Available online at www.sciencedirect.

com

Procedia Engineering 14 (2011) 2187–2195

The Twelfth East Asia-Pacific Conference on Structural Engineering and Construction

Performance Assessment of Buckling Restrained Braces

H. Y. CHANG1a and C. K. Chiu2
1
Department of Civil and Environmental Engineering, National University of Kaohsiung, Taiwan
2
Department of Construction Engineering, National Taiwan University of Science and Technology, Taiwan

Abstract

In the company paper, the global drift capacity and demands of a 6-story steel office building adopted with buckling
restrained braces (BRBs) have been evaluated. This paper evaluates the seismic performance of the building from a
local perspective. In detail, the capacity and demands of the BRBs were assessed by test data and response analysis.
Then, the level of confidence was computed against the potential of BRB yielding, buckling and fracture failures. The
result shows that the BRBs can provide a high level of confidence, ensuring the building to achieve the performance
objectives of immediate occupancy and life safety. But in meeting the performance objective of collapse prevention,
the confidence level may be far smaller than the 50%-value recommended by FEMA 351. It suggests a necessity of
more carefully assessing the seismic vulnerability of braced steel frames, especially for collapse prevention and from
a local perspective.
Keywords: Confidence level, performance evaluation, local failures, buckling restrained brace.

Keywords: Confidence level, performance evaluation, local failures, buckling restrained brace.

a
Corresponding author: Email: [email protected]

1877–7058 © 2011 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
doi:10.1016/j.proeng.2011.07.275
2188 H.Y. CHANG and C.K. Chiu / Procedia Engineering 14 (2011) 2187–2195

1. INTRODUCTION

Toward the development of performance-based engineering, there has been a necessity of assessing
structural performance on a reliability basis. FEMA 351, for example, recommends design solutions to
provide a 50 % level of confidence against the potential of local failures. In the company paper, the global
drift demand and capacity of a 6-story steel office buildings adopted with buckling restrained braces
(BRBs) have been evaluated (Chang 2009). This paper assesses the level of confidence against BRB
yielding, buckling and fracture failures. All that helps gain a better understanding about the global and
local performance of steel buildings adopted with BRBs.

2. STUDIED BRB MEMBERS

2.1. Performance objectives

Table 1 gives an example illustrating the structural performance levels and damage of braced steel
frames. As can be seen, BRB deformations at first yield, buckle and fracture are important to justify the
performance levels of braced steel frames.

Table 1: Structural performance levels and damage (FEMA 356)

Structural performance levels

Elements Type
Collapse Prevention (C.P.) Life Safety (L.S.) Immediate Occupancy (I.O.)
Extensive yielding and Extensive braces yield or Minor yielding or buckling of
buckling of braces. Many buckle but may not totally fail. braces.
Primary
braces and their connections Many connections may fail.
Braced may fail.
steel
frames Secondary Same as primary Same as primary Same as primary
2% transient or permanent 1.5% transient; 0.5% transient; negligible
Drift permanent
0.5% permanent

2.2. BRB mechanism

(a) Double-cored BRB member (b) BRB core plate

Figure 1: A steel frame adopted with a double-cored BRB and the core plate of the BRB
 H.Y. CHANG and C.K. Chiu / Procedia Engineering 14 (2011) 2187–2195 2189

Beam

Axial force
Column

BRB BRB

Deformation

Figure 2: Configuration of a chevron BRBF, hysteresis loops and BRB strain and story drift

To reduce the length and number of bolts in the brace-to-gusset connection, the double-cored buckling
restrained braces (DCBRBs) have been developed and extensively tested in Taiwan in the past few years.
Figure 1 gives the details about the application of a double-cored BRB to a steel frame and the details of
the BRB core plate.
Ac is the area of core plate and Fy is steel nominal yield stress. : and : h respectively denote over-
strength factor and strain hardening factor ( : =1.5 for A36 steel, : =1.1 for A572 Gr. 50 steel; : h =1.5
for A36 steel, : h =1.3 for A572 Gr. 50 steel). E is an adjustment factor for the difference between
tensile and compressive strength (~1.1).The yield strength and ultimate strength of a BRB can be
estimated as follows

Py :Ac Fy (1)

Pmax : h EPy (2)

E is Young’s modulus. At and A j respectively denote the areas of transition segment and connection
segment. Lc , Lt and L j respectively represent the length of core plate (yield segment), transition
segment and connection segment. The stiffness of a BRB can be evaluated using the following equation.

1 1 1 1 Lc At A j  Lt Ac A j  L j Ac At
  (3)
K eff Kc Kt K j EAc At A j
D is energy concentration factor ( D =0.5) and Lwp is the distance between work points (see Figure
1(b)). The core strain H c can be evaluated using the following equations.

