Professional Documents
Culture Documents
Rafael Sabelli1, SE, Warren Pottebaum1, PE, J. Cressica Brazier1, PE, and Walterio López2, SE,
1
DASSE Design Inc., 33 New Montgomery, San Francisco, CA 94105.
2
Rutherford & Chekene, 427 13th Street, Oakland, CA 94612.
Abstract
The paper illustrates a design procedure for Buckling-Restrained Braced Frames using the current draft
versions of the 2005 ASCE 7 and AISC 341 standards. The system is new to this cycle of codes, and presents
Downloaded from ascelibrary.org by PENN STATE UNIV on 06/11/14. Copyright ASCE. For personal use only; all rights reserved.
many performance and design advantages over conventional braced frames. The illustration of a design
procedure is intended to aid engineers in understanding the requirements of the provisions as they pertain to
buckling-restrained braced frames, and to convey the intended performance of the system under severe seismic
loading. The procedure illustrated optimizes the design, and is consistent with the required capacity-design
methodology without requiring numerous calculations to supplement an elastic analysis. A five-story building in
San Francisco is designed to meet the provisions, and the final design information is described.
Braced frames have long been known to exhibit undesirable modes of post-elastic behavior
due to the loss of strength and stiffness resulting from buckling of braces (SEAOC, 1999).
The development of design and detailing procedures aimed at extending the ductility of
buckling braces has been a significant avenue of research (Astaneh and Goel, 1984; Goel,
1992). Another avenue has been the development of braced-frame systems that do not rely
upon brace buckling for ductility, such as the eccentrically braced frame (Popov et al., 1989).
Buckling-restrained braced frames fall into this latter category. The source of ductility for
this system is the buckling-restrained brace, which permits ductile yielding of steel in both
compression and tension distributed along a large portion of its length. Buckling-restrained
braces resist axial forces in a steel core, which is restrained from transverse buckling by
means of a sleeve or other confining element. For more information, see Uang and
Nakashima (2003), Aiken et al. (2000), Prasad (1992) and the AISC 341 Commentary.
The buckling-restrained braced-frame system was introduced to US design practice in 1999.
Since then, design provisions have been under development by the Structural Engineers
Association of California, the American Institute of Steel Construction, and the Building
Seismic Safety Council (Sabelli and Aiken, 2004). Draft provisions and guidelines have been
published at numerous points during the development process (Sabelli and Aiken, 2001;
BSSC, 2003; Sabelli, 2004).
This development effort has led to the inclusion of design provisions for the system in the
2005 edition of the AISC Seismic Provisions for Structural Steel Buildings (AISC 341).
Provisions for the system are also under review for Supplement Number 1 to the 2005 ASCE
Minimum Design Loads for Buildings and Other Structures (ASCE 7). Thus with the 2005
editions of these standards, designers will have a coordinated set of provisions that have gone
through public review and are suitable for use on building projects.
The buckling-restrained braced frame provisions follow a basic capacity-design
methodology. The required strength of braces is dependent on a minimum system strength
dictated by the building code (e.g., ASCE 7). The required strength of connections is
computed from the brace capacity, adjusted for various sources of overstrength. These
include compression overstrength, which reflects the fact that most brace designs allow for
minor transfer of compression forces between the steel core and the sleeve; strain-hardening,
These values indicate a general range of variation; brace manufacturers can give more
precise information concerning their products. Values of compression and strain-hardening
factors are dependent on the deformations to which the braces are subjected. Most brace
manufacturer’s can maintain a fabrication tolerance near the lower end of this range.
Like that of bracing connections, the required strength of beams and columns is computed
based on the capacity of braces, adjusted by the overstrength factors listed above. This is
equivalent to design provisions for eccentrically braced frames, in which link strength is the
basis of the required strength of the other elements.
Because the performance of buckling-restrained braces depends on proper design and
interaction of the separate steel core and buckling-restraining mechanism, the design
provisions for this system in AISC 341 require that the braces used be qualified by testing.
Qualification testing must include a test that imposes the expected end rotations on braces in
addition to the axial deformations (Sabelli, 2004). Qualification testing is also used to
establish the forces that braces are expected to impose on the rest of the system at large
displacements (i.e., the strain-hardening and compression overstrength factors listed above).
In current industry conditions, designers specify buckling-restrained braces that are designed
and furnished by specialty manufacturers (López and Sabelli, 2004). Each manufacturer has a
particular design and has performed at least some of the qualification testing required by
AISC 341. These tests may not be applicable to allbuilding de signs.
