Sei sulla pagina 1di 20

Received: 13 July 2017 Revised: 22 November 2017 Accepted: 21 December 2017

DOI: 10.1002/tal.1469

RESEARCH ARTICLE

Frame‐type load bearing system for long‐span cantilevers


Seyed Omid Sajedi* | Seyed Rasoul Mirghaderi | Farhad Keshavarzi

School of Civil Engineering, University of


Tehran, Tehran, Iran Summary
Correspondence Cantilevers experience high risks of vulnerability against progressive collapse and vertical ground
Seyed Omid Sajedi, School of Civil motion effects. In addition, despite common engineering practice that regards cantilevers
Engineering, University of Tehran, 16th Azar dominated by vertical loads, it is shown that while subjected to lateral forces, they might
Street, Enghelab Square, Tehran, Iran.
undergo large deflections due to the formation of plastic hinges in the supporting structural
Email: ssajedi@buffalo.edu
subassemblage; even if cantilevers satisfy deflection limits proposed by design codes. To
overcome such vulnerabilities, successive cantilever beams can be coupled in the height of the
structure using vertical elements to develop a framed cantilever system. Frame behavior in
cantilevers is formulized using spring models and by employing optimization procedures,
economic efficiency is compared to the conventional method. The optimization results for 6
2‐D steel moment frames showed that employing framed cantilever system has the potential
to reduce required material weight up to 40%. Furthermore, nonlinear pushdown and incremental
dynamic analyses were conducted to extract capacity and fragility curves. The results reveal the
superior performance of framed cantilevers in both lateral and vertical loads while offering better
resistance against the progressive collapse.

KEY W ORDS

cantilever coupling, frame behavior, incremental dynamic analysis, progressive collapse, pushdown
analysis, structural optimization

1 | I N T RO D U CT I O N

Cantilever beams are frequently used as structural support for projected elements such as balcony and loggias, and they are prominent among mod-
ern and old architecture, mainly as parts of the building's façade. The façade is of paramount significance in terms of building energy performance
and esthetics. The proper functioning of the supporting structure is fundamental to the performance of cladding and the building as a whole. Due to
the fact that nonstructural components are much prone to the cantilevers' deflections, the satisfactory design is of much importance to achieve

Notation list: Aopt,Bopt,…,Eopt, polynomial coefficients to find optimum spring location; Ai, ith element's cross‐sectional area; C, penalized cost of structure; Dall, allowable
story's drift; DCRi,j, ith element, criterion j's demand/capacity ratio; DL, uniform dead static load; DRi, displacement ratio of assemblage CBi over CB1; DMi, damage
measure, criterion i; Du, ultimate story's drift; E, modulus of elasticity; Fy, steel material's yield stress; H, horizontal ground motion; Icb, cantilever beam's moment of
inertia; Icce, cantilever coupling element's moment of inertia; KL, portal frame's stiffness against nodal support displacement; KE, elastic slope in pushdown curves; ki,
ith spring's rotational stiffness; KP, penalty constant; KT, tangent slope in pushdown curves; Lcb, cantilever beam's length; Lcce, cantilever coupling element's length;
Li, ith element's length; LL, uniform live load; M, bending moment function; M1, first spring's moment in CB2 assemblage; M2, second spring's moment in CB2
assemblage; M3, spring's moment in CB3 assemblage; MA, point A's moment; ME, maximum bending moment in FCS caused by lateral drift; MCBi , maximum bending
moment of assemblage CBi; MRi, moment ratio of assemblage CBi over CB1; n, spring's relative stiffness constant; Nelem, total number of 2‐D frame's elements; P,
cantilever's concentrated tip load; PMRu, moment and axial force interaction vector; QD, linear distributed dead load; QL, linear distributed live load; U, total
substructure's energy; Ucb, cantilever beam's internal energy; USpring, spring's internal energy; V, vertical ground motion; Vcce, shear force in coupling element caused
by lateral drift; MESS, maximum expected shear strength; Vall, allowable shear force; Vu, ultimate shear force; w, uniform distributed linear load; WTB, width of
tributary area; x, horizontal distance from cantilever beam's support; α, relative coupling element's location; α1, first spring's relative location in CB2; α2, second
spring's relative location in CB2; α3, spring's relative location in CB3; θi, ith spring's rotation; δCBi , maximum displacement in assemblage CBi; δD, point D's deflection;
δD/A, deviation of the tangent at point A with respect to the tangent at point D; δh, floor's lateral displacement; Δ1, maximum deflection in separate cantilevers; Δ2,
maximum deflection in coupled cantilevers; Δall, allowable deflection value; Δu, ultimate deflection value; ρ, material density; ψi, ith element's total penalty function;
ψi,j, ith element, criterion j's penalty function.
*
Current affiliation: Graduate Teaching Assistant, School of Civil, Environmental and Structural Engineering, University at Buffalo, New York, USA.

Struct Design Tall Spec Build. 2018;e1469. wileyonlinelibrary.com/journal/tal Copyright © 2018 John Wiley & Sons, Ltd. 1 of 20
https://doi.org/10.1002/tal.1469
2 of 20 SAJEDI ET AL.

desired performance levels. Although engineers can design their own cantilevers with existing code specifications, authors have presented case
studies that show when facing irregular long spans, common design provisions might fail to guarantee the desired performance of such members.
In other words, the criteria proposed by building codes such as International Building Code (IBC) or American Society of Civil Engineers (ASCE)
might not advise safe design procedures for cantilevered elements, especially in unusual span lengths. One of the sources of this problem is prob-
ably the fact that common design procedures are limited to elastic static analyses, which simplify calculations but cannot predict certain phenom-
ena that will be discussed furthermore.
Certain types of long‐span structures such as bridges are studied in the literature to evaluate code provisions[1]; however, specific research to
evaluate long‐span cantilever extensions underpinning their seismic interaction with the main body structure seems to be missing from the litera-
ture. Many studies are conducted to investigate the optimum shape of cantilever beams under various static and dynamic loading conditions by
means of analytical and numerical approaches.[2–6] However, most of them do not present a practical outlook with regards to building industry
and common construction methods. In other words, many researchers have utilized cantilever as a single determinate structural element to confirm
numerical results and cantilevers' performance as a part of the frame structure is not investigated. In a nutshell, the first objective of this study is to
demonstrate the fact that structural proportioning of long‐span cantilevers requires special provisions to be considered in future codes. In addition,
a structural system has been proposed to enhance the performance of hanging elements in terms of structural performance and cost efficiency.
This article tends to elaborate on these issues and present a solution. It should be noted that the main emphasis in the following investigations
is to evaluate “long‐span” elements. Further studies will show that cantilevers longer than story's height could be assigned to this category.

2 | C HA LL E N G E S I N D E S I G N O F LO N G‐ S P A N C A N T I L E V E RS

Prior to elaborate on the proposed solution, major challenges in the design procedure of long or heavy‐loaded cantilevers can be listed as follows.

