EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2008; 37:1099–1115

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eqe.803

Pseudo-dynamic test of a full-scale CFT/BRB frame—Part II:

Seismic performance of buckling-restrained braces
and connections

Keh-Chyuan Tsai1, 2, ∗, †, ‡, § and Po-Chien Hsiao2, 3, ¶ , 

1 Department
of Civil Engineering, National Taiwan University, Taipei, Taiwan
2 NationalCenter for Research on Earthquake Engineering (NCREE), Taipei, Taiwan
3 Department of Civil and Environmental Engineering, University of Washington, Seattle, WA, U.S.A.

SUMMARY
This paper is Part II of a two-part paper describing a full-scale 3-story 3-bay concrete-filled tube (CFT)/
buckling-restrained braced frame (BRBF) specimen tested using psuedo-dynamic testing procedures. The
first paper described the specimen design, experiment, and simulation, whereas this paper focuses on the
experimental responses of BRBs and BRB-to-gusset connections. This paper first evaluates the design of
the gusset connections and the effects of the added edge stiffeners in improving the seismic performance
of gusset connections. Test results suggest that an effective length factor of 2.0 should be considered
for the design of the gusset plate without edge stiffeners. Tests also confirm that the cumulative plastic
deformation (CPD) capacity of the BRBs adopted in the CFT/BRBF was lower than that found in typical
component tests. The tests performed suggest that the reduction in the BRB CPD capacities observed in
this full-scale frame specimen could be due to the significant rotational demands imposed on the BRB-
to-gusset joints. A simple method of computing such rotational demands from the frame inter-story drift
response demand is proposed. This paper also discusses other key experimental responses of the BRBs,
such as effective stiffness, energy dissipation, and ductility demands. Copyright q 2008 John Wiley &
Sons, Ltd.

Received 9 March 2007; Revised 16 January 2008; Accepted 23 January 2008

KEY WORDS: buckling-restrained brace; gusset plate; cumulative plastic deformation; rotational demand;
energy dissipation; ductility demand

∗ Correspondence to: Keh-Chyuan Tsai, National Center for Research on Earthquake Engineering (NCREE), 200,
Sec. 3, HsinHai Rd., Taipei 106, Taiwan.
†
E-mail: [email protected], [email protected]
‡
Professor.
§ Director.
¶ Former Research Assistant.
 Ph.D. Student.

Contract/grant sponsor: National Science Council of Taiwan

Contract/grant sponsor: Nippon Steel Company

Copyright q 2008 John Wiley & Sons, Ltd.

 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1100 K.-C. TSAI AND P.-C. HSIAO

1. INTRODUCTION

As described in Part I [1], a two-phase test of a full-scale 3-story 3-bay concrete-filled tube (CFT)/
buckling-restrained braced frame (BRBF) specimen was conducted. The input ground motions for
the pseudo-dynamic testings (PDTs) were chosen from the 1999 Chi-Chi and the 1989 Loma Prieta
earthquakes and scaled to represent 50, 10, and 2% in 50-year seismic hazard levels. Following
the PDTs in each phase, quasi-static loads were applied to cyclically push the frame to larger
inter-story drifts up to the failure of the braces [2]. The investigation allows for the evaluation of
the experimental story shears resisted by the BRBs and the hysteretic energy absorbed by the BRB
at the structural system level. These tests have also allowed for the investigation of seismic design
issues of BRBF for the BRB members and their connections. This paper first presented the detailed
design of the BRB members installed in the CFT/BRBF. By instrumentation and calibration of the
BRBs before the tests, the measured BRB axial forces were used to investigate the responses of
the BRBs and the brace-to-gusset connections noted above. The paper then reviews the design of the
gusset plates, which buckled in the PDTs and cyclic loading tests as described in Part I of this paper,
and evaluates the effects of the added edge stiffeners in improving the seismic performance of the
gusset connections. Finally, this paper introduces a simple method of computing BRB end rotational
demands from the frame inter-story drift response demands and concludes with the discussion of the
BRB cumulative plastic deformation (CPD) capacity, energy dissipation, and peak ductility demand.

2. DETAILED DESIGN OF BRB COMPONENTS

2.1. Compositions of BRBS

The BRB is primarily composed of a ductile steel core, a buckling-restraint mechanism, and an
unbonding mechanism, as shown in Figure 1(a). The steel core consists of three segments: a
restrained yielding segment, a restrained non-yielding segment, and an unrestrained non-yielding
segment [3]. Only the steel core is supposed to supply stiffness to the BRB. The effective stiffness,
K eff [4], of the BRB can be calculated as follows:
1
K eff = (1)
Lc Lt Lj
+ +
E Ac E At E Aj
where L c , L t , and L j are the length of the restrained yielding, restrained non-yielding, and unre-
strained non-yielding segments, respectively, and Ac , At , and Aj are the cross-sectional areas of the
restrained yielding, restrained non-yielding, and unrestrained non-yielding segments, respectively.
These dimensions are shown in Figure 1(b). The buckling-restraint mechanism is made up of one
or multiple steel tubes, which may or may not be filled with mortar, to restrain the steel core from
buckling. The unbonding mechanism, intended to minimize the transfer of axial force from the
steel core to the buckling-restraining parts, can be achieved by a layer of unbonding material or a
very small air gap between the steel core and the buckling-restraining parts.

