Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Seismic Design and Performance Evaluation

of Long-Span Special Truss Moment Frames


Chatchai Jiansinlapadamrong 1; KyoungSub Park 2; John Hooper, M.ASCE 3;
and Shih-Ho Chao, M.ASCE 4

Abstract: The need for a very-long-span structure in sporting and industrial venues puts steel moment frames and braced frames at a
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

disadvantage. In this regard, special truss moment frames (STMFs) can accommodate a large span by utilizing truss girders to provide high
lateral stiffness. However, current seismic provisions for structural steel buildings do not allow STMFs’ span and truss depth to exceed 20 and
1.8 m, respectively. Moreover, when a very long span is used, the high axial forces induced by the gravity load could cause considerable
ductility reduction of the chord members in the special segment. This paper presents a study on seismic behavior of long-span STMFs with
double-channel truss members, a span length of 27.4 m, and a truss depth of 3.05 m. Plastic hinge models of double-channel sections
considering the effect of high axial forces were developed based on experimental and nonlinear finite-element analysis (FEA) for both design
basis earthquakes (DBEs) and near-collapse earthquakes or maximum considered earthquakes (MCEs). A design procedure for long-span
STMFs using nonlinear pushover analysis is presented. The seismic performance of long-span STMFs was verified using the developed
plastic hinge models as well as DBE and MCE ground motions through nonlinear time-history (NTH) analyses. According to the FEA results,
a modification to the axial load limit in current seismic provisions for the chord members is recommended. DOI: 10.1061/(ASCE)ST.1943-
541X.0002340. © 2019 American Society of Civil Engineers.
Author keywords: Special truss moment frame; Long span; AISC; Double channel; Near collapse.

Introduction of an STMF without diagonal web members within the SS is shown


in Fig. 1 (Chao and Goel 2008).
Long-span structures are commonly seen in sporting and industrial Key dimensional requirements for STMFs are as follows (AISC
venues due to the need for large openings, which cannot be pro- 2016):
vided by steel braced frames. Steel moment frames, although able 1. Span length between columns cannot exceed 20 m;
to provide open space, become very flexible and require large 2. Overall depth of the truss cannot exceed 1.8 m;
sections to control their drifts when a long span is used. Therefore, 3. Length of the SS shall be between 0.1 and 0.5 times the truss
moment frames are impractical when a span exceeds approximately span length; and
12 m (Hamburger and Malley 2016). A special truss moment frame 4. Length-to-depth ratio of any panel in the special segment shall
(STMF) is a seismic force-resisting system (SFRS) that can provide neither exceed 1.5 nor be less than 0.67.
both a large open space between columns and a long span by using The upper-bound length-to-depth ratio is to control the lateral
a steel truss girder rather than a beam. Furthermore, the open web stiffness of the truss, whereas the lower-bound length-to-depth ratio
of the trusses allows the utility and mechanical ducts to go through, is to limit the rotational demand of the chord members in the SS.
thereby providing greater architectural freedom. STMFs dissipate These requirements are based on considerations of the overall stiff-
earthquake energy through flexural yielding of the chord members ness of STMFs and the rotational demand of the chord members
and intermediate vertical web members within the middle section of within the SS. A shorter SS length increases the rotational demand
the truss girder, called special segment (SS), whereas members out- of the chord members, and a longer SS length decreases overall
side of the SS are designed to remain elastic. When diagonal web stiffness of the STMF. However, recent research shows that the
members are present within an SS, axial yielding and buckling of length-to-depth ratio of the SS could be as high as 2.48 without
these diagonals also help dissipate energy. The yielding mechanism affecting the integrity and performance of STMFs (Chao et al.
forthcoming; Simasathien et al. 2017).
1 The goal of this study was to determine if STMFs can serve
Project Engineer, AG&E Structural Engenuity, 15280 Addison Rd.,
as SFRS for long-span (more than 20 m) structures. In general,
Addison, TX 75001.
2
Postdoctoral Research Fellow, Dept. of Civil Engineering, Univ. of typical span-to-depth ratios for long-span trusses range from
Texas at Arlington, Arlington, TX 76019. approximately 8 (for vibration-controlled situations) to 10 (for
3 strength-controlled situations). The span length of three prototype
Director, Earthquake Engineering, Magnusson Klemencic Associates,
Seattle, WA 98101. STMFs used in this study is 27.4 m, which violates the current
4
Professor, Dept. of Civil Engineering, Univ. of Texas at Arlington, limits in AISC 341 (AISC 2016). A 27.4-m-long span is the typical
Arlington, TX 76019 (corresponding author). ORCID: https://orcid.org module size for convention centers, exhibit halls, and facilities that
/0000-0003-2679-7364. Email: shchao@uta.edu
would potentially utilize STMFs. The depth of the truss girders is
Note. This manuscript was submitted on February 25, 2018; approved
on November 16, 2018; published online on April 26, 2019. Discussion 3.05 m, which is deeper than AISC’s limit (1.8 m) in order to in-
period open until September 26, 2019; separate discussions must be sub- crease overall stiffness of the long-span STMFs. These dimensions
mitted for individual papers. This paper is part of the Journal of Structural give a span-to-depth ratio of the truss equal to 9.0. The length-to-
Engineering, © ASCE, ISSN 0733-9445. depth ratio of the SS in one of the STMFs is 3.0, which falls outside

© ASCE 04019053-1 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


Plastic hinges and Goel (2008) suggested that the intermediate vertical mem-
F2 bers be treated as secondary members; hence, most of the truss
strength and energy dissipation capacity is provided by the
Special segment chord members. In addition, plastic hinges must be avoided
in the chord members except at chord ends; therefore, the mo-
F1 ment capacity of the intermediate vertical members must be lim-
ited so that the bending moment in the chord members at
sections adjacent to the intermediate vertical members is less
than the chord members’ moment capacity when the intermedi-
ate vertical members reach their maximum capacity. Chao and
Fig. 1. Yielding mechanism of STMF. (Adapted from Chao and Goel (2008), therefore, recommended the plastic moment capa-
Goel 2008.) city of the intermediate vertical member, M pv , be calculated as
Mpc =m, where M pc is the plastic moment capacity of the chord
member, and m is number of intermediate vertical members in
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

the SS. For an SS with two intermediate vertical members,


the limitation specified by AISC 341. Moreover, in a typical STMF, Mpv =Mpc ¼ 0.5. A recent full-scale STMF experiment shows
axial forces in the SS chord members are negligible. In a long-span that, for an SS designed with this moment capacity ratio, after
STMF, however, the gravity load could induce high axial forces in the intermediate vertical members failed, the strength of the
the SS chord members, which could have a detrimental effect on STMF specimen drastically dropped, and its hysteretic behavior
their rotational capacity. The effect of axial force on rotational followed that of an STMF without intermediate vertical mem-
capacity of double-channel sections was investigated in this study. bers (Chao et al. forthcoming). As a result, the intermediate ver-
This paper presents the design procedure for long-span STMFs, and tical member cannot be too strong in order to avoid a drastic
their seismic performance evaluation was carried out by nonlinear drop in the strength of the STMF. In this study, member sizes
time-history (NTH) analyses. with Mpv =M pc ¼ 0.16 were selected.
3. STMF-3 has a single Vierendeel panel within a 3.05-m-long
special segment. This configuration provides very high stiff-
ness to the truss girder but could also cause high rotational de-
Prototype Structure and Special Truss Moment
mands to the chord members due to the short SS (Simasathien
Frames
et al. 2017).
Fig. 2 shows the floor plan of a 1-story building where two The dimensions and elevation of the three STMFs are shown
27.4-m-long STMFs are used for the SFRS along the east-west in Fig. 3. Note that the STMFs serve simultaneously as the gravity
(E-W) direction. Three STMFs with different layouts of SS are load–carrying system and lateral load–resisting system. Because of
investigated (Fig. 3): the long special segments of STMF-1 and STMF-2, lateral bracing
1. STMF-1 has one Vierendeel panel within a 9.14-m-long special within the SS is needed to prevent premature buckling of the SS
segment. This layout provides a wide uninterrupted special seg- due to substantial axial compressive forces induced by the gravity
ment for architectural usage. and lateral loads. Fig. 2(a) shows the floor plan for STMF-1 and
2. STMF-2 has three Vierendeel panels made of two intermediate STMF-2 where the majority of the gravity loads are carried by the
vertical members within a 9.14-m-long special segment. Inter- two STMFs along the E-W direction. A series of steel joist girders
mediate vertical members are typically the first ones to yield and along the north-south (N-S) direction are used to transfer the grav-
can be easily replaced after a moderate earthquake (Chao et al. ity loads to the two STMFs at various truss joints. Additional brac-
forthcoming). Full-scale STMF experimental testing shows that ing members are also provided within the SS to reduce the
adding intermediate vertical members can considerably increase unbraced length [also see Figs. 3(a and b)]. Fig. 2(b) shows the
the lateral stiffness of an STMF (Chao et al. forthcoming). Chao floor plan for STMF-3, which has a similar layout of steel joists

