Experimental and theoretical studies of corrosion fatigue in a high strength steel
The present work consists of both experimental and theoretical aspects of corrosion fatigue in a high strength steel. Fatigue tests were conducted under fully reversed shear loading in an aerated 0.6M NaC1 solution at pH6, for a silico-manganese spring steel (BS251A58) having a yield strength around 1400 MPa. The fatigue crack evolution process can be defined sequentially in terms of a pit, a short crack in stage I and a crack in stage II. It was observed that in the early stages pits developed at Mn-rich sulphide inclusions, from which short cracks developed and propagated in stage I; the crack growth rate of such cracks was dominated by microstructural features. Stage II, microstructure-independent crack growth was observed following a transition from stage I crack growth. In addition consideration was given to the influences of cyclic frequency, the effect of cathodic polarisation and the effect of electrolyte compositions on corrosion fatigue lifetime and crack growth behaviour. Mechanical properties, metallurgical and electro-chemical properties of the material used were also investigated. The failed specimens were examined using SEM and X-ray spectra analysis in order to study the failure mechanism. Using a dislocation based Navarro-de los Rios model to represent the crack and its associated plastic zone, the crack growth process is characterised by the interaction of dislocations associated with micro-cracks, or local damage, and microstructural barriers (grain boundaries, precipitates, dispersed particles, phase interfaces, etc.). Some modifications to the model were made to incorporate the corrosive effect on the early stages of crack initiation (i.e. pit growth), the transition of stage I crack growth to stage II crack growth, and strain hardening of a parabolic form in order to achieve a more accurate prediction. The validity of the model was justified by the good agreement achieved between predictions based on the model and the experimental results. Based on the present study the following conclusions were made: 1. The fatigue strength of BS251A58 steel, subjected to 107 cycles in a 0.6M NaC1 solution, at pH6 and at a frequency of 5Hz, is 98 MPa, and is significantly lower than the value of the fatigue limit in air, i.e. 457.5 MPa. 2. Corrosion fatigue crack propagation in the high strength BS251A58 steel can be described sequentially in terms of corrosion-induced pits, microcracks from pits, cracks in stage I (shear type cracks), and cracks in stage II (tensile cracks). 3. The number of cycles in the stage I-to-stage II transition, in relation to the whole fatigue life, varies from 20% to 60%, increasing as the applied load decreases. 4. Crack coalescence may occur in both the stage I and stage II regimes. In the stage I regime, microcracks with an effective tip-to-tip distance less than 1 to 2 prior austenite grains will coalesce, while those with distances greater than 2 to 3 prior austenite grains will change to stage II propagation. In the stage II regime, crack coalescence °ems after about 70% of the fatigue life Nf. 5. The lower the cycling frequency, the shorter will be the corrosion fatigue life. The combination of a high stress level and a low frequency can eliminate the in-air-fatigue micro structural barrier effect. 6. Cathodic polarisation (E=-1280mv (SCE)) can prevent the surface of the specimen from pitting, delay the stage I-to-stage II transition, and decelerate the early stage II crack growth rate. However, as a stage II crack grows, cathodic polarisation will accelerate the crack growth rate. 7. In a 0.6M NaC1 solution, the final stage I crack length increases from 50ium to 300 gm with increasing cyclic shear stress range (from 224 MPa to 926 MPa). This distance equals approximately 2 to 10 prior austenite grain diameters. The final stage I crack length is much longer under cathodic polarisation than for a free corrosion potential under the same stress range. There is no significant frequency effect on the final stage I crack length when the frequency is in the range of 2Hz to 12.51-1z. 8. The dislocation based Navarro-de los Rios model was employed in the present study to describe crack growth behaviour and to predict fatigue life. The pit growth behaviour, the crack transition from stage I-to-stage II, and a parabolic strain hardening law, were incorporated into the model to reflect more closely the actual fatigue behaviour of the material. The validity of the model was justified by the good agreement achieved between predictions based on the model and the experimental results.