Hc H wp / D (4)
2190 H.Y. CHANG and C.K. Chiu / Procedia Engineering 14 (2011) 2187–2195

D Lc / Lwp (5)

Figure 2 shows the configuration of a chevron BRBF, hysteresis loops and BRB strain and story drift.
For a chevron BRBF, if the beam deformation is ignored, the BRB average strain H wp can be estimated
by using the equation of story drift angle T and brace slope I
T
H wp sin(2I ) (6)
2

2.3. Deformation criteria

From equations (1) and (3), BRB deformations at first yield can be estimated as

Dy Py / K eff (7)

BRB members tend to develop the maximum strength when buckling failures occur. From
experimental statistics, the ratio of post-yield stiffness to elastic stiffness was found to have an average of
0.05. BRB deformations at buckling can therefore be calculated as

Dm Dy  ( Pmax  Py ) /(0.05K eff ) (8)

A regression analysis has been made on the result of fracture tests on 26 BRB members. The
maximum deformation D f ,which a BRB member can develop before fracture, can predicted by the
following equations,

Df 2.78( Dave / D y ) 0.65 ˜ D y (9)

Dave CPD / 4 N f (10)

In the above equation, Rain Flow Theory has been used to count the number of plastic loading
cycles N f . In addition, the average plastic deformation Dave is defined as the ratio of accumulative
plastic deformation CPD to 4 N f .

2.4. BRB properties

The analyzed BRB members were originally designed to adopt in a 6-story steel office building
(Chang 2009). Figure 3 depicts the floor plan and elevation of the building. The building uses A572
Grade 50 steel built-up box columns and braces, and A36 steel H-shaped beams. Table 2 summarizes the
member sizes.
The earthquake response of the building has been simulated via a platform of inelastic structural
analysis for 3D systems (Lin and Tsai 2006). The building has a fundamental period of 0.99 sec and it
translates in the Z-direction at the first vibration mode. In the analysis, the frame was excited in the Z-
direction. The beams and columns were simulated utilizing plastic hinge models with bi-linear stress-
strain relation. The interaction surface of axial force and bending moment were also considered in the
columns.
The BRBs were replicated utilizing truss elements with the two-surface plastic hardening rule. The
BRB average strain analyzed was used to estimate the BRB core strain with equations (4) and (5). The
 H.Y. CHANG and C.K. Chiu / Procedia Engineering 14 (2011) 2187–2195 2191

largest deformation demands were found on the BRBs of the 2nd-story structure. In the following, the
seismic performance of the BRBs was selected to study in detail.

Table 2: Member size of example building

Stor
Story Column BRB_X BRB_Z Girder_X Girder_Z
y
H488×300×11×1
6F ƶ-450×19 Ac=25 cm2 Ac=35 cm2 RF H386×299×9×14 H386×299×9×14
8
H488×300×11×1
5F ƶ-450×19 Ac=25 cm2 Ac=35 cm2 6F H386×299×9×14 H386×299×9×14
8
H488×300×11×1 H390×300×10×1 H400×300×11×1
4F ƶ-450×22 Ac=50 cm2 Ac=55 cm2 5F
8 6 8
H488×300×11×1 H390×300×10×1 H400×300×11×1
3F ƶ-450×22 Ac=50 cm2 Ac=55 cm2 4F
8 6 8
H582×300×12×1 H390×300×10×1 H400×300×11×1
2F ƶ-450×25 Ac=50 cm2 Ac=55 cm2 3F
7 6 8
H582×300×12×1 H390×300×10×1 H594×302×14×2
1F ƶ-450×25 Ac=60 cm2 Ac=70 cm2 2F
7 6 3

9m
P #P

9m

9m

9m 9m 9m 9m 9m
(a) Floor plan (bold line: braced frame) (b) Elevation

Figure 3: Floor plan and elevation of example building

3. CONFIDENCE ASSESSMENT

A confidence parameter, O , has been used to associated with the probability that a structure will
satisfy a definite performance objective for a specific hazard (Yun et al 2002),
JJ a D
O (11)
IU IRC
In equation (1), C and D are median estimates of structural capacity and demand. J is demand
uncertainty factor that principally accounts for uncertainty inherent in prediction of demand arising from
variability in ground motion and structural response to that ground motion; J a is analysis uncertainty
factor that accounts for bias and uncertainty associated with specific analytical procedures used to
2192 H.Y. CHANG and C.K. Chiu / Procedia Engineering 14 (2011) 2187–2195

estimate structural demand; IU and IR are resistance factors that accounts for the uncertainty and
randomness inherent in prediction of structural capacity.
The C.L. can be determined using the O value by a back calculation to obtain the standard Gaussian
variate K x associated with probability of x not being exceeded found in conventional probability tables.