2.1 | Cantilever's interaction with the main structure


Despite common engineering practice that regards cantilevers as members dominated by gravity load combinations, support's rigidity, and defor-
mations are critical to the maximum deflection in singly‐supported beams. Assuming a moment frame's structural subassemblage depicted in
Figure 1, the formation of plastic hinges in the seismic beams or columns will reduce cantilever's support rigidity, which is accompanied with resid-
ual story drifts. Although the flexural deflection (ΔF) is constant and a function of cantilever's properties, additional rigid body displacement (ΔR)
causes extra deflection values that might be significant for long‐span cantilevers. In conclusion, the structure as a whole might present a satisfac-
tory seismic performance; however, this phenomenon produces significantly larger deflections than what is expected from linear analyses and
design. Such deflections could be the main cause of unwanted nonstructural damage especially in glass façade types such as glazing systems.
To better demonstrate this phenomenon, an 8‐story 2‐D steel frame with 5‐m cantilevers has been proportioned to satisfy code requirements
(model properties are explained furthermore as frame A). In addition, ground motion records have been properly scaled in accordance with code
provisions.[7] All members satisfy recommended code requirements regarding strength and service design criteria; however, nonlinear time‐history
analyses, even without considering vertical ground motion data, indicate that cantilevers experience significantly larger deflections after the struc-
ture has reached a motionless state. Demand‐capacity ratios for the critical member has been calculated and presented in Figure 2. This parametric
case study shows that deflections can go way beyond the allowable limit‐state with the consideration of lateral seismic effects.

2.2 | Lack of structural redundancy


Catastrophic structural failures such as Ronan Point apartment collapse (1968), Kansas City Hyatt Hotel (1981), Mianus River bridge (1983) inci-
dents, and balcony collapse at the University of Virginia (1997) are all tragedies, which similarly involve the absence of a redundant method of

FIGURE 1 Cantilevers residual deflection due to the deformations in the structural subassemblage
SAJEDI ET AL. 3 of 20

Max/Allowable Deflection
Nonlinear Dynamic Analysis
4 Extra Deflection Ratio

3 Linear Static Analysis

Ratio
Code Design Deflection Ratio
2

0
Imperial Kobe Landers Chi Chi Imperial Irpinia Erzican
Valley 1116 900 1244 Valley 292 821
169 181

FIGURE 2 Comparison of maximum deflection ratios after nonlinear dynamic analyses

support.[8–11] Cantilever beams while being supported at only one end have higher chances of collapse compared to beams restrained at both ends
since the element will become structurally unstable by the loss of any supporting restraints. Nevertheless, the collapse of a specific cantilever might
impose extra massive loads on lower ones and initiate a progressive collapse mechanism like the Ronan Point incident.

2.3 | Greater internal forces and demands


Based on simple concepts of structural equilibrium, restraining a beam at merely one end compared to other alternatives, produces the maximum
bending moment and shear force alongside elevated values of deflection at the free end of the beam. Such demands force structural designers to
use deep‐height sections in order to satisfy code requirements. Nevertheless, stress values at the fixed end of cantilever beams can be more critical
than what is expected from simple theories of solid mechanics.[12] Since architectural requirements limit structural element's dimensions, designed
cantilever section might not abide by the desired limitations.

2.4 | Vertical ground motion effects


Multiple studies on damage parameters of buildings in historic earthquakes such as Northridge (1994) and Kobe (1995) point towards vertical
ground motion as a direct cause of failure, especially in near‐field locations. There are several motion‐data analyses, which indicate to the fact that
vertical to horizontal peak ground acceleration ratios (V/H) can exceed 1 and reach values as high as 2.21 (Chi‐Chi earthquake, 1999).[13] Cantile-
vers are considered to be among the most vulnerable elements in structures against the upward ground motion. In many design specifications, an
equivalent static load is suggested to predict such consequences on cantilevered elements, but these simplifications might lead to underestimation
of demand in near‐field site locations.[1]

2.5 | Limited column orientation


For the purpose of seismic design, H‐shaped columns are oriented in accordance with the lateral load resistant system. Cantilever beams also
require rigid connections at their fixed end. For this reason, only H‐shaped columns with unusual thick members or complex connection details
can provide sufficient rigidity for fixed‐end beams in both principal axes of an H‐column section. Hot‐rolled steel sections, which are common
in many countries for the construction practice, lack such dimensional proportions.

3 | P R O P O SE D M E T H O D

This article investigates a structural technique known as framed cantilever system (FCS) for construction purposes such as balcony and loggias or
any “long‐span” cantilevered elements that are extended from the structures. It is believed that FCS has various advantages over conventional
practice and can address most challenges discussed before. Spring models based on solid mechanics' theory are used to study the behavior of
this system and proper equations have been presented to aid designers. Authors have verified the potential economic efficiency of FCS in
two stages. A brute‐force attack algorithm has been utilized to design FCS substructural model to evaluate optimum configuration and assess
preliminary cost efficiency. Furthermore, with the purpose of achieving an accurate and more realistic design procedure, the genetic algorithm
has been utilized as an optimization method. Multiple 2‐D steel moment frames with and without modification of cantilever beams are then
designed and employed for further nonlinear studies. To assess the performance of FCS under excessive vertical loading, a nonlinear static pro-
cedure known as pushdown is applied to obtain capacity curves. In the final step, nonlinear behavior of models utilizing framed cantilevers has
been studied with a series of incremental dynamic analyses (IDAs) to obtain fragility curves for several damage parameters. Considering multiple
near‐ and far‐field ground motion records, the behavior of extended substructure and its interaction with the frame as a whole has been
4 of 20 SAJEDI ET AL.

thoroughly investigated. It should be noted that the term cantilever usually refers to a horizontal beam with merely one fixed support while
coupled cantilevers in FCS behave quite differently; however, throughout this paper, the current terminology will be used for the coupled
cantilevers as well.
Limiting deformations for cantilever beams as determinate elements is a vital criterion for service design levels. Magnitudes of deformation and
vibration under the action of unfactored loads are often ahead of ultimate strength limit states for long‐span cantilevers. In order to maintain
structural stiffness at desired levels, there are two common procedures for providing sufficient stiffness: improving section properties or supplying
adequate redundancy. Although the first option could be applicable in some cases as the common solution, it leads to bulky structural elements
such as high‐depth beams that not only contrast with architectural requirements but also question cost efficiency. On the other hand, while the
second procedure could not be gained by direct modification of boundary conditions, it might be achieved by applying supplementary elements
including struts and cables. This article aims to introduce a new practical alternative by adding elements, which might conflict the least with archi-
tectural purposes. The main configuration of this idea is based on vertical coupling elements in successive cantilevers commonly serving as a part of
supporting system for balconies or protrusions in building structures. These vertical coupling members with rigid connections at the ends could
possibly mobilize moment frame action. Employing the aforementioned frame behavior should potentially provide an enhanced performance for
lateral and vertical load resistance. Continuous or discrete coupling of cantilever beams with vertical elements proposes a significant level of redun-
dancy and leads to greater stiffness in elements carrying vertical loads. To simplify FCS behavior, one might consider the simple structural
subassemblage depicted in Figure 3. Properties related to cantilever beams and cantilever coupling elements will be respectively referred to by
cb and cce indices.
Assuming that supporting structural subassemblage provides sufficient rigidity for cantilevers, general FCS behavior can be investigated similar
to a portal frame. Being subjected to vertical loads, coupling elements introduce enhanced moment distribution in cantilevers accompanied with
double curvature deformation, which results up to twice bending strength and 4 times more stiffness in comparison with separate cantilevers
(Δ1 = 4Δ2). In addition, constrained cantilevers have the potential to contribute to the story's lateral load resistance. In order to monitor such effects,
lateral story drifts can be modeled as nodal displacements imposed on supporting nodes (δH1 and δH2 ). FCS's lateral stiffness (KL) against relative
story drift can be calculated using the slope‐deflection method as expressed in Equation 1. KL reaches an upper limit as EIcce approaches infinite
values (Equation 2).

