CORRELATION OF REDUCED-ORDER MODELS OF A THREADED FASTENER IN MULTI-AXIAL LOADING

Fasteners are represented with varying degrees of fidelity in finite element models in order to meet solver constraints and accuracy requirements. Three reduced-order models are evaluated to quantify how well they perform in terms of calibration effort and accuracy. To baseline their performance, a new series of test data are developed by pulling NAS1351-3-20P screws at angles between pure tension (0 degrees) and pure shear (90 degrees) in 15 degrees increments. First, a nonlinear elastic spring element joins two portions of the fastener shank. Unique load-displacement curves for tension and shear are taken directly from the test data, making it easy to calibrate yet fairly accurate. Second, the fastener is modeled with a solid cylindrical shank. The material properties are only calibrated to the tensile test results; consequently, this model is less accurate for predicting the ultimate shear load, though it comes reasonably close to the ultimate displacement in shear and combined loading. Third, a solid cylinder is divided into tensile and shear regions, and the material properties in each region are calibrated to the corresponding test data. This model deviates further from the ultimate displacement but effectively predicts the ultimate load in combined tension and shear. Each model has advantages, but the first is easiest to implement and most accurate overall.