You are on page 1of 9

Structures Congress 2018 21

Seismic Evaluation of Dual Steel Frames with Brace Buckling Consideration


H. Ataei, Ph.D., P.E., P.Eng., F.ASCE1; and K. Kalbasi Anaraki, S.M.ASCE2
1
Dept. of Civil and Materials Engineering, Univ. of Illinois at Chicago, 842 West Taylor St.,
ERF 3091, Chicago, IL 60607. E-mail: hataei@uic.edu
2
Dept. of Civil and Environmental Engineering, College of Engineering and Computer Sciences,
Syracuse Univ., 151 Link Hall, Syracuse, NY 13244. E-mail: kkalbasi@syr.edu
Downloaded from ascelibrary.org by University of Sunshine Coast on 05/02/18. Copyright ASCE. For personal use only; all rights reserved.

ABSTRACT
This paper investigates the seismic behavior of short dual systems–integrated moment and
braced frames-considering the braced frame failure. Modeling of beams, columns, braces, and
probable types of braces buckling are discussed in detail. Multi-story 3-bay moment frames were
designed according to AISC-ASD 89. To ensure the shear performance of dual frames, moment,
and shear displacement of frames are calculated and compared. To evaluate shear displacement
of dual frames, shear displacement of moment, and braced frames are calculated separately and
added together. These frames can be considered as a cantilever beam so moment displacement of
dual frames is dependent on column sections. Therefore, the moment-area theory was used to
determine the moment displacement. Buckling of the X-braces can be occurred in two modes
depending on the lateral constraint of middle connection that is supplied by moment stiffness of
tension braces and middle connections. When moment stiffness of tension brace is considerable
and middle connection of X-braces has adequate rigidity, the buckling of braces occurs in second
mode. Otherwise first mode overcomes the brace buckling. In this study, the results are obtained
by the consumption of adequate moment stiffness of tension braces and middle connection
rigidity. In order to consider the stiffness degradation that is caused by gravity load, P-∆ effect,
corotational theory is used and pushover curves are compared considering the ultimate and
residual strengths. Braced frames buckling includes elastic and inelastic depending on the
slenderness ratios. Euler and Belayesh relations were used to evaluate critical strength of braces
in elastic and plastic regions respectively. Nonlinear static analysis was conducted on the frames
considering braces failure using OPENSEES platform with displacement control pushover
analyses on all designed frames. In other words, according to consumption of rigid diaphragm of
all stories, one of the roof nodes was pushed up to 3% of the frame height. The load pattern of
pushover analysis was chosen adverse triangular. The results indicate that brace failures occur at
2% drift of corresponding stories. In this paper during the pushover analyses, 16, 64, 144, and
1024 pushover analyses were done on 2-, 3-, 4-, and 5-story dual frames subsequently. Pushover
curves corresponding with the same story braces failure in each frame are averaged. The results
of pushover analyses on short dual frames indicate that the best performance of frames,
considering the ultimate and residual strengths, are obtained with brace failures occurring in
higher stories.

INTRODUCTION
Steel moment frames are generally exposed to large lateral deformations during sever ground
motions. Therefore, special requirements should be provided to limit the damage and detriment
of nonstructural elements, which is mostly due to the P-Δ effect. This may lead to the design
engineer’s increased tendency in using braces as economic solutions for such.
Detriments of braced frames in earthquake events such as Mexico City (1985), Loma Prieta

© ASCE

Structures Congress 2018


Structures Congress 2018 22

(1989) and Northridge (1994) brought up many concerns regarding ultimate deformation
capacity of these types of structures [1].
Braced elements have limited ductility capacity in cyclic (hysteresis) loading. Hysteretic
behavior of braces are not equal in tension and compression: braces experience severe reduction
in stiffness in compression compare to when in tension. Due to the complicated behavior of
braces, real load distribution and internal deformations in these members may sometimes be
completely different from what considered for design purposes. The application and
Downloaded from ascelibrary.org by University of Sunshine Coast on 05/02/18. Copyright ASCE. For personal use only; all rights reserved.