 
1 12EI
KL ¼   : ; (1)
6EI L3 cce
L
1 þ   cce
EI
L cb

FIGURE 3 Extreme limits for frame behavior while (a) EIv → ∞ and (b) EIv → 0
SAJEDI ET AL. 5 of 20

2EIcb
lim K L ¼ : (2)
EIcce →∞ L2 cce Lcb

It is clear that in the absense of a coupling element (EIcce = 0), cantilevers do not contribute to the lateral story's stiffness and strength. Coupling
element's shear force (Vcce) and bending moments (ME) while subjected to lateral displacements can be expressed as follows:

V cce ¼ K L :ðδh1 −δh2 Þ; (3)

K L V cce Lcce
ME ¼ : (4)
2

The proposed simplification of FCS behavior demonstrate the potential of constrained cantilevers to enhance different aspects of structural
performance; however, more accurate investigations have been conducted in this manuscript by means of advanced analysis methods. Neverthe-
less, by relocating the coupling elements closer to side column, one might be able to create more efficient and complex structural arrangements
(Figure 4). FCS's main ambition is to improve cantilevers' performance by including them as a part of the frame action discussed before. This cou-
pling system is supposed to constrain the out of plane rotational degrees of freedom (major bending moment) in cantilever beams. Constraining
horizontal beams increases the number of turning points based on the arrangement of joining elements and reduces deformation magnitudes sig-
nificantly. It should be noted that although vertical coupling elements might appear as column‐alike members, they do not share the same behavior
and performance of main columns. Coupling elements, unlike columns, are not connected to the base and due to the nature of the extended
structure, possess minimal axial forces. In other words, they are not axial load collectors and instead, cause a much better moment distribution
in cantilevers. Diagonal bracing also provides a suitable structural alternative to cover long‐span cantilevers but can violate many architectural
aspects and might not be applicable for many purposes; for this reason, FCS proves to be a useful and efficient replacement in many cases.

4 | S P RI N G M O D E L 'S F OR M U L A TI O N

Flexural behavior in cantilever coupling elements is dominant since the axial deformation is negligible when adjacent members carry the same loads.
By this assumption, vertical constraining elements can be replaced with equivalent torsional springs shown in Figure 5.
To satisfy equilibrium in nodes B and C, an axial load is generated to balance shear forces at the extreme ends of vertical elements that will be
carried by constrained parts of the cantilever beam (AB and BC) in the absence of a rigid diaphragm. With the assumption that a rigid diaphragm
exists and due to the relatively higher length of cantilever beams, axial and shear deformations are negligible and total energy of the subassemblage
(U) may be expressed as Equation 5:
   
U ¼ Ucb þ USpring 1 þ USpring 2
L ½MðxÞ2 1 1 L ½MðxÞ
2
1 M1 2 1 M2 2 (5)
¼ ∫0 dx þ k1 θ1 2 þ k2 θ2 2 ¼ ∫0 dx þ þ ;
2EIcb 2 2 2EIcb 2 k1 2 k2

(a) (b) (c)

(d) (e) (f)


FIGURE 4 Two‐dimensional frames with (a) simple and (b)–(f) framed cantilevers
6 of 20 SAJEDI ET AL.

FIGURE 5 (a) Framed cantilever's substructure, (b) simplified spring model, (c) bending moment diagram

where k1 and k2 = torsional spring's stiffness, α1 and α2 = relative spring's location, θ1 and θ2 = spring's rotation, E = modulus of elasticity, and
Icb = cantilever beam's moment of inertia.
Using least energy method as a result of Castigliano's second theorem, spring moments M1 and M2 can be derived by solving the following
system of equations:

8
> ∂U α1 Lcb ½PLcb ðα1 −2Þ þ 2ðM1 þ M2 Þ M1
>
< ¼ þ ¼ 0;
∂M1 4EIcb k1
>
> ∂U ¼ Lcb ½Pα2 Lcb ðα2 −2Þ þ 2ðα1 M1 þ α2 M2 Þ þ M2 ¼ 0;
:
∂M2 4EIcb k2

−Pα1 k1 Lcb 2 ½EIcb ðα1 −2Þ þ α2 k2 Lcb ðα1 −α2 Þ


M1 ðα1 ; α2 Þ ¼ h i; (6)
2 ðEIcb Þ2 þ EIcb Lcb ðα1 k1 þ α2 k2 Þ−α1 k1 k2 Lcb 2 ðα1 −α2 Þ

−k2 PLcb 2 ½EIcb α2 ðα2 −2Þ−α1 k1 Lcb ½α1 ðα1 −2Þ−α2 ðα2 −2Þ
M2 ðα1 ; α2 Þ ¼ h i; (7)
2 ðEIcb Þ2 þ EIcb Lcb ðα1 k1 þ α2 k2 Þ−α1 k1 k2 Lcb 2 ðα1 −α2 Þ

MA ¼ PLcb −M1 −M2 : (8)

Maximum deflection at the tip of the cantilever by moment‐area's second theorem can be written as

L MðxÞ
δD ¼ δD=A ðα1 ; α2 Þ ¼ ∫0cb ðLcb −xÞdx
EIcb
(9)
2PLcb 3 þ 3Lcb 2 ½α1 M1 ðα1 −2Þ þ α2 M2 ðα2 −2Þ
→ δD ¼ :
6EIcb

In specific cases where k1 ≠ 0 and k2 = 0, a closed‐form polynomial equation can be obtained by finding the root of Equations 10a and 10b pre-
sented in Table 1 which provides the optimum location of the spring (cantilever coupling element) to minimize deflection in different load patterns.
To investigate frame action in FCS arrangements, three beams shown in Figure 6 have been studied. All three have one fixed support and a free
end. CB1 will be the basis structure as a conventional cantilever beam. CB2 and CB3 are simplified models of FCS, in which cantilever coupling ele-
ments are replaced with equivalent torsional springs with different arrangements. Both springs are assumed to have same stiffness (k), which are
merged into a single one with 2k stiffness in CB3.
SAJEDI ET AL. 7 of 20

