Speed-up of isothermal forging processes of titanium aluminides by microstructure-adapted control of ram speed

金属间钛铝化物（TiAl）的锻造，例如用于现代航空发动机的涡轮叶片，目前是两阶段等温锻造过程，温度至少为1200°C，每冲程的加工时间为几分钟。这与漫长的工艺时间、极其昂贵的钼基模具上的高负荷和高制造成本有关，这阻碍了TiAl的广泛使用。该项目的目的是缩短TNM-B1的等温锻造过程，这种过程在工业上以恒定的冲压速度进行，通过有针对性地控制冲压运动。第一个资助期的结果可以通过实验和模拟证明，通过先前的热处理可以显著加速成形过程，而不会对材料造成更大的损害。到目前为止，只考虑了材料特有的硬化/软化行为，而没有考虑材料的高方差、工件几何形状的影响或变形过程中材料的复杂结构和损伤发展。为了包含这些因素，将开发一种结合机器学习（ML）算法、现有FEM模型和损伤模型的过程控制。采用有限元法对成形过程中材料的变化特性和工件的不同几何形状进行了模拟。将Gurson-Tvergaard-Needleman损伤模型应用于钛铝材料，并与有限元模型相结合。然后，机器学习算法将根据工件几何形状、材料方差和预测的损伤行为来调整冲压速度。为了更深入地了解TNM-B1在热成形过程中的微观组织发展，将进行现场压力测试并附带组织和相分析（DESY - Petra III）。结果将提供有关确切相组成和再结晶起始的信息。还期望详细说明为什么热处理导致屈服应力降低，无论测试条件如何。在较大部件（如涡轮叶片）的热处理过程中，会形成温度梯度。靠近表面的区域比部件内部冷却得快，这导致部件横截面上不同的球状、片层和珠光体/细胞微观结构。为了评估微观结构梯度，锻造坯料将通过整个工艺链，包括恒定和加速测试条件。此外，将开发后热处理以获得所需的最终显微组织和机械性能。

The forging of intermetallic titanium aluminides (TiAl), used e.g. for turbine blades in modern aero engines, currently occurs as a two-stage isothermal forging process, at temperatures of at least 1200°C and process times of several minutes per stroke. This is associated with long process times, high loads on the extremely expensive molybdenum-based dies and high manufacturing costs, which prevents the wider use of TiAl. The aim of this project is to shorten the isothermal forging process of TNM-B1, which industrially takes place with constant ram speed, by targeted control of the ram movement. Results of the first funding period could demonstrate, both, experimentally and simulatively, that the forming process can be accelerated significantly, without causing increased damage to the material, by a previous heat treatment. So far, only the material-specific hardening / softening behavior has been taken into account and not the high variance of the material, the influence of the workpiece geometry or the complex structure and damage development of the material during deformation. In order to include these factors, a process control will be developed that combines a machine learning (ML) algorithm, the existing FEM model and a damage model. FEM is used to simulate the variable material behavior and the different workpiece geometries during the forming process. A Gurson-Tvergaard-Needleman damage model will be adapted to titanium aluminides and coupled with the FEM model. The ML algorithm will then adjust the ram speed according to the workpiece geometry, the material variance and the predicted damage behavior. In order to gain more in-depth knowledge about the microstructure development of TNM-B1 during hot forming, in-situ pressure tests with accompanying structure and phase analysis (DESY - Petra III) will be carried out. The results will provide information about the exact phase composition and the initiation of recrystallization. Details of why the heat treatment results in decreased yield stress, regardless of the test conditions, are also expected. During the heat treatment of larger components, e.g. Turbine blades, a temperature gradient is formed. The areas close to the surface cool faster than the inside of the component, which leads to different globular, lamellar and pearlitic / cellular microstructures over the component cross-section. In order to evaluate the microstructure gradient, a forging blank will pass the entire process chain, both, for constant and for accelerated test conditions. In addition, a post-heat treatment will be developed to achieve the desired final microstructures and mechanical properties.