中文摘要：

在法向与切向荷载的共同作用下,节理剪切应力–位移曲线宏观表现为峰前呈剪切硬化、峰后呈剪切软化现象：当微凸体爬坡效应占主导地位时,宏观表现为剪切硬化;当微凸体在剪切过程中磨损累积到一定程度时,宏观表现为剪切软化。剪胀硬化、磨损软化共同控制节理的剪切力学行为。在详细分析已有的剪切应力–位移本构模型的局限性的基础上,提出硬化–软化全剪切本构模型,采用单个函数反映节理剪切位移曲线的变化特征,其优点为：（1）避免分段讨论剪切位移曲线特征、分段求取拟合参数的麻烦;（2）在整个讨论区域内存在一个峰值且峰值处斜率为0;（3）当剪切位移足够大时,强度趋于残余值,体现出节理面材料的力学特征。对不同类型的剪切位移曲线进行分析,该模型均表现出较强的适应性。同时分析现有的剪胀模型,其主要缺陷是不能反映初始剪缩现象与体现真正的剪胀起始点。在分析典型剪胀曲线基本特征的基础上,采用分段函数考虑剪胀特征,能较好地拟合剪胀曲线。

英文摘要：

Laboratory direct shear tests show that shear stress-displacement curve usually follows the pre-peak hardening and post-peak softening rule. When the asperities overriding effect is dominant, the curve shows stress hardening： and while damage and wear is accumulated to a certain extent, it will result in stress softening behavior. The hardening and softening effects together control the joint shear behavior. A new generalized hardeningsoftening constitutive model is put forward after detailed analysis of the existing model＇s limitations. One function is used to describe the features of the pre- and post-shear curves and its advantages are as follows： （1） Don＇t need discuss the characteristics of shear displacement curves for several segments and its fitting parameters separately. （2） Throughout the entire shear displacement range, this function has only one peak and at this point, its slope is zero. （3） Shear stress tend to residual strength when shear displacement is large enough which reflects the basic features of the joint material. The new model is suitable for analysis three types of shear displacement stress curves, which arc shear stress hardening curve, hardening-softening curve and curve with an initial shear stress. Meanwhile, the limitations of the existing dilation models are that they cannot reflect negative dilation and the real point where the joint starts to dilate. So, a new normal-shear displacement coupling model is derived which takes two types of functions to express the different phenomena during direct shear test. At last, validity of the two proposed constitutive models is verified through applications to several existing laboratory test results. There is a high fitting accuracy between fitting results and experimental results.