Mahin, Deierlein等，针对地震模拟与减灾的SimCenter计算框架

针对地震模拟与减灾的SimCenter计算框架

SimCenter Computational Framework for Simulating and Mitigating the Effects of Earthquakes

Mahin, S.A., Deierlein, G.G., Govindjee, S., Lowes, L.N., Kareem, A., Crittenden, C., McKenna, F., Schoettler, M.

Proceedings of the 11th National Conf. in Earthquake Engrg., Earthquake Engineering Research Institute, Los Angeles, CA. 2018

       美国国家科学基金会资助的自然灾害工程研究基础设施（NHERI）SimCenter正在研发系列技术。该技术能提供计算建模和仿真软件工具包、用户支持和必要的教育材料，以提升（科研工作者）模拟自然灾害对土木基础设施影响的能力。SimCenter的方法是开发一系列软件技术，该技术可以完美集成诸多拥有高性能计算平台和数据存储库的仿真软件。本文简要回顾了SimCenter的计划和方法——通过开发计算工作流试验平台来模拟7级地震对旧金山湾区大量建筑物的破坏效应。

(The NSF-funded Natural Hazards Engineering Research Infrastructure (NHERI) SimCenter is developing technologies to provide access to computational modeling and simulation software tools, user support, and educational materials needed to advance the capability to simulate the impact of natural hazards on civil infrastructure. The SimCenter’s approach is to create software technologies that can seamlessly integrate a broad array of simulation software with high performance computing platforms and data repositories. This paper briefly reviews the SimCenter’s plans and approach through the development of a computational workflow testbed to simulate the damaging effects of a M7 earthquake scenario on a large inventory of buildings in the San Francisco Bay Area.)

       计算机仿真模拟是地震工程研究和实践的重要组成成分，可用于评估和减轻地震对社区的破坏效应。SimCenter作为美国国家科学基金（NSF）NHERI计划的一部分，于2016年秋创立。SimCenter提供计算机模型和仿真软件工具包、用户支持和必要的教育材料，旨在提高（科研工作者）模拟自然灾害对建成环境和社区的影响的能力。该中心在认识到这项挑战涉及面广，多学科交叉且涵盖了范围广泛的多种灾害后，SimCenter的解决方案是创造一系列计算工作流技术，该技术可以完美集成诸多已有的拥有高性能计算平台和数据库的仿真软件，从而可以利用现有的软件平台。本文和此次演讲将通过对旧金山湾区的地震仿真模拟开发的计算工作流程来介绍这些理念。关于SimCenter更广泛的研究和教育工具开发的详情，可以访问http://simcenter.designsafe-cl.org

(Computational simulation is as an essential component of earthquake engineering research and practice to assess and mitigate the damaging effects of earthquakes on communities.  Supported as part of the National Science Foundation’s Natural Hazards Engineering Research Infrastructure (NHERI) program, the SimCenter was created in fall 2016 to access computational modeling and simulation software tools, user support, and educational materials needed to advance the capability to simulate the impact of natural hazards on the built environment and communities. Recognizing the challenge as broad and multi-disciplinary, encompassing multiple hazards across a wide range of scales, the SimCenter’s approach is to leverage existing software platforms by creating computational workflow technologies that can seamlessly integrate a broad array of simulation software with high-performance computing  platforms and  data repositories. This paper and presentation will illustrate these ideas through a computational workflow developed for an earthquake simulation of the San Francisco Bay Area. Further details on the broader research and educational tool development by the SimCenter are available at: https://simcenter.designsafe-ci.org.)

       如图1所示，计算区域模拟将单个设施与基础设施网络的详细性能分析和地震地面变形与震动的高分辨率仿真相结合。此方法的整体概念与HAZUS以及其他区域性损失评估软件没有什么不同。只是所提出的方法旨在巨大的（dramatically）提高地震灾害仿真模拟的分辨率和真实感，并对建筑物和土木基础设施的震后的损坏和性能进行分析，并与城市规划工具相整合。借助（through）这些高分辨率的模型，上述仿真模拟除可以进行财务损失评估外，还可以为制定政策、采取工程措施和其他步骤提供参考信息，从而提高社区的韧性。

(As illustrated in Figure 1, the computational regional simulations combine detailed performance based analyses of individual facilities and infrastructure networks with high-resolution simulations of earthquake ground deformations and shaking. The overall concept is not unlike that of HAZUS or other regional loss assessment software, except that the proposed methods aim for dramatically higher resolution and realism in earthquake hazard simulations, analyses of damage and performance of buildings and civil infrastructure, and integration with urban planning tools. Through these higher resolution models, these simulations can move beyond financial loss assessment to informing the development of policies, engineering measures and other steps to improve community resilience.)

