Uncertainty analysis of a shape memory alloy model for dynamic analysis

Abstract

Superelastic shape memory alloys (SMAs) are capable of recovering large inelastic deformation and dissipating energy under loading reversals, and therefore present promising application in vibration control of engineering structures subjected to dynamic loads. Properties of SMAs are often optimised based on experimental results and treated as deterministic for dynamic analysis of structures with SMA-based devices. This study applies the Metropolis–Hasting algorithm to characterise the uncertainties within an SMA material model and to provide insight into its parameter uncertainty. Cyclic tests of SMA bars are first conducted and the experimental data are analysed using the Markov chain Monte Carlo method. The statistical properties of SMA model parameters are calculated based on posterior parameter distributions. The influence of SMA model parameter uncertainty is further explored on energy dissipation capacity under cyclic tests. The first four L-moment method is applied to transform the posterior parameter distributions into standard normal distribution to further evaluate the influence of model parameter uncertainty on energy dissipation in dynamic analysis.