The 3rd generation high-strength Al-Li alloy, AA2099 alloy has high specific strength and hardness, strong corrosion resistance, favorable service characteristics at ultra-low temperatures and good damage tolerance. Especially, its fine-grained compact fiber structure which does not have any internal defects provides high thermal creep strength and good surface quality after hot working. These advantages make AA2099 alloys an alternative to the traditional high-strength aluminum alloys in aerospace fields.
Like other high-strength Al-Li alloys, AA2099 alloys have such weak plasticity at room temperature that the cold plastic working methods are not applicable to making machine parts or workpieces which require complicated or large deformation. Therefore, hot working is the main choice. Accurate prediction of the flow stress of AA2099 alloys in hot working process is of primary importance in economizing Al-Li alloys widely used in aerospace industry, designing rational hot working processes such as hot rolling and forging, and increasing the processing efficiency.
During hot working process, the plastic deformation exerted on the material causes changes in the microstructure and mechanical properties, which are shown as the change in flow stress. Therefore, the development of constitutive material models is of major theoretical and practical significance in predicting the flow stress at different deformation temperatures, strain rates and strains to provide important theoretical basis for the selection of hot working equipment and engineering design.
The constitutive models developed so far include phenomenological models, physical models and ANN models. Phenomenological constitutive models include Johnson-Cook (J-C), Arrhenius (AH) models, etc. which are simple to develop, but have low prediction accuracy. In contrast, physical models such as microstructure-based models (MS-B) have very high prediction accuracy, but they are complex to express and difficult to develop. On the other hand, ANN constitutive models require a large number of experimental data to train the neural network, so it costs a lot to develop them. In a word, each model has its own advantages and disadvantages.
Han Yun Hyok, a student at the Faculty of Material Science and Technology, has developed phenomenological models (modified J-C and strain compensated Arrhenius (SC-AH) model) and a physical model (MS-B model) for predicting the hot flow stress behavior in AA2099 alloy, and identified a constitutive model that can predict the flow stress of this alloy most accurately by comparing their predictability.
He estimated the prediction accuracy by means of the average absolute relative error and correlation coefficient. The accuracy was 5.09% and 0.992 8 for the modified Johnson-Cook model, 6.12% and 0.9820 for the strain compensated Arrhenius model, and, 1.58% and 0.9994 for the microstructure-based model.
The accuracy of the proposed MS-B model was the highest with 1.58% and 0.9994 and the flow stress predicted by the proposed microstructure-based model was in good consistent with the experimental results. This indicates that the proposed microstructure-based model can accurately describe the hot deformation behavior of AA2099 Al-Li alloy.
You can find the details in his paper “Constitutive Analysis on Hot Deformation Behavior of AA2099 Alloy for Modeling of Flow Stress” in “Mechanics of Solids” (SCI).