Alloy 718 ingot with a diameter of 300mm was made by the vacuum melting process; VIM (vacuum induction melting) followed by VAR (vacuum arc re-melting). The VIM/VAR ingot was heat-treated for homogenization, and casting structure of the ingot was broken down for uniform microstructures and mechanical properties by controlled cogging processes using a hydraulic press. The VIM/VAR-processed ingot contains three different microstructure zones along radial direction, i.e. surface chill zone, intermediate columnar gain zone, and central equiaxed zone, because the local solidification procedure varies depending on locations within the ingot. To understand the local deformation behavior and microstructure evolution, compression tests were conducted on samples collected from different zones of the ingot in wide temperature and strain rate ranges, i.e. 900~1150 o C and 0.01 s -1 . The existence of different microstructures within the ingot resulted in different compression behaviors, which was attributed to the preferred orientation in the columnar grain zone, in comparison with the equiaxed grains in the central region. At large strains, the initial difference in microstructure eventually disappeared due to dynamic, meta-dynamic, and static recrystallizations. Constitutive relations were established for the simulation of microstructure evolution, which was applied to the billet cogging process.

An artificial neural network (ANN) model was applied to simulate the phase volume fraction oftitanium alloy under isothermal and non-isothermal hot forging condition. For isothermal hot forging process, equilibrium phase volume fraction at specific temperature was predicted. For this purpose, chemical composition of six alloy elements (i.e. AI, Y, Fe, 0, N, and C) and specimen temperature were chosen as input parameter. After that, phase volume fraction under non-isothermal condition was simulated again. Input parameters consist of initial phase volume fraction, equilibrium phase volume fraction at specific temperature, cooling rate, and temperature.The ANN model was coupled with the FE simulation in order to predict the variation of phase volume fraction during non-isothermal forging. Ti-6AI-4Y alloy was forged under isothermal and non-isothermal condition and then, the resulting microstructureswere compared with simulated data.