Probabilistic Fracture Mechanics Analysis of Friction Stir Weld Fatigue Performance

Recent studies have suggested that friction stir welding (FSW) represents a transformative technology, which could have a significant influence on the durability and economics of extruded aluminum decks for vehicular bridge construction and retrofitting applications. Design codes in Canada are only just starting to acknowledge FSW as a viable welding process for structural applications. Existing international codes (e.g. ISO, AWS) include quality control (QC) provisions, which appear to be based on “best practice”, rather than a correlation with expected fatigue performance. With this in mind, a recent project was undertaken, which included the following tasks: 1) FSW specimen preparation and metallurgical analysis; 2) fatigue testing of FSW specimens; and 3) fracture mechanics analysis of the test results. A novel aspect of the project was the fatigue testing and analysis of welds with varying levels of quality (i.e. various defect types with varying degrees of severity intentionally introduced). In this paper, a fracture mechanics model developed previously to analyze the test results will be used along with Monte Carlo simulation (MCS) to establish probabilistic S-N (stress-life) curves for FSW joints with various defect types and degrees of severity. It is believed that curves like these can provide a basis for the establishment of improved guidelines and code provisions for the QC and fatigue design of FSW joints.