O e  EUT ( K x kEUT / 2b ) (12)

EUT ¦E
i
i
2
(13)

In equations (12) and (13), EUT is associated with the uncertainty about estimating structural demand
and capacity (see the details in 3.2 and 3.3). k is the logarithmic slope of the hazard curve at the desired
hazard level (refer to 3.1). b represents the change in demand as a function of ground motion intensity
(refer to 2.4).
As mentioned, the assessment of C.L. (or the calculation of O ) requires estimating site-specific
seismic hazard, structural demand and capacity. The following sections steps through the assumptions
used to calculate the coefficients in the two equations for the studied steel frame and brace members.

3.1. Site-specific Seismic Hazard

0.6 1.2
Spectral acceleration (g)

Achy024ew 1
Acceleration (g)

0.8
0 0.6
0.4
0.2
-0.6 0
0 10 20 30 40 50 60 70 80 90 Time (sec) 0 0.5 1 1.5 2 Period (sec)

Figure 4: Acceleration time history and elastic response spectra (5% damping ratio)

A set of 20 artificially generated ground motions was adopted in the analysis. Figure 4 gives an
example illustrating the acceleration time history and elastic response spectra. The phase contents of the
ground motions were simulated by using the data actually recorded around the building site. Throughout
iterative procedures, the amplitude spectra of the ground motions were proportionally fitted to that of the
maximum considered earthquake (MCE), corresponding to the 2% in 50-year probability of exceedance.
After that, the ground motions were scaled by multiplying scalar factors of 0.80 and 0.33, respectively
corresponding to the 10% and 50% in 50-year probability of exceedance.
The logarithmic slope of the hazard curve at the desired hazard level, k , is used in the evaluation of
the resistance factors I , demand factors J and confidence levels (C.L.). The hazard curve is a plot of the
probability of exceedance of a spectral amplitude value, H Si , versus the spectral amplitude for a given
response period, Si .

H Si ( Si ) k0 Si k (14)
 H.Y. CHANG and C.K. Chiu / Procedia Engineering 14 (2011) 2187–2195 2193

In equation (3), k0 is a constant. Given the spectral acceleration values at 10%/50 year and 2%/50 year
exceedance probabilities, for example, the value of k can be calculated as
H s1(10 / 50 )
ln( )
H s1( 2 / 50 ) 1.65
k | 7.40 (15)
S S
ln( 1( 2 / 50 ) ) ln( 1( 2 / 50 ) )
S1(10 / 50 ) S1(10 / 50 )

3.2. Structural Capacity and Resistance Factors

The median estimates of deformation capacities and resistance factors for BRB members tested using
the AISC loading protocol are summarized in Table 3. Structural capacity and resistance factors needs
evaluating for a specific limit state. For the studied BRB members, first yield, buckling and fracture are
the three limit states to consider. The deformation capacity C and uncertainty parameter EUC have been
evaluated by experimental statistics and theoretical prediction (Chang and Huang 2010). It was found that
the deformation capacities of BRB members, C , could be theoretically predicted. For example, The BRB
capacities against fracture failures were evaluated using equations (9) and (10). In the equations, the
number of load cycle and BRB average strain were estimated by nonlinear time history analysis (see the
details in 2.4 and 3.3).
EUC is the standard deviation of the natural log of BRB deformation capacity due to uncertainty. For
first yield and fracture failure, as illustrated by Table 3, EUC is 0.10 and 0.15, respectively. The ratios of
post-yield stiffness to elastic stiffness were found to have a large variation, and significantly increased the
uncertainty about predicting the deformation at buckling. For simplicity, EUC is assumed to be 0.4 for
buckling. Following that, the resistance uncertainty factor IU is calculated as follows,
2
IU e  kEUC / 2b (16)

E RC is the standard deviation of the natural log of BRB deformation capacity due to
randomness. E RC can be assumed to be 0.20. Following that, the resistance variability
factor IR is calculated as
2
IR e  kE RC / 2b (17)

Table 3: BRB deformation capacities and resistance factors ( b =0.813, k =7.40)

Failure mode C (%) EUC IU E RC IR
Yield 0.19 0.10 0.96 0.20 0.83
Buckling 1.82 0.40* 0.48 0.20 0.83
Fracture 0.77 0.15 0.90 0.18 0.86
2194 H.Y. CHANG and C.K. Chiu / Procedia Engineering 14 (2011) 2187–2195

Table 4: BRB deformation demands and demand factors ( b =0.813, k =7.40)

