PSI - Issue 17

Zizhen Zhao et al. / Procedia Structural Integrity 17 (2019) 555–561 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

559

5

respectively. The fraction of fatigue and creep damage in CRF tests is shown in Fig. 5, which also includes the damage envelope for 2.25Cr1Mo steel in ASME-NH, the diagonal line, and the envelope when the sum of D f and D c is one. All data fall outside the envelope given by ASME-NH, and most data are within the envelope when the sum of D f and D c is one. D f is a bit higher than D c at short holding periods. When the hold period is longer than 5s, D c becomes dominant, and D f is almost negligible when the hold period extends to 60s. This diagram indicates that creep damages created during stress holding periods lead to shortened fatigue lives in CRF tests. 4. Fatigue life prediction The LDS rule assumes the accumulation of time-dependent damage and time-independent damage govern the fatigue life of a specimen. It uses fraction rule to evaluate the fatigue damage and creep damage produced in each cycle, and failure occurs when total damage reaches a utility, 1/ ( 1 ) fp f R t N N t  = + (3) where N f is the fatigue life in continuous cycling, and N fp is the predicted fatigue life. For prediction in CRF tests, fatigue life of the corresponding RF tests is taken as N f . Predicted fatigue lives from the LDS rule for CRF tests are compared with observed ones in Fig. 6 (a), with scatter band of 1.5 and 2 being given. Obviously, the LDS rule provides better predictions for peak holding mode, as all the predictions fall within a scatter band of 1.5, but predictions for double hold is not as satisfactory. The inconsideration of unbalanced cycling in the LDS rule and the underestimate of creep damage in valley holding may be responsible. To compensate for this deficiency, stress ratio denoted by R was incorporated in the evaluation of creep damage per cycle as below. (4) A new model has been developed by substituting Eq. (4) into Eq. (3), and the predicted results compared with the observed ones in Fig. 6(b). The new model keeps satisfactory predictions for peak holding mode, while the predicted fatigue lives for double holding mode are improved as they fall within a scatter band of 1.5. c R 1 ( ) h t = + h d t R c

Fig. 5 The influence of stress holding periods on (a) fatigue life, (b) creep and fatigue damage