The design procedure for buckling-restrained braced frames is similar to that employed for
other systems. First the required strength of braces is calculated based on the design base
shear. Drifts must not exceed code-specified limits, and the deformations of braces
themselves must not exceed the range of testing (which may vary from manufacturer to
manufacturer); these criteria may lead to revision of brace sizes. Once brace sizes are
In order to illustrate this procedure, a five-story building located in San Francisco will be
designed using the AISC 341 and ASCE 7 standards. The building description is taken from
the nine-story building in the SAC project (the building information is presented in a report
by Sabelli [2001]). The building is regular and has two braced bays on each perimeter line of
framing. A typical plan of the building is shown in Figure 1.
5 x 30’ = 150’
5 x 9.1m = 45.7m
5 x 9.1m = 45.7m
5 x 30’ = 150’
For purposes of this example, it is assumed that buckling-restrained braces will be provided
by a hypothetical specialty manufacturer. This manufacturer provides braces with A36 core
steel material with a minimum yield stress of 38.0 ksi (262 MPa) and a maximum of 42.0 ksi
(290 MPa). This manufacturer also maintains a tolerance of +3%/-0% on the core area.
Further information from the brace manufacturer is necessary; the description of the design
process will include a hypothetical consultation with the brace manufacturer in order to more
fully illustrate design procedures under current conditions.
(with values of Ct and x from AISC 341 Appendix R). Thus the design base shear is 10.9% of
the seismic weight. As the seismic weight is estimated to be 13,300 kips (59,400 kN), the
resulting base shear is 1450 kips (6470 kN).
Next, the base-shear force is distributed vertically over the building height. For the
preliminary design, this is done using ASCE 7 Equation 12.8-11 and 12.8-12. For better
performance it is recommended that a dynamic analysis be conducted to refine the design,
preferably using a site-specific response spectrum. (Such a dynamic analysis cannot, of
course, be performed before preliminary sizes of members are selected.)
Table 2 shows the story forces corresponding to this preliminary analysis. In order to qualify
for a redundancy factor ( ) of 1.0, those stories resisting 35% or more of the base shear have
to meet either of two sets of criteria given in ASCE 7 12.3.4.2. This building, which is regular
and has two bays of bracing on each side, meets the latter; a value of 1.0 can be used.
Table 2. Preliminary analysis story forces.
Diaphragm Story Force Brace Story Shear % of Total
Level kip Level kip Base Shear
(kN) (kN)
504 504
Roof 5 34.7%
(2242) (2242)
360 864
5 4 59.4%
(1602) (3844)
278 1142
4 3 78.6%
(1237) (5081)
196 1338
3 2 92.0%
(871) (5953)
116 1454
2 1 100.0%
(515) (6467)
In order to calculate required brace forces, the story shears have to be distributed to the
frames. For this design, the diaphragms, which have concrete fill over metal deck, are treated
as rigid. Accounting for accidental eccentricity, it is determined that each of the two frames
in each line is designed for 26.3% of the story forces at each level, assuming equal stiffness
of all frames.
Brace forces and areas from this preliminary stage are presented in Table 3.
At this stage, the designer consults with the brace manufacturer, or reviews the brace
manufacturer’s literature. One reason for this is that the designer must verify that the brace
sizes in the design do not fall outside of the range of the manufacturer’s applicable tests.
AISC 341 requires two types of tests: a subassemblage test (in which brace end rotations are
included), and an axial test. Some subassemblage test set-ups permit the test to qualify as an
axial test as well (Merritt et al., 2003). For subassemblage tests, the specimen’s capacity must
be at least that of the largest prototype brace (AISC uses the term “prototype” to refer to the
actual construction); downward extrapolation is not limited. For the preliminary brace sizes,
this capacity is 299 kips (1328 KN). For the axial test, the specimen’s capacity is permitted to
vary up to 50% of that of the prototype. The range of applicable axial tests for each prototype
brace is shown in Table 3.
Table 3. Preliminary design of braces.
Brace Brace Angle From Required Brace Core Area Applicable Axial Test
Level Horizontal Strength Pu Asc Range
kip in.2 kip
deg. (kN) (mm2) (kN)
87.7 2.56 44 — 131
5 40.9
(390) (1653) (195) (585)
150 4.39 75 — 225
4 40.9
(668) (2835) (334) (1003)
199 5.81 99 — 298
3 40.9
(884) (3747) (442) (1325)
233 6.80 116 — 349
2 40.9
(1035) (4390) (518) (1553)
299 8.73 149 — 448
1 50.2
(1328) (5631) (664) (1992)
Tables 4 and 5 show the capacities of braces tested by the hypothetical manufacturer; the
former showing axial tests and the latter showing subassemblage tests. The tables also show
the qualification range for each brace based on the AISC 341 extrapolation limits. These
limits are presented in AISC 341 (and Table 3) as limiting applicable test capacities based on
prototype capacity; for the manufacturer’s purposes, prototype sizes are limited based on the
qualifying tests performed, and the inverse of the limiting ratio is used (i.e., 1/50% = 200% and
1
/150% = 67%). As the preliminary-design brace capacities are within the manufacturer’s
qualification range, the design does not need to be revised on that basis.