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Bracing
STMF STMF
A A

Bracing Beam Bracing Bracing


3@9.14 m = 27.4 m

3@9.14 m = 27.4 m

Beam
Braced Frame

Braced Frame
Braced Frame

Braced Frame

Beam B B
Steel Joist Girder Steel Joist Girder
Beam

Beam

Beam C C

Beam N
Bracing Beam Bracing Bracing
D E W D
STMF Bracing STMF
S
9@3.05 m = 27.4 m 9@3.05 m = 27.4 m
Tributary Area for Gravity Load, P Tributary Area for Gravity Load, P
(a) (b)

Fig. 2. Floor plan of (a) STMF-1 and STMF-2; and (b) STMF-3.

© ASCE 04019053-2 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


Detail-1 Detail-2 Detail-3
3.05 3.05 3.05 9.14 3.05 3.05 3.05 3.05 3.05 3.05 9.14 3.05 3.05 3.05

3.05
3.05

9.14
9.14
3.05 3.05 3.05 3.05 3.05 3.05

6.1
6.1
designates the locations of lateral bracing within SS designates the locations of lateral bracing within SS
and at the ends of SS and at the ends of SS

27.4 27.4
(a) (b)

3.05 3.05 3.05 3.05 3.05 3.05 3.05 3.05 3.05


Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

3.05
9.14
6.1
designates the locations of lateral bracing at the ends
of SS

27.4
(c)

Fig. 3. Dimensions and elevation (in meters): (a) STMF-1; (b) STMF-2; and (c) STMF-3.

girders. The joist girders near the ends of the SS also serve as brac- members. This weld-free gusset plate can prevent the members
ing members for the top chord member in the SS [Fig. 3(c)]. from lateral-torsional buckling, which, in turn, enhances the duc-
Lateral bracing of the bottom chord members can be done by using tility of the double-channel member (experimental results are dis-
steel members extending from the beams [Fig. 2(a)] or joist girders cussed subsequently). In this region, a horizontal stitch is used to
[Fig. 2(b)]. Figs. 3(b) and 4 show typical details for truss-to- prevent the channels from separating. In addition, the vertical
column connection, end joint of the SS, and intermediate vertical members at the end of the SS cannot be welded to the chord mem-
member–to–chord member connection within the SS. These de- bers because it will prevent plastic flow and cause the chord
tails have been verified by recent full-scale STMF testing and members to fracture early (Jiansinlapadamrong 2017; Chao et al.
component testing (Jiansinlapadamrong et al. 2018; Chao et al. forthcoming). In Detail-3, the intermediate vertical members can
forthcoming). Detail-1 shows the truss-to-column connection, be welded to the chord members because plastic hinges will not
which can be done by welding truss members to a gusset plate form in the chord members at that location. The cutout is used to
and the gusset plate to a column flange. Detail-2 and Detail-3 show increase the weld area to provide sufficient strength.
the end joint of the SS and intermediate vertical member–to–chord Jiansinlapadamrong (2017) tested member 2C150 × 19.3, with
connection within the SS, which demonstrate some special fea- the proposed details representing an intermediate vertical member
tures that enhance the rotational capacity of the double-channel in a full-scale STMF. The member exhibited steady response and a
member. One of the features is a center gusset plate with a large rotational capacity [Fig. 5(d)]. Note that asymmetrical loading
weld-free portion extended into the plastic hinge regions of the of the last two cycles was caused by load-capacity limitations in
chord members in the SS and the intermediate vertical web different directions of the hydraulic actuator.

Weld-free Horizontal stitch


gusset plate Web cutout

Not butted up
Butted up Continuity
against column
against chord plate
Weld-free
not butted up gusset plate
against chord
Horizontal stitch
Not butted up
against column

Detail-1 Detail-2 Detail-3


Fig. 4. Typical details of the connections at the end of SS and intermediate vertical members [Fig. 3(b)]. (Reprinted from Jiansinlapadamrong
2017.)

© ASCE 04019053-3 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

(a)

(b)

400 200
Experiment
300 FEA
100
Moment (kN-m)

200
100
0 0
-100
-200 -100
-300
-400 -200
-0.12 -0.08 -0.04 0.00 0.04 0.08 0.12
Member Rotation (rad)

(c) (d)

Fig. 5. (a) Symmetric loading protocol derived from AISC loading protocol and hysteretic behavior, top view; bottom views of (b) 2C310 × 30.8 with
no special connection details; (c) 2C310 × 30.8 with special connection details; and (d) 2C150 × 19.3 with special connection details.

Effect of Axial Force in Double-Channel Flexural 2C310 × 30.8 [Figs. 5(b and c)] and the intermediate vertical
Members member 2C150 × 19.3 [Fig. 5(d)] as part of a prototype full-scale
STMF with a typical span length of 9,698 mm. Conventional con-
To obtain the modeling information, a series of double-channel flexu- nection details (i.e., no extended gusset plate with a weld-free zone)
ral members with no axial load were first tested (Jiansinlapadamrong are shown in Fig. 5(b), whereas Fig. 5(c) gives special connection
et al. 2018). These flexural members represented chord member details featuring an extended gusset plate and horizontal stitches.

© ASCE 04019053-4 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


The loading protocol for chord or vertical member component tests reduced integration and hourglass control techniques. The gusset
in this study was derived from AISC loading protocol as specified plate and stitches were modeled using an eight-node solid element.
by story drifts ratio (SDR) for testing moment frame beam-to- Welds connecting the gusset plate to the specimen and horizontal
column connections [AISC 341 (AISC 2016)]. The SDR history stitches to the flange of the specimen were done by combining the
was first applied to the prototype full-scale STMF. The tests were hexahedral and tetrahedral solid elements. Tie constraints were
done according to the corresponding member rotations at the used at the contact surfaces to model weld bonding. The total num-
specified SDR of the prototype STMF. The loading protocol for ber of elements used was approximately 170,000, which was suf-
the chord member is shown in Fig. 5(a). Note that due to limitation ficient to produce accurate numerical results without the issue of
on the maximum stroke of the actuator, which could only reach a mesh sensitivity.
displacement corresponding to 2.75% SDR, it was decided to Material nonlinearities including the classical metal plasticity
terminate the tests at this SDR, but two additional cycles were theory based on the Von Mises yield criterion, associated flow rule,
added at 2.38% SDR. The new detailing of the double-channel and isotropic/kinematic hardening assumption were considered.
flexural section proved capable of effectively preventing lateral- Based on uniaxial coupon tests for both the flange and web from
torsional buckling. The new detailing also enhanced the rotational the C310 × 30.8 sections, their strain–stress curves were modeled
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

