抄録
The main purpose of this study is to verify the ability and the limitations of the multiplicative viscoplastic constitutive model and the finite element method for the prediction of time-dependent creep behavior of SM490 steel at elevated temperatures.
In the previous paper1), the uniaxial multiplicative viscoplastic model was selected for the time-dependent mechanical behavior model of structural steel SM490 at elevated temperatures, and the material constants of the model were identified using only the stress-strain-strain rate data of uniaxial tensile tests under two different tensile speeds. Also, based on the comparison of calculated creep curves under constant loads from viscoplastic model and tested creep ones, it appears that one can predict the creep response under constant loads of structural steels reasonably well using the viscoplastic constitutive model.
In this paper, the developed uniaxial viscoplastic model was implemented into a general-purpose finite element program, and the simulations of uniaxial creep tests and bending creep tests of SM490 structural steel specimens were carried out. The analysis of the behavior of viscoplastic structures under realistic conditions requires the adoption of an adequate numerical framework capable to generate approximate solutions within reasonable accuracy. In this paper the approximate solution to such problems is addressed within the context of the Finite Element Method.
In summary, the following results were obtained from the finite element analysis:
1) In the case of uniaxial tensile tests under two different tensile speeds, due to the calculated results with finite element method were almost identical to those of tested results, it was confirmed that the effect of strain rates on the stress-strain curves of tensile tests was expressed reasonably well with the use of the multiplicative viscoplastic model and the viscoplastic finite element procedure.
2) In the case of uniaxial creep tests, the simulated creep curves at the highest applied stress level corresponded with the actual creep responses, and the predicted creep curves at low strain levels gave a little variation. This discrepancy suggests the necessity to develop a more appropriate mathematical form of viscoplastic model.
3) In the case of bending creep tests, the numerical results and the experimental ones showed relatively good agreement except for the 450°C case. In this case, numerical results seemed to range from 1.2 to 1.3 times larger than the experimental ones. These discrepancies mainly depended on the lack of the materials constant data at 450°C, and suggest the necessity to develop a more appropriate mathematical form of viscoplastic model.
