Advanced Constitutive Modelling of Soils for Characterization of Seismic Hazard through Nonlinear Dynamic Simulations

加拿大东部的地质条件使地震危险性评估具有挑战性。基岩上的软土层，如敏感粘土，可以产生很大的应变，显著地放大地面运动，并导致运动比其他地质环境传播得更远。例如，在1988年的萨格内地震中，尽管距离震中300公里，但蒙特利尔东部城市酒店（Montreal-Est Hotel de Ville）遭到破坏，不得不拆除。建筑物下面的粘土层厚达17米，这被认为是建筑物倒塌的主要原因。地震波还可以在敏感粘土中引起地震诱发的滑坡，因为它们容易受到循环软化的影响，其特征是由循环加载引起的刚度损失。通过模拟（场地响应分析）或经验场地因子来估计场地放大或减弱地面运动的趋势。模拟可以使用线性、等效线性或非线性方法进行。在剪切应变大于0.1%时，非线性模拟已被证明比等效线性模拟更精确，但它们需要复杂的本构模型，并且对计算的要求更高。经验场地因子通常基于记录的地面运动和等效线性模拟的组合。对于软土剖面，由于等效线性模拟的限制以及地面运动记录数据库中缺乏软土剖面，这些场地因子可能是不准确和不安全的。这些类型的场地因素只能捕获一维场地效应，但多维场地效应，如盆地效应和地形放大也可以控制场地响应。这种复杂的效应只能通过二维或三维非线性模拟来捕捉，但后者是复杂的，因为现有的本构模型往往需要复杂的校准过程，而且往往不能正确地模拟土壤的动态行为，特别是在大应变下。第一个短期目标是为现场响应分析和涉及循环软化的问题创建两个本构模型，以适应从业者的需要，工程师可以输入所需的行为，而不是复杂的参数。第二个短期目标是通过结合数据和非线性模拟，提供一维和二维场地因素的回归模型，这将减少不确定性，从而纠正对软土剖面的偏见。第三个短期目标是在现有案例和模拟的基础上，推导出敏感粘土中边坡在地震荷载作用下破坏概率的回归模型。长期目标是更好地了解地震危害，并为行业提供更好的工具，以促进地震危害的评估。

The geological conditions in Eastern Canada make the assessment of seismic hazard challenging. Soft soil layers, such as sensitive clays, atop the bedrock can develop large strains, significantly amplify ground motion, and cause the motion to propagate much further than other geological settings. For example, in the 1988 Saguenay earthquake, the Montreal-Est Hotel de Ville was damaged and had to be torn down despite being situated 300km away from the earthquake's epicenter. The clay layer beneath the building was 17m thick, which was assessed to be the main cause of the building's demise. The seismic waves can also cause seismically induced landslides in sensitive clays because they are susceptible to cyclic softening, which is characterized by a loss of stiffness induced by cyclic loading.The tendency of a site to amplify or de-amplify ground motion is estimated either with simulations (site response analyses) or empirical site factors. Simulations can be performed using linear, equivalent linear, or nonlinear methods. Nonlinear simulations have proven to be more precise than equivalent linear simulations at shear strains greater than 0.1%, but they require complex constitutive models and are more computationally demanding. Empirical site factors are generally based on a combination of recorded ground motions and equivalent linear simulations. For soft soil profiles, these site factors can be inaccurate and unsafe because of the limitation of equivalent linear simulations and the lack of soft soil profiles in the database of recorded ground motions. These kinds of site factors are derived to capture only 1D site effects, but multidimensional site effects such as basin effects and topographic amplification can also govern site response. Such complex effects can only be captured through 2D or 3D nonlinear simulations, but the latter are complicated because existing constitutive models often require a complex calibration process and often do not model the dynamic behaviour of soils properly, especially at large strains.The first short term objective is to create two constitutive models for both site response analysis and problems involving cyclic softening adapted to the need of practitioners, where engineers can input a desired behaviour, instead of complex parameters. The second short term objective is to provide regression models for 1D and 2D site factors by combining data and nonlinear simulations, which will reduce uncertainty and thus correct the bias toward soft soil profiles. The third short term objective is to derive a regression model, based on existing cases and simulations, for the probability of the failure of slopes in sensitive clays under earthquake loading. The long term objectives are to gain a better understanding of seismic hazard and to provide better tools to the industry to facilitate the assessment of seismic hazard.