capacity of double-channel members as evident in Figs. 5(b–d) as shown in Figs. 7(a and b). To account for the Bauschinger effect
(Jiansinlapadamrong et al. 2018). and cyclic hardening under cyclic loading, the nonlinear kinematic
A series of nonlinear finite-element analyses (FEAs) were car- and isotropic hardening components C, γ, Q∞ , and b were as-
ried out to investigate the effect of the axial force on the rotational signed, in addition to the modulus of elasticity (E) and the yield
ductility of built-up double-channel sections. The finite-element stress (Fy ), where C is the initial kinematic hardening modulus,
(FE) model was first calibrated using the experimental results from γ is the rate at which C decreases with cumulative plastic deforma-
the 2C310 × 30.8 specimen without axial force [Fig. 5(c)]. FE tion (εpl ), Q∞ is the maximum change in the size of the yield sur-
analyses were further carried out on the double-channel sections face, and b is the rate at which the size of the yield surface changes
with various axial load demand-to-capacity ratios, α, from 0.1 to as plastic deformation develops. These material parameters must
0.3 for both compression and tension. The axial load demand in this be calibrated from cyclic test data, but because no such data for
study is defined as the axial load induced by gravity loading on the material were available, these cyclic plastic behavior parameters
STMF, Pu . The axial capacity is defined as Ry Fy Ag , where Ry is the were assumed through the parametric studies based on Crisan
ratio of the expected yield stress to the specified minimum yield (2016) as follows: C ¼ 2,500 MPa; γ ¼ 20; Q∞ ¼ 44 MPa; and
stress. Fy is the specified minimum yield stress, and Ag is the gross b ¼ 11. The strain–stress curve of weld was based on Chao et al.
area. Therefore, α ¼ Pu =Ry Fy Ag . (2006), and that of the gusset plate and stitches were based on
ASTM A572M, Gr. 345 (ASTM 2018), as shown in Figs. 7(c
and d), respectively. In this study, geometric nonlinearities were
Finite-Element Model, Analyses, and Results
also considered through a large-strain, large-deformation formu-
The finite-element model used in this study is shown in Fig. 6 using lation without assigning initial imperfections. Fig. 5(c) shows
the ABAQUS version 6.14 program. The beam area of the model the cyclic response of the calibrated FE model and experimental
was discretized using a hexahedral solid element that includes results of a specimen without axial force, as well as the trilinear

Fig. 6. FE model and boundary conditions.

497 506 552 448


Stress (Mpa)

Stress (Mpa)

Stress (Mpa)

Stress (Mpa)

378 386 480 345

E=210 GPa E=210 GPa E=200 GPa E=200 GPa

0.0018 0.1 0.00184 0.1 0.0024 0.1 0.00173 0.1


(a) Strain ( ) (b) Strain ( ) (c) Strain ( ) (d) Strain ( )

Fig. 7. Material stress–strain model of (a) beam flange; (b) beam web; (c) weld; and (d) gusset plate and stitch.

© ASCE 04019053-5 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Deformed shapes at 0.02-rad member rotation: (a) Model 1: 2C310 × 30.8 without axial load; and (b) Model 2∶ 2C310 × 30.8 with α ¼ 0.3.

(a) (b)

(c)

Fig. 9. FEA results and trilinear backbone model of 2C310 × 30.8 with axial compression per capacity ratio: (a) α ¼ 0.1; (b) α ¼ 0.2; and
(c) α ¼ 0.3.

backbone model. Figs. 8(a and b) show the deformed shape of has a higher impact on the ductility of double-channel sections than
2C310 × 30.8 flexural members without axial load (Model 1) axial tension does. Moreover, in both cases, the ultimate moment
and with an axial compression demand-to-capacity ratio of 0.3 capacities of the sections also decreased when the axial load
(Model 2), respectively. Both analyses were carried out by follow- increased.
ing the AISC loading protocol, and both images show the defor- It is worth noting that AISC 341, Section E.4.5b, requires that
mation shapes at 0.02 rad. As shown, high axial compression the axial strength of the chord members in a special segment be
force caused severe web and flange buckling, which, in turn, re- determined in accordance with ϕPn ¼ ϕFy Ag (ϕ ¼ 0.9), which
duced ductility of the built-up double-channel section. should be equal to or greater than 2.2 times the required axial
The analysis results of three axial compression cases are strength, Pu . In other words, Pu in the chord members shall not
shown in Figs. 9(a–c) along with the trilinear backbone models. exceed 0.45 times ϕFy Ag . That is
Figs. 10(a–c) show the effect of axial tension on the hysteresis be-
havior of the 2C310 × 30.8 member. As shown, axial compression Pu ≤ 0.45ϕPn or ϕPn ≥ 2.2Pu ð1aÞ

© ASCE 04019053-6 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


(a) (b)
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

(c)

Fig. 10. FEA results and trilinear backbone model of 2C310 × 30.8 with axial tension per capacity ratio: (a) α ¼ 0.1; (b) α ¼ 0.2; and (c) α ¼ 0.3.

Pu analyses were then carried out with coefficient of determination


or ¼ α ≤ 0.37 ð1bÞ
Ry Fy Ag (R2 ) ranges between 0.96 and 1, and the relationships between
a versus α and b versus α are as follows:
where Ry ¼ 1.1 for Gr. 345 steel. Axial compression case
a ¼ −0.21α þ 0.03 when 0 < α < 0.1 ð2aÞ
Plastic Hinge Models of Truss Members
¼ 0.009 when 0.1 < α < 0.2 ð2bÞ
Moment versus plastic rotation (PR) trilinear relationship in most
commercial programs only requires PRs and corresponding mo- ¼ −0.08α þ 0.025 when 0.2 < α < 0.3 ð2cÞ
ments at four points if strength degradation is considered, as shown
in Fig. 11. These moment parameters are specified as yielding mo-
b ¼ 1.1α2 − 0.42α þ 0.04 when 0 < α < 0.2 ð3aÞ
ment, M y (¼Ry Fy Z), ultimate moment, M u , and residual moment,
M r . Based on extensive experiments on various double-channel
¼ 0 when 0.2 < α < 0.3 ð3bÞ
sections, a generalized moment–rotation relationship for double-
channel sections without axial force was given in Table 1 (Chao
et al. forthcoming). To generalize the PR parameters, Du , Dl , Axial tension case
and Dr for various axial demand-to-capacity ratios, parameter a a ¼ 0.03 when 0 < α < 0.2 ð4aÞ
and b, as shown in Fig. 11, were determined from the trilinear mod-
els of hysteresis behavior of 2C310 × 30.8 in Figs. 5(c), 9(a–c), and ¼ −0.12α þ 0.05 when 0.2 < α < 0.3 ð4bÞ
10(a–c) for various axial demand-to-capacity ratios, α. Regression
b ¼ −0.08α þ 0.04 when 0 < α < 0.3 ð5Þ

Table 1. Plastic hinge modeling parameters of double-channel sections


u
without considering the effect of axial force
y
Modeling parameter Recommended value
r My Ry Fy Z
Mu 1.4M y
u l r x Mr 0.1Mu
Du (rad) 0.03
Fig. 11. Generalized moment–plastic rotation model for plastic hinge Dl (rad) 0.07
with and without strength degradation. Dr (rad) 0.09

© ASCE 04019053-7 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


Besides the rotational capacity, FEA also indicated that the Proposed Design Procedure and Design Example
extent of the axial load-to-capacity ratio, α, affects the overstrength
factor, β, (defined as M u =M y ). β ¼ 1.4 when α ¼ 0, which was Design Procedure
the average value based on a series of testing on double-channel
sections (Jiansinlapadamrong et al. 2018). Moreover, this ratio Three prototype STMFs were designed based on ASCE’s Minimum
decreases when axial compression (or tension) increases. The over- Design Loads for Buildings and Other Structures (ASCE 7) (ASCE
strength factor, β, when the axial load-to-capacity ratio α ¼ 0.1, 2010) for a site in San Francisco, California. A generic Site class C
0.2, and 0.3 was obtained from FEA. Thus, the relationship be- was used. Detailed design parameters are summarized in Table 3.
tween β (overstrength factor) and α (axial load-to-capacity ratio) 1. Equivalent lateral force (ELF) procedure. Design of the long-
was derived from a regression analysis, as follows: span STMF starts from determining initial member sizes. All
Axial compression case STMF members are designed as beam-column elements, based
on elastic analysis under lateral earthquake force specified by
β ¼ −0.88α þ 1.4 when 0 < α < 0.3 ð6Þ building code and gravity loading according to standard load
combinations. The load combination for gravity loading (shown
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

subsequently) includes the vertical seismic load effect.