2.2. Design of BRB specimens

As noted in the CFT/BRBF Phase-1 tests, three types of BRBs were installed, including the single-
core UB, the double-core BRB, and the all-metal BRBs. The cross-sectional areas of the BRBs

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PSEUDO-DYNAMIC TEST OF A FULL-SCALE CFT/BRB FRAME 1101

were determined as described in Part I [1]. Figures 2(a)–(c) show the cross-sectional schemes of
the BRBs in Phase-1 from the 1st- to the 3rd-stories, respectively. The two BRBs in the 3rd-story
were double-core constructed using cement mortar infilled in two rectangular tubes [4]. The two
single-core unbonded braces (UBs), each consisting of a steel flat plate in the core, were donated
by Nippon Steel Company and installed in the 2nd-story. The two all-metal double-core BRBs with
detachable features [5] were installed in the 1st-story. Table I and Figure 1 give the dimensions
of the Phase-1 BRBs. After the Phase-1 tests, six new BRBs were installed in the original frame

Concrete
Core Steel
(Mortar)
Member

Tab
Plate
Unbonding
Materials

Steel Tube Buckling

(Buckling Restrained Part) Restrained
(a) Brace

Lj/2 Lt/2 Lc Lt/2 Lj/2

W.P. Lwp W.P

(b) .

Figure 1. Construction and dimensions of buckling restrained braces.

(a) (b) (c)

flat and U-shape and

opened closed casing
casing
Before modification After modification
(as used in Phase-1) (as used in Phase-2)
(d)

Figure 2. Cross sections of three types of BRBs: (a) all-metal-detachable BRB in 1st-story; (b) single-core
UB in 2nd-story; (c) double-core BRB 3rd-story; and (d) the modification of the all-metal BRBs.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1102 K.-C. TSAI AND P.-C. HSIAO

Table I. The dimensions of the Phase-1and Phase-2 BRBs (and UBs).

Phase-1 Phase-2
Specimen no. 1BRB 2UB 3BRB 1BRB 2BRB 3BRB
Py (tonf) 127 102 56 107 89.2 53.5
Material A572 Gr.50 SN490 B A572 Gr.50 A572 Gr.50 A572 Gr.50 A572 Gr.50
Core 2@T16×64×30 PL16×157 2@PL12×63 2@PL12×108 2@PL12×90 2@PL12×54
Tube 4@80× 250×250×6 2@200× 2@200× 2@200× 2@200×
80×4 100×6 100×6 100×6 100×6
Bolts
(F10T M24) 10 8 6 10 (welded) 6

0.57 0.49 0.60 0.50 0.50 0.50
Ac 3016 2512 1510 2592 2160 1296
At 5632 0 3840 5280 4800 4320
As 0 76.8 0 0 0 0
Aj 7680 8352 5520 6000 6000 6000
Lc 2965 2628 3220 2605 2658 2660
Lt 520 0 600 880 1050 1160
Ls 0 20 0 0 0 0
Lj 1698 2667 1514 1698 1607 1514
L wp 5183 5315 5334 5183 5315 5334
Units are in mm.

specimen for the Phase-2 tests. It included two modified all-metal double-core BRBs used in the
1st-story and four concrete-filled double-core BRBs used in the 2nd- and the 3rd-stories. Figure
2(d) shows the key difference between the original (used in Phase-1) and the modified (used in
Phase-2) all-metal BRBs in the 1st-story. After modification, the buckling-restraint mechanism of
the BRBs (by using a U-shape and closed casing) was stiffer than the original one where a flat
and open casing was used. In particular, welded connections were used at the bottom ends of the
2nd-story BRB-to-gusset connections; all other BRB connections used bolted details. Table I and
Figure 1 also give the dimensions of the Phase-2 BRBs and its connections.

3. EXPERIMENTAL RESPONSES OF BRBS IN THE 3-STORY FRAME SPECIMEN

3.1. Calibration of BRBS before the pseudo-dynamic frame tests

In order to measure the BRB axial forces during the frame tests, four strain gauges were placed
at each unrestrained non-yielding segment of the BRB, as shown in Figure 3. Before the BRBs
were installed in the frame specimen in each of the two phases, a least squares linear relationship
between each BRB force (up to 60% tensile yield capacity of BRB) and the average strain-gauge
reading was established. These coefficients of the linear relationships of all BRBs were used to
compute the BRB axial forces during the frame tests.