图1 区域地震模拟的组成部分

Fig. 1 Components of Regional Earthquake Simulation

       传统建模软件的计算过程通常在单个软件内部，与之不同的是，现代工作流结构可以实现多种软件模块之间的链接和数据交换，并且能够实现分布式计算。为了开发计算工作流，SimCenter利用Pegasus系统。该系统将高级工作流描述，映射到分布式计算资源(https://pegasus.isi.edu/)。模块化工作流允许为区域仿真的不同组成部分提供代替模型和数据库。例如：特定场地的地震灾害和地面运动包括：（1）基于物理的地震模拟得到的地震运动（比如SCEC的Cybershake平台），（2）能直接使用或作为目标来选择和调幅地面运动记录或模拟地震运动目标的地面运动烈度，该参数与空间位置有关，一般基于经验方法获得，（3）地震时记录的数据（例如：USGS ShakeMaps）或者4）其他数据集。同样，建筑物的性能评估可以基于以下不同的数据库和模型：（1）详细的非线性时程分析和单个建筑物的性能评估（例如：使用FEMA P58），（2）计算建筑损伤和经济损失的经验函数（例如，使用HAZUS类型的函数），或（3）详细分析和经验函数相结合。对于分布式的交通网络和基础设施网络组件与系统可以采用类似的组合。最后，性能评估的结果可以以其他方式进行处理、解释和可视化，以支持风险明确的决策，包括将地震性能指标与城市规划工具联系起来。

(In contrast to conventional modeling software, where the computations are typically performed within a single software program, modern workflow architectures facilitate the linking and data exchange between multiple software modules and distributing the calculations across multiple computing resources. To develop the computational workflows, the SimCenter is utilizing the Pegasus system, which maps high-level workflow descriptions onto distributed computing resources (https://pegasus.isi.edu/). The modular workflow allows alternative models and datasets for various components of the regional simulations. For example, the site specific earthquake hazard and ground motions can include: (1) seismographs from physics-based earthquake simulations (e.g., SCEC’s Cybershake platform), (2) spatially correlated empirical ground motion intensity measures that may be used directly or as targets to select and scale recorded or simulated seismographs, (3) data from recorded earthquake events (e.g., USGS ShakeMaps), or (4) other data sets. Similarly, building performance evaluations can be based on variety of data sets and models, ranging from (1) detailed nonlinear dynamic analyses and performance assessments of individual buildings (e.g., using FEMA P58), (2) empirical building damage and loss functions (e.g., using HAZUS type functions), or (3) combinations of detailed analyses and empirical functions. Similar combinations can be envisioned for distributed transportation and infrastructure network components and systems. Finally, the results of the performance evaluations can be processed, interpreted and visualized in alternate ways to support risk-informed decision making, including linking of the seismic performance metrics into urban planning tools.)

       为了促进计算工作流程的开发和演示应用，SimCenter开发了旧金山湾区的区域地震模拟的试验平台应用（见图2）。实验平台初步实施包括对旧金山湾区内180万建筑的分析，这些建筑受到来自Hayward断层7级模拟地震波引起的地面运动的影响。地震的仿真由劳伦斯利弗莫尔国家实验室的Rodgers，Pitarka和Peterson使用USGS 3D北加州地球模型实行，最终在该地区范围内2km的网格上产生地震动。基于分块级别解析度的建筑属性数据是从UrbanSim城市规划平台的一个数据集中获取的(http://www.urbansim.com/)。180万建筑物中的每栋建筑物都采用简化的（有限自由度的）OpenSees模型进行分析，其性质可以通过Xiong等[1]提出的方法，根据建筑物几何信息，建造日期以及建筑使用功能推断出来。基于OpenSees模型计算得到的峰值层间位移角和楼面加速度后，即可依据FEMA P58确定建筑物损坏，维修成本和停工时间。

(To facilitate development and demonstrate application of the computational workflows, the SimCenter developed a testbed application of a regional earthquake simulation for the San Francisco Bay Area (see Figure 2). The initial implementation of the testbed included analyses of 1.8 million buildings in the SF Bay Area, subjected to ground motions from a simulated M7.0 earthquake on the Hayward Fault. The earthquake simulation, run by Rodgers, Pitarka and Peterson from Lawrence Livermore National Lab using the USGS 3D Northern California Earth Model, produced seismograms on a grid spacing of 2 km across the region. The building inventory, described at parcel-level resolution, is based on a dataset from the UrbanSim platform for urban planning (http://www.urbansim.com/). Each of the 1.8 million buildings in the inventory are analyzed using a simplified (limited degree of freedom) OpenSees model, properties of which are inferred from the building geometries, construction date and occupancy using an approach developed by Xiong et al.[1]. Using peak drifts and floor accelerations from the OpenSees models, the building damage, repair costs and repair downtime are determined based on an implementation of FEMA P58.)