Axial tension case 2. Design of members outside of SS. The design of members
outside of the SS is then modified based on the capacity design
β ¼ −0.73α þ 1.4 when 0 < α < 0.3 ð7Þ approach. After the structure was designed using the elastic
analysis result in Step 1, a nonlinear pushover analysis was
performed until the chord members within the SS and intermedi-
Note the developed moment–rotation relationships (Fig. 9)
ate vertical members (if used) fully yielded and were strain-
based on isolated cantilever FE models can be overly conservative
hardened. To simplify the design procedure, axial load effect
because in an actual STMF the truss members, columns, and other
was not considered in the plastic hinge model of yielding truss
elements can provide restraints to prevent large local buckling as
members. In other words, the plastic hinges of the chord mem-
shown in the FE models (Jiansinlapadamrong et al. 2018).
bers in the SS and intermediate vertical members are modeled
This conservative model is justified because the study STMFs
using values in Table 1 or 2 (α ¼ 0). This gives a conserva-
are the only gravity load–carrying system in the prototype building;
tive design for nonyielding members outside of the SS because
therefore, any strength degradation due to lateral loads could cause
the strength of the chord members is greatest when α ¼ 0. Non-
collapse.
yielding truss members could be modeled with elastic-perfectly-
plastic material properties. Plastic hinges of the columns are
modeled using the values in Table 2. If the pushover analysis
Plastic Hinge Models of Columns shows that any yielding occurs outside of the SS, except at the
base of the columns, such nonyielding member needs to be re-
Seismic Evaluation and Retrofit of Existing Buildings, ASCE/SEI
designed to ensure that it remains elastic. Although AISC 341
41 (ASCE 2017), recommends a conservative plastic rotational
(Commentary Section C2) suggests that nonyielding members
capacity model for flexural members subjected to axial compres-
sion. However, recent column experiments indicate that steel col-
umns with certain compactness ratios could sustain much larger PR Table 3. Design parameters for 1-story STMF convention hall at a site in
than that specified in ASCE/SEI 41 (Newell and Uang 2008; Zargar San Francisco, California
et al. 2014). Newell and Uang (2008) tested columns with bf =2tf
Values according Reference
and h=tw ratios less than 7.2 and 17.7, respectively. Their tests Parameter to ASCE 7 in ASCE 7
showed that steel wide-flange columns with stocky webs subject
to an axial load, ranging from 35% to 75% of nominal axial yield Risk category III Table 1.5-1
Importance factor 1.25 Table 1.5-2
strength, could achieve an interstory drift capacity of 0.07–0.09 rad
Site class C N/A
before their strength degraded below 90% of maximum bending Ss 1.5g Fig. 22-1 and
strength. On the other hand, Ozkula et al (2017) investigated the Section 12.8.1.3
effect of axial compression using a series of deep wide-flange beam S1 0.78g Fig. 22-2
columns (W610). These sections have bf =2tf and h=tw ratios of Fa 1.0 Table 11.4-1
4.81–8.5 and 28.7–54.6, respectively. The results showed that most Fv 1.3 Table 11.4-2
strong axis-bending specimens could not sustain more than 0.03- SMS ¼ Fa Ss 1.5g Eq. 11.4-1
rad PR. Conservatively, wide-flange columns used in this study had SM1 ¼ Fv S1 1.014g Eq. 11.4-2
flanges and webs with bf =2tf and h=tw ratios equal to or smaller SDS ¼ ð2=3ÞSMS 1.0g Eq. 11.4-3
than those of the lightest columns tested by Newell and Uang SD1 ¼ ð2=3ÞSM1 0.676g Eq. 11.4-4
Seismic design category E Section 11.6
(2008), thereby ensuring large rotational capacity. The plastic hinge
Building height 9.14 m N/A
model of these columns is shown in Table 2. TL 12 s Fig. 22-12
Cu 1.4 Table 12.8-1
Ta 0.256 s T a ¼ 0.0488h0.75
n
Table 2. Plastic hinge modeling parameters for yielding members consid- (Eq. 12.8-7)
ering the effect of axial force T ¼ Cu × T a 0.359 sa Sec. 12.8.2
Response modification factor, R 7 Table 12.2-1
Du Dl Dr
Total building weight, W 2,235 kN N/A
Members My Mu Mr (rad) (rad) (rad)
Cs ¼ V=W (DBE) 0.179 Eqs. 12.8-2, 12.8-3,
Truss members Ry Fy Z βM y 0.1M u a Du þ b Dl þ 0.02 12.8-5, 12.8-6
Columns Ry Fy Z 1.1M y 0.45Mu 0.06 0.061 0.08 a
Fundamental periods for STMF-1, STMF-2, and STMF-3 are 0.428,
Note: Parameters β, a, and b are determined from Eqs. (2)–(13). 0.355, and 0.179 s, respectively, based on modal analysis.

© ASCE 04019053-8 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


are forced to behave elastically to determine their required included in each member. Because gravity loads are entirely carried
strength, our analysis shows that this can lead to an unrealisti- by STMFs, no additional P-Δ columns are used. A 2% Rayleigh
cally large column. It is nearly impossible to completely prevent damping (combination of mass and stiffness proportional damping)
yielding in the columns; therefore, the column base is suggested was used for all modes in the NTH analyses.
to be allowed to yield up to a PR of approximately 0.005 rad.
Alternatively, the members outside of the SS can be designed
using a “column tree” approach (Commentary Section C2 in Design Example (STMF-1)
AISC 341) according to the expected vertical shear strength of
the special segment, V ne , given in AISC’s seismic provisions Lateral Loads
(AISC 341) or a proposed equation by Chao et al. (forthcoming). The prototype structure was assumed to be a 1-story, one-bay con-
3. Check axial compressive demand-to-capacity ratio. Due to the vention hall site in San Francisco, California. The design param-
detrimental effect of the axial compressive load on the rota- eters were obtained according to ASCE 7 (ASCE 2010) and are
tional capacity of double-channel sections in the SS shown in shown in Table 3. Based on a dead load of 2.39 kPa, a live load
Figs. 9(a–c), the ratio of axial compression demand to axial of 0.57 kPa, and a tributary area of 27.4 × 27.4 m, the resulting
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