TABLE 1 Optimization equations for coupling element's location to minimize deflection values
Uniformly distributed load Concentrated point tip load

dδD ðαÞ dδD ðαÞ


¼ Aopt α4 þ Bopt α3 þ Copt α2 þ Dopt α þ E ¼ 0 (10a) ¼ Aopt α3 þ Bopt α2 þ Copt α−1 ¼ 0 (10b)
dα dα
5 9 1
Aopt ¼ Icb Lcce ; Bopt ¼ ðIcb Lcce −18Icce Lcb Þ; Aopt ¼ − EIcce Lcb ; Bopt ¼ − ðEIcb Lcce −12EIcce Lcb Þ;
 24    4 2
5 27 9 4 3
Copt ¼ − Icb Lcce − Icce Lcb ; Dopt ¼ Icb Lcce − Icce Lcb ; Copt ¼ ðIcb Lcce −2Icce Lcb Þ
6 5 8 3 2
1
Eopt ¼ − Icb Lcce
2

FIGURE 6 Spring model of cantilever coupling beams

Equations 5–10b present useful information about the efficiency of cantilever coupling element's location (αi) and stiffness constant (EIcce/Lcce).
By using beam element's stiffness matrix, equivalent spring's stiffness (ki) can be presented as follows:

6EIcce EIcb
ki ¼ ¼n ; (11)
Lcce Lcb

where n = constant number representing relative spring's stiffness.


Due to the complexity of equations mentioned before, finding a closed‐form solution for optimum location of springs is not practical in most
cases; hence, numerical methods with computer programming are utilized to find the optimum location of springs for each criterion. To do so,
maximum moment and deflection values for all three cases have been calculated while α1, α2, α3, and n are set as variables. The output results
of analysis have been normalized by division of each result in CB2 and CB3 over their corresponding value in CB1 so that the normalized information
presented in Figures 7 and 8 can be generalized to every structural properties and length. The essential key point in Figure 7 is that the ideal
position for springs varies for deflection and bending moment optimization, depending on the value of constant n; however, in both structures,
optimum location converges to a unified answer (α1 = .4, α2 = .8, and α3 = .67) for n > 500. The efficiency of employing cantilever coupling elements
for the three structures has been demonstrated in Figure 8.
Graphs summarized in Figure 8 demonstrate FCS's level of structural efficiency to the extent that it can reduce the amount of bending moment
and maximum deflection respectively up to 80% and 94%. In other words, for n > 500, cantilever beams will experience infinitesimal rotations at the
location of coupling element's connection, which causes a great reduction in design demands and displacements. Better performance of two
torsional springs compared to single spring with twice the magnitude of stiffness is also apparent in the results by approximately 13% less bending
moment and 7% more reduction in deflection. Current results, obtained from simplified models, indicate that coupling elements with separate
locations might provide elevated stiffness for the extended structure against vertical loads.

5 | DESIGN VERIFICATION

In the previous section, structural advantages of FCS in reducing demands of cantilever beams were discussed. One of the benefits produced by
FCS could be lighter structural components over the conventional way of supporting a beam in merely one end. This claim has been investigated
in two stages. First, a brute force algorithm has been adopted to find the optimized sections for a simple single‐story cantilever frame with absolute
fixed supports. Design guide charts and graphs are presented as well. Furthermore, the authors have utilized a design optimization method based
on genetic algorithm for multiple 2‐D steel moment frames with and without FCS modifications to compare the results.
8 of 20 SAJEDI ET AL.

FIGURE 7 Optimum location for each spring to minimize maximum demand in CB2 and CB3

1
Demand Reduction Ratio

0.8
MR2
0.6
DR2
0.4
MR3
0.2 DR3

0
0.05 1 20 400
n

FIGURE 8 Demand reduction in framed cantilever system based on variation in n values, where MR2 ¼ MCB2 =MCB1 ; DR2 ¼ δCB2 =δCB1 ,
MR3 ¼ MCB3 =MCB1 ; DR3 ¼ δCB3 =δCB1 , MCBi ¼ maximum bending moment through beam' s length, δCBi ¼ maximum deflection at free end of the beam

5.1 | Single‐story cantilever coupling


Due to the relative higher stiffness and strength of columns, it is safe to assume that beams have perfectly fixed support. Figure 9 shows two single
(a) and framed (b) steel cantilevers. Constant parameters of both structures are dead static load, DL = 50 kN/m2; live load, LL = 30 kN/m2; material
density, ρ = 7850 kg/m2; span length, Lcb = 5 m; story height, Lcce = 3.3 m; width of tributary area, WTB = 8 m; modulus of elasticity, E = 2 × 108 kN/
m2; and steel's yield stress limit, Fy = 2.35 × 105 kN/m2. Steel design will be in accordance with common codes of practice using Load and Resis-
tance Factor Design (LRFD) method.[7,14,15]
A standard database of 42 hot‐rolled European IPE and IPB (HE‐B) sections has been used in this study. Similar section properties will be
assigned to AC and DE elements. To find the exactly best‐designed sections, all possible choices for each element have been investigated by a com-
puter program developed by the authors in MATLAB[29] since the number of possible solutions is limited to 422. It should be noted that even
though some commercial computer programs conduct design procedure, there is no guarantee for the design to be the most optimum. In addition,
an efficient computer code can be used as a strong tool to evaluate and build the thousands of structural models. The optimum (lightest) structural
design has been reported for different values of α with .05 increments. Results are presented in Table 2.
It can be observed that proper selection of cantilever beam and coupling element's section can reduce the total weight of the substructure by 40%.
The same procedure has been repeated for different span lengths, and the results are summarized in Figure 10. In cases where Lcb/Lcce > 1, FCS has
shown promising results but for span length less than story's height, separated and stronger cantilever beams result in lighter structural weight.
More importantly, using framed cantilevers made it possible to design up to 12‐m span lengths while the strongest IPB section cannot solely
satisfy code requirements in spans larger than 9 m. As mentioned in the previous section, the interaction between moment and axial force
reduces beam capacity and the optimum values in Figure 10 are slightly less than what is obtained from Figure 7 diagrams. In other words, can-
tilever coupling elements will be closer to supports in order to reduce critical bending moment and provide axial force capacity in P‐M interaction.

