Resilient Steel Plate Shear Walls: Analysis of Performance Using OpenSees and TeraGrid

Resources

Patricia M. ClaytonUniversity of Washington

Jeffrey Berman (PI)Laura Lowes (Co-PI)

NEES-SG: SPSW Research

• Tasks:– Develop a resilient SPSW– Develop performance based

design tools for SPSW– Develop a new model for SPSW

web plates– Explore the behavior of coupled

SPSWs and develop design recommendations

Jeff Berman and Laura Lowes

Michel Bruneau

Larry Fahnestock

K.C. Tsai

Sponsored by NSF through the George E. Brown NEES Program

Jeff Dragovich

Rafael Sabelli

What is a Resilient Steel Wall?• Combines benefits of Steel Plate Shear Walls (SPSWs) with self-

centering technologies• SPSW provides:

– Ease of construction– High strength and initial stiffness– Ductility– Yielding over many stories– Replaceable energy dissipation elements (steel plates)

• Post-Tensioned (PT) Connection provides:– Self-centering capabilities– Quick return to occupancy after earthquake

Conventional SPSW Behavior• Resists lateral load through development of

Tension Field Action

Courtesy of Berman and Bruneaudiagonalfolds

tensilestresses

lateralload

a

angle ofinclination

HBE

HBE

VBE

Web plate

Conventional SPSW Behavior• Idealized hysteretic behavior of SPSW with simple

HBE-to-VBE connections:

1st Cycle2nd Cycle

VSPSW

D

Plate yieldsUnloading

Low Stiffness

PT Connection Behavior• Provides self-centering capabilities

•Connection is allowed to rock about its flanges•PT remains elastic to provide recentering force

•Requires some energy dissipation•Examples from previous research:

•Yielding angles (Garlock, 2002)•Friction devices (Iyama et al., 2009; Kim and Christopoulos, 2008)

Garlock (2002) Iyama et al. (2009)

PT Connection Behavior• Nonlinear elastic cyclic behavior of PT connection:

1st Cycle

VPT

D

ConnectionDecompression

qr 2nd Cycle

Combined System: Resilient SPSWVSPSW

D

1st Cycle2nd Cycle

VPT

D

VR-SPSW

D

Plate yields

Unloading

ConnectionDecompression

ConnectionRecompression

Plates Unloaded

Performance-Based Design

V

D

V10/50

D10/50

First occurrence of:·PT yielding·Frame yielding·Residual drift > 0.2%

REPAIR OF PLATES ONLY

V2/50

D20/50

First occurrence of:·PT rupture·Excessive PT yielding·Excessive frame yielding·Excessive story drifts

COLLAPSEPREVENTION

D50/50

V50/50 Plate yielding

NO REPAIR

VwindConnection decompression

Prototype Building Designs

• Based on 3- and 9-story SAC buildings in LA• Vary number of R-SPSW

bays in building• 2 design types:

• Plates designed for V50/50

• Plates designed for V10/50/R

Analytical Model• Nonlinear model in OpenSees• SPSW modeled using strip method:

• Tension-only strips with pinched hysteresis• Strips oriented in direction of tension field

Analytical Model (cont.)

• PT connection model:

HBE

VBE

Rocking about HBE flangesCompression-only springs at HBE flanges

Rigid offsets

Shear transferDiagonal springs

PT tendons Truss elements with initial stress (Steel02)

Analytical ModelPhysical Model

Dynamic Analyses• Each model subjected to 60 LA SAC ground motions representing

3 seismic hazard levels• 50% in 50 year• 10% in 50 year• 2% in 50 year

• Used OpenSeesMP to run ground motions in parallel on TeraGrid machines

Using TeraGridOpenSeesMP .tcl scripts

Ground acceleration records

Batch submission script

#!/bin/bash#$ -V#$ -cwd#$ -N jobName#$ -o $JOB_NAME.o$JOB_ID#$ -e $JOB_NAME.err$JOB_ID#$ -pe 16way 64#$ -q long#$ -l h_rt=48:00:00#$ -M myemail@u.washington.edu#$ -m be

set –x

ibrun $HOME/OpenSeesMP $WORK/OSmodel.tcl Ranger

Abe

Ranger

Processor = 0

Processor = 1

Processor = n-1R-SPSW model

Run all models and ground motions simultaneously using OpenSeesMP

Abe

Using TeraGrid

Using TeraGrid

Ranger

AbeOpenSees recorder & output files

All results in the time it takes to run one ground motion.

• Example of Response during 2% in 50 year EQ– System Response

Response History Results

– Connection Response

Response History Results• Statistical results from all 60 ground motions

• Performance Objectives:– No plate repair (Story drift < 0.5%) in 50/50

(this example designed using V10/50/R; plates not explicitly designed to remain elastic)

– Recentering (Residual Drift < 0.2%) in 10/50– Story drift < 2.0% in 10/50 (represents DBE)– Limited PT, HBE, and VBE yielding in 2/50All performance objectives met !!!

Comparing Designs

R-SPSW designed using V50/50

• Plates designed to remain elastic in 50% in 50 year EQ

R-SPSW designed using V10/50/R

• Plates designed using reduced “DBE” forces

• Larger plate thicknesses & frame members • Improved response

o Recentering at all hazard levelso Smaller peak drifts

Conclusions• Preliminary design procedure developed for R-SPSW• Dynamic analyses show R-SPSW can meet proposed

performance objectives– including recentering in 10% in 50 year EQ

• Highly nonlinear model significant computational effort

• Use of TeraGrid resources reduced computational time by more than 90%

• Experimental studies on R-SPSW currently taking place

Thank You