Project Title/ID Number 3D Simulation of Seismic Ground Deformation in OpenSees —2202002
Start/End Dates 10/1/02—9/30/03
Project Leader Ahmed Elgamal (UCSD/Faculty)
Team Members Zhaohui Yang (UCSD/Post Doc), Jinchi Lu (UCSD/Grad Student)
Project goals and objectives

This project complements other ongoing efforts related to the open computational simulation platform OpenSees (led by Professor Greg Fenves). Specifically, the project is focused on:

  1. adding new geotechnical simulation capabilities
  2. calibration of these capabilities using available data sets
  3. advancing the spatial simulation capabilities
  4. integrating the Soil and Structural simulation components
  5. exploring the seismic response of combined large-scale soil-structure systems
  6. advancing the PEER Humboldt Bay Bridge Testbed efforts.

Recent work has been addressing:

  • Further development of the three-dimensional (3D) analysis capabilities for nonlinear soil-structure systems
  • Related Parallel-processing challenges (in collaboration with Prof. Kincho Law, Prof. Greg Fenves and the San Diego Supercomputing Center)
  • Computational developments related to spatial definition of ground motion (in collaboration with Prof. Jacobo Bielak and Prof. Joel Conte).
Role of this project in supporting PEER’s vision

Large-scale seismic 3D modeling of soil-structure systems and inclusion of spatial variation of input excitation allow for more refined prediction of ground/foundation response, which is currently one of the most uncertain factors in Performance-Based Earthquake Engineering analyses and design. Parallel-processing capabilities are crucial for conducting such large-scale earthquake simulations. Combined, these capabilities will help significantly reduce the uncertainties involved in seismic assessment investigations.

Generally, the conducted research contributes to the overall PEER mission towards development of reliable simulation tools as an important component of the performance-based decision-making framework. In this regard, development, calibration, and testing of the PEER OpenSees computational platform is a cornerstone.

Methodology employed

Further development and refinement of geotechnical and numerical algorithms within the OpenSees platform. Continued calibration of the geotechnical modeling capabilities remains a primary concern (based of available full-scale and other relevant physical model data sets).

Brief description of past year’s accomplishments and more detail on expected Year 6 accomplishments

In Year 5, a solid-fluid fully coupled plane-strain element has been added into OpenSees. Combined with the soil constitutive models developed earlier (Year 4), the new element provides capabilities of simulating generation/dissipation of pore pressure and liquefaction-related large shear deformations. This allows for modeling a broad range of soil and soil-foundation-structure interaction scenarios. The above numerical developments are playing a major role in the PEER Testbed studies, and are being employed by many PEER researchers at UCD, UCSD, and elsewhere.

In Year 6, a preliminary 3D model of the Humboldt bay, Middle-Channel bridge has been developed using OpenSees (Figure 1), and some initial results are obtained for the cases of uniform base excitation. It was found that combined longitudinal and transverse shaking may increase the demand on various bridge components, compared to longitudinal only shaking (as in the 2D case). It is anticipated that by the end of Year 6, detailed 3D simulation results will be obtained and a comprehensive comparison with the 2D results will be carried out.

Figure 1. 3D model of the Humboldt bay, Middle-Channel bridge
Larger View

In addition, to the above, the domain-reduction method developed by Prof. Jacobo Bielak (Carnegie Mellon Univ.) has also been implemented and tested in OpenSees for plane-strain elements. This method allows for more realistic representation of spatially varying input excitation. It is expected that by the end of Year 6, the effects of spatially varying input motion on bridge and soil response will be investigated.

The current project is a natural extension of earlier efforts. Many of the existing/upcoming modeling capabilities of OpenSees are integrated into the Humboldt Bay Bridge model to simulate 3D large-scale seismic wave propagation and local site effects. Results from these simulations will provide useful estimates for Seismic Demand, a critical element in Performance-Based Earthquake Engineering (PBEE) applications. By continuing to work closely with Prof. Fenves and Dr. Frank McKenna, the main developers of the OpenSees platform, we will:

  1. continue adding the latest research results in computational geomechanics into the OpenSees (e.g., implementing the Domain-Reduction Method), and
  2. help test and debug many existing and new components of OpenSees (e.g., parallel computing algorithms) through modeling of the Humboldt Bay Bridge, and other large-scale soil-structure systems.

Extensive collaboration is also ongoing with Prof. Joel Conte. His current 2D model of the Humboldt Bay Bridge includes all developed soil nonlinearity algorithms and transmitting boundary effects.

Other similar work being conducted within and outside PEER and how this project differs

Large-scale 3D parallel/distributed computer simulation of seismic wave propagation and nonlinear site response effects are only at the very early stage of development. Related work is being conducted at UC Berkeley, UC Davis (Professor Boris Jeremic) and Carnegie Mellon Univ. for linear and mildly nonlinear scenarios. A main goal of this project is to more accurately model soil-structural interaction response ranging from linear to highly nonlinear scenarios (e.g., liquefaction).

Plans for Year 7 if this project is expected to be continued

In Year 7, the 3D bridge model will be extended in the direction of super-computer applications. By then, more refined large-scale simulations can be conducted in a reasonable amount of time.

Describe any instances where you are aware that your results have been used in industry

question skipped

Expected milestones
  • 02/01/03: Obtain preliminary 3D simulation results for Humboldt Bay Bridge site
  • 04/01/03: Implement Domain-Reduction Method (developed by Prof. Jacobo Bielak at Carnegie Mellon Univ.) or similar for defining spatially varying input excitation
  • 07/01/03: Analyze Humboldt Bay Bridge model in OpenSees incorporating spatially varying input excitation
  • 09/01/03: Documentation
  • Throughout project: Assist PEER users in conducting computational simulations using the developed geotechnical algorithms. Collaborate with UC Davis (Professor Boris Jeremic) towards integration of geotechnical capabilities and comparative studies.

Calibrated and verified 3D OpenSees models of the Humboldt Bay Bridge Testbed, and other large-scale soil-structure systems. Use of such models in PEER Probabilistic and Reliability Analysis frameworks.