3.2. Experimental responses of gusset plates

3.2.1. Buckling of the gusset plates. Before Phase-1 testing, no edge stiffeners were installed in
any of the gussets. However, the gusset at the 1st-story buckled out of plane in Phase-1 test

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PSEUDO-DYNAMIC TEST OF A FULL-SCALE CFT/BRB FRAME 1103

Strain
Guage

BRB UB
(1FL,3FL) (2FL)

Figure 3. Location of strain gauges at the ends of each brace.

no. 1, as noted in Part I. After the buckled gusset was heat-straightened, edge stiffeners were
added onto three gussets at the 1st- to the 3rd-stories brace-to-beam connections as shown in
Figures 4(a)–(c), respectively. In addition, a 12 mm thick vertical stiffener has been welded under
the bottom flange of the existing lateral supporting beam and connected to the web stiffener of
the main girder. Subsequently, five PDTs were successfully conducted, without failure of any
gusset or BRB. However, in the Phase-1 final cyclic loading test, the bottom gusset (with no edge
stiffener) near the 3rd-story south BRB buckled. In the Phase-2 tests, six new BRBs were installed
after all the remaining unstiffened gussets were straightened and edge-stiffened for the 1st- to
3rd-stories as shown in Figures 4(d)–(f), respectively. After that, Phase-2 PDTs and cyclic loading
tests were conducted successfully without experiencing any gusset or BRB buckling. It is apparent
that stiffening the gussets, thus improving the stability of the gusset, was very effective in ensuring
the superior performance of the BRB frame systems. The effects of these stiffeners in improving
the seismic performance of the connections are carefully evaluated in the following section.

3.2.2. Review of gusset plate designs. The design procedures for the gusset plates were adopted
after following the investigations of [6, 7]. Based on Whitmore’s investigation, the magnitude of
the maximum stress, max , in the gusset plate can be estimated as
Pbrace
max = (2)
be ×t
where Pbrace is the brace force and t is the gusset thickness. The Whitmore effective width, be , is
defined as the distance between two lines radiating outward at 30◦ angles from the first row of bolts,
along a line running through the last row of bolts, as shown in Figure 5. In addition, the peak stress
is expected to occur at the location of the Whitmore effective width. Thornton [7], investigating
the compressive strength of steel gusset plates, proposed an intuitive and lower bound approach.
This was based on the buckling capacity of unit strips of lengths L 1 , L 2 , and L 3 below the line
of the Whitmore effective width, also shown in Figure 5. The most critical unbraced length L c of
the unit strip was taken as the maximum of lengths L 1 , L 2 , and L 3 , and an effective length factor
of 0.65 was recommended in Thornton’s research. The gusset plate would not buckle where the
buckling strength of the critical unit strip was greater than the compressive stress on the Whitmore
effective width. If
h ,
, and AFy represent the strain hardening, the material over-strength, and

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1104 K.-C. TSAI AND P.-C. HSIAO

the nominal yield capacity of the BRB, respectively, the maximum BRB tension and compression
strengths can be estimated from the following equations [8]:

Pmax T =
h
AFy (3)

Pmax C = 
h
AFy (4)

Figure 4. (a)–(c) The brace-to-beam gusset joint details after adding stiffeners in Phase-1 and (d)–(f) the
brace-to-column gusset joints after stiffening before Phase-2 from the 1st- to the 3rd-stories.

Pbrace
t
be
30

L2

L1
L3

Figure 5. Schematic diagram of the gusset under brace forces.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PSEUDO-DYNAMIC TEST OF A FULL-SCALE CFT/BRB FRAME 1105

The  factor, which commonly ranges from 1.1 to 1.3 in BRB component tests, indicates the ratio
between the BRB peak compressive and tensile capacities; it reflects the quality of the unbonding
mechanism. Thus, in the BRBF system, the capacity design procedures of gussets would require
that the gusset yield strength be greater than the BRB maximum tension force Pmax T and the
gusset buckling strength be greater than the BRB maximum compressive capacity force Pmax C :

Py = Fy be t
Pmax T for tension (5)

2 E
Pcr = be t
Pmax C for compression (6)
(K L c /r )2

where Pcr is the gusset buckling strength and L c is the most critical unbraced length of the
unit strips. Note that the Euler buckling formula has been applied in this study as it follows the
original research made by Thornton. Before the Phase-1 tests, all the gussets met the requirements
prescribed in Equations (5) and (6) using an effective length factor of 0.65 as suggested. However,
the gussets buckled out of plane in Phase-1 test no. 1 at a brace force of 805 kN, which was much
less than the Pcr of 7975 kN computed from Equation (6). Apparently, the critical compressive
capacity of the gusset connections requires further investigation.