5.2 | FCS performance in 2‐D steel moment frames


In the previous subsection, FCS was modeled with the assumption that cantilever beams acquire rigid supporting restraints. This hypothesis,
though suitable for single‐story cantilever coupling and preliminary design, might not be accurate in some cases since supporting structural
subassemblage does not provide perfectly rigid support for the cantilever beam. This part of the article investigates the efficiency of FCS for
the six different arrangements of framed cantilevers in an eight‐story steel moment frame (see Figure 4). The main purpose of this investigation
SAJEDI ET AL. 9 of 20

FIGURE 9 Two‐dimensional (a) separated and (b) framed cantilevers evaluated for structural efficiency

TABLE 2 Structural weight and design results


CB CCE Equivalent Icb Icce Structural Relative cost
Structural assemblage section section load (kg/cm) (cm4) (cm4) α weight (kg) of FCS structure

Simple cantilever beam (a) IPB 500 – 55.9 10,7200 – – 1,870 –


Optimum arrangement IPE 450 IPE 550 55. 33,740 67,120 0.35 1,126 0.60
Optimum framed cantileverdesign for IPE 600 IPE 500 55.8 92,080 48,200 0.1 1,519 0.81
different α values (b) IPE 600 IPE 330 55.5 92,080 11,770 0.2 1,382 0.74
IPE 500 IPE 500 55.5 48,200 48,200 0.3 1,206 0.65
IPE 500 IPE 450 55.4 48,200 33,740 0.4 1,163 0.62
IPE 550 IPE 400 55.5 67,120 23,130 0.5 1,279 0.68
IPE 600 IPE 220 55.4 92,080 2,770 0.6 1,306 0.70
IPE 600 IPE 200 55.4 92,080 1,940 0.7 1,294 0.69
IPE 600 IPE 200 55.4 92,080 1,940 0.8 1,294 0.69
IPE 600 IPE 200 55.4 92,080 1,940 0.9 1,294 0.69
IPE 600 IPE 200 55.4 92,080 1,940 1.0 1,294 0.69

Note. FCS = framed cantilever system.

is to evaluate the efficiency of cantilever coupling strategy in realistic and practical case studies. All structural elements have been proportioned to
perform satisfactorily on both seismic and gravity load combinations in the way that will be explained furthermore. Structural model's properties
have been defined as DL = 50 kN/m2, LL = 20 kN/m2 for each middle span, and 30 kN/m2 for cantilevers, WTB = 8 m; steel material and available
design sections will be the same as previous. Dimensions are shown typically for Frame A depicted in Figure 11; however, all six models share the
general model properties. Joints in each story are assumed to be horizontally constrained by a rigid diaphragm. Design and loading criteria are sim-
ilar to the previous section, adopted for intermediate steel moment frames.
10 of 20 SAJEDI ET AL.

FIGURE 10 Variation of relative cost in framed cantilever system for different span lengths and α values

FIGURE 11 General modeling parameters

At this stage, structural elements have been proportioned using a discrete binary genetic algorithm (GA) that is based on Darwinian natural
selection and survival of the fittest. In the field of structural optimization, GA application has proven to be successful.[16,17] The ability to define
discrete design variables and applying multiple constraints on a single‐objective function makes GA an ideal design strategy over many other opti-
mization methods. The authors, based on the concepts of the simple genetic algorithm,[18] have developed a computer program named GAD.[19–21]
Optimization's objective function and GA operators are explained briefly in what follows:

5.2.1 | GA formulation
Because cost is mostly a function of total structural weight, the optimization problem can be expressed as

Nelem
Minimize C ¼ ∑ ρi Ai Li ; (12)
i¼1

subject to
fPMRu g≤1;
fV u g≤fV all g;
fΔu g≤fΔall g;
fDu g≤fDall g;
SAJEDI ET AL. 11 of 20

where ρi, Ai, and Li = material density, area cross section, and length of element i, respectively, PMR = moment and axial force interaction vector (AISC
360‐10 equation H1‐1), V and Δ respectively contain values of shear and deflection, and D represents story drift vector. Indices u and all represent
ultimate and allowable values for each criterion, and Nelem = total number of beam‐column elements in the frame. A penalty coefficient (ψ) has been
utilized to include multiple constraint conditions, which is dependent on demand/capacity ratio (DCR) of each element and can be defined as

4
ψi ¼ ∏ ψi; j ; (13)
j¼1

(  2
1 þ K p DCRi; j −1 if DCRi; j > 1;
ψi; j ¼ (14)
1 if DCRi; j ≤ 1;

where j = PMR, V, Δ, and drift criteria. Because each beam and column's stiffness in a specified story affects the drift parameter, all the mentioned
elements will be penalized with the same ψj value, Kp = penalty constant set as 5.
Utilizing Equation 13, optimization problem can be modified to a problem with single‐objective cost function:

Nelem
Minimize C ¼ ∑ ψi ρi Ai Li : (15)
i¼1

Detailed weight report for Frames A to E has been reported in Table 3. Final results as predicted show promising results for application of FCS
as an alternative vertical load bearing system. Total structural weight has been reduced by up to 43% for projected part of the structure and 23% as
a whole. Designed cantilever beam sections in FCS often have considerably less depth compared to case A because frame action provides
significant vertical stiffness and strength. Structural analysis results for modified cantilevers in the these eight‐story frames indicate that FCS's
critical design criterion, unlike simple cantilevers, is strength in most cases. By this assumption, cantilever beams and coupling elements with greater
bending strength (or section modulus) and less section height are more suitable especially for architectural purposes, in which beam‐column
element's size is restricted. Such characteristics can be easily achieved by H‐shaped or built‐up wide flange steel sections. Arrangements B and
D seem to be the most cost‐efficient solutions although architect will make the final call on coupling elements' arrangement. Relative reduction
in weight of required steel material compared to frame a can also be observed in Table 3.
Rigid support assumption might not be viable for beams located in higher stories because lighter beam and column sections are proportioned at
these elevations; however, using graphs and equations provided by spring models can be still helpful for preliminary design stages. Based on these
investigations, FCS arrangements such as F require stronger and stiffer sections for higher elevation cantilevers in building structures. Also, as
previously mentioned, story joints are constrained together and cantilever beams do not bear any axial loads in FCS modeling. For this reason,
diaphragms should be controlled as well.

6 | NONLINEAR BEHAVIOR

A brief summary of the nonlinear behavior of cantilevers was presented in Section 1. Some characteristics of FCS and separate cantilevers cannot
be studied without considering what happens beyond the elastic state. This section investigates cantilevers with two different approaches. In the
first stage of nonlinear analyses, a pushdown procedure will be used to evaluate nonlinear load capacity of framed and simple cantilevers as a
criterion for vertical progressive collapse. Furthermore, with a series of nonlinear time‐history analyses, strength and service limit states will be
investigated using fragility curves. This stage of the article provides a significant database, in which structural behavior will be evaluated most
accurately to provide a better perspective and to understand the interactions between FCS and the main structure.

TABLE 3 Detailed GAD design report


Steel material's weight (kg) Reduction in steel material
Frame Structure Cantilever coupling elements Cantilever beams FCS Frame Projected structure

A 49011 0 14762 – – –
B 37571 3074 5328 8384 23.3% 43.2%
C 39994 2131 8497 9980 18.4% 32.4%
D 39742 2114 7360 9208 18.9% 37.6%
E 39355 2746 7107 9853 19.7% 33.3%
F 38646 2449 7012 9461 21.1% 35.9%

Note. FCS = framed cantilever system.


12 of 20 SAJEDI ET AL.