capacity, α, of the chord member within the SS is advised to seismic weight is 2,235 kN (all live loads are conservatively in-
be 0.15 or lower. As shown in Fig. 9, the rotation before strength cluded). Since there are two STMFs sharing the design base shear,
degradation is approximately 0.025 rad for α ¼ 0.15. Based on each STMF carries 200 kN of lateral force. As shown in Fig. 12(a),
extensive NTH analyses with various lower α values (not shown the design base shear was applied discretely to the top joints of the
in this paper), α ¼ 0.15 is sufficient to provide a conservative truss according to tributary ratio of masses.
and satisfactory performance for the prototype building inves-
tigated in this study. If α is greater than 0.15, the size of the Gravity Loads
chord members within the SS should be increased and repeated The gravity load at each truss joint was determined according to
in Step 2. its respective tributary area and load path. An example of the
tributary area used for load P1 is shown in Fig. 2. The gravity
loads for STMF-1 are: dead load (DL) at column (PCD), Joints
Computer Modeling
1 and 2 (P1D and P2D), and Joint 3 (P3D) are 50, 100, and
PERFORM-3D (CSI 2006) was used for the nonlinear pushover 200 kN, respectively. Live load (LL) at column, (PCL), Joints 1 and
and NTH analyses. To simplify the analyses, a two-dimensional 2 (P1L and P2L), and Joint 3 (P3L) are 12, 24, and 48 kN,
STMF was considered to carry half of the gravity and lateral loads. respectively.
Half of the seismic masses were attributed to each STMF. Loads The critical gravity load combination of ð1.2 þ 0.2SDS ÞDL þ
and seismic masses at each joint were determined according to the 0.5LL was used according to ASCE 7 (ASCE 2010). The design
respective tributary area. Columns were modeled as a beam-column loads [Fig. 12(a)] used in the design of STMF-1 at column (PC ),
element with a lumped plastic hinge at each end. Strain hardening Joints 1 and 2 (P1 and P2 ), and Joint 3 (P3 ), are 76, 152, and
and strength loss, as well as axial force and bending moment in- 304 kN, respectively.
teraction, were included in the plastic hinge model. Due to sizeable
gusset plates and the weld connection used at each truss joints, each Initial Design of STMF-1
end of the truss members is rigid. As a result, truss members are Elastic analysis according to the ELF procedure for all study
also modeled as a beam-column element with a lumped plastic STMFs were performed using the software program RISA-3D
hinge at each end. End portions of the truss members, which are version 15. The initial design resulted in the members shown in
welded to the gusset plate, were modeled as elastic rigid elements Table 4. The small code design lateral force resulted in a light
to represent the high rigidity of the gusset plates. P-δ modeling is column and truss members.

Diagonal 3 Diagonal 3 Diagonal 3 Diagonal 3


Intermediate
Intermediate

Diagonal 2 Diagonal 2 Diagonal 2 Diagonal 2


Vertical SS
Vertical SS

Vertical SS
Vertical SS

Vertical 1

Vertical 2
Vertical 1

Vertical 2

Vertical 2

Vertical 1
Vertical 2

Vertical 1

Top SS Top SS
l1
l1

D
D

na
na

ia
ia

go
g

go
go

on

Bottom SS
na
ia
ia

al

D
D

l1
1

Chord 1 Chord 2 Chord 3 Chord 3 Chord 2 Chord 1 Chord 1 Chord 2 Chord 3 Bottom SS Chord 3 Chord 2 Chord 1

STMF-2
STMF-1
Diagonal 4 Diagonal 4
STMF-1
P3 P3 Diagonal 3 Vertical SS Diagonal 3
P1 P2 P2 P1 Diagonal 2 Diagonal 2
Pc Pc
Vertical 3

Vertical 3

Top SS
Vertical 1

Vertical 2

Vertical 2

Vertical 1
l1

11.1 22.2 22.2 44.4 44.4 22.2 22.2 11.1


D
na

ia
go
go

na
ia
D

l1
Bo

Chord 1 Chord 2 Chord 3 Chord 3 Chord 2 Chord 1


Ch

Ch
tto
o rd

ord
m
4

4
SS

STMF-1 STMF-3

(a) (b)

Fig. 12. (a) Member designation and loadings of STMF-1 (kN); and (b) member designation of STMF-2 and STMF-3.

© ASCE 04019053-9 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


Table 4. Initially designed (equivalent lateral force procedure) members of at 2.1% SDR, and the SDR when yielding occurred. Fig. 13 shows
STMF-1, STMF-2, and STMF-3 that when the first member (intermediate vertical members) of
Member STMF-1 STMF-2 STMF-3 STMF-2 yields at 0.36% SDR, the base shear is 1,215 kN, which
is much larger than the design base shear. It should be noted that
Column W310 × 143 W310 × 143 W310 × 143
Panel zone W310 × 143 W310 × 143 W310 × 143
in STMFs, intermediate vertical members yield before the chord
SS 2C310 × 45 2C250 × 45 2C250 × 37 member because of their higher rotational demands (Chao et al.
Vertical SS 2C310 × 45 2C250 × 45 2C250 × 37 forthcoming).
Chord 1-4a 2C310 × 45 2C250 × 45 2C250 × 37 Similar to STMF-1 and STMF-2, the chord member of STMF-3
Diagonal 1-4a 2C230 × 30 2C230 × 30 2C230 × 30 increased in size to keep α under 0.15. Table 5 summarizes the final
Vertical 1-3a 2C150 × 19.3 2C150 × 19.3 2C150 × 19.3 design members. The chord members in the SS were fully strain
Intermediate — 2C130 × 13 — hardened when SDR reached 1.1% (Table 8). At this SDR, none
a
STMF-1 and STMF-2 only have Chord 1-3, Diagonal 1-3, and of the members outside of the SS yielded except column bases.
Vertical 1-2. The maximum PR of the columns was 0.005 rad. First, yielding
occurred in the top and bottom chords in the SS at 0.52% SDR,
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

and the corresponding base shear was 6,637 kN (Fig. 13). The three
STMFs’ yield drift ratios (YDRs) were between that of concentri-
Design of Members Outside of the SS of STMF-1
cally braced frames (YDR approximately 0.3%) and moment
The final designed members of STMF-1 are shown in Table 5. The
frames (YDR approximately 1%). The elastic stiffness of STMF-1,
chord members were increased in size to reduce α to be less than
STMF-2, and STMF-3 was 25,697, 37,107, and 140,741 kN=m, re-
or approximately equal to 0.15. A nonlinear pushover analysis was
spectively. The results show that the SS layouts used in STMF-3
performed up to 2.1% SDR using the program PERFORM-3D (CSI
provide much higher lateral stiffness than that of STMF-1 and
2006) because at this SDR, all four plastic hinges at the end of the
STMF-2.
chord members in the SS reached 1.4My (or Mu ). The sizes of non-
yielding members outside of the SS were selected so that yielding
did not occur except at the columns. The maximum PR of the col- Seismic Performance Evaluation of the Studied
umn bases at 2.1% SDR was 0.005 rad, as shown in Table 6. Fig. 13 STMFs
shows the pushover curve, and Table 6 shows the yielding sequence
of STMF-1. The first yielding of STMF-1 occurred at the top and
bottom chord in the SS at 0.89% SDR. At this point, the base shear Ground Motions
was 2,099 kN, which was 10.5 times the code design base shear NTH analyses were conducted using two suites of 20 SAC ground
(200 kN). motions (Somerville et al. 1997) oriented in the direction of
STMFs. The first suite has a return period of 475 years (10% prob-
Design of STMF-2 and STMF-3 ability of being exceeded in 50 years) and is equivalent to design
STMF-2 and STMF-3 were designed in the same manner as basis earthquakes (DBEs). The second suite has a return period of
STMF-1. The elastic designed members of STMF-2 as designated 2,475 years (2% probability of being exceeded in 50 years) and is
in Fig. 12(b) are shown in Table 4. The details of loading used in equivalent to maximum considered earthquakes (MCEs). Typically,
the design of STMF-2 and STMF-3 can be found elsewhere for NTH analysis, the average response spectrum of a series of
(Jiansinlapadamrong 2017). The designated yielding members ground motions is scaled over a certain period range to envelope
(DYMs) in STMF-2 are the SS chord members and intermediate the code-specified DBE or MCE spectrum (Maison and Speicher
vertical members. Similar to STMF-1, the final size of the chord 2016). The average accelerations of the 20 DBE-level and 20 MCE-
member was increased to have an α of approximately 0.15. The level ground motions at the design period (0.359 s, Table 3) were
final designed members are shown in Table 5. STMF-2 was 1.32 and 1.97g, respectively. These values are much higher than the
pushed to 2.1% SDR when the chord members were fully strain code-specified DBE and MCE spectral accelerations of 1.0 and
hardened. At this story drift, all members outside of the SS were 1.5g (Table 3), respectively. In this study, however, the ground mo-
elastic except the columns, which minimally yielded with a PR of tions were not scaled down to obtain more conservative rotational
0.006 rad. Table 7 lists the PR of yielding members and columns demands of the yielding members.

