Cyclic Vs. Monotonic Loading in Determining Fatigue Crack Growth Exponents

General log linear (GLL) regression is able to model the combined effects of mechanical stress, temperature, and nominal factors on product lifetime. This might provide more accurate predictions of dental restoration lifetimes than modeling with traditional Weibull analysis. Objectives: To determine whether the fatigue crack exponent estimated by a GLL model is equivalent to that estimated by paired rank regression Weibull analysis for a dental ceramic. Methods: Rectangular beam specimens of a fluorapatite glass-ceramic (IPS e.max ZirPress, Ivoclar-Vivadent) were fabricated by pressing. The polished specimens (15 μm diamond) were subjected to rapid monotonic loading (oil, 1 MPa/s, N=30) or cyclic loading between 4.5 and 45 MPa at 2 Hz or 10 Hz (deionized water, 37°C, N=30 each). Tests were performed using a fully articulating four-point flexure fixture. Cyclic loading data were fit to the GLL model using ALTA Pro software (Reliasoft), ln η= α₀ + α₁f + α₂ln s + α₃f ln σ, where η is the characteristic lifetime, f is the cycling frequency, and s is the peak stress. Cyclic and monotonic loading data were paired for determining fatigue crack parameters according to Munz & Fett (1999). Results: Paired rank regression provided similar estimates for the fatigue exponent (n) regardless of cyclic frequency (21.7 for 2 Hz; 19.1 for 10 Hz). GLL provided a similar fatigue exponent (-α₂=21.2 (9.73, 31.9)) as paired rank regression. GLL showed no effect of frequency on lifetime (α₁=-3.16 (-15.1, 6.30)) and no interaction between frequency and stress-corrosion (α₃=0.820 (-1.59, 4.02)). The preceding results are given as point estimate (95% confidence interval). A coefficient with interval overlapping zero indicates no effect. Conclusion: GLL and traditional Weibull analysis provided equivalent fatigue exponent estimates for the ceramic studied, and GLL correctly detected a lack of frequency effect on lifetime.