3.2.3. Buckling analysis of gusset plates. Figure 6 shows the possible flexural buckling modes of
the BRB-to-gusset joints. Because the cross section near the two BRB ends outside the restrain-
casing of BRBs is smaller and more flexible than adjacent segments, the reduced cross section
can be represented by a hinge with a spring (shown in Figure 6(a)). Based on the buckling mode
observed in the experiment, the buckled shape shown in Figure 6(b) (K = 2.0) was more agreeable

K < 2.0 K= 0.65

(a) (b) (c)

Figure 6. Buckling shapes and boundary conditions of the gusset plates.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1106 K.-C. TSAI AND P.-C. HSIAO

Table II. Analytical capacities of gussets and braces in the two phases.
Analytical gusset capacity Analytical brace capacity
Phase-1 Phase-2
Pcr (kN) Pcr (kN)
Story Py (kN) L c (mm) K = 0.65 K = 2.0 Pmax T (kN) Pmax C (kN) Pmax T (kN) Pmax C (kN)
1F 2287 Top 288 7975 842 1558 1713 1312 1443
Bot. 226 12 939 1367
2F∗ 1841 Top 270 6695 707 — — 1094 1203
Bot. Welded connection
3F 1397 Top 297 5709 603 686 755 656 722
Bot. 268 7036 743
∗ Only considered for Phase-2 PDTs.

than that shown in Figure 6(c) (K = 0.65). Table II shows the gusset critical length L c , gusset
buckling capacities Pcr considering the 0.65 and 2.0 effective length factors Py , the gusset yield
strength, analytical BRB peak tension Pmax T , and compression Pmax C forces using Equations (5)
and (6) of all the brace-to-gusset joints in the two phases. Note that the end cross section of the
2nd-story UBs is cross-shaped; therefore, the gusset was stiffened along the centerlines of the
end ribs of the UB (Figure 4(b)). These gusset stiffeners also serve to transfer the brace forces
into the brace-to-beam or column joints. The effects of these centerline stiffeners on the gusset
are not discussed in this paper. Considering the buckled gusset in the Phase-1 test no. 1 with an
effective length factor K = 2.0 and a critical length L c of 288 mm, it was found that the analytical
gusset-buckling capacity of 842 kN more or less matched the experimental gusset-buckling force
of 805 kN. Further analytical results obtained from the ABAQUS finite element analyses [9] also
suggested that the buckling capacity of the gusset joint without the presence of edge stiffeners
was 750 kN. These stated analyses help to explain why the 1st-story gusset buckled. Table II
also explains why the 3rd-story south-BRB-bottom-end gusset, which has a buckling capacity of
743 kN when an effective length factor of 2.0 was considered, buckled during the Phase-1 cyclic
loading test when the BRB peak compression force of 755 kN had developed. Based on these
experimental observations and the stated analyses, it appears that an effective length factor of 2.0
is more appropriate than 0.65 for the design of double-core BRB gusset plates, when no edge
stiffener is present in the gusset connections.
After the gussets were stiffened at the edges, the boundary conditions and the buckling shapes
of the gusset plates should be close to those shown in Figure 6(c). In this case, when the effective
length factor K = 0.65 is considered, the stiffened gussets’ buckling capacities would be much
greater than the BRBs’ experimental peak compression forces. This helps to explain why there
was no gusset or BRB buckling failure after stiffening the gussets. From these tests and analyses,
it appears that the buckling capacity of the buckled gussets can be estimated using the simplified
approach stated above. However, additional research is required to gain further insights into the
seismic design of gusset connections for various types of BRB connections. For these purposes, a
full-scale 2-story BRBF specimen has been tested using substructure PDT procedures considering
bi-directional earthquake ground motions [9].

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PSEUDO-DYNAMIC TEST OF A FULL-SCALE CFT/BRB FRAME 1107

3.3. Experimental responses of BRBS

It was found that the Phase-1 BRBs started to develop plastic deformation at the 1st- and 2nd-
stories in the Phase-1 test no. 2 ( 50
50 event). Except for the buckling of the gusset noted above, the
BRBs performed satisfactorily, without any failure evident, during the PDTs of the two phases.
Figure 7 shows the measured south-BRBs force versus the axial deformation relationships for the
2
1st- to 3rd-stories in the 50 events of the two phases. The BRB had full hysteretic responses and
dissipated a significant amount of energy. Table III presents the average ratio of the story shear
resisted by braces to the total story shear before braces yield. It was found that the ratios of each
story were about 80%, which is in close agreement with the design target.