6.1 | Pushdown capacity curves


Aside from cost advantages, FCS can be designed to provide a more efficient performance in extreme loading conditions such as intensive vertical
ground motions and blast loads; on the other hand, progressive collapse is a disastrous structural problem for cantilever systems because of its
determinate nature. For example, local structural failures might damage other members (in this case, lower cantilevers). It is believed that FCS
arrangements provide a greater energy capacity for such unfortunate events due to their higher level of redundancy. This statement can be inves-
tigated using a pushdown analysis by following the formation of plastic hinges. Pushdown is widely used to predict the progressive collapse with
different mechanisms in structures.[22–24] This method is a simplified nonlinear static analysis similar to seismic pushover, but load and monitored
displacement definitions are characterized differently and in accordance with general FCS behavior. Monitored displacements are set to be at the
tip of cantilevers and total shear force will be the sum of shear in all cantilever beams. The nonlinear static procedure has been performed by
assigning plastic hinges at both ends of all frame elements based on the definition of backbone curve for different structural steel elements.[25]
Frame A, which was previously designed, is assumed to be the benchmark structure, and coupling elements with unified sections are added and
located at the tip of cantilevers. Stiffness values for vertical coupling beams were gradually increased while coupling elements get the same section
as cantilever sections remained the same in each model. FCS's nonlinear performance regarding progressive collapse was evaluated by the maxi-
mum expected shear strength (MESS), elastic (KE) and secant stiffness (KS), which were extracted from the pushdown capacity curves shown in
Figure 12. By evaluating the rates of enhanced behavior (ΔMESS, ΔKE, and ΔKS) compared to the increase in section properties, the efficiency of
frame behavior is evident (See Table 4). Due to the fact that common design procedure leads to smaller column sections in higher stories, the stiff-
ness of supporting columns is reduced, which will cause increase in deflection values of cantilevers. In order to overcome such problems in long
cantilevers, higher moment of inertia is required for cantilevers, which lead to stronger section properties as well. This phenomenon explains slight
multistep drops presented in the pushdown curves, in which every sudden drop implicitly indicates the collapse of a cantilever beam. In other
words, after the first drop in the benchmark structure, total collapse might be inevitable. In contrast, while coupling cantilevers, multiple plastic
hinge formations in horizontal and vertical elements offer considerable energy dissipation capacity before the total structural collapse. Since the
addition of coupling elements might increase total costs, principal design objectives shall prioritize between economic efficiency and better energy
capacity for FCS arrangements as the dominant proportioning criteria.

6.2 | IDA and fragility curves


The main purpose of this article is to demonstrate the real potential in coupling successive cantilevers. Through advanced means of analysis, all the
five frames that have been proportioned with regard to FCS arrangements offer lighter design compared to model A while satisfying common code

FIGURE 12 Capacity curves for steel moment Frames A to F subjected to pushdown analysis

TABLE 4 Pushdown analyses summary


CB section CCE section Icce/Icb Zcce/Zcb MESS (kN) ΔMESS % KE (kN/m) ΔKE % KS (kN/m) ΔKS %

HE500B – – – 2,358 – 29,556 – 15,420 –


HE500B HE100B 0.4% 2.1% 2,453 4.0% 31,711 7.3% 16,374 6.2%
HE500B HE140B 1.4% 5.1% 2,571 9.0% 33,980 15.0% 17,195 11.5%
HE500B HE200B 5.3% 13.3% 2,907 23.3% 40,453 36.9% 19,549 26.8%
HE500B HE240B 10.5% 21.7% 3,228 36.9% 45,773 54.9% 23,113 49.9%
HE500B HE300B 23.5% 38.3% 3,861 63.7% 52,753 78.5% 30,351 96.8%
HE500B HE340B 34.2% 49.6% 4,269 81.0% 55,976 89.4% 34,059 120.9%
HE500B HE400B 53.8% 66.8% 4,565 93.6% 59,518 101.4% 36,635 137.6%
HE500B HE450B 74.5% 82.5% 4,676 98.3% 62,062 110.0% 37,035 140.2%
HE500B HE500B 100.0% 100.0% 4,835 105.0% 63,719 115.6% 35,083 127.5%
SAJEDI ET AL. 13 of 20

requirements. However, as discussed in Section 1, cantilevers may not meet the desired performance while plastic hinges form in the main struc-
tural components. In addition, common engineering practice considers horizontal cantilevers as elements that only contribute to the vertical load
bearing of any building structure. One of the most significant characteristics of FCS is to include cantilevers in lateral seismic resistance as well as
vertical. In other words, coupled beams have the potential to increase story's stiffness and reduce drift values as it was mentioned before.
Conducting multiple nonlinear time‐history analyses, all these issues could be properly investigated. Doing so requires an appropriate method of
evaluation. IDA is one of the most advanced and state‐of‐the‐art methods to study structural performance and reliability.[26] This method utilizes
a combination of nonlinear dynamic analyses under a multiply scaled suite of ground motions.
In order to make a sensible comparison for all six frames, the main structural framework, which consists of seismic beams and columns, has
been set the same and the six models differ merely in the design of cantilevers or FCS components. Employment of the coupling elements will
reduce the bending moment caused by cantilevers in the support connection and could possibly reduce the size of supporting side columns;
however, to monitor the variations in lateral drift caused by FCS modification, main structural components were assumed to be similar for all
models. General section assignments for all the six frames that will be subject to IDAs have been presented in Appendix A. Due to the fact that
each earthquake represents unique frequency content, proper selection of ground motion records is of utmost importance to interpret the results.
A set of 35 ground motion data including 15 far‐field and 20 near‐field record data has been chosen based on the recommendations made by Fed-
eral Emergency Management Agency (FEMA) document on the quantification of building seismic performance factors.[27] Time‐history analyses[28]
presented here include several historic ground motion records including different data stations for Northridge, Loma Prieta, and Kobe events.
These ground motion records include several frequency contents to evaluate the structure and hanging members in different acceleration magni-
tudes. In addition, vertical component of ground acceleration for each event is considered in time‐history analyses to better monitor the behavior
of structure, during and after each analysis.
It should be noted that coupling elements have been modeled as nonlinear beam‐column elements to monitor axial load effects; however,
nonlinear analyses demonstrated that coupling elements as previously mentioned will not carry much axial loads and represent dominant flexural
behavior. A summary of record characteristics has been expressed in Appendix B. Record data has been obtained from NGA‐West2 Online data-
base.[29] According to the recommendations proposed by Vamvatsikos and Cornell, three criteria have been adopted to consider total structural
collapse.[30] Maximum story drift of 10%, dynamic instability of models, and the point where local tangent reaches 20% of the elastic slope will
be considered as limit states where IDA curves reach the flat line. In accordance with several provisions for structural seismic regulations,[25,31,32]
the seismic performance level of structures could be assessed considering maximum interstory drifts shown in Table 5.
Probability of exceedance (POE) from a specific damage level (DMi) can be calculated with the logarithmic normal distribution function (Equa-
tion 16). Figure 13 displays IDA curves for 35 record sets with respect to the vertical ground motion effects. Fragility curves for different seismic
performance levels of frame A is also demonstrated in Figure 14. Graphs fit properly to the extracted data from IDA curves and justify the number
of record sets adopted.

