Nonlinear

Analysis of Concrete
Wall Buildings Using OpenSees

Acknowledgements
The research presented here was
accomplished in collabora6on with
Joshua Pugh, EDG Inc.
Dawn Lehman, University of Washington

With funding provided by

The Na6onal Science Founda6on
The Applied Technology Council and the Na6onal Ins6tute
for Tes6ng and Standards
The Charles Pankow Founda6on

Why Concrete Walls

Damage to Chilean Walled Buildings

Compression-
controlled exural
failure for walls with
poorly conned
boundary elements

CM: 18+2 stories,

residential building,
2006 construction,
Concepcion, Chile

AH: 15+2 stories, mixed-use, 2009

construction, Concepcion, Chile (J. Moehle)

PR: 12-story, residential building, 2006

construction, Concepcion, Chile

Damage to Chilean Walled Buildings

Compression and
shear damage in
lightly reinforced
walls

Diagonal cracking

Compression
failure

Plan View
PR: 12-story, residential, 2006 construction, Concepcion, Chile

Elevation View

Damage to Chilean Walled Buildings

Compression and shear
damage

Plan View
Elevation View

PR: 12-story, residential, 2006

construction, Concepcion, Chile

Damage to Walled Buildings in

Christchurch, NZ
Damage to modern walls
Modern walls exhibi6ng
compression-controlled
exural failure and shear-
compression failure

(Figures from Kim, Pampanin and Elwood 2011)

Earthquake Damage to Walled

Buildings in the Past
Consider earthquake reconnaissance data
from 22 earthquakes around the world going
back as far as the 1957 Mexico City
earthquake.
Reports document damage to 97 walled
buildings.

Failure Mode versus Building Height

Damage to Slender Walls Tested in

the Laboratory

Drift Capacity

Approximately 50% of ACI-compliant walls fail in

compression.
ACI-compliant walls fail in compression even if steel strain at
Mn is well in excess of 0.005.
Dri[ capacity not correlated with connement ra6o or s/db

All walls
BE confinement

All walls
s/db for BE confinement

ACI-compliant
walls
steel tensile strain at Mn

Implica6ons for the Analyst

Concrete walls, both modern and older, may
exhibit bri]le compression-controlled failure
at rela6vely low dri[ levels.
Thus, assessment of wall performance
requires accurate simula6on of this failure
mechanism.

Research Ques6ons
1. For walls and walled buildings designed using current
US Codes and standards of prac6ce?
What is the expected failure mode for a wall? Flexure or
shear? Compression- or tension-controlled exure?
What is the collapse risk for a walled building for various
levels of earthquake demand?

2. How can we improve design to achieved desired

performance:
Achieve desired failure mechanism.
Achieve acceptable collapse risk.

Research Process
1. Develop a numerical modeling approach for slender
concrete walls that enables accurate simula6on of
response through failure, including accurate
simula;on of failure mode and dri< capacity.
2. Use this model to
Evaluate the earthquake performance of concrete walled
building designed using current US design codes.
Develop recommenda6ons to improve wall design.

Using OpenSees to
Simulate Wall
Response

Modeling the Earthquake Response of

Concrete Wall Buildings
Objec6ves

Accurate simula6on of response

Computa6onal eciency and robustness
Script-based input to facilitate parameter studies

Models considered
Con6nuum (Abaqus, VecTor2,ATENA): too computa6onally demanding
for system analyses, not numerically robust (Abaqus), no script-based
input (VecTor2).
PERFORM 3D Fiber Wall Element: too computa6onally demanding, no
script-based input.
Line-element models with exure/shear interac6on: inaccurate/not
calibrated for a range of wall designs OR not readily available for use.
OpenSees ber-type beam-column elements with distributed
plas6city: greatest poten6al to meet all modeling objec6ves

Modeling the Earthquake Response of

Concrete Wall Buildings
Objec6ves

Accurate simula6on of response

Computa6onal eciency and robustness
Script-based input to facilitate parameter studies

Models considered
Con6nuum (Abaqus, VecTor2,ATENA): too computa6onally system
analyses, not numerically robust (Abaqus), no script-based input
(VecTor2).
PERFORM 3D Fiber Wall Element: too computa6onally demanding, no
script-based input.
Line-element models with exure/shear interac6on: inaccurate/not
calibrated for a range of wall designs OR not readily available for use.
OpenSees ber-type beam-column elements with distributed
plas6city: greatest poten6al to meet all modeling objec6ves