2 2 2
TCU082EW (2/50) TCU082EW (2/50) TCU082EW (2/50)
0.622g 0.622g 0.622g
Axial Force (103kN)

Axial Force (103kN)

Axial Force (103kN)

1 Py 1 1
Py Py
Phase 1
Phase 1
0 0 0

-1 -1 -1
Test No.5
Test No.5 1BRBS Test No.5 2UBBS Phase 1 3BRBS
-2 -2 -2
-60 -30 0 30 60 -60 -30 0 30 60 -60 -30 0 30 60
(a) Displacement (mm) (b) Displacement (mm) (c) Displacement (mm)

2 2 1.5
LP89g04NS (2/50) LP89g04NS(2/50) LP89g04NS (2/50)
Axial Force (103kN)

Axial Force (103kN)

Axial Force (103kN)

0.61g 0.61g 1 0.61g

1 Py 1 Py
Py 0.5

0 0 0
-0.5
-1 -1
Test No.2 Test No.2 -1 Test No.2
Phase2 1BRBS Phase2 2BRBS Phase2 3BRBS
-2 -2 -1.5
-60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60
(d) Displacement (mm) (e) Displacement (mm) (f) Displacement (mm)

Figure 7. (a)–(c) Hysteretic responses of the 1st- to 3rd-stories south BRBs in the Phase-1 tests and (d)–(f)
2
in the Phase-2 tests (all in the 50 events).

Table III. Ratios of the brace shear to the story shear in the two phases.
Average ratio of the brace shear to the story shear (%)
Story 1 2 3
Phase-1 80.8 82.4 79.6
Phase-2 74.8 81.6 85.4

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1108 K.-C. TSAI AND P.-C. HSIAO

3.4. Energy dissipation capacity of BRBS

Hysteretic energy dissipations can be computed by direct integration of the force versus displace-
ment of the BRBs or of the entire system. Figure 8 showsthe energy dissipated by the BRBs and
by the entire structural system in each story during the 50 10 2
50 , 50 , and 50 events of the two phases
Evidently, the BRBs dissipated most of the energies during the PDTs. The north and south braces
dissipated energy uniformly in each story. Test results confirmed that the BRBs could effectively

50 10 2
Figure 8. Energy dissipations of the BRBs in Phase-1 tests, (a) 50 , (b) 50 , (c) 50 events
and in Phase-2 (d) 10 2
50 , (e) 50 events.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PSEUDO-DYNAMIC TEST OF A FULL-SCALE CFT/BRB FRAME 1109

dissipate seismic input energy in various seismic hazard levels, while the moment resisting frame
would need to dissipate more energy only during strong earthquakes.

3.5. CPD capacity of BRBS

The CPD was calculated for all the BRBs using the following equation:

Ed
CPD = (7)
2Py y

where Py and y are the yield force and deformation of the brace, respectively, and E d is the total
amount of energy dissipated by the BRB. Figure 9 presents the CPD of all the BRBs, ranging
from about 200 to 350, computed from the PDTs and the cyclic loading test before brace or
gusset failures occurred in the two stated phases of study. All the BRBs had met the minimum
required CPD capacity, 200, prescribed in the AISC specification [10]. Also, before applying the
cyclic loads in each of the two phases, the CPDs of the BRBs ranged from about 70 to 140 never
reaching one half of the total CPDs imposed during the entire phases. It was confirmed that the
CPD demands imposed by any single earthquake were much less significant than a single large
magnitude cyclic loading test The BRBs had the ability to resist the combined earthquake load
effects.
In the Phase-1 tests, as noted in Part I, the failure mode of the 1st-story all-metal BRBs was
high-mode buckling, and the failure mode of the 3rd-story BRBs was gusset out-of-plane buckling.
Only the 2nd-story UBs were tested to fracture in the final cyclic loading test. Thus, the 1st- and
3rd-story braces did not develop their CPD capacity before the brace or gusset buckling failure.
Unlike the situation in Phase-1, all Phase-2 BRBs fractured near the end of Phase-2 cyclic loading
test without brace or gusset buckling failure and reached their CPD capacities. Test results indicate
that the CPD capacities of the fractured UBs or BRBs adopted in the CFT/BRBF were lower than
those found from the individual component tests in other studies [11]. In a typical BRB component
test, only cyclic axial loads were applied and no rotational demand was imposed at the brace ends.

Figure 9. CPDs of the BRBs (a) in Phase-1 and (b) in Phase-2 tests.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1110 K.-C. TSAI AND P.-C. HSIAO

3.6. Rotational demands imposed on the BRB-to-gusset joints

As the BRBs were either bolted or welded to the gusset plate, the BRB ends could not be
considered as entirely pin-connected to the structure. In order to gain insight into the rotational
demand imposed on both ends of a BRB, three tiltmeters were installed on each BRB, one on the
mid-span of the steel casing, and one on the rib stiffeners of each end of the BRB. Assuming that
the steel casing is rigid, brace end rotations (BERs) can be computed by subtracting the mid-span
tiltmeter reading from each of the corresponding end tiltmeter readings. The idealized BERs, top
and bottom , are shown in Figure 10(b), and the relationships between the maximum BER and the
peak inter-story drift angle were investigated in this study. The experimental average ratios of the
BER to the inter-story drift in the two phases are presented in Columns 7 and 8 (EXP values)
of Table IV in terms of percentage of story drift (PSD). Because the tiltmeter on the 2nd-story
BRB casing was found to be non-functional after tests, experimental 2nd-story BERs were not
obtained. In Table IV, the positive values ofBERs mean counterclockwise rotations. It was found
that the PSD values were quite significant and vary from positive 96 to negative 30. Furthermore,
having different rotational directions at two ends of the BRBs, the braces could deform into one
of the two shapes, an S-shape or a C-shape as shown in Figures 10(d) and (e), respectively. Based