Seismic hazard D (%) Ea Ja E RD J

50%/50 year 0.46 0.17 1.14 0.20 1.20
10%/50 year 0.59 0.23 1.27 0.20 1.20
2%/50 year 0.80 0.24 1.30 0.20 1.20

Table 5: Calculation of confidence levels for BRB members ( b =0.813, k =7.40)

Performance level Failure mode O EUT Kx C.L.
Yield 4.16 0.28 -3.82 <1%
I.O.
Buckling 0.87 0.48 2.47 99.3%
Fracture 1.51 0.34 0.34 63.1%
L.S.
Buckling 1.24 0.50 1.85 96.8%
Fracture 2.10 0.35 -0.53 29.9%
C.P.
Buckling 1.72 0.51 1.26 89.6%

3.3. Structural Demand and Demand Factors

The median estimates of deformation demands and demand factors for BRB members are summarized
in Table 4. Structural demands and demand factors needs evaluating at a definite hazard level. For the
studied steel frame and BRB members, the 2%, 10% and 50% in 50-year probability of exceedance are
the three levels of seismic hazard to consider. The deformation demand D and variability parameter
E RD have been evaluated using nonlinear time history analysis (Chang 2009). Since the example
building is not located near a known fault, the variability in excitation orientation is not considered.
The E RD value is calculated by taking the standard deviation of the log of the maximum story drifts
evaluated for each of the 20 accelerograms referred above. Therefore E RD only reflects the variability in
acceleration. Following that, the demand factor J is calculated as
2
J e kE RD / 2b (18)

The parameter E a considers the uncertainties about analysis procedures, bias factor, damping, live
load and material properties. The uncertainties about damping, live load and material properties are
negligible, when compared to those about the analysis procedure and bias factor. The uncertainty about
analysis procedures is the extent that the benchmark, nonlinear time history analysis represents actual
physical behavior. The bias factor for each procedure is calculated as the ratio of the median demand
resulting from nonlinear time history analysis divided by that from the other analysis procedure. Based on
the judgment and understanding of the relative importance of strength degrading, P-delta effects, and
phenomena not well considered for in the analysis, E a is assumed to be 0.20 for the studied steel frame
and BRB members. Following that, the demand uncertainty factor J a is calculated as
2
Ja e kEa / 2b (19)
 H.Y. CHANG and C.K. Chiu / Procedia Engineering 14 (2011) 2187–2195 2195

4. SUMMARY AND CONCLUSION

The calculation of confidence levels are summarized in Table 5. For IO, as illustrated by Table 1, the
BRBs are allowed to yielding or buckling. The level of confidence therefore has been calculated against
the potential of BRB yielding and buckling failures. Accordingly, for LS and CP, the level of confidence
has been calculated against the potential of buckling and fracture failures.
From Tables 3 and 4, it is clear that the BRBs have enough capacity against buckling failures when the
building approaches the IO limit state. It can also be found that the BRBs have large capacity against
fracture failures when the building approaches the LS limit state. Therefore, as can be seen in Table 5, the
BRBs can provide a high level of confidence that ensure the building to satisfy the desired IO and LS
performance.
When the building approaches the CP limit state, however, the level of confidence against the potential
of BRB fracture failures may become far smaller than the 50%-value recommended by FEMA 351 (see
Table 5). It means the BRBs cannot provide an appropriate level of confidence in meeting the
performance objective of CP. The result has suggested a necessity of more carefully assessing the seismic
vulnerability of braced steel frames, especially for collapse prevention and from a local perspective.

ACKNOWLEDGMENTS

The author greatly thank National Center for Research on Earthquake Engineering (NCREE) in
Taiwan for providing the BRB test data and artificially ground motions analyzed in this paper.

REFERENCES

[1] FEMA 356 (2000). Prestandard and Commentary for the Seismic Rehabilitation of Buildings, Federal Emergency
Management Agency, Washington, D.C.
[2] Chang HY. (2009). Seismic fragility assessment of buckling restrained braced frames. Proceedings of the 11th Taiwan-
Korea-Japan Joint Seminar on Earthquake Engineering for Building Structures (SEEBUS), Kyoto, Japan.
[3] Chang HY and Huang YJ. (2010). An investigation of BRB experimental performance. Proceedings of the National
Conference on Structural Engineering, Taoyuan, Taiwan (in Chinese).
[4] Yun SY, Hamburger RO, Cornell CA and Foutch DA. (2002). Seismic performance evaluation for steel moment frames.
Journal of Structural Engineering, Vol. 128, No.4, pp. 534-545.
[5] Lin BZ and Tsai KC. (2006). Platform of inelastic structural analysis for 3D systems - PISA3D R2.0.2 users’ manual. Taipei,
National Center for Research on Earthquake Engineering (NCREE).