Evalua6ng Fiber-Type Beam-Column

Elements for Modeling Wall Response
Force-Based Element:

Displacement-Based Element:

Assume linear moment

Assume linear curvature
Typical Test Specimen
distribu6on and constant
distribu6on and constant axial
axial load (along the length of
strain
(along the length of the
Applied Shear,
the element).
element).
Axial Load and
Moment
Intra-element solu6on to
No Possibly
intra-element
solu6on reqd.
determine sec6on strains and
Use mul6ple elements per story;
curvatures that sa6sfy
elements can have fewer sec6ons.
compa6bility reqts.
Add single shear sec6on at each
Aggregate exure and shear
story.
Fixed Base
sec6ons.
Use one element per story;
Nonlinear
fiber-type flexural section
each element has
~5 sec6ons.
Linear elastic shear section

Fiber Sec6on:
Concrete 02 model used for concrete

Unconfined Fibers:

Confined Fibers:

Saatcioglu & Razvi

(1992)

Fiber Sec6on:
Steel 02 used for reinforcing steel

Experimental Data Used for Model

Evalua6on,
Calibra6on & Valida6on

19 rectangular, 3 barbell, 6 c-shape, 4 t-shaped specimens

from 10 test programs
All walls are slender with (M/V)/lw > 2
All walls exhibit exural failure mechanisms
Crushing of boundary-element concrete, buckling and/or rupture of
long. reinforcement
Walls exhibi6ng web crushing (barbell walls) not included

All wall have scale = tw/12 in. > 1/3

Axial load ra6os: 0.01fcAg - 0.16fcAg
Shear stress demands: 1.06.0

psi

Quan66es Used for Model Evalua6on,

Calibra6on & Valida6on
Typical Test Specimen
Applied Shear,
Axial Load and
Possibly Moment

Fixed Base

Force-Based Distributed-Plas6city
Beam-Column Element:

Evalua6on, Calibra6on and
Valida6on

Model Evalua6on

No. of
I.P.

No. of
Specs

Mean

COV

Mean

COV

23

0.98

0.07

0.90

0.27

23

0.97

0.08

0.90

0.27

23

0.97

0.08

0.90

0.27

Mean

COV

Mesh
Dependent

Localiza6on of Damage / Deforma6on

Inelastic Localization

Specimen WSH4
(Dazio et al. 2009)

0.63%

No Localiza6on Prior to Strength Loss

Specimen WSH4
(Dazio et al. 2009)

To Achieve Mesh-Objec6ve Results

Regularize material response using a mesh-dependent length
Typically done in con6nuum analysis
Coleman and Spacone (2001) propose this for beam-column
elements;
To regularize
Concrete: Use experimental data to dene energy under post-peak por6on
of the stress-deforma6on curve & convert stress-deforma6on to stress-
strain using integra6on-point length, LIP
Steel: Use experimental data to dene stress-strain response and adjust
post-peak strength strain response based on ra6o of laboratory gage
length to integra6on-point length, LIP

Note that regulariza6on of steel hardening response reqd

because deforma6on localizes to so[ening sec6on

Concrete Tensile Fracture Energy

Tensile fracture energy, Gf , commonly used to regularize material

response for con6nuum-type nite element analysis
Several standard approaches for dening Gf (e.g., RILEM 50-FMC)
Gf 75-150 N/m (Wong and Vecchio, 2006)
RILEM 50-FMC Lab Test
Laboratory Test Data

- used in analysis

Concrete Material Regulariza6on Using Gf

Has essen6ally no impact; therefore ignore

No mesh sensitivity in range of

demands in which concrete cracking
occurs
Thus, material regularization has no
impact

Plain Concrete Crushing Energy

Jansen and Shah, 1997

Material Regulariza6on: Plain

Concrete

Crushing energy, Gfc = ~20 N/mm per Jansen

and Shah (1997)
LIP,1

LIP,2
LIP,1

3-I.P. Element

Determine Required Gfc

Use experimental data for two planar walls constructed of unconned
concrete and exhibi6ng exural failure due to concrete crushing
Gfc = 60 80 N/mm = 2fc with fc in MPa
Note that increase in Gfc above Jansen and Shah 20 N/mm for plain concrete
cylinders is a]ributed to the presence of longitudinal steel
Specimen WSH4
(Dazio et al. 2009)