Table 5. Final designed members of STMF-1, STMF-2, and STMF-3


STMF-1 STMF-2 STMF-3
Member Final design α Final design α Final design α
Column W360 × 818 — W360 × 900 — W920 × 970 —
Panel zone W360 × 818 — W360 × 900 — W920 × 970 —
Top SS 2C380 × 74 0.16 2C380 × 74 0.16 2C380 × 74 0.15
Bottom SS 2C380 × 74 0.14 2C380 × 74 0.14 2C380 × 74 0.10
Vertical SS 2C380 × 74 — 2C380 × 74 — 2C380 × 74 —
Chord 1-2 2C380 × 74 — 2C380 × 74 — 2C380 × 74 þ 25 mm thick side platesa —
Chord 3-4b 2C380 × 74 — 2C380 × 74 — 2C380 × 74 —
Diagonal 1 2C250 × 30 2C230 × 45 2MC310 × 52 —
Diagonal 2-4b 2C230 × 30 — 2C230 × 45 — 2MC310 × 52 —
Vertical 1-3b 2C150 × 19.3 — 2C150 × 19.3 — 2C150 × 19.3 —
Intermediate — — 2C130 × 13 — — —
a
Side plate is 289 mm deep and welded to the web of the member.
b
STMF-1 and STMF-2 only have Chord 1-3, Diagonal 1-3, and Vertical 1-2.

© ASCE 04019053-10 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


Nonlinear Time History Analysis under Design Basis ground motions. Chord members in the SS of STMF-1 yielded
Earthquakes under three ground motions, and the average maximum PR was
Plastic hinge models of the chord members in the SS of the three 0.027 rad. One of the ground motions caused one of the four plastic
prototype STMFs were modeled after suggested values in Table 2. hinges at the ends of the chord members in the SS to reach the point
Gravity loading induced high compression and tension in the top where it lost strength. This resulted in a PR of 0.06 rad; however,
the system did not collapse. Chord members in the SS of STMF-2
and bottom chord members within the SS, respectively. As a result,
only yielded under two ground motions, and the average maximum
parameters a, b, and β needed for the plastic hinge model of the top
plastic ground motion was 0.006 rad. On the other hand, intermedi-
chord member were obtained from Eqs. (2), (3), and (6), whereas
ate vertical members yielded under 17 ground motions, and the
those for the bottom chord member were obtained from Eqs. (4),
average maximum PR was 0.01 rad. Nevertheless, the proposed
(5), and (7). The average maximum SDR of STMF-1, STMF-2,
design procedure could safely provide seismic resistance for the
STMF-3 were 0.71%, 0.53%, and 0.14%, respectively (Fig. 14).
three long-span STMFs.
Table 9 shows the average maximum PRs of DYMs. These aver-
ages were calculated based on the number of ground motions that
caused yielding in any of the members. Neither the STMF-3 mem-
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

Maximum Considered Earthquake


bers nor the STMF-1 and STMF-2 columns yielded under any DBE
Near-Collapse Loading Sequence
MCE-level ground motions are used for NTH analyses to check the
Table 6. Story drift ratio at yielding and plastic rotation at 2.1% story drift collapse prevention of structures. Note that nearly all component
ratio (fully strain-hardened chord members in SS) of members of STMF-1
test data are based on a symmetric cyclic loading protocol with
Story drift ratio Plastic rotation at many cycles and two to three cycles per deformation level. This is
Yielding when yielding 2.1% story drift not representative of the ratcheting effect that leads to structural
sequence Member occurs (%) ratio (rad)
1 Top chord in SS 0.89 0.034
1 Bottom chord in SS 0.89 0.032 Table 8. Story drift ratio at yielding and plastic rotation at 1.1% story drift
2 Right column 1.57 0.002 ratio (fully strain-hardened chord members in SS) of members of STMF-3
3 Left column 1.84 0.005
Story drift ratio Plastic rotation at
Yielding when yielding 1.1% story drift
sequence Member occurs (%) ratio (rad)
9000 2000 1 Top chord in SS 0.52 0.030
8000 1 Bottom chord in SS 0.52 0.031
3
2
2 Right column 0.66 0.005
7000 STMF-1 3 Left column 0.75 0.004
1500
Base Shear (kips)

STMF-2
Base Shear (kN)

6000 1
STMF-3
5000
6 1000
4000 5 2.0
3000 3 4
Maximum Story Drift Ratio (%)

2 STMF-1 Avg STMF-1


3
2 500 STMF-2 Avg STMF-2
2000
1 1.5 STMF-3 Avg STMF-3
1
1000
0 0
0 0.5 1 1.5 2 2.5 1.0
Story Drift Ratio (%)

Fig. 13. Pushover curves of STMF-1, STMF-2, and STMF-3.


0.5

Table 7. Story drift ratio at yielding and plastic rotation at 2.1% story drift 0.0
0 4 8 12 16 20
ratio (fully strain-hardened chord and intermediate vertical members in SS) DBE Ground Motion No.
of members of STMF-2
Story drift ratio Plastic rotation at Fig. 14. Maximum story drift ratio of STMF-1, STMF-2, and STMF-3
Yielding when yielding 2.1% story drift under 20 DBE ground motions.
sequence Member occurs (%) ratio (rad)
1 Left intermediate 0.36 0.090
vertical member
1 Right intermediate 0.36 0.088 Table 9. Average maximum plastic rotations under DBEs
vertical member Average maximum plastic rotation (rad)
2 Top left chord in SS 0.82 0.034
Member STMF-1 No. of DBEs STMF-2 No. of DBEs
3 Top right chord in SS 0.88 0.029
a
3 Bottom left chord in SS 0.88 0.033 Chord in SS 0.027 3 0.006 2
4 Bottom right chord in SS 0.94 0.028 Intermediate N/A N/A 0.01 17
5 Right column 1.46 0.006 a
Under one ground motion, one plastic hinge failed which had a high plastic
6 Left column 1.69 0.003
rotation of 0.06 rad.

© ASCE 04019053-11 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


0.06 0.15
Near Collapse

Member Rotation (rad)


Plastic Rotation (Rad)
0.04 0.10 Rainflow Technique

0.05
0.02

0.00
0.00 MCE 2 MCE 18
-0.05
-0.02
0 10 20 30 40 50 60 0 5 10 15 20
(a) Time (s) (b) Cycle Number

Fig. 15. (a) Plastic rotation of chord member in SS under MCEs; and (b) near-collapse loading protocol.
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