simplifications in design procedures may require heavier sections for a story. These variations in
story capacities and buckling of a brace element before another will result in concentrated
damages in a specific story that ultimately leads to severe reduction of overall stiffness of the
whole structure. Moreover, buckling of a brace element will cause serious damage to
nonstructural connected elements.
Using braces within moment frames will improve many of the frame properties. In a dual
system, if buckling occurs in braces, the other lateral-resistance elements of the frame (i.e. beams
and columns) can help reduce the harm of losing braces. Furthermore, through wisely using dual
systems, the designer could take advantage of both the moment frame and braced frame in the
meantime [2].
Modeling Braces: Many analytical models have been introduced to simulate the behavior of
braces. In general, these models can be categorized in three groups:
1. Behavioral models (macro models) that are based on simple and most efficient
computational methods. In these analytical models, usually the behavior of one-
dimensional truss elements are used and calibrated based on prior experiments and
experiences. Hysteretic characteristics in these models are only based on empirical rules
regardless of physical theorem govern the situation. Some of these models are capable of
predicting the stiffness reduction. However, there are major deficiencies as per the
following: (1) A lack of experimental databases for different situations especially given
the fact that these models require many parameters. (2) The shape of hysteretic cycle is
generally modeled to simulate the reality but many details are not predictable where some
of them could adversely affect the results such as negative stiffness after yielding and/or
concavity of hysteretic curves at the beginning of unloading of compression. (3) Usually
modeled as brace element with hinges at the ends, so, the in-plane flexural stiffness is not
considered [3, 4].
2. Models that are based on physical theories where most of these methods apply an elastic
beam-column with non-elastic hinges at the ends. In this case, the geometrical
nonlinearity is directly considered. These models could overcome some of the
shortcomings of the macro models especially in terms of less-dependency on empirical
parameters. Generally, the inputs for these models are geometry of braces, fiber behavior
and consistent material characteristics. These models are capable of capturing the
interaction between bending and axial forces, although, more computational effort is
required.
3. Finite Element models where a more accurate analysis of larger displacements could
provide a more precise prediction on behavior of the brace elements with least depending
on empirical parameters. However, use of the FE models is not a common practice for
frame and brace buckling analyses due to the inherent complications in FE models in
addition to the already-existing heavy computational costs.

© ASCE

Structures Congress 2018


Structures Congress 2018 23

Uriz and Mahin (2008) suggested a model (that is used in this study) which is based on the
physical theorem. As stated previously, this model could provide an acceptable predict of vast
variety of bracing cross-sections with various boundary condition behaviors. As illustrated in
Figure 1, to model the buckling in this model, an initial imperfection is required in the middle of
the brace.
Downloaded from ascelibrary.org by University of Sunshine Coast on 05/02/18. Copyright ASCE. For personal use only; all rights reserved.

Figure 1. Illustrated model of imperfection


[Uriz and Mahin, (2008)].
The initial response of the element is obtained using small deformation theory. The
geometric nonlinearity due to the large deformations will then be considered with force and
deformation transformation from local to global coordinates. Inelastic beam-columns, using
small deformations, are employed with corotational principal to capture the large displacement
and deformations. To achieve this goal, the brace elements should be divided into at least two
sections. In order to incorporate the interaction of axial and flexural forces along the braces,
beam-column elements are assigned to braces instead of truss elements. In addition, shear
deformation is neglected. The steel material that is used in this study is based on Menegotto-
Pinto Hysteretic material model -steel02. The initial imperfection considered for braces is 0.005
of the length of braces and the fibers in the sections - as illustrated in Figure 2. This is based on
the assumption that the buckling will occur around the z direction.
A sensitive analysis is performed to make a decision on the number of fibers in cross-section
and the number of subdivided brace elements.

