Session: 25-02 Deformation Modeling

Submission Number: 177143

Construction of Chaboche Kinematic Hardening Models Based on Cyclic Ramberg-Osgood Stress-Strain Curves 

The demand for higher efficiency and power output continue to grow due to competitive markets, rising fuel cost and decarbonization targets.  The integration of intermittent renewable energy sources has led to more cyclic operation to support load following or peak demand.  The need for accurate prediction of thermomechanical fatigue (TMF) life is critical to avoid costly outages. Low cycle fatigue (LCF), typically defined as less than 100,000 cycles, is generally expected to occur with strain ranges that include some level of plasticity.  For decades, analysts have applied shakedown techniques based on cyclic strain-controlled tests to estimate plastic strain ranges from linear elastic stress analyses.  However, the application of empirical shake-down formulas like Neuber’s rule or Glinka’s rule (elastic strain energy density) to regions of high local stress introduces inaccuracies.  Advances in computer processing power now make it economical to include plasticity in finite element analysis (FEA).  Implicit solutions have already been used in FEA,for a number of years to predict creep strains, but there are relatively few valid combinations of plasticity models that are compatible with implicit creep. Ideally, the plastic model should be able to mimic the behavior displayed in uniaxial strain-controlled or load-controlled tests by using the cyclic Ramberg-Osgood stress-strain curves already generated from those tests.  This paper presents a methodology for constructing third- to fifth-order Chaboche Kinematic Hardening models that fit Ramberg-Osgood curves across multiple temperatures using MATLAB. Existing strain versus life models are used to determine the maximum strain range for curve fitting Chaboche parameters.  The initial yield stress is based on an assumed plastic strain, selected to minimize stress error in the region between 0 and the 0.2% strain offset.  A 2000-point stress versus plastic strain curve is generated at each time. The fitting method uses constrained nonlinear minimization of the sum of squared stress errors across the curve, with carefully chosen bounds for the gamma and C parameters. The behavior across temperature curves is examined to ensure that interpolated constants remain within the bounds of the fitted curves. The kinematic shift in maximum stress is compared to expectations based on pseudo-monotonic stress-strain curves.  Simulations of the stress evolution over multiple cycles are compared, cyclic testing and to FEA results with the same parameters.

Presenting Author: William Day Power Systems Mfg., LLC

Presenting Author Biography: Is a Chief Engineer at Power Systems Manufacturing and a former Chair of the MMM Committee

Authors:

William Day Power Systems Mfg., LLC
Ali Gordon University of Central Florida
Adedotun Banjo University of Central Florida