Conventional strain-based numerical prediction assumes that failure occurs when ductility is exhausted or accumulation of creep strain reaches the critical failure strain. Due to instability at the onset of rupture, the failure strain value appears to be scattered and leads to the erroneousness in prediction. In this paper, a new local constraint-based damage model incorporating the Monkman–Grant ductility, as a measure of strain during uniform creep deformation stage, was implemented into a Finite Element (FE) model to predict the creep damage and rupture of Grade 92 steel under uniaxial and multiaxial stress states. The prediction was applied on plain and notched bar specimens with various notch acuities. The uniaxial stress-dependent Monkman–Grant (MG) failure strain was adopted in the FE to simulate the influence of the constraints which were induced by the creep damage. The implication of reduced failure strain in long-term creep time on the rupture prediction is discussed. The multiaxial MG failure strain of the notched bar, which has a lower value than uniaxial failure strain due to the geometrical constraint, was estimated based on the linear inverse relationship between normalised MG failure strain and normalised triaxiality factor. It was found that the results obtained from the proposed technique were in good agreement with the experimental data within the scatter band of ± factor of 2. It was shown that MG failure strain can be used as an alternative to strain at fracture. MG strain outweighed strain at fracture because the determination of its value only required short-term testing to be performed. In most cases considered in the present investigation, the rupture-type fracture was predicted, however, there was evidence that under high constraint and low stress, stable crack propagation occurred before fracture. The location of the maximum creep damage was found to be dependent on the creep time, geometry or acuity level of the specimen. For sharp notch specimen, the failure was initiated near the notch root, however, as the notch radius increased, the initiation location moved further away towards the specimen centre.

传统的基于应变的数值预测假设在延展性耗尽或蠕变应变累积达到临界破坏应变时发生破坏。由于破裂开始时的不稳定性，破坏应变值看起来很分散，导致预测错误。在本文中，一种新的基于局部约束的损伤模型结合了 Monkman-Grant 延展性，作为均匀蠕变变形阶段应变的量度，被实施到有限元 (FE) 模型中，以预测 92 级的蠕变损伤和破裂钢在单轴和多轴应力状态下。该预测应用于具有各种缺口锐度的普通和缺口棒材试样。在有限元中采用单轴应力相关的 Monkman-Grant (MG) 破坏应变来模拟蠕变损伤引起的约束的影响。讨论了长期蠕变时间中失效应变降低对破裂预测的影响。由于几何约束，缺口杆的多轴 MG 失效应变值低于单轴失效应变，根据归一化 MG 失效应变和归一化三轴度因子之间的线性反关系进行估计。发现从所提出的技术获得的结果与±因子2的散射带内的实验数据非常一致。表明MG失效应变可以用作断裂应变的替代。MG 应变超过断裂应变，因为确定其值只需要进行短期测试。在目前的调查中考虑的大多数情况下，破裂型断裂是预测的，然而，有证据表明在高约束和低应力下，裂纹在断裂前发生稳定扩展。发现最大蠕变损伤的位置取决于试样的蠕变时间、几何形状或敏锐度水平。对于尖锐缺口试样，破坏开始于缺口根部附近，然而，随着缺口半径的增加，起始位置向试样中心移动得更远。有证据表明，在高约束和低应力下，裂纹在断裂前稳定扩展。发现最大蠕变损伤的位置取决于试样的蠕变时间、几何形状或敏锐度水平。对于尖锐缺口试样，破坏开始于缺口根部附近，然而，随着缺口半径的增加，起始位置向试样中心移动得更远。有证据表明，在高约束和低应力下，裂纹在断裂前稳定扩展。发现最大蠕变损伤的位置取决于试样的蠕变时间、几何形状或敏锐度水平。对于尖锐缺口试样，破坏开始于缺口根部附近，然而，随着缺口半径的增加，起始位置向试样中心移动得更远。