TABLE 5 Seismic performance levels limit states

Performance level OPa IOb DCc LSd CPe

Maximum interstory drift % 0.5% 1.0% 1.5% 2.0% 2.5%


a
Operational.
b
Immediate occupancy.
c
Damage control.
d
Life safety.
e
Collapse prevention.

FIGURE 13 Frame A, incremental dynamic analysis graphs. PGA = peak ground acceleration
14 of 20 SAJEDI ET AL.

FIGURE 14 Frame A, fragility curves (story drift). PGA = peak ground acceleration

P½DMi jPGA ¼ Φ½LnðPGAÞ−μ=σ (16)

6.2.1 | Cantilevers' performance


This section presents graphs and diagrams that compare deflection values of extended parts in six models. Two different limit states have been
considered for cantilevers (Lcb/120 and Lcb/240). Maximum and average values for the critical deflection of cantilevers at each side of the frames
have been monitored, and the results are expressed in Figure 15.
Table 6 represents POE values obtained from previous graphs. It is clear that FCS behavior as expected, due to a significant level of
redundancy, offers better performance against vertical deflections.

6.2.2 | Lateral seismic performance


Evaluating the impact of FCS behavior on lateral load resistance of the main structure is of great importance. By the first look, one might consider
vertical coupling elements as columns or axial force collectors while in contrast, they behave in a completely different manner. These auxiliary
members contribute to reducing (redistributing) flexural moments in cantilever beams that increase the stiffness of the projected structure by
inducing a multiple curvature deformation pattern. This reduction in demand will cause the neighboring beam in cantilever joint to experience less
gravity moments (or the moment caused by any vertical excitations). As a result, it can be expected that using same body structure, projected
structures that utilize FCS might increase structural resistance while being subjected to lateral loads such as earthquake. In addition, separate
cantilevers, will not contribute to story's lateral stiffness while in contrast, coupled cantilevers will resist against such deformations as shown in
Equation 4. In a nutshell, it could be predicted that framed cantilevers (B‐F) will offer better seismic performance compared to the commonly con-
structed cantilevers (A). To study the interaction between the main body and extended structure, dynamic analyses explained before are exploited
to extract fragility curves regarding lateral story drifts. Damage levels including Operational (OP), Immediate Occupancy (IO), Damage Control (DC),
Life Safety (LS), and Collapse Preventions (CP) have been used as limit states to evaluate total structural performance under seismic loads with uni-
formly scaled ground accelerations. Table 7 summarizes the results from fragility curves in Figure 16 and proposes numeric data for drift assess-
ment of all six models in three peak ground acceleration values. As expected, all FCS arrangements represent superior seismic behavior in terms
of lateral drift. Frames E and C also offer best performance in these case studies.
Nonlinear dynamic analysis performed in different increments of maximum scaled acceleration indicate better performance of framed
cantilevers compared to the conventional method. It should be noted that all models subjected to time histories were designed to comply with
code; however, coupled cantilevers offer modest enhanced performance in terms of lateral drift and significant improvement in restraining vertical
deflection of cantilevers as demonstrated in damage parameters (Tables 6 and 7).

7 | S P E C I A L RE M A R K S

This research tends to investigate the problems associated with building structures with long‐span cantilevers. Considering the various
architectural requirements, hanging and the main body structure should be able to provide desired service and ultimate performance levels. Hence,
structural engineers should consider nonlinear response of buildings beyond most code prescriptions. In addition, a thorough investigation of
structural components under combined vertical and horizontal ground motion excitements as well as possible wind load effects might be necessary.
It should be noted that cantilevered components with or without utilizing FCS should be designed considering the interactions with the main body,
especially when structures are experiencing nonlinear deformations.
SAJEDI ET AL. 15 of 20

1 1

0.8 0.8

0.6 0.6
POE

POE
Frame A
0.4 0.4 Frame B
Frame C
Frame D
0.2 0.2 Frame E
Frame F
0 0
0 0.1 0.2 0.3 0.4 0 0.2 0.4 0.6 0.8 1
PGA PGA
(a) Lcb/240 limit-state, Maximum critical deflection measure (b) L cb/120 limit-state, Maximum critical deflection measure
(All Frames) (All Frames)
1 1

0.8 0.8
Frame A-Max L
0.6 0.6 Frame A-Mean L

POE
POE

Frame A-Max R
0.4 0.4 Frame A-Mean R
Frame E-Max L
Frame E-Mean L
0.2 0.2 Frame E-Max R
Frame E-Mean R
0 0
0 0.1 0.2 0.3 0.4 0 0.2 0.4 0.6 0.8 1 1.2
PGA PGA
(c) Lcb/240 limit-state, Mean and maximum critical deflection (d) Lcb/120 limit-state, Mean and maximum critical deflection
measure (Frames A&E) measure (Frame A&E)

FIGURE 15 Fragility curves for cantilevers deflection criteria. PGA = peak ground acceleration

TABLE 6 Sample POE (%) values for deflection criteria


Limit state
Frame ID Lcb/240|PGA = 0.1 Lcb/120|PGA = 0.35

A 68.5 72.3
B 45.3 64.3
C 44.7 58.0
D 55.7 69.3
E 44.5 57.9
F 55.0 72.6

Note. PGA = peak ground acceleration.

For the purpose of design, utilizing normalized graphs shown in Figures 7 and 8 could be helpful in preliminary design stages, giving an insight
about optimum location and stiffness of coupling elements. Structural engineer might also reconsider in the relative structural stiffness of
beams and columns at the supporting joints to reduce demands on cantilevers and coupling elements. An iterative procedure might be required
to obtain optimum results. In the end, hanging parts of the structure should also be evaluated for seismic effects. Additional considerations such
as capacity design approaches might be required for certain design objectives. Pushdown method could be a suitable tool for initial estimations
of FCS capacity in this regard.
Construction sequence could also affect member forces and deformations of these types of buildings like many others. A staged construction
method could be considered in this case to eliminate a proportion of internal forces in cantilevered elements due to column shortening effects in
gravity loads. Models included in this study did not experience significant change from this phenomenon; however, this might not be always
the case, especially in high‐rise construction. Coupling cantilevers after that the main frame has been erected might be an appropriate solution
in such cases.
16 of 20 SAJEDI ET AL.

TABLE 7 Sample POE values (%)—drift criteria


PGA = 0.25 g PGA = 0.35 g PGA = 0.45 g
Frame ID OP IO DC LS CP OP IO DC LS CP OP IO DC LS CP

A 62.7 12.1 2.1 0.8 0.4 85.4 33.3 10.1 4.8 2.8 94.5 54.7 23.7 13.2 8.7
B 60.0 10.5 2.1 0.8 0.3 83.4 29.9 9.5 4.5 2.4 93.4 50.6 22.0 12.3 7.4
C 60.2 10.5 1.6 0.4 0.2 83.6 29.7 7.8 2.9 1.4 93.5 50.3 19.0 9.0 5.0
D 62.4 11.6 2.1 0.8 0.3 85.4 32.3 9.8 4.8 2.4 94.6 53.7 23.0 13.0 7.6
E 58.1 9.2 1.4 0.4 0.1 82.2 27.4 7.3 2.7 1.3 92.8 47.6 18.1 8.7 4.7
F 61.8 11.1 2.1 0.9 0.4 84.9 31.6 10.2 5.1 2.8 94.3 53.0 24.0 13.7 8.6

Note. PGA = peak ground acceleration.