Tables 4 and 5 also show other information necessary for design: the deformation to which
each of these tests was performed, which must be larger than the expected brace deformation
(this will be addressed further in the design procedure); the cumulative ductility achieved in
the test (a minimum value of 200 times the yield deformation, bm, is required by AISC 341);
and values of strain-hardening and compression adjustment factors ( and ) corresponding
to the maximum deformation. It is generally more convenient to use these maximum values
for design, although lower values can be extracted from test results if the expected prototype
brace deformation is significantly below the test maximum deformation.
Another reason that the designer should consult the manufacturer at this point is to develop
estimates of brace stiffness. The designer has established preliminary cross-sectional areas of
the yielding segment of the steel core. The non-yielding zones and the connection zones are
considerably stiffer, and the effective brace stiffness depends on both the area and length of
these zones with respect to the yielding zone. These vary with brace manufacturer, brace
capacity, and overall brace length. The brace manufacturer can give guidance as to the
effective brace stiffness, which is typically in the range of 130%-200% of the stiffness
corresponding to a brace with the cross-sectional area of the yielding segment that extends
Yi No
e n
Zo ldin -
ne g
Downloaded from ascelibrary.org by PENN STATE UNIV on 06/11/14. Copyright ASCE. For personal use only; all rights reserved.
Yi Non
e
Zo ldin -
ne g
With the additional information from the brace manufacturer, the preliminary brace designs
can be included in a model in order to size the framing members and check drift. An elastic
model is constructed, using the brace cross-sectional areas and stiffness (including the
stiffness modification factors recommended by the brace manufacturer). Beam-column
connections of the braced frame in the model are rigid, reflecting the fully-fixed connections
that will be used in the design. (The presence of a strong frame is beneficial to the
performance of the system [Uang and Kiggins, 2003].)
The seismic forces are applied according to the vertical distribution in ASCE 7 Equation
12.8-11 and 12.8-12, or, preferably, using a distribution based on dynamic analysis. The
model is analyzed using three sets of load combinations: the first for checking the brace area,
the second for the framing members, and the third for drift.
The first set of load combinations consists of the basic seismic load combinations (ASCE 7
Equations 5 and 7 in Section 12.4.2.3), and is used to verify the adequacy of the braces, and
possibly reduce their cross-sectional area due to a fraction of the story shear being resisted in
the frame columns.
The second set of load combinations, consisting of the Special Seismic Load combinations
(ASCE 7 Equations 5 and 7 in Section 12.4.3.2), is used to verify the adequacy of the frame
columns and beams. In lieu of the system overstrength factor o, a frame-specific
overstrength factor can be constructed by multiplying the various sources of overstrength.
(This approach is appropriate if all braces are to be designed to precisely their required
strength, or to a uniform overstrength; where brace demand-to-capacity ratios vary, this
approach can be used with the largest value of o, but the results may overestimate required
strength considerably.)
The brace design strength ( FyAsc) has been set equal to the required strength (Pu).
Rearranging Equation (1) to solve for Pu and substituting into Equation (2) yields Equation
(3). Values for this design are shown in the equation.
Ry CTol 1.14 × 1.41×1.11× 1.03
o = = = 2.03 (3)
0.900
For this design, o is calculated to be 2.03. With this factor, the model is analyzed using the
Special Seismic Load combinations and the framing members are designed. The system
overstrength factor computed in this way is often lower than the factor (2.5) given in AISC
341 (Appendix R). However, if braces are sized to reduce drift, the resulting higher
overstrength factor must be used in the design of framing members; the overstrength factor
given in AISC 341 is not permitted to be used in the design of these frames.
For V-braced frames, the required strength of the beams may be affected by the vertical
unbalance forces resulting from brace compression overstrength (AISC 341, Section 16.4).
These forces can be computed separately and applied to the model. Table 6 shows the
vertical forces based on the preliminary brace designs; these forces are moderate.
Furthermore, for the configuration used, they are all in the upward direction (counteracting
gravity forces). The vertical forces can safely be neglected at this stage.
Table 6. Vertical forces on beams from preliminary brace designs.