图2  旧金山湾区Hayward断层地震模拟

Fig. 2 Illustration of San Francisco Bay Area Hayward Fault Earthquake Simulation

       由此产生的建筑破坏和建筑和性能指标可从以下几个方面来解释。如图2所示，结果可以使用标准地理信息系统进行映射，例如使用开源系统QGIS (https://www.qgis.org/)，或者可以使用3D城市多边形建模工具[2]将建筑响应做成动画（如图2）。为了促进与城市规划者和社会学家的交流，SimCenter与Paul Waddell和UrbanSim开发人员合作，将高分辨率建筑性能数据整合到UrbanSim平台中。如图3所示，此集成使得使用者能利用城市规划工具，进行多种类型的查询，从而获得区块级别的建筑物损伤、损失和修复信息。

(The resulting damage and performance measures of the buildings can be interpreted in a few ways. As shown in Figure 2, the results can be mapped using standard geographic information systems (GIS) software, such as the open source system, QGIS (https://www.qgis.org), or building response can be animated (inset graphic in Figure 2), using a 3d urban polygonal modeling tool [2]. To facilitate integration with urban planners and socialscientists, the SimCenter is collaborating with Paul Waddell and UrbanSim developers, to integrate the high-resolution building performance data into the UrbanSim platform. As shown in Figure 3, this integration enables various types of queries to interrogate the parcel-level building damage, loss and recovery information with urban planning tools.)

图3 Urbansim模拟结果的地块级别查询

Fig. 3 Parcel Level Interrogation of Simulation Results in UrbanSim

       本文描述的区域地震仿真试验台旨在作为概念证明，可以帮助开发工作流工具以及展示工作流在实际地震场景中的应用。这个测试平台是一个手段，目的是帮助增强地震工程、城市规划和社会科学研究人员的计算工作流工具，使他们能够进行有意义的多学科研究，有助于提高社区韧性。

(The regional earthquake simulation testbed described in this paper is intended as a proof of concept to help develop the workflow tools and demonstrate their application to a realistic earthquake scenario. This testbed is one means to an end to help empower earthquake engineering, urban planning and social-science researchers with computational workflow tools that will enable meaningful multi-disciplinary research that will contribute to improving community resilience.)

       SimCenter得到了美国国家科学基金会（No.1612843）的资助。除了NSF的财务支持，作者还要感谢NSF CMMI项目主管Joy Paushke博士的支持和鼓励，以及SimCenter和NHERI活动中许多参与者的贡献。特别是，作者感谢以下团队对区域试验台的贡献：来自劳伦斯利弗莫尔国家实验室的A.Rodgers，A.Pitarka和N.A.Petersson，来自清华大学的陆新征教授和来自加州伯克利分校P. Waddell。

(The SimCenter is supported through a grant from the National Science Foundation (#1612843). In addition to NSF’s financial support, the authors gratefully acknowledge the support and encouragement of the NSF CMMI program manager, Dr. Joy Paushke, along with the contributions of the many participants in the SimCenter and NHERI activities. In particular, the authors acknowledge contributions from the following groups to the regional testbed: A. Rodgers, A.Pitarka, and N.A.Petersson from Lawrence Livermore National Lab, X. Lu from Tsinghua University, and P. Waddell from U.C. Berkeley.)

参考文献

[1] Xiong, C., Lu, X., Lin, X., Xu, Z., Ye, L. (2017), “Parameter Determination and Damage Assessment for THA-Based Regional Seismic Damage Prediction of Multi-Story Buildings,” Jl. Of Eq. Engrg., V 21, Iss.3, pp. 461-485.

[2] Xiong, C., Lu, X., Hori, M., Guan, H., Xu, Z. (2015), “Building seismic response and visualization using 3D urban polygonal modeling,” Automation in Construction, V55, pp. 25-24.

本文翻译：郑哲