points there is clearly stiffening of the specimen, which would generally be attributed to a small degree of work hardening in the specimen (near the crack tip). However, it should be recognized that under some circumstances, this could also be an anomaly of the test fixtures, as the specimen and load string align themselves fully; generally tests are started with a short series of “bedding in” load-unload cycles at a low, purely elastic, force level, with a view to avoiding this. Although neither curve is perfect in this regard, the reloading data appears to be less severely affected.