Self-piercing riveting process: Prediction of joint characteristics through finite element and neural network modeling

• A calibrated, stress-state based SPR process FE model is developed. The joining feasibility and joint quality of a wide range riveted connections investigated. High rivet leg inner radius values are shown to increase the quality. Low central tip depths relate high probabilities successful formation. Machine learning techniques can classify for riveting scenarios. Developing robust numerical simulation tools investigate self-piercing critical, since feasibility, strength relies on formation mechanical interlocks between sheet materials. In present study, we aluminum alloy dual-phase steel sheets (AA7075-F/DP600 AA6016-T4/DP600) over thicknesses, as function die geometries employed. More specifically, study dependence probability formation, defined ratio number acceptable total test cases, depth. Towards this, use experimentally calibrated Hosford-Coulomb fracture surfaces each deformable part, incorporated in dedicated axisymmetric 2D finite element (FE) models. predictions validated through comparisons with obtained joints. Moreover, analyze relation achieved employed, deriving practice-relevant conclusions respect most favorable design parameters. particular, show that while decreases upon increasing radius, quality, terms effectuated interlock distance per mean residual equivalent plastic strain, increases. Furthermore, machine-learning be employed remarkable accuracy possible scenarios, accounting both die-related geometrical attributes, well thickness metal sheets.