collapse. Indeed, analytical studies carried out by Maison and net excursion of 0.04 rad measured from −0.01 rad to þ0.03 rad
Speicher (2016) confirmed that the time history response of the was selected. As can be seen, this maximum PR of 0.03 rad from
building interstory drifts under MCE ground motions was better the rainflow analysis is much less than the rotational capacity of
described by single-sided cyclic and/or monotonic loading rather the double-channel sections under the AISC loading protocol.
than fully reversed cyclic loading. As a result, the nonlinear behav- Therefore, to be conservative, more rigorous amplitudes were se-
ior of the double-channel members obtained from the fully reversed lected where the symmetric fully reversed cycles started from one
symmetric loading protocol with a high number of repeating cycles cycle each at 0.00375, 0.005, 0.0075, and 0.01 rad total member
such as the AISC loading protocol can be unrealistic to represent rotation. Then, the amplitude was increased at a 0.01-rad interval to
the time history response of STMF members subject to MCEs. 0.06 rad, which is approximately 0.05-rad PR or 1.7 times the aver-
Therefore, a more realistic loading protocol for a near-collapse sit- age maximum PR of the chord member under 20 MCE ground mo-
uation needed to be developed. To develop the loading protocol, a tions. Then, nonreversed cycles were added as shown in Fig. 15(b).
rainflow counting technique was first used. Krawinkler et al. (2000) The nonreversed cycles then progressed in an ever-increasing man-
used this technique to determine excursion ranges, which are the ner (Jiansinlapadamrong 2017).
amplitude of the response measured from either valley to peak or Behavior of Double-Channel under Near-Collapse Loading
vice versa. Due to the nature of the response spectra of structures and Plastic Hinge Model
under MCE ground motions, they created a loading protocol for a Double-channel member 2C310 × 30.8 was tested with the same
moment frame using the three largest excursion ranges. After that, special connection details used in the specimen shown in Fig. 5(c),
the loading protocol has pulselike amplitudes cycling around a to compare with the test results of the double-channel under the
residual drift. The total number of cycles is only eight. In this study, AISC loading sequence. The specimen was 1,346 mm (53 in.)
nonlinear time history analyses of a prototype long-span STMF long from the loading point to the plastic hinge location. Special
with the same configurations as STMF-1 under 20 SAC MCEs connection details featured extended gusset plate and horizon-
were carried out by using a plastic hinge model without considering tal stitches. The detailed drawing of the test specimen is shown
the loss of strength (which was obtained from fully reversed cyclic in Fig. 16. The specimen under near-collapse loading protocol
testing) (Fig. 11) to capture the rotational demand of chord mem- in Fig. 17(a) was cyclically loaded until a very large rotation of
bers. Typical PR responses of the chord member in the SS are 0.14 rad, at which the tip of the specimen nearly came off the lateral
shown in Fig. 15(a). By employing the rainflow technique used by bracing frame. At this point, the test was terminated. The hysteretic
Krawinkler et al. (2000), the plot of a loading protocol is shown in behavior, shown in Fig. 17(b), was stable up to 0.14 rad, at which
Fig. 15(b). Because the largest excursion of the average maximum the force slightly increased due to the horizontal stitch hitting the
plastic rotational demand from the NTH analyses was 0.036 rad, a center gusset plate.

Fig. 16. Experimental test specimen dimensions (millimeters).

© ASCE 04019053-12 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


400 300
300
200
200

Moment (kN-m)

Moment (kip-ft)
100
100
0 0
-100
-100
-200
-300 Experiment -200
FEA
-400 -300
-0.05 0.00 0.05 0.10 0.15
Member Rotation (rad)
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

Fig. 17. (a) Specimen at 0.13 rad; and (b) hysteretic behavior, FEA result, and trilinear model.

(a) (b)

(c)

Fig. 18. FEA results under NC loading protocol and trilinear model of 2C310 × 30.8 with axial compression to axial capacity ratio: (a) α ¼ 0.1;
(b) α ¼ 0.2; and (c) α ¼ 0.3.

To consider the axial load effect on the PR under near-collapse- 2C310 × 30.8 under AISC loading sequence, the relationships be-
type loads, similar to the procedure for ground motions with multi- tween a and α, b and α, and β and α were determined by regression
ple fully reversed cycles (DBEs), FEA was carried out and the analyses as follows:
model was calibrated so that the result was in good agreement with Axial compression case
the experimental result as shown in Fig. 17(b). A trilinear backbone
model is shown using a thick line in this figure. The FE model was a ¼ −0.1α þ 0.03 when 0 < α < 0.2 ð8aÞ
then used to analyze flexural behavior of the 2C310 × 30.8 section
under combined bending and various axial compression and ten- ¼ 0.011 when 0.2 < α < 0.3 ð8bÞ
sion forces. Figs. 18(a–c) show hysteresis behavior and trilinear
backbone models of 2C310 × 30.8 with axial compression-to- b ¼ −0.7α þ 0.1 when 0 < α < 0.1 ð9aÞ
capacity ratio, α, of 0.1, 0.2, and 0.3. Figs. 19(a–c) show those
of 2C310 × 30.8 with axial tension. Similar to the results of ¼ 0.03 when 0.1 < α < 0.2 ð9bÞ

© ASCE 04019053-13 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


(a) (b)
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

(c)

Fig. 19. FEA results under NC loading protocol and trilinear model of 2C310 × 30.8 with axial tension per capacity ratio: (a) α ¼ 0.1; (b) α ¼ 0.2;
and (c) α ¼ 0.3.

¼ − 0.3α þ 0.09 when 0.2 < α < 0.3 ð9cÞ 3.0


Maximum Story Drift Ratio (%)

STMF-1 STMF-2 STMF-3


β ¼ −0.76α þ 1.4 when 0 < α < 0.3 ð10Þ Avg STMF-1 Avg STMF-2 Avg STMF-3

Axial tension case 2.0

a ¼ 0.03 when 0 < α < 0.3 ð11Þ

b ¼ 0.1 when 0 < α < 0.1 ð12aÞ 1.0

¼ − 0.4α þ 0.14 when 0.1 < α < 0.3 ð12bÞ

β ¼ 1.4 when 0 < α < 0.13 ð13aÞ 0.0


0 4 8 12 16 20
MCE Ground Motion No.
¼ − 0.7α þ 1.4 when 0.13 < α < 0.3 ð13bÞ
Fig. 20. Maximum story drift ratio of STMF-1, STMF-2, and STMF-3
under 20 MCE ground motion.
NTH Analyses under Maximum Considered Earthquakes
Similar to the modeling for DBEs, plastic hinge models of SS chord
members were modeled using the values listed in Table 2. Param-
eters a, b, and β needed for the plastic hinge model of the top chord Table 10. Average maximum plastic rotations under MCEs
member were obtained from Eqs. (8)–(10), whereas those for the Average maximum plastic rotation (rad)
bottom chord member were obtained from Eqs. (11)–(13). Average
maximum SDRs of STMF-1, STMF-2, and STMF-3 were 1.11%, Member STMF-1 No. of MCEs STMF-2 No. of MCEs
0.73%, and 0.16%, respectively (Fig. 20). Although the SDRs in- Column 0.004 3 0.003 1
crease by 14%–56% compared with those of STMFs under DBEs, Chord in SS 0.01 17 0.005 8
the analyses show that the structures are far from collapse. None of Intermediate N/A N/A 0.018 19
the members in STMF-3 yielded. Table 10 shows the average maxi-
mum PRs of both the DYMs and columns of STMF-1 and STMF-2.
Notably, STMF-1 and STMF-2 columns only slightly yielded members more than 0.01 rad, and the absolute maximum PR was
under three and one MCE ground motions, respectively. The chord 0.042 rad. Note that 0.01 rad PR is the maximum rotational capac-
members in the SS of STMF-1 yielded under 17 ground motions ity of a double-channel section when α ¼ 0.3 [Fig. 18(c)]. This
with the average maximum PR of 0.01 rad. Among 17 ground means, if STMF-1 had been designed based on α ¼ 0.3, it could
motions, five ground motions caused maximum PR of the chord have collapsed under several MCE events. Average maximum PR

© ASCE 04019053-14 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


of intermediate vertical members caused by 19 ground motions was and were strain hardened. This procedure led to overdesign of
0.018 rad. Chord members of STMF-2 slightly yielded under eight the nonyielding members. Further optimization of the design
ground motions. Analyses indicated that the earthquake-induced procedure is warranted for a more economic design.
axial forces in the chord members did not completely cancel out 6. Based on nonlinear FEA for double-channel sections and NTH
the gravity-induced axial forces. analysis of three prototype STMFs, it is shown that if the axial
compression demand-to-capacity ratio (Pu =Ry Fy Ag ) is kept un-
der 0.15, the built-up chord members have sufficient rotational
Summary and Conclusions ductility for STMFs to maintain satisfactory performance under
DBE and MCE ground motions (only 1 of 120 analyses had
This paper presents a design procedure, recommended plastic hinge chord member failure). Note the developed moment–rotation re-
models of designated yielding members and nonlinear analyses lationships based on isolated cantilever FE models can be overly
of the seismic performance for STMFs with a very long span of conservative because in an actual STMF the truss members, col-
27.4 m, approximately 40% longer than the currently allowed umns, and other elements can provide restraints to prevent large
span length specified in AISC seismic provisions (AISC 341). local buckling as shown in the FE models (Jiansinlapadamrong
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

