Collaborative Research: Validation of Constitutive and Numerical Modeling Techniques for Soil Liquefaction Analysis

在地震期间，土壤液化是一个普遍存在的问题，它对美国和世界其他地区的民用基础设施造成了重大而昂贵的破坏。实践工程师和学者一直在使用计算模型和数值模拟工具来设计和分析涉及可液化土壤的地质系统。这些工具是非常复杂和精密的，但在过去的许多大地震之后，它们经常被修改，以纳入从事件后观察和调查中吸取的教训。因此，这些工具的有效性还有待充分确定。液化实验和分析项目（LEAP）是一个国际合作实验室，利用离心机测试产生一套高质量的实验数据，然后利用这些数据建立现有计算模型和模拟程序的有效性和适用性范围，用于土壤液化分析。合作实验室包括加州大学戴维斯分校、伦斯勒理工学院、剑桥大学（英国）、京都大学（日本）、北洋理工大学（台湾）、浙江大学（中国）、IFSTTAR（法国）、KAIST和KWater（韩国）以及香港科大（香港）的离心机设施。该项目利用这些设施的实验资源来补充参与的美国机构的测试能力，以调查一些具有挑战性的问题，其中土壤液化起着突出的作用。三个连续的项目（LEAP-2017， -2018和-2019）计划用于解决三个重要问题：可重复性和不确定性；土壤结构效应；以及双轴地震动对土壤液化的影响。LEAP-2017将采用改进的质量控制措施，以最大限度地减少不确定性，并将设计和实施广泛的测试程序，以表征相对简单的横向扩散离心机测试的灵敏度和不确定性。LEAP-2018将产生高质量的离心机数据，以评估预测织物和分层影响的计算工具的适用性。LEAP-2019将利用合成地震和真实地震运动来评估数值工具在一系列双轴震动离心机实验中的性能。在每次LEAP之后，将举行一个研讨会，讨论实验和数值模拟的结果。该项目产生的数据将使用新的数据库发布机制进行组织、存档和公开发布。有组织的数据将有助于(1)未来的验证工作，(2)其他正在进行的液化研究，以及(3)岩土建模教育。许多研究生都参与了这个项目，并将获得独特的培训机会，超出任何一个机构单独的能力。将开发一个验证驱动框架（Validation Driver Framework， VDF），以允许建模者和从业者有效地验证他们的代码。

Soil liquefaction is a pervasive problem during earthquakes that has caused significant and costly damage to civil infrastructure in the US and other parts of the world. Practicing engineers and academics have been using computational models and numerical simulation tools in the design and analysis of geosystems involving liquefiable soils. These tools are highly sophisticated and refined, but they often have been revised following many major past earthquakes to incorporate the lessons learned from post-event observations and investigations. Thus, the validity of these tools is yet to be fully established. The Liquefaction Experiments and Analysis Projects (LEAP) is an international collaboratory to produce a set of high quality experimental data using centrifuge testing, and then to use these data to establish the validity and range of applicability of existing computational models and simulation procedures for soil liquefaction analysis. The collaboratory includes centrifuge facilities at the University of California at Davis, Rensselaer Polytechnic Institute, Cambridge University (UK), Kyoto University (Japan), NCU (Taiwan), Zhejiang University (China), IFSTTAR (France), KAIST and KWater (Korea), and HKUST (Hong Kong). The project leverages the experimental resources at these facilities to complement the testing capabilities in the participating US institutions for investigating a number of challenging problems in which soil liquefaction plays a prominent role.Three consecutive projects (LEAP-2017, -2018, and -2019) are planned which will be used to address three important issues: reproducibility and uncertainty; effects of soil fabric; and effects of biaxial ground motion on soil liquefaction. LEAP-2017 will use improved quality control measures to minimize uncertainties, and an extensive testing program will be designed and conducted to characterize the sensitivity and uncertainty for relatively simple lateral spreading centrifuge tests. LEAP-2018 will produce high quality centrifuge data to enable the assessment of the applicability of computational tools for predicting the influence of fabric and layering. LEAP-2019 will assess the performance of numerical tools against a series of biaxial shaking centrifuge experiments using synthetic and real earthquake motions. Following each LEAP, a workshop will be held to discuss the results of experiments and numerical simulations. The data resulting from this project will be organized, archived and publically released using a new database publication mechanism. The organized data will be useful for (1) future validation efforts, (2) other ongoing liquefaction research, and (3) in geotechnical modeling education. A number of graduate students are involved in the project and will receive unique training opportunities beyond the capabilities of any one institution alone. A Validation Driver Framework (VDF) will be developed to allow modelers as well as practitioners to efficiently validate their codes.