L
a 1 w.p.
top L
L
a2
bottom
w.p.
(a) (b) (c)

(d) (e)

Figure 10. Schematics of (a) the BRB frame, (b) concentration of brace end rotations, (c) definition of
the brace end and the casing lengths, (d) S-shape, and (e) C-shape deformations of the BRBs.

Table IV. Computation of the BRB end rotations in the two phases using the simplified procedures.
3 4 top bottom Exp. top Exp. bottom top bottom
(PSD) (PSD) a1 a2 (PSD) (PSD) (PSD) (PSD) Error (%) Error (%)
Phase-1 1F 7 35 0.294 0.293 94 52 96 58 −2 −10
2F −9 97 0.403 0.413 56 −32 — — — —
3F −4 99 0.212 0.278 57 −38 63 −30 −9 27
Phase-2 1F 13 44 0.287 0.286 99 42 86 36 15 15
2F 4 108 0.258 0.266 68 −44 — — — —
3F −4 105 0.218 0.285 55 −46 60 −39 −8 17
PSD, percentage of story drift.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PSEUDO-DYNAMIC TEST OF A FULL-SCALE CFT/BRB FRAME 1111

on these observations, the BER demands can be significant and should be properly incorporated
into the seismic design of BRBF. Unfortunately, the BER demands measured from the tests are
very difficult to compute using traditional inelastic frame response analysis computer programs
where BRBs are represented by truss elements. Thus, a simplified procedure for estimating the
BER demands is proposed as follows.
In this model, it is assumed that the BRB-to-gusset connections are rigid, and only the BRB’s
steel-casing-end might be subjected to rotational demand, as shown in Figure 10(b). The BER is
affected by the story drift angle , beam rotation 3 , and column rotation 4 , as shown in Figure 11.
First, if given a story drift  to the frame (Figure 11(a)) and keeping the ends of BRB near the
beam and column without rotating as shown in Figure 11(b), the BERs, top and bottom , will then
equal 1 as follows:

1 =  sin2 (1+a1 +a2 ) (8)

where a1 and a2 represent the length factors measured from the work point to the ends of the BRB
steel-casing (Figure 10(c)). Similarly, after the effects of beam and column rotations, 3 and 4 ,
as shown in Figures 11(c) and (d), the final top and bottom BERs, top and bottom (Figure 11(e)),
can be computed from the following equations:

top = 1 +2 +3 −5 = 1 +3 (1+a1 )−a2 4 (9)

bottom = 1 +2 −4 −5 = 1 +a1 3 −4 (1+a2 ) (10)

Figure 11. Contributions of frame joint rotations on the BRB end rotations, assuming all BRB-to-gusset
joints are rigid and BRB end rotations occur at the ends of steel casing.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1112 K.-C. TSAI AND P.-C. HSIAO

where 2 = a1 3 and 5 = a2 4 . Thus, if the story drift angle , the beam rotation 3 , and the
column rotation 4 are given, the BERs can be determined using Equations (8)–(10). In order to
examine the effectiveness of this simplified procedure in estimating the BER, the experimental , 3
(obtained from tiltmeters mounted on the beam web), and 4 (obtained from tiltmeters mounted on
the CFT column web immediately above the floor slab or footing) are substituted into the equations.
As listed in Table IV, the computed results (columns 5 and 6) predicted the experimental BER
demands (columns 7 and 8). The errors in applying Equations (9) and (10) to estimate the BERs
range from 2 to 27%. As noted previously, the tiltmeter on the 2nd floor brace casing was not
functioning. The values of the computed 2nd-story BERs suggest that its demands were similar to
those of the 3rd-story. Based on these analyses, it appears that the proposed method was effective
in estimating the BERs. This method can be conveniently implemented for structural engineering
practice of BRBF design as the story drift angle , the beam rotation 3 , and the column rotation
4 are available from the output of typical nonlinear frame response analyses.
Moreover, from Table IV it is evident that the beam rotation 3 in each story was only a rather
small percentage of the story drift, and the column rotation 4 was almost equal to the story drift in
the 2nd- and the 3rd-stories. Thus, it can be assumed that the beam rotation 3 is zero in each floor,
and the column rotation 4 is equal to the story drift  for typical aboveground floors. Based on the
experimental responses observed from the test frame, 4 was assumed to be about one-third of the
1st-story drift. Equations (11) and (12) show the results of incorporating the above assumptions
into Equations (9) and (10):

a2
top = 1 −  for 1st story
3
= 1 −a2  for other stories (11)

bottom = 1 − (1+a2 ) for 1st story
3
= 1 −(1+a2 ) for other stories (12)

Table V presents the predicted BERs for the two phases after applying Equations (11) and (12).
The errors of the prediction are less than 50%. Thus, this is a simple method to estimate the BERs
with using only the inter-story drift demands.