© ASCE

Structures Congress 2018


Structures Congress 2018 24
Downloaded from ascelibrary.org by University of Sunshine Coast on 05/02/18. Copyright ASCE. For personal use only; all rights reserved.

Figure 2. Fibers distribution in the brace sections.


INITIAL IMPERFECTION SENSITIVITY
Without implementing the initial imperfection in OpenSees, the brace elements will
demonstrate ideal behavior without any buckling. Therefore, to investigate the effect of the
initial imperfection on the results, the values 0.001, 0.005, 0.01, 0.02 and 0.03 are introduced and
used for performing the sensitivity analysis. The results are illustrated in Figure 3.

Figure 3. Initial imperfection effect.


Based on Figure 3, the initial imperfection magnitude is selected to be 0.05. In addition, as
seen also in Figure 3, regardless of the initial imperfection amount, after 5 cm displacement, the
post buckling force is almost the same in all investigated cases.
Number of the brace subdivision: To perform the sensitive study on the number of brace
element subdivisions, 2, 4, 10 and 30 elements were assigned as illustrated in Figure 4. It is seen
that 10 elements seem to be the optimal number for this parameter and assigning more elements
than that will not affect the results. To model the columns, the beam-column elements with 10
integration points are used.
The beams have to be capable of capturing the accurate response of structure during the
severe seismic event. They also require relatively simple calculations so to monitor the real
response of structure during the selected many ground motions.

© ASCE

Structures Congress 2018


Structures Congress 2018 25
Downloaded from ascelibrary.org by University of Sunshine Coast on 05/02/18. Copyright ASCE. For personal use only; all rights reserved.

Figure 4. Subdivision number of brace element effect.


This study employs the “One-Component Model” that is introduced by Luca Martinelli et al.
(2013). The literature suggest that the critical region in moment frames are near the connection
of the beams to columns. Therefore, the effects such as low-cyclic fatigue and local buckling will
prevent developing of plasticity along the beams. As illustrated in Figure 5, nonlinear effects –
plasticity - due to pure bending is not developing along the element but, concentrate at the ends
of the beam with zero length where plastic hinges are and formed [2].

Figure 5. Nonlinear beam element model.


As Figure 5 demonstrates, the beam is consisted of a central component whose length is
equal to the beam length and has an elastic behavior. Central component is connected to the
supports through rotational moment springs. Springs have nonlinear behaviors with formation of
plastic hinges that has the equal moment-curvature as the real beams. The moment-curvatures
can be verified with finite element models against experimental results. The flexural matrix [f] is
obtained through the superposition of the [f] matrix for the beam as well as the [f] for the springs
as illustrated in Equation 1 where the axial stiffness values are neglected.
f f  f 0
 f    fb    f s    f b11 f b12    0s1 f  (1)
 b 21 b 22   s2 

Then, the local stiffness matrix [k ] is obtained by inverting the flexural stiffness matrix and
by multiplying those with the appropriate transformation matrix - Jacobian Matrix [ j ] .
Hence, the stiffness matrix [ K ] - in global coordinates - is obtained as per Equation 2:
[ K ]  [T ]T [k ][T ] (2)

© ASCE

Structures Congress 2018


Structures Congress 2018 26

MODEL PROPERTIES AND ANALYSIS


Structure properties: The frame has been designed based on AISC-ASD89 provision
assuming that the height of each story and spans are 3.2 m and 5 m, respectively. The dead load
and live load amounts are 17650.8 kN / m and 5883.6 kN / m , correspondingly. Allowable
bending stress in beams is 0.6 of the yield stress. Furthermore, it is assumed that the buckling
will occur on the short length of the beam because of beams being buried in concrete. Beams to
columns connections are considered to be fixed.
Downloaded from ascelibrary.org by University of Sunshine Coast on 05/02/18. Copyright ASCE. For personal use only; all rights reserved.

Figure 6. Sample push over curve.

Figure 7. Comparison of push over curves for two-story structure.


