Project Title/ID Number PEER Analysis Platform for Demand Simulation—4102002
Start/End Dates 10/1/02—9/30/03
Project Leader Gregory Fenves (UCB/Faculty)
Team Members Filip Filippou (UCB/Faculty), Frank McKenna (UCB/Post Doc), Silvia Mazzoni (UCB/Post Doc), Michael Scott (UCB/Grad Student), Asfin Saritas (UCB/Grad Student)
Project goals and objectives

The goal of this project is to advance the research for simulation capability of structural and geotechnical systems for performance-based earthquake engineering. The simulation software developed in the project utilizes advances in information technology and computing methods, and the development is collaborative, with an open-source methodology, that engages a large number of researchers in PEER and many outside PEER. The primary products of the project are:

  1. new models and methods for simulation of earthquake performance
  2. open-source software framework, OpenSees, that incorporates the latest research and computing technology
  3. a committed group of developers and users in the research community

Simulation of structural and geotechnical systems is a key step in PEER PBEE methodologies. The PEER assessment methodology relies on simulation to compute engineering demand parameters and in some cases damage measures. PEER has embarked on software development because existing simulation software is inadequate for a number of reasons: models of behavior are too simplified, particularly for reinforced concrete; little support for soil-foundation-structure interaction; closed or antiquated software architecture makes it difficult to research new simulation methods; generally slow adoption of information technology; does not expose students to modern software engineering and computer science.

The OpenSees website is

Role of this project in supporting PEER’s vision
Many projects in Thrust Areas 2, 3, 4, and 5 are directly related to this project because of the central role in PBEE. In the past year all of the testbed projects have utilized OpenSees for simulations, demonstrating the broad use OpenSees has developed within PEER. Depending on the testbed project, models were validated, simulation methods were tested for scalability and robustness, or data were developed for experimental testing.
Methodology employed

The project has two distinct, but closely related, components. The first is the software development and support of OpenSees (Open System for Earthquake Engineering Simulation) in PEER. The second is the development of improved models and solution methods for simulation and their implementation in OpenSees. The first component is based on modern software development approaches, suitable for a research environment and collaborative development effort. The second is primarily focused on beam-column components with nonlinear material and geometry and on scalable, robust solution methods. 

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

Software Framework Development
Version 1.5 of OpenSees was released on the website in the past year. In addition to bug fixes and extensions of current functionality, the major change was the inclusion of the reliability computation modules (developed under a separate PEER project by A. Der Kiureghian).

User and Developer Support
Support was provided by a research staff of two post-doctoral researchers. The staff provided assistance to researchers on model and algorithm development for the testbed projects and coordinated with the integration of software from the other Thrust Area 4 projects. A workshop was held in September 2002 for users and developers and the first edition of the online documentation was released and is maintained on the website.

Sensitivity Analysis
Sensitivity analysis—computing gradients of response with respect to model parameters—is the a key step in optimization used for reliability, system identification, or design. Whereas gradients are straightforward to computer for displacement formulation elements, there is no direct procedure for force formulation. Since the robust models use the force formulation, this was a severe limitation. A new and very efficient procedure for sensitivity computation of force formulation has been developed and implemented.

Robust Force-Formulation
A generalized variational framework for all types of force and mixed-formulations has been developed based on the Hu-Washizu principle. It places all work to date within a consistent variational framework. Another advantage stems from the fact that all types of numerical implementations of force and mixed frame elements can be directly derived from this formulation (iterative, direct, incremental with several history variables or with a very small number of history variables but larger number of computations).

The force formulation has been extended to account for the coupling between shear, flexure and axial force at the material level. The element has been tested with great success in the context of a steel beam using a J2 cyclic plasticity model. Preliminary extensions of the theory to reinforced concrete members using the compression field theory to represent the interaction between shear and normal stress at the section level has been completed. Under monotonic loading conditions and we are able to trace crack formation and evolution in a shear critical beam with as much success as some plane stress finite element models

Robust Nonlinear Equation Solution
A new quasi-Newton solution method based on a Krylov subspace has shown promise for significant efficiencies and stabilizing solutions for degrading systems.

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

MCEER is continuing to support limited development of IDARC, and there have been preliminary discussions on developing inter-operability with some components of OpenSees. MAE appears to be supporting development of a PC based program for static and possibly dynamic analysis of buildings. Commercial software development for PBEE is either very simplified or lagging behind research in PEER.

This project is fully integrated with an NSF supported project on Seismic Performance of Urban Regions (SPUR). OpenSees is providing the computation for building simulation and in the next year will include nonlinear soils in the SFSI models and input motion from the regional ground motion simulation.

Discussions have been underway in using OpenSees for NEES and developing Grid-based version of OpenSees to support experimental and computational needs in NEES using community-based software.

Plans for Year 7 if this project is expected to be continued
OpenSees framework will move to Version 2, the model development for reinforced concrete will be completed, and the solution strategies will be examined for scalability on parallel computers. This will also aid SPUR and NEES applications. OpenSees will be extended to provide additional interfaces with the scientific visualization methods under development, possibly using the database technology that is being completed in the thrust area.
Describe any instances where you are aware that your results have been used in industry
Industry has followed the developments of OpenSees as indicated at the PEER Annual Meeting. Most firms are waiting for the results of the testbeds, and user interfaces, before utilizing the technology in OpenSees and the software itself. Interactions will NEES provide a good opportunity to pursue portal and interface development, possibly using the NEES CHEF standard.
Expected milestones

Release of Version 2; released of updated beam-column models; benchmarks for parallel processing for solution methods; user workshops and other support for PEER, SPUR, and NEES projects.


Validated models for reinforced concrete, robust solution strategies for highly nonlinear systems, updated OpenSees version for serial and parallel computers, updated user documentation, increased community involvement, particularly through NEES.