Table V. Computation of the BRB end rotations in the two phases by the simplified procedures
neglecting the beam joint rotation.
3 4 top bottom Exp. top Exp. bottom top bottom
(PSD) (PSD) a1 a2 (PSD) (PSD) (PSD) (PSD) Error (%) Error (%)
Phase-1 1F 0 33 0.294 0.293 85 34 96 58 −11 −41
2F 0 100 0.403 0.413 43 −32 — — — —
3F 0 100 0.212 0.278 49 −38 63 −30 −22 28
Phase-2 1F 0 33 0.287 0.286 68 52 86 36 −21 44
2F 0 100 0.258 0.266 49 −35 — — — —
3F 0 100 0.218 0.285 49 −38 60 −39 −19 −2
PSD, percentage of story drift.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PSEUDO-DYNAMIC TEST OF A FULL-SCALE CFT/BRB FRAME 1113

Figure 12. (a) Locations of tiltmeters in each brace and connection details adopted for (b)
single-core UBs and (c) double-core BRBs.

Table VI. Relative rotation between two work points and two brace ends for the BRBs of the two phases.
Brace type total (deg.) brace (deg.) Difference (%)
Phase-1 1F All-metal BRB 42 38 10
2F UB 88 65 26
3F BRB 95 93 2
Phase-2 1F All-metal BRB 57 50 12
2F BRB 112 107 4
3F BRB 101 99 2

In addition, the rotations that occurred within the two gusset connections can be computed from
the installed tiltmeters, in Positions 1–5 on a BRB body and connections, as shown in Figure 12(a).
Figures 12(b) and (c) schematically show connection details adopted for BRBs and UBs. The
rotational difference between Positions 1 and 5 represents the relative rotation of two work points,
defined as total . The rotational difference between Positions 2 and 4 represents the relative rotation
of the two ends of the brace, defined as brace . The difference between total and brace represents the
total relative rotation between the brace and the gussets, including the deformations of the gussets
and bolt slippage in the bolted joints. Table VI, column 4, shows the magnitude of the differences
(in terms of total ) in the two phases. It is apparent that the total relative rotation between the
Phase-1 2nd-story UB and its gussets was as large as 26% of total , whereas it was only about 4%
of total for Phase-2 2nd-story BRB. This suggests that the UB bolted connection details adopted
in Phase-1 (Figure 12(b)) were more flexible than the welded BRB to edge-stiffened gusset joint
adopted in Phase-2.

3.7. Pseudo-dynamic peak ductility demands on BRBS

The peak ductility of BRBs was considered as the maximum plastic deformational demand imposed
during the PDTs. Figure 13 shows the tensile and compressive ductility of all BRBs in the two
phases, assuming that the yield strain of the BRBs was 0.002. Based on the PDT results, the peak
axial ductility demands of the BRBs imposed in the 10 2
50 and 50 events were about 5.0 and 10.0,
respectively, and the 2nd-story BRBs had the highest ductility demand in both the phases.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1114 K.-C. TSAI AND P.-C. HSIAO

3 3
50/50-Ten. 50/50-Ten.
50/50-Com. 50/50-Com.
10/50-Ten. 10/50-Ten.
10/50-Com. 10/50-Com.
2/50-Ten. 2/50-Ten.

2/50-Com. 2/50-Com.

Floor
Floor

2 2

Phase 1 Phase 1
1 1
0 5 10 15 0 5 10 15
(a) North Brace Ductilit y (b) South Brace Ductility

3 3
Phase 2 Phase 2

10/50-Ten. Floor 10/50-Ten.

Floor

10/50-Com. 10/50-Com.
2 2 2/50-Ten.
2/50-Ten.
2/50-Com. 2/50-Com.

1 1
0 5 10 15 0 5 10 15
(c) North Brace Ductility (d) South Brace Ductility

Figure 13. Tensile and compressive ductilities, in the (a) north, (b) south BRBs in Phase-1 tests, (c) north,
and (d) south BRBs in Phase-2 tests.