Specimen WR0
(Oh et al. 2004)

LIP,1

fc/fcc

Material Regulariza6on: Conf. Concrete

Gfcc

fcc

LIP,2

/co
0.2fcc

LIP,1
fc/fcc

co

3-I.P. Element
/co

c20u

Determine Required Gfcc

Use experimental data for eight planar walls w/ conned
concrete exhibi6ng exural failure due to concrete crushing
Gfcc appears to be a func6on of connement detailing, but
insucient data for model calibra6on
(Gfcc/fcc)Mean 2.7

Material Regulariza6on: Steel

Required despite steel hardening because deforma6ons localize to
single so[ening sec6on
Gfs / lgage determined from material tests
Regularized steel stress-strain response determined by LIP
Regulariza6on results in adjusted tensile rupture strain
To simulated strength loss due to buckling, include compressive
failure strain equal to strain at which concrete loses 80% of
compressive strength

Response from lab data

Regularized response model

FBBC: Regularized Results for Planar Walls

Failure Mode

Mean

COV

Mean

COV

Mean

COV

Crushing
(9 specimens)

0.93

0.04

0.83

0.26

0.96

0.15

Rupture/Buckling
(6 specimens)

0.95

0.05

1.01

0.33

1.12

0.21

Rupture
(2 specimens)

0.98

0.03

0.94

0.02

1.08

0.04

Out of Plane
(2 specimens)

0.98

0.03

0.94

0.28

1.31

0.08

All Flexure

0.95

0.07

0.90

0.28

1.06

0.22

Regularized Results: Planar Walls

Good results: WSH4
Dazio et al.

Not so good results:

PW4 Lowes et al.

Regularized Results: C-Shaped Walls

Apply regulariza6on method calibrated for
planar walls to C-shaped walls:
Specimen

Loading

UW1 (Lowes et al.)

Strong Axis

1.01

1.13

1.20

W1 (Ile and Reynouard)

Strong Axis

0.90

0.85

1.00

W2 (Ile and Reynouard)

Weak Axis

0.94

0.87

0.77

W3 (Ile and Reynouard)

Bi-Direc;onal

0.93

1.10

0.70

TUA (Beyer at al.)

Bi-Direc;onal

1.06

0.90

1.04

TUB (Beyer et al.)

Bi-Direc;onal

1.08

1.15

1.06

0.99 (0.08)

1.00 (0.14)

0.96 (0.20)

Mean (COV)

Regularized Results: C-Shaped Walls

Good: TUA Beyer et al.

Not so good: W3 Ile and

Reynouard

Regularized Results: T-Shaped Walls

Apply regulariza6on method calibrated for
planar walls to T-shaped walls:
Specimen

Loading

TW1
Uni-direc;onal
(Thomsen and Wallace)

1.25

2.4

0.42

TW2
Uni-direc;onal
(Thomsen and Wallace)

1.00

1.6

0.45

NTW1
(Brueggen et al.)

Bi-Direc;onal

1.00

1.14

0.86

NTW2
(Brueggen et al.)

Bi-Direc;onal

0.95

1.05

0.82

1.05/0.13

1.55/0.40

0.64/0.37

Mean/COV

Regularized Results: T-Shaped Walls

Good: NTW1 Brueggen et al. Not so good: Thomsen and Wallace
Data show plane sec6ons do not remain
plane, so strain distribu6on is not
correctly simulated

Displacement-Based Distributed-
Plas6city Beam-Column Element:

Evalua6on, Calibra6on and
Valida6on

Model Evalua6on: Mesh Renement

Study
Load-displacement
response

Axial load at the sec6on

(formula6on assumes
constant axial strain not
force )

Impact of Axial Load Varia6on

Cri6cal (i.e. so[ening) sec6on is located above the base
of the wall and is not the sec6on with highest demand:
Fiber sec6on at the base of the wall has an axial load that is larger than the
applied axial load; this results in increased exural strength.
Fiber sec6on above the base of the wall has an axial load that is smaller
than the applied axial load; this results in reduced exural strength.

Accurate simula6on of dri[ capacity requires modied

Gfc and Gfcc to account for error in sec6on axial load:
Unconned: Gfc_DBBE = 0.28Gfc_FBBE
Conned: Gfcc_DBBE = 0.28Gfcc_FBBE

Force-based element is preferred over disp.-based

element to achieve accurate simula6on of failure.

