CUREE Banner

Projects : CUREE-Kajima Joint Research Program


Category II: Analysis of Structural Component Failure

Researcher: Stephen Mahin (University of California, Berkeley)

TASK LIST - In the following, each of these tasks is briefly described.

Task 1 - Coordination
It is expected that the work in this project will be performed on a collaborative basis with researchers at Kajama Corporation, Stanford University and the University of Buffalo. As such, continuous collaboration in the form of exchanges of email, data files, and so on would be expected throughout the project. To get the project off to a good start, it is expected that special efforts will be undertaken during the first few months of the project, including discussions regarding selection and capabilities of computer programs, development of case study buildings, coordination of activities, and scheduling of future milestones. A web site will be established to facilitate other investigators finding information and models developed in this part of the investigation.

Task 2 - Classification of Members and Behavior Modes and Cataloging of Available Relevant Test Data

In this part of the investigation, various components will be identified for future studies and potential failure modes identified. It is expected that this effort will focus on beams, columns, braces, connections, and other members that comprise steel braced frame structures. Work at Stanford will focus on steel moment frames. Behavior modes will include not only ones associated with material nonlinearity at various critical locations along the various components, but also ones associated with local, lateral-torsional and global buckling of the components, and their failure due to fracture and low cycle fatigue. Test data available to the investigation team will be compiled and synthesized to assist in calibrating and validating analytical models.

Task 3 - Review and Evaluate Software and Analytical Models for Simulating Component Behavior, including Deterioration and Failure
Various software platforms will be reviewed and their capabilities will be compared. These will include commercial and non-proprietary finite element programs such as Abaqus, LS-Dyna, OpenSees, and SAP. Where applicable, pre- and post-processing software (TrueGrid, Hypermesh, FEMB, etc.) will also be assessed.

Numerical models for predicting member behavior will be summarized and evaluated. These models will focus mainly on distributed plasticity models based on fiber representations of critical sections, or finite element models based on small or large displacement shell and solid elements. The ability of these models to track large displacements and various forms of instability that may occur as axially loaded elements are deformed will be assessed from the perspective of accuracy of the simulations, user convenience in establishing the required initial conditions and finite element mesh, and computational stability and effort.

The capabilities of available material and damage models will be evaluated from three perspectives: (1) predicting the onset of various failure modes (fracture initiation, onset of buckling, etc.); (2) predicting the initiation and evolution of various types of deterioration, including fracture and low-cycle fatigue; and (3) computational effort. In some cases, it may be sufficient to predict whether particular details are susceptible to failure relative to other details, and simpler and more computationally efficient models might be used. In the case of modeling propagation of failure, several approaches are possible, ranging from simplified cumulative damage models that mimic deterioration of elements by removing them from the model, to ones that simulate the actual physical failure modes and modify the finite element mesh as the failure progresses.

Test results involving various types of components (braces, net reduced sections, gusset plates, shear yielding, column splices in tension and bending, columns under high axial load and bending, etc.) and failure modes will be used to assess the various combinations of programs, elements and material/damage models.

Task 4 - Develop and Validate Optimal Models
Test data will be used to calibrate preferred analytical models and establish recommendations for modeling. In this case, only one or two programs, and a limited selection of element types and material models will be considered. Literature will be examined to find recommendations for modeling recommendations by others.

Test results of complete systems will be analyzed to assess the ability of the models to track the redistribution of forces and deformations and progression of damage among an assembly of components.

Results will be examined to identify opportunities for reducing the vulnerability of conventional details to premature failure modes.

This work will be coordinated with similar efforts by investigators at Stanford, Buffalo and Kajima.

Task 5 - Evaluate Case Study Building and Actual Buildings using Optimal Models
The models will be used to examine the response of the case study building (or other actual buildings) under various types of loading scenarios (earthquake loading, removal of one or more elements, collision, etc.). To the extent possible within the time and budget, means of changing the details or proportions of the system to avoid progressive collapse will be examined.

Task 6 - Prepare and Submit Final Report
A report summarizing the overall accomplishments of this part of the overall investigation will be prepared and submitted to CUREE and Kajima.

The project will be undertaken according to the following approximate schedule.

CUREE-Kajima Project

CUREE-Kajima: Phase 6
Project Overview
Project Team Rosters
Test Videos
Preliminary Reports

Prior Research
CUREE-Kajima: Phase 5
CUREE-Kajima: Phase 4
CUREE-Kajima: Phase 3
CUREe-Kajima: Phase 2
CUREe-Kajima: Phase 1
Funded by:
Kajima Corp. Logo

© 2000-2008 CUREE. All rights reserved.
Consortium of Universities for Research in Earthquake Engineering
last updated 02.01.06