Brace Core Adjusted Strength Vertical Force Diaphragm Net
Level Area Unbalance Location Level Vertical
Compression Tension Force Force
Asc RyFyAscCTol RyFyAscCTol
in.2 kip kip kip kip
(mm2) (kN) (kN) (kN) (kN)
2.56 178 156 14 14
5 Top Roof
(1653) (793) (695) (64) (64)
4.39 306 268 -25
4 Bottom 5 0
(2835) (1359) (1192) (109)
5.81 404 354 32 8
3 Top 4
(3747) (1796) (1576) (144) (35)
6.80 473 415 -38
2 Bottom 3 0
(4390) (2104) (1846) (169)
8.73 607 532 57 19
1 Top 2
(5631) (2700) (2368) (255) (86)
example), the braces may have larger forces due to gravity loading. Thus either the brace
areas must be modified to achieve uniform demand-to-capacity ratios equal to unity or the
overstrength factor o must be modified to account for the overstrength associated with
brace demand-to-capacity ratios lower than unity; the former approach is followed in the
example. This may affect not only the drift, but possibly the required member strengths as
well, because a change in brace stiffness will affect the percentage of shear resisted by the
frame. (It should be noted that if dynamic analysis is used with this iterative procedure, a
software program that can perform the dynamic analysis automatically is most convenient.)
Once all three sets of load combinations produce satisfactory results (the first showing that
the brace designs are adequate, the second that framing member sizes are adequate, and the
third that drift limits are not exceeded), the brace capacities in the model can be considered
final. The final frame design is presented in Figure 3 and Table 7. Brace sizes at the top level
have been reduced slightly from the preliminary design; those at the lower levels have been
increased due to gravity forces.
The brace capacities must once again be compared to the tests from the brace manufacturer,
as well as the expected deformations ( bm). While brace deformations can be determined by
an elastic analysis, it is more convenient to calculate them from the yield displacement. The
This design deformation (2.0 bm) should not be assumed to be less than that corresponding
to 2% drift, according to Appendix T of AISC 341. Drift-based deformation can be safely
estimated as shown in Equation 5.
bm = m cos( ) (5)
Where m is the expected story drift.
The AISC 341 provisions require that for V- and inverted-V braced frames, brace axial
deformations include the effects of beam flexibility. This additional displacement is the brace
axial component of the vertical beam deflection due to unbalance loads, such as those
presented for the preliminary brace designs in Table 8; the values in the table are based on
non-composite beam stiffness. The brace axial component of vertical beam deflection can be
calculated as shown in Equation 6.
b = v sin( ) (6)
Where b is the brace axial deformation resulting from the vertical beam deflection v.
11 0.14
4
(48) (3.4)
3 0 0
24 0.29
2
(105) (7.5)
Likewise, brace-end rotations at 2.0 bm must be calculated. These can safely be assumed to
be less than or equal to twice the story drift angle corresponding to bm. Alternatively,
relative end rotations can be extracted from the elastic model and amplified by 2.0Cd. Like
axial deformations, these must not be less than those corresponding to 2% drift.
Displacement
Brace Story Elastic Maxi- Rotation Rotation Maxi- Vertical Brace Total
Level Height Defor- mum Angle Angle mum Dis- Defor- Defor-
mation Defor- Defor- placement mation mation
mation mation
H be 2 bm 2 m/H 2 bm v b
in. in. in. in. in. in. in.
(mm) (mm) (mm) (mm) (mm) (mm) (mm)
156 0.21 2.06 2.36 0.23 0.15 2.51
5 1.74% 2.00%
(3962) (5.2) (52.3) (59.9) (5.8) (3.8) (63.7)
156 0.20 2.02 2.36 0.14 0.09 2.45
4 1.71% 2.00%
(3962) (5.1) (51.2) (59.9) (3.4) (2.2) (62.1)
156 0.20 2.04 2.36 0.14 0.09 2.45
3 1.73% 2.00%
(3962) (5.2) (51.7) (59.9) (3.4) (2.2) (62.1)
156 0.18 1.83 2.36 0.29 0.19 2.55
2 1.55% 2.00%
(3962) (4.6) (46.4) (59.9) (7.5) (4.9) (64.8)
216 0.24 2.41 2.77 0.29 0.23 2.99
1 1.75% 2.00%
(5486) (6.1) (61.3) (70.2) (7.5) (5.7) (76.0)
With the final brace designs, the design can proceed to the next phase. This includes the
design and detailing of connections. The design of the gusset plate and its connection to the
frame is outside the responsibility of the brace manufacturer. The brace manufacturer may
give certain limitations, such as a certain gusset thickness and a minimum size to
accommodate the connection detail. If the brace connection includes bolts, pins, or splice
plates, typically those are not designed by the brace manufacturer, although the manufacturer
may provide guidance or restrictions in these areas. Connections are designed for 110% of
the adjusted brace strength in compression (i.e., 1.1 RyFyAscCTol).
Conclusion
A buckling-restrained braced frame has been designed using the 2005 ASCE 7 and AISC 341
draft standards. An iterative procedure for optimizing the design was presented, and the
interaction between designer and brace manufacturer illustrated. The ability to use capacity
design for buckling-restrained braced frames without performing numerous calculations was
demonstrated.