Applica6on of the Model

to Advance Design of
Walled Buildings

Applica6on of the Model

Use FEMA P695 Methodology to evaluate the earthquake
performance of walled buildings designed using US codes.
Approx. 2000 dynamic analyses: ITHA (Incremental 6me-history analyses) of 8
building designs using suite of 44 ground mo6on records.

To improve performance, 1) develop capacity-design procedure

for shear and 2) recommend demand envelope for exural
design.
Approx. 4500 dynamic analyses: ITHA of 64 buildings using suite of 7 synthe6c
ground mo6ons AND dynamic analysis of 96 building designs using suite of 14
synthe6c ground mo6ons.

Use FEMA P695 Methodology to develop strength reduc6on

factors (ASCE 7 R-factors) to achieve desired collapse risk.
Approx. 1600 dynamic analyses: ITHA of 6 buildings for suite of 44 ground mo6on
records.

FEMA P695 Used for Evalua6on

T1= Cu Ta

Determines
1. Probability of collapse in
the MCE, and
2. If the design procedure (Rfactor, etc.) is acceptable.

SMT
ST1

Collapse

= Margin Ratio

Evalua6on of Current
Design Procedures

Building Designs
Design 8 walled bldgs.
16, 20, 24 and 30 stories
Core-wall buildings
Only uncoupled loading dir. considered
loading direction
considered

EQ demands per ASCE 7 (2010)

SDC D (highest eq. demand category)
Strength reduction factor, R = 6
Both ELF procedure and MRSA used;
MRSA demands scaled up to meet ELF
base shear.

Walls sized to achieve

Seismic weight = 8.1 kPa (170 psf)

Gravity weight = 9.1 kPa (190 psf)
Wall axial load at base = 0.1fcAg

/ 16, per building inventory review

Size for shear per NIST (2011):
=0.20.3 ()
=24 ()

Wall capacity and detailing per ACI 318

(2011)

Numerical Model Used for Evalua6on

Force-based ber-type beam-
column element used to model
walls:

Contribu6on of gravity system to

lateral s6ness is ignored.
P-delta eects included.
2% Rayleigh damping employed.

Core Wall

P- Column

Nonlinear exural response is

simulated using ber-sec6on model.
Flexure-shear interac6on is ignored.
Elas6c shear response is assumed
(shear s6ness = GAcv).
1 element w/ 5 ber sec6ons per
oor.

ag(t)

Nonlinear Analysis Results for ITHA

Using FEMA P695 Far-Field Mo6ons
Answer to question #1:

Ground Motion Intensity Ratio = ST1/SMT

MCE

DBE

Vn,pr

Walled buildings in the US are

likely to exhibit shear failure.

shear failure
flexural failure

Capacity Design for

Shear in Walls

Capacity Design for Shear

Shear demand used for design must account for

Flexural over-strength
Dynamic amplica6on

Current US design method (ASCE 7 and ACI 318)

does not account for either:
Vn Vu with = 0.6 and Vu from elas6c analysis

Capacity design
Vn Vu with Vu = voVu
Dyn. Amplification
Flex. Overstrength

Capacity design approach for shear has

been adopted by

New Zealand Standard NZS-3101 (1995)

Canadian Standard CSA-A23.3 (2004)
Eurocode 8 (2004)
SEAOC (2008)

To Determine a Capacity-Design
Method for Shear
Design and analyze a set of prototype buildings
Compare maximum shear demand from ITHA
(VITHA) with design shear (Vu) using suite of
synthe6c mo6ons
Building designs represent larger design space:
64 Buildings
Building heights: N = 6 24 stories
Fundamental building periods: T1 = 0.08N 0.20N
ASCE 7 force reduc6on factors: R = 2,3,4

Idealized Buildings

loading
direction

N = 6, 8, 12 stories

N = 16, 20, 24 stories

Seismic weight = 8.1 kPa (170 psf)
Gravity weight = 9.1 kPa (190 psf)
Wall axial load at base = 0.1fcAg

Suite of Synthe6c Mo6ons Used for ITHA

7 synthe6c ground mo6on records
Mo6ons provide demands that are consistent with
the design spectrum used for design

Ra6o of Maximum Simulated Shear to

Design Shear

Recommended Capacity Design

Procedure for Shear
Proposed Shear Demand

o = 1.4 to account for exural

over-strength
Vu determined using Modied
MRSA method:

