Fracture Fatigue Life Prediction of Geometrically Varied Multi-scaled 5083 H116 Welded Aluminum Structures: A Numerical Analysis using Extended Finite Element Method (XFEM) and Experimental Verification using Digital Image Correlation (DIC)

Abstract

Numerical computational methods to predict fracture fatigue life of functional materials are cost effective when compared to experimental-based prediction approaches. However, the validity of a numerical fracture fatigue solution for welded aluminum structures depends on its agreement with experimentally obtained results. Obtaining physical experimental results is a challenge for fracture fatigue life at the structural scale because it requires the identification, isolation, and testing evaluation of various factors influencing crack growth, such as material properties, environmental conditions, loading spectrum, residual stresses, etc. There are well established testing standards (e.g., ASTM) designed to quantify the effect of these factors; however, the appropriate scaling of benchtop results to a suitable structural scale is not well understood. Hence, it is common practice to physically test full scale structures (e.g., ships) to gain knowledge of the fracture fatigue life of the components. This can be cost prohibitive. The aims of the work presented here are to: 1) develop and verify a digital image correlation (DIC) based crack length measurement technique for fracture fatigue life testing and 2) develop an application of the DIC based measuring technique for use in the investigation of the effect of geometry on fracture fatigue life response of aluminum 5083 H116. The initial investigation of the concept for using DIC as a fracture fatigue life measurement technique demonstrated the feasibility of the DIC approach, i.e., the construction of a plot of crack length as a function of loading cycle for a Monitored Aluminum Health Investigation (MAHI) component. In addition, the technique can continuously measure crack growth to give insight regarding crack length growth at a finer time scale. The verification of DIC and the ability to measure strain at low service loads were confirmed. Strain results obtained via DIC were in good agreement with measurements of strain obtained from electrical resistance strain measurements. DIC?s capability to measure the full field strains was employed to investigate the effect of welds and crack-like flaws, and established unique fingerprint-like characteristics that can be used to identify and inspect structural details of welds and cracks. The aluminum fracture fatigue experiments were conducted at constant amplitude with a stress ratio (R) of 0.1 on six different welded aluminum 5083 H116 specimen types with a through thickness crack at the center. DIC established the crack growth as a function of the number of cycles (i.e., the crack growth rate). Extended finite element mothed (XFEM) was able to predict the general profile of the crack length growth when compared to DIC results. However, there were differences between the predicted and experimentally obtained results, ranging from 17 % to 251 %. The groups with no stiffener showed a 36 % reduction in life when compared to the groups with stiffeners. Samples with stiffener removed and no buttweld had the lowest life with 13,185 cycles (� 661 standard deviation), and samples with stiffeners with buttweld had the highest life (60,000+). This work has developed an effective fracture fatigue life experimental design approach and measuring technique to improve predictive capabilities to assess damage in structures.