深圳地区原状花岗岩残积土硬化土模型参数的试验研究

硬化土（hardening soil）模型是岩土工程数值分析中常采用的模型。为研究加载方式和排水条件对原状花岗岩残积土剪切行为的影响，对深圳地区原状花岗岩残积土进行了常规三轴固结排水、常规三轴固结不排水、固结排水侧向卸载和固结不排水侧向卸载4组试验，并由试验结果确定出不同加载方式和排水条件下的硬化土模型参数。结果表明：在常规三轴试验中，试样在剪切过程中表现为先剪缩后剪胀的特性，在侧向卸载试验中，试样始终表现为剪胀的特性；模型参数值与加载路径密切相关，固结不排水侧向卸载试验（CDLU）测得的有效内摩擦角 较常规三轴固结不排水试验（CD）试验大20%，而有效黏聚力 小了46%；CDLU试验得到的三轴压缩试验的参考割线模量 为CD试验的2.9倍，卸载再加载参考割线模量 为CD试验的1.8倍；在相同加载路径下，模型参数值也受排水条件的影响，由CU试验得到的 与CD试验的结果相近，但CU试验得到的 要明显大于CD试验的结果，CU试验得到的 为CD试验的2倍， 为CD试验的3.8倍。因此，在岩土工程数值分析中应根据工程的实际情况确定和选用模型参数值。

Experimental study on parameters of the hardening soil model for undisturbed granite residual soil in Shenzhen

Hardening soil model is commonly used in geotechnical numerical analyses. In order to study the effect of loading path and drainage condition on the shearing behaviors of undisturbed granite residual soil. Four series of triaxial tests, including the conventional consolidated drained tests, the conventional consolidated undrained tests, the consolidated drained laterally unloading tests and the consolidated undrained laterally unloading tests have been conducted on the undisturbed samples of granite residual soil in Shenzhen. The values of model parameter for different conditions were also calibrated from the test results. It was observed that the specimens first contracted and then dilated in the shearing stage of the conventional triaxial test, but always dilated in the laterally unloading tests. The values of model parameters were closely related to the loading path. The effective internal friction angle measured by the CDLU tests was 20% higher, and the effective cohesion was 46% smaller than those by the CD tests. The reference secant modulus determined from the CDLU tests was 2.9 times, and the unloading-reloading reference secant modulus was 1.8 times those from the CD tests. The values of model parameters were also affected by the drainage conditions. The effective internal friction angle obtained from the CU tests was similar to that from the CD tests, but the effective cohesion obtained from the former was significantly larger than that from the latter. The reference secant modulus determined from the CU tests was twice that from the CD tests, and the unloading-reloading reference secant modulus obtained from the former was 3.8 times that from the latter. Hence, the values of parameters should be determined in accordance with the real loading and drainage conditions for geotechnical numerical analyses.