Indian Journal of Engineering & Materials Sciences Vol. 21, April 2014, pp.179-188 A comparative study of fatigue life prediction of 7020 Al-alloy under load ratio effect J R Mohanty a *, H C Das b & A C Mohanty c a Department of Mechanical Engineering, Veer Surendra Sai University, Burla, Sambalpur 768 018, India b Department of Mechanical Engineering, Institute of Technical Education & Research, SOAU, Bhubaneswar 751 030, India c Shapoorji Pallonji Engineering Procurement & Construction Division, SPEPCD, Mumbai 400005, India Received 8 August 2013; accepted 10 February 2014 Most of engineering structures and components come across complicated fatigue loading during their service lives. From economical point of view it is essential to predict residual life in order to avoid catastrophic failure by scheduling suitable inspection intervals. In the present investigation, fatigue life of 7020 T7 Al alloy under constant amplitude loading with load ratio effects has been predicted by adopting an ‘Exponential Model’. The performance of the proposed model has been compared with artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS). It is observed that the fatigue life predicted by exponential model is comparatively better with maximum percentage deviation of -0.65% in comparison to ANN and ANFIS which are -4.37% and -2.48% respectively. Keywords: Exponential model, Fatigue crack growth, Load ratio, ANN, ANFIS Cracks, in the form of defects/imperfections are inherently present in almost all engineering structures. Under fatigue loading, these cracks grow to a critical size leading to catastrophic failure. Hence, it is important to predict residual fatigue crack propagation life in order to schedule suitable inspection intervals thereby suggesting a preventive measure for repair/ replacement of damaged parts before failure. Several prediction models have been developed till date which correlates fatigue crack propagation rate with the corresponding stress intensity factor range (K). All such models can be broadly classified into two basic types: empirical models and analytical models. An empirical model is based on the analysis of actual experimental data but not necessarily on the first principle (i.e. on the mechanism of fatigue); whereas, an analytical model is generally based upon the physics that govern the problem but always involves simplifying assumptions to reduce the mathematical complexity. The power law relationship of Paris and Erdogan 1 has been considered as the most fundamental and popular differential model relating crack growth rate (da/dN) with stress intensity factor range (K). However, the crack driving force K is considerably affected by various factors such as load ratio, frequency, environment, alloy composition microstructure etc. Out of them the main factor is the load ratio (R) that influences the trends of da/dN – K curves. Observing the trends, it can be seen that with the increase in applied K (for higher R values), differences in the corresponding crack growth rate values can either increase (“diverging” da/dN – K), or be constant (“parallel” da/dN – K), or decrease (“converging” da/dN – K) 2 . Earlier, the ‘crack closure’ mechanism proposed by Elber 3 was able to account for the influences of load ratios on crack growth process and was accepted by some researchers. Later it was criticized because of some reasons, e.g., difficulty in measuring crack opening load 4 , its dependence on heat treatment 5 , and insignificance of plasticity-induced closure in high vacuum 6 . Kujawski 7 introduced two parameters (K max and K) approach using positive part of stress intensity factor range K + and K max to consider R-ratio effect on crack growth under constant amplitude loading by avoiding disputable crack closure concept. The influences of the R-ratio on fatigue crack growth have been critically analyzed by Kalnaus et al. 8 Fatigue crack propagation is a continuous physical process of material damage. The test results obtained from measurement on laboratory specimens reveal that there is an increase in crack length with number of loading cycles. This increase in crack length, although contains large number of scatter, is observed —————— *Corresponding author (Email: guddy95@gmail.com)