Different SS layouts were investigated for long-span STMFs by et al. 2018). This conservative model is justified because the
considering the overall lateral stiffness, rotational demand of the study STMFs are the only gravity load–carrying system in the
DYMs, available open space in the SS, and lateral bracing require- prototype building; therefore, any strength degradation due to
ments. The design was done by first determining the DYMs lateral loads could cause collapse. For a long-span STMF or an
(members within the SS) by conventional ELF procedure. Then, STMF with high axial forces in chord members, to minimize
nonyielding members outside of the SS were designed by using the effect of axial forces, it is suggested that the current axial
a capacity design approach when DYMs are fully strain hardened. load limit in AISC 341 be changed to Pu ≤ 0.18ϕPn (or
A new detailing featuring an extended gusset plate and a weld-free ϕPn ≥ 5.5Pu ), assuming Ry ¼ 1.1.
zone at the end of the double-channel sections (Jiansinlapadamrong 7. The research results suggest that STMFs with greater dimen-
et al. 2018) provides high rotational capacity and prevents the truss sions than those in the current AISC seismic provisions (AISC
members from experiencing strength degradation under strong 341) can be confidently used in high seismic areas. The length
earthquakes. The effect of axial compression and tension on the of STMFs can be extended from 20 to 27.4 m, the total depth
rotational ductility of built-up double-channel sections was inves- can be extended from 1.8 to 3.05 m, and the length-to-depth
tigated by nonlinear finite-element analysis. Recommended plastic ratio of a panel in the SS can be extended from 1.5 to 3.0 without
hinge models for DYMs of STMFs for nonlinear analysis under compromising the stiffness or strength of STMFs.
both DBE and MCE level ground motions are proposed for both
design and seismic performance analysis purposes. The following
conclusions can be drawn from this study: Acknowledgments
1. When using nonlinear pushover analysis to design nonyielding
members, it is suggested to allow the column base to develop a This research was partially supported by the US National Science
plastic rotation up to approximately 0.005 rad in order to obtain Foundation under Award No. CMMI-0936563 and the AISC.
a realistic design.
2. Under pushover analyses, the first yielding of STMF-1, STMF-2,
and STMF-3 occurred at 0.89%, 0.36% (at intermediate vertical References
members), and 0.52% SDRs, respectively, which is between the
AISC. 2016. Seismic provisions for structural steel buildings. ANSI/AISC
typical steel braced frames and steel moment frames. STMF-3
341. Chicago: AISC.
with a single short Vierendeel special segment provides a very
ASCE. 2010. Minimum design loads for buildings and other structures.
high lateral stiffness, which considerably reduces the chord ASCE/SEI 7. Reston, VA: ASCE.
rotations under either DBE or MCE ground motions. ASCE. 2017. Seismic evaluation and retrofit of existing buildings. ASCE/
3. The average maximum SDRs of long-span STMFs under DBE SEI 41. Reston, VA: ASCE.
ground motions in this study were between 0.14% and 0.71%, ASTM. 2018. Standard specification for high-strength low-alloy
and the members outside of the SS did not yield. columbium-vanadium structural steel. A572/A572M. West Consho-
4. NTH analyses according to MCE-level ground motions indi- hocken, PA: ASTM.
cate that the DYMs’ time history response is predominately Chao, S.-H., and S. C. Goel. 2008. “A modified equation for expected
single-sided and very different from the response based on fully maximum shear strength of the special segment for design of special
reversed cyclic loading. A near-collapse loading protocol was truss moment frames.” Eng. J. 45 (2): 117–125.
developed according to analytical results. Component testing Chao, S.-H., C. Jiansinlapadamrong, S. Simasathien, and T. Okazaki.
results show that members can sustain much larger plastic Forthcoming. “Full-scale testing and design of steel special truss
rotational demands under MCE-level loading sequence. moment frames for high seismic areas.” J. Struct. Eng., in press.
5. Using the nonlinear models of the plastic hinge based on near- Chao, S.-H., K. Khandelwal, and S. Ei-Tawil. 2006. “Ductile web fracture
collapse loading protocol, the three prototype STMFs’ maxi- initiation in steel shear links.” J. Struct. Eng. 132 (8): 1192–1200.
https://doi.org/10.1061/(ASCE)0733-9445(2006)132:8(1192).
mum SDRs under MCE ground motions ranged between 0.16%
Crisan, A. 2016. “Material calibration for static cyclic analyses.”
and 1.11%. The members outside of SS, other than the columns,
Intersections/Intersectii 13 (2): 43–58.
did not yield. This suggests that the design procedure and SS CSI (Computers and Structures). 2006. “PERFORM-3D–Nonlinear analy-
layouts provide conservative and satisfactory seismic collapse sis and performance assessment for 3D structures.” In User guide,
prevention performance for long-span STMFs. Note that while Version 4. Berkeley, CA: Computers and Structures.
SDRs obtained from NTH are small, the members outside of SS Hamburger, R. O., and J. Malley. 2016. “Seismic design of steel special
were designed based on pushover analysis where the structure moment frames: A guide for practicing engineers.” NEHRP Seismic
was displaced to a lager SDR until the chord members within Design Technical Brief No. 2, NIST GSR 09-917-3. Gaithersburg, MD:
the SS and intermediate vertical members (if used) fully yielded NIST.

© ASCE 04019053-15 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053


Jiansinlapadamrong, C. 2017. “Seismic performance of double-channel Newell, J. D., and C.-M. Uang. 2008. “Cyclic behavior of steel wide-flange
and double-HSS sections special truss moment frame with columns subjected to large drift.” J. Struct. Eng. 134 (8): 1334–1342.
buckling-restrained braces, and long span special truss moment https://doi.org/10.1061/(ASCE)0733-9445(2008)134:8(1334).
frame.” Ph.D. dissertation, Dept. of Civil Engineering, Univ. of Texas Ozkula, G., J. Harris, and C.-M. Uang. 2017. “Observations from cyclic
at Arlington. tests on deep, wide-flange beam-columns.” Eng. J. 54 (1): 45–59.
Jiansinlapadamrong, C., B. Price, and S.-H. Chao. 2018. “Cyclic behavior Simasathien, S., C. Jiansinlapadamrong, and S.-H. Chao. 2017. “Seismic
of steel double-channel built-up components with a new lateral- behavior of special truss moment frame with double hollow structural
torsional-buckling prevention detail.” J. Struct. Eng. 144 (8): 04018127. sections as chord members.” Eng. Struct. 131 (2017): 14–27. https://doi
https://doi.org/10.1061/(ASCE)ST.1943-541X.0002125. .org/10.1016/j.engstruct.2016.10.001.
Krawinkler, H., A. Gupta, R. Medina, and M. Luco. 2000. Development Somerville, P. G., M. Smith, S. Punyamurthula, and J. Sun. 1997. Develop-
of loading histories for testing of steel beam-to-column assemblies. ment of ground motion time histories for Phase 2 of the FEMA/SAC steel
Rep. Prepared for SAC Steel Project. Palo Alto, CA: Dept. of Civil project. Rep. No. SAC/BD-97/04. Sacramento, CA: SAC Joint Venture.
and Environmental Engineering, Stanford Univ. Zargar, S., R. A. Medina, and E. Miranda. 2014. “Cyclic behavior of
Maison, B. F., and M. S. Speicher. 2016. “Loading protocols for ASCE 41 deep steel columns subjected to large drifts, rotations, and axial loads.”
backbone curves.” Earthquake Spectra 32 (4): 2513–2532. https://doi In Proc., 10th National Conf. in Earthquake Engineering. Anchorage,
.org/10.1193/010816EQS007EP. AK: Earthquake Engineering Research Institute.
Downloaded from ascelibrary.org by Auburn University on 04/26/19. Copyright ASCE. For personal use only; all rights reserved.

© ASCE 04019053-16 J. Struct. Eng.

J. Struct. Eng., 2019, 145(7): 04019053

You might also like