In numerical simulations of components subjected to low-cycle fatigue loading, the material cyclic plasticity behavior must be modelled until complete stabilization, which occurs approximately at half the number of cycles to failure. If the plastic strain per cycle is small, a huge number of cycles must be simulated, which results into a huge and thus unaffordable simulation time. Acceleration techniques for shortening this time are useful, although their accuracy needs to be checked. This work aims to compare different approaches (nonlinear kinematic with "initial" and "stabilized" parameters and combined nonlinear kinematic and isotropic with the speed of stabilization fictitiously increased). It considers two benchmarks taken from the literature, in which the material has opposite cyclic behaviors (hardening, softening). A plane finite element model can be used in both benchmarks, thus permitting a simulation up to complete stabilization. Results confirm that the common approach of considering only the kinematic model (calibrated on "initial" or "stabilized" material state) from the very first cycle could lead to relevant errors. The acceleration technique based on a fictitious increase in the speed of stabilization leads to accurate results. Guidelines for calibrating this technique on a material's hardening or softening behavior are, finally, proposed.