4. CONCLUSIONS

Based on the tests and analyses, summaries and conclusions are made as follows:
• Adding edge stiffeners onto the gussets was effective in preventing the out-of-plane buckling
of the gussets.
• The 1st- and the 3rd-story BRB gusset buckling capacities can be satisfactorily predicted
by assuming an effective length factor K = 2.0 before the gusset edges were stiffened. Test
results suggest that an effective length factor of 2.0 should be considered for the design of
the gusset plate without edge stiffeners.
• All BRBs exhibited rather full hysteretic responses and dissipated a significant amount of
energy. Within the elastic limit, BRBs resisted about 80% of the total story shear as designed.
• Tests confirmed that BRBs dissipated most of the energies imposed on the structural system
for various levels of earthquakes, while the moment resisting frame would need to dissipate
energy only during severe earthquakes.
• All BRBs possessed a large CPD capacity, ranging from 250 to 350. Tests confirmed that the
CPD capacity of the BRBs adopted in the CFT/BRBF is lower than that found in the typical
component tests.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe
 10969845, 2008, 7, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.803 by National Taiwan University, Wiley Online Library on [14/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PSEUDO-DYNAMIC TEST OF A FULL-SCALE CFT/BRB FRAME 1115

• Test results confirmed that the BRB end rotation demands were rather significant. These
demands ranged from 30 to 96% of the inter-story drift demand. The rotational demands can
be estimated from the frame inter-story drift response demand by using the proposed method.
• Test results indicated that the peak axial ductility demands of the BRBs imposed during the
10 2
50 and 50 events are about 5.0 and 10.0, respectively.

ACKNOWLEDGEMENTS
The National Science Council of Taiwan provided the financial support for this experimental research
program. Nippon Steel Company donated two unbonded braces, which were installed in the 2nd floor of
the frame specimen. Nippon Steel Company also provided some financial support for the construction of
the frame specimen. Valuable suggestions provided by many U.S., Japanese, and Taiwanese researchers
on this joint effort are gratefully acknowledged. In particular, Professors Lap-Loi Chung, Hsieh-Lung
Hsu, Cheng-Chih Chen, Ker-Chun Lin, Subhash Goel, Stephen Mahin, and Charles Roeder assisted in the
design of the frame specimen. Mr Kung-Juin Wang was in charge of the networked hybrid test software
development. Drs Yuan-Tao Weng and Min-Lang Lin, Mr Chui-Hsin Chen, Juin-Wei Lai, Sheng-Lin Lin,
Shiang-Jung Wang, and Wei-Choung Cheng assisted in the preparation and execution of the tests. The
laboratory support provided by NCREE is very much appreciated. A detailed list of other participants is
given at the web site http://cft-brbf.ncree.gov.tw. All these supports are gratefully acknowledged.

REFERENCES
1. Tsai K-C, Hsiao P-C, Wang K-J, Weng Y-T, Lin M-L, Lin K-C, Chen C-H, Lai J-W, Lin SL. Pseudo-dynamic
tests of a full-scale CFT/BRB frame. Part 1: specimen design, experiment and analysis. Journal of Earthquake
Engineering and Structural Dynamics 2008; DOI: 10.1002/eqe.804.
2. Tsai KC, Hsiao PC. Pseudo dynamic performance and analysis of a full scale CFT-BRB composite frame. Report
No. NCREE-05-001, National Center for Research on Earthquake Engineering, Taipei, Taiwan, 2005 (in Chinese).
3. Uang CM, Nakashima M, Tsai KC. Research and applications of buckling-restrained braced frames. International
Journal of Steel Structures, Korean Society of Steel Construction 2004; 4(4):301–313.
4. Tsai KC, Hwang YC, Weng CS, Shirai T, Nakamura H. Experimental tests of large scale buckling restrained
braces and frames. Proceedings, Passive Control Symposium, Tokyo Institute of Technology, Tokyo, December
2002.
5. Tsai KC, Lin SL. A study of all metal and detachable buckling restrained braces. Report No. CEER/R92-03,
Center for Earthquake Engineering Research, National Taiwan University, 2003 (in Chinese).
6. Whitmore RE. Experimental investigation of stresses in gusset plates. Bulletin No.16, Engineering Experiment
Station, University of Tennessee, 1952.
7. Thornton WA. Bracing connections for heavy construction. Engineering Journal (AISC) 1984; 21(3):139–148.
8. Tsai KC, Weng YT, Lin SL, Goel S. Pseudo-dynamic test of a full-scale CFT/BRB frame: part 1—performance
based specimen design. Proceedings of 13th World Conference on Earthquake Engineering, Vancouver, BC,
Canada, 1–6 August 2004; Paper No. 750.
9. Lin ML, Tsai KC, Hsiao PC, Tsai CY. Compressive behavior of buckling-restrained brace gusset connections.
The 1st International Conference on Advances in Experimental Structural Engineering (AESE), Nagoya, Japan,
19–21 July 2005.
10. AISC (American Institute of Steel Construction). Seismic Provisions for Structural Steel Building. AISC: Chicago,
IL, 2005.
11. Tsai KC. Research and application of double-core buckling restrained braces. Proceedings of the 7th Pacific
Structural Steel Conference, Long Beach, March 2004.

Copyright q 2008 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2008; 37:1099–1115
DOI: 10.1002/eqe