1 1

0.8 0.8
Frame A-OP
Frame A-IO 0.6
0.6 Frame A-DC

POE
POE

Frame A-LS
Frame A-CP Frame A
0.4 0.4 Frame B
Frame B -OP
Frame B-IO Frame C
0.2 Feame B-DC 0.2 Frame D
Frame B-LS Frame E
Frame B-CP Frame F
0 0
0 0.5 1 1.5 2 2.5 0 0.1 0.2 0.3 0.4 0.5 0.6
PGA PGA
(a) Frames A&B performance comparison (b) Operational (OP)
1 1

0.8 0.8

0.6 0.6
POE

POE

Frame A
0.4 0.4 Frame B
Frame C
Frame D
0.2 0.2 Frame E
Frame F
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.2 0.4 0.6 0.8 1 1.2 1.4
PGA PGA
(c) Immediate occupancy (IO) (d) Damage control (DC)
1 1

0.8 0.8

0.6 0.6
POE

POE

Frame A
Frame B
0.4 0.4
Frame C
Frame D
0.2 0.2 Frame E
Frame F
0 0
0 0.3 0.6 0.9 1.2 1.5 1.8 0 0.3 0.6 0.9 1.2 1.5 1.8 2.1
PGA PGA
(e) Life safety (LS) (f) Collapse prevention (CP)

FIGURE 16 Fragility curves for lateral drift limit states. PGA = peak ground acceleration
SAJEDI ET AL. 17 of 20

8 | S U M M A R Y A N D C O N C L U S I ON

This article investigated the performance of long‐span cantilever beams with accurate means of analysis and design to show certain limitations
proposed by common codes. To overcome certain challenges in the design of conventional cantilevers, a structural system named as FCS has been
introduced as an alternative to design extended part of building structures. The following conclusions can be derived by studying multiple factors in
terms of formulation, economic efficiency, and nonlinear behavior:

1. Unlike conventional engineering practice that leads to cantilever design regardless of the inelastic supporting structure's deformations, forma-
tion of plastic hinges in neighboring beams and columns under lateral loads can make cantilevers undergo excessive deflection values. The
design of such structural elements, especially for longer spans, requires further elaboration in monitoring the nonlinear behavior of the
structure. Additional code requirements might be necessary to ascertain satisfying cantilever's performance in different service design levels.
2. Due to the higher level of redundancy, framed cantilevers can be connected with hinged connections to the supporting structure in contrast to
the definition of conventional cantilever beams. This will benefit structural engineers to design cantilevered and projected extensions regard-
less of detailing limitations for moment connection of cantilever to the supporting column.
3. In FCS arrangements with rigid supports, for single‐story coupled cantilevers, values of maximum flexural demands and deflection could be
reduced up to 80% and 94%, respectively, depending on the location of the coupling element and relative stiffness of members.

4. Design procedure with a brute‐force algorithm indicates a 40% reduction in total weight of designed projected structure with rigid supporting
restraints and single‐story coupled cantilevers depicted in Figure 9b. The most economic arrangement happens where α = .35.
5. Optimized design for eight‐story intermediate steel moment frames suggests up to 43% reduction in weight of the cantilevered system and
23% less in total weight. All the arrangements represent more economic design compared to the conventional methods.

6. In the last stage of the investigation, capacity curves for multiple arrangements have been extracted through nonlinear static pushdown anal-
yses. It can be observed that in unexpected and extreme load conditions such as intensive near‐field vertical ground motions or progressive
collapse, framed cantilevers offer significantly more energy capacity compared to separate cantilevers. Nevertheless, more predictable and
safer performance due to the structural redundancy is a characteristic of FCS structures.
7. Utilizing incremental dynamic analysis and obtaining fragility curves, the frames modified for FCS behavior offer superior behavior by lowering
the probability of exceedance in several limit states.
8. Using cantilever coupling elements not only improved vertical load resistance of projected structure in terms of cost and safety but also
contributed to the seismic performance of all five aforementioned frames against lateral loads. Continuous arrangements shown in Frames
C and E proved to be the most effective to reduce the values of maximum story drift.

Although even optimized FCS models presented better performance in nonlinear analyses compared to the conventional method, the
structural engineer will decide how to proportion FCS elements and maintain a balance among different design objectives including architectural
requirements, serviceability, cost efficiency, and progressive collapse criteria.

ORCID
Seyed Omid Sajedi http://orcid.org/0000-0002-4552-4794

RE FE R ENC E S
[1] L. Di Sarno, A. S. Elnashai, G. Manfredi, Eng. Struct. 2011, 33(5), 1514.
[2] B. K. Lee, T. E. Lee, Y. S. Jung, KSCE J. Civil Eng.s 2012, 16(1), 169.
[3] G. V. Rao, R. Narayanaswami, Comput. Struct. 1980, 12(6), 843.
[4] P. K. Roy, N. Ganesan, Comp. Struct. 1993, 49(2), 275.
[5] A. J. Stanley, N. Ganesan, Comput. Struct. 1994, 53(1), 209.
[6] I. Takewaki, Comput. Struct. 1997, 62(1), 107.
[7] American Society of Civil Engineers, SEI/ASCE 2010, 7.
[8] G. P. Luth, J. Perform. Constr. Facil.s 2000, 14(2), 51.
[9] J. S. O'Keefe, J. N. Kheir, M. L. Martin, L. F. Leslie, J. G. Neal, R. F. Edlich, J. Emerg. Med. 1999, 17(2), 293.
[10] N. Delatte, Eng. Struct. 2010, 32(7), 1952.
[11] C. Pearson, N. Delatte, J. Perform. Constr. Facil. 2005, 19(2), 172.
[12] S. R. Ahmed, M. R. Khan, K. M. S. Islam, M. W. Uddin, Comput. Struct. 1998, 69(3), 329.
[13] A. J. Papazoglou, A. S. Elnashai, J. Earthquake Eng. Struct. Dyn. 1996, 25(10), 1109.
[14] American institute of steel construction. "Seismic provisions for structural steel buildings". AISC/ANSI 341–10, Chicago (IL) 2010a.
[15] American institute of steel construction. "Specifications for structural steel buildings". AISC/ANSI 360–10, Chicago (IL) 2010b.
18 of 20 SAJEDI ET AL.

[16] C. Camp, S. Pezeshk, G. Cao, J. Struct. Eng. 1998, 124(5), 551.


[17] S. Rajeev, C. Krishnamoorthy, J. Struct. Eng. 1992, 118(5), 1233.
[18] D. E. Goldberg, Genetic Algorithms in Search, Optimization and Machine Learning, Addison‐Wesley Longman Publishing Co. 1989.
[