(
)

Vn o Vu

=(1/)2+
22+32+
1>2
=12+(
2/)2+32contribu6ons
+ control.
2>1

Design for Flexure

Flexural Design Envelopes

MRSA/ELF:
Mn Mu
over the height

Constant:
Mn Mu
at base

Mu Mn

MRSA Moment Envelope

Mu

Mn

Paulay/Priestley (1992)
Mn Mu at base
Mn > Mu elsewhere

Dual Hinge
(Panagiotou and
Restrepo, 2009)
Mn Mu at base and
Mn Mu at mid-height,
Mn > Mu elsewhere

0.5H

Mu Mn

Mu

Mn

Impact of Flexural Design Envelopes

Designs employ R=3
Analyses are done for
MCE intensity level
Curvature Demand =

=
/

Recommenda6ons for Wall Design

Shear Design
Vn o Vu
o = 1.4 to account for exural over-strength
Vu dened by Modied MRSA Method to account for dynamic amplica6on:

=(1/)2+22+32+
1>2=12+(2/)2+3
2+ 2>1
Flexural Design Envelope
Paulay/Priestley or Dual Hinge
(RASCEl oca6ons,
4.0)
detailing is provided
Yielding is limited to expected
wASCE
here 7d-10:
uc6le
RASCE = 5,6
(RASCE 5.0)

Flexural Strength Reduc6on Factor (per FEMA P695)

Planar walls: R 2.5
Core walls: R 3.5

Design to Achieve Acceptable Collapse Risk

1. Elas6c analysis to determine demands:
Modal response spectrum analysis (MRSA) with wall exural s6ness equal to
50% of gross-sec6on s6ness.

2. Shear design:

Capacity design approach to prevent shear failure: Vn ovVu

Vn = factored shear capacity per ACI 318
0 = exural over-strength factor = 1.4
v Vu = shear demand with dynamic amplica6on determined using the Modied
MRSA Method in which only the elas6c response mode that contributes most to
base shear is reduced by the ASCE 7 R-factor

3. Flexural design:
Flexural demands determined using envelope by

Paulay & Priestley (1992), for which demands are increased above the base to ensure yielding only at the base, OR
Dual-hinge method (Panagiotou and Restrepo 2009), for which demands are increased everywhere except the base
and approx. mid-height to achieve two regions of yielding.

Planar walls: R 2.5 (this is approx. a 50% increase is demand per ASCE 7)
Core walls: R 3.5 (this is approx. a 20% increase is demand per ASCE 7)

Design to Achieve Acceptable Collapse Risk

1. Elas6c analysis to determine demands:
Modal response spectrum analysis (MRSA) with wall exural s6ness equal to 50% of gross-sec6on
s6ness.

2. Shear design:

Capacity design approach to prevent shear failure: Vn ovVu

Vn = factored shear capacity per ACI 318
0 = exural over-strength factor = 1.4
v Vu = shear demand with dynamic amplica6on determined using the Modied MRSA Method in which
only the elas6c response mode that contributes most to base shear is reduced by the ASCE 7 R-factor

3. Flexural design:
Flexural demands determined using envelope by
Paulay & Priestley (1992), for which demands are increased above the base to ensure
yielding only at the base, OR
Dual-hinge method (Panagiotou and Restrepo 2009), for which demands are increased
everywhere except the base and approx. mid-height to achieve two regions of
yielding.

Planar walls: R 2.5 (this is approx. a 50% increase is demand per ASCE 7)
Core walls: R 3.5 (this is approx. a 20% increase is demand per ASCE 7)

Conclusions
Modeling Slender Concrete Walls
Regulariza6on of material response is required for predic6on of dri[
capacity because response is compression controlled with localized
so[ening.
OpenSees ber-type force-based beam-column elements with
regularized material models provide accurate and precise simula6on
of s6ness, strength and dri[ capacity for planar and c-shaped walls.

Design of Slender Concrete Walls

Current US code design underes6mates shear demand in walls.
An over-strength factor, 0 = 1.4 and the Modied MRSA method, in
which only the 1st or 2nd mode contribu6on to base shear is reduce,
must used to es6mate shear demand in walls.
Force reduc6on factors, R-factors, that are smaller than currently
specied in ASCE 7 are required to limit exural damage at MCE.