考虑时效损伤的硬岩非线性蠕变本构模型研究

Nonlinear creep intrinsic modeling of hard rock considering aging damage

为探究硬岩蠕变全过程中各阶段的变形特征,对硬质凝灰岩开展恒围压、分级升轴压的室内蠕变试验。通过不同应力路径下硬岩蠕变特征分析,结合扫描电镜(SEM)测试探讨岩石细观损伤演化机制与宏观破坏特征之间的关系;基于蠕变过程中岩石力学性能随时间的劣化效应,引入损伤变量构建能描述硬岩蠕变全过程的损伤本构模型;引入优化的BP神经网络模型,改善模型参数的识别与预测效果。研究结果表明:随偏应力水平提高,硬质凝灰岩表现出较为明显的脆–延性变形转化特征,且岩石蠕变现象更加明显;在较高偏应力持续作用下,岩石损伤累积与力学性能劣化更加充分,进而引起破坏阶段硬质凝灰岩的加速蠕变特征随围压增加而突显,硬岩破坏变得更加剧烈。改进后的损伤本构模型与Burgers模型相比能更好地描述硬岩蠕变全过程,基于优化后的BP神经网络模型,能准确预测高应力水平下硬岩的加速蠕变曲线,提高了模型的适用性。研究成果不仅丰富了硬岩蠕变本构模型的理论研究,还可为硬岩地层中工程岩体长期稳定性设计提供理论支持。

In order to investigate the deformation characteristics of hard rock during the whole creep process,this study carried out indoor creep tests on hard tuff with constant peripheral pressure and graded ascending axial pressure.Through the characterization of hard rock creep under different stress paths,combined with scanning electron microscope(SEM) tests,the relationship between the fine-scale damage evolution mechanism and macroscopic damage characteristics was explored.Based on the introduction of damage variables into the deterioration effect of rock mechanical properties over time during the creep process,an intrinsic model of damage creep was constructed that can describe the whole process of hard rock creep.At the same time,by optimising the BP neural network model,the identification and prediction of the model parameters were improved.The results show that,with increasing the bias stress level,the hard tuff shows more obvious brittle-ductile deformation transformation characteristics,and the rock creep phenomenon is more obvious.Under the continuous action of a higher bias stress,the accumulation of rock damage and the deterioration of mechanical properties are more adequate,which leads to the accelerated creep characteristics of the hard tuff during the damage stage with increasing the peripheral pressure,and the damage of the hard rock becomes more intense.Compared with Burgers,the improved damage constitutive model can better describe the whole process of hard rock creep,which verifies the reasonableness of the model.Based on the optimised BP neural network model,the accelerated creep curve of hard rock under high stress level can be accurately predicted,which improves the applicability of the model.The results of this study not only enrich the theoretical study of hard rock creep constitutive model,but also provide theoretical guidance for the design of long-term stability of engineering rock bodies in hard rock formations.