Nonlinear static analysis: Considering the fact that the story diaphragms are solid, by
selecting a node on the roof level of each structure and applying the reversed triangular load
distribution to the structure, the structure is pushed to obtain the nonlinear behavior of the frame.
Figure 6 further illustrates the sample response of the structure.

© ASCE

Structures Congress 2018


Structures Congress 2018 27

For each structure, multiple scenarios are considered in terms of brace sections. In each
scenario, specific story brace is designed in a way to buckle first so to investigate the effect of
brace buckling sequencing and priority. The following figures (Figures 7, 8, 9 and 10)
demonstrate this brace-buckling priority effects for 2, 3, 4 and 5-story buildings, respectively.
while Figure 8 does the same for a three-story structure. Figures 7 to 10 illustrate that if braces
are designed in such a way that initial buckling in braces occur in higher story levels, it helps to
reduce and postpone the drop of resistance of the whole structure.
Downloaded from ascelibrary.org by University of Sunshine Coast on 05/02/18. Copyright ASCE. For personal use only; all rights reserved.

Figure 8. Comparison of push over curves for three-story structure.

Figure 9. Comparison of push over curves for four-story structure.


To further investigate this conclusion, hysteretic analyses have been performed on the
structures. The typical loading protocol and sample results of the analyses are illustrated in
Figure 11 (Left) and 11 (Right), respectively.

© ASCE

Structures Congress 2018


Structures Congress 2018 28
Downloaded from ascelibrary.org by University of Sunshine Coast on 05/02/18. Copyright ASCE. For personal use only; all rights reserved.

Figure 10. Comparison of push over curves for five-story structure.

Figure 11. (Left) Loading protocol on the structures; (Right) Sample hysteretic result
obtained.
As seen in Figure 11 (Right), the overall shape of the hysteretic plots is matching the
monotonic pushover results. So, as expected, the absorbed energy increases as the story level of
the buckled brace increased. The values of absorbed energy for each multi-story building in each
scenario are provided in Table 1.

Table 1. Absorbed energy in each scenario


1st 2nd 3rd 4th 5th
2 Story 970 kN 1220 kN - - -
3 Story 1300 kN 1210 kN 1550 kN - -
4 Story 1540kN 1790 kN 1870 kN 2120 kN -
5 Story 1580 kN 1630 kN 1990 kN 2230 kN 2560 kN

SUMMARY AND CONCLUSION


Based on nonlinear static analysis of multi-story building structures, implementing the
diagonal braces could increase the ultimate resistance of the whole structure. In addition, as this

© ASCE

Structures Congress 2018


Structures Congress 2018 29

study suggests, in dual steel frames with brace buckling action, losing structural performance of
the braces in higher story levels - in the structure - is less risky to the integrity of the structure
than that of in the lower levels. In other words, braces in lower levels carry a more important role
on performance and integrity of the entire structure and hence, as a result, the amount of their
absorbed energy will increase.

REFERENCES
Downloaded from ascelibrary.org by University of Sunshine Coast on 05/02/18. Copyright ASCE. For personal use only; all rights reserved.

[1] Mahin. S, “Lessons from damage to steel buildings during the Northridge earthquake”,
Earthquake Engineering, Vol. 20, pp. 261–270, 1998.
[2] Martinelli. L, Mulas. M and Perotti. F, “The seismic response of concentrically braced
moment- resisting steel frames”, 2013, Italy.
[3] Uriz. P, Mahin S, “Toward earthquake-resistant design of concentrically braced steel frame
structures”, Pacific Earthquake Engineering Research Center, Berekley, California, 2008, 72–
104.
[4] Bose. S, “Cyclic inelastic buckling of steel braces”, International Journal Of Civil And
Structural Engineering, Vol. 2, No. 4, 2012.
[5] American Institute of steel construction, Inc. (1989) Allowable Stress Design and Plastic
Design, Specification for Structural Steel Buildings, Chicago, IL

© ASCE

Structures Congress 2018

You might also like