Fatigue-crack growth in Ti-6Al-4V-0.1Ru in air and seawater: Part I. Design of experiments, assessment, and crack-grown-rate curves

Abstract

Fatigue-crack growth rates for different simulated ocean environments and loading conditions have been investigated for beta-annealed Ti-6Al-4V-0.1Ru (extra-low interstitials, (ELI)), a candidate material for oil production risers; the focus was on uncovering whether certain combinations of conditions could produce unexpectedly high crack growth rates. A two-level, one-quarter-fraction factorial design-of-experiments (DOE) approach was used to ascertain which testing variables and environmental conditions warranted further study. This study used eight different combinations of variables: parent/deformed material, 27 °C/85 °C temperature, 2 Hz/20 Hz loading frequency, 0.1/0.6 load ratio (R=σ min/σ max, where σ min is the minimum and σ max is the maximum stress during a fatigue cycle), and aerated/deaerated seawater. Comparisons were based on crack growth rates at ΔK=17 MPa $$\sqrt m $$ , roughly the middle of the Paris portion of the da/dN vs ΔK curves. The da/dN vs ΔK curves were also examined, and conclusions based upon these data were compared with those from the DOE. Consideration of the microstructure’s influence on the crack path is postponed until Part II of this article. Samples tested at the higher load ratio showed a statistically significant increase in the crack propagation rate compared to those tested at R=0.1; the same was true of specimens tested at 20 Hz vs those tested at 2 Hz, but the level of significance was lower. The parent material had somewhat higher crack growth rates than the deformed samples. Changes in environmental conditions other than frequency produced little effect on the crack growth rate. Comparison of crack growth rates over the ΔK range measured revealed details that would have not been uncovered in comparisons at a single ΔK value. The Paris exponent ranged between 3.7 and 6.7, and the only systematic variation observed was an increase in the exponent with increasing test frequency. In seawater, cold work (a 5 pct reduction in thickness by rolling) reduced fatigue-crack growth rates by a factor of 2 (compared to the parent material) at intermediate and high ΔK values. There was a crossover of crack growth rates for low ΔK values: below 10 MPa $$\sqrt m $$ , growth rates were lower for the parent material than for the cold-rolled material, suggesting a higher ΔK th for the parent material, while above this value, fatigue cracks grew more rapidly in the parent material than in the cold-rolled material. Crack growth rates were slightly higher in seawater than in air, but only slightly more than the sample-to-sample variation of crack growth rates, and cold work reduced fatigue-crack growth rates in air by about the same amount as in seawater. Somewhat more scatter was observed for the R=0.1 tests than for the R=0.6 tests. Differences in temperature (27 °C, 53 °C, and 85 °C) do not appear to affect fatigue-crack growth rates. For ΔK<20 MPa $$\sqrt m $$ , crack growth rates were similar for 0.2 and 2 Hz but were higher for 20 Hz; above 20 MPa $$\sqrt m $$ , the crack growth rates were similar for all three frequencies. One explanation for the unusual frequency dependence relies on the possibility that the environment produces different amounts of closure for different test frequencies. According to this view, closure is effective in air and in seawater at 0.2 and 2 Hz but not at 20 Hz: perhaps the higher loading rate breeches the passive layer at a rate more rapid than it can reform. Because the crack growth rate appeared independent of temperature, it is unlikely that there is a significant influence of thermally activated corrosion-fatigue mechanisms for the conditions tested. The results demonstrate that beta-annealed Ti-6V-4Al-0.1Ru (ELI) possesses a robust response to the combinations of environment and loading expected in oil production riser service. The value of the DOE approach was clear, and supplementary tests verified the main effects predicted by the DOE results. Comparison of the single-value DOE results with the da/dN curves reveals a limitation in the former: different slopes of the Paris curves and crossover effects are or would be missed for DOE comparing crack growth rates derived from constant ΔK tests. The use of constant Δσ data and a second level of interrogation, following DOE analysis and based on the da/dN curves, addressed this limitation effectively. A DOE comparison based, for example, on three ΔK values (the lower, middle, and upper portions of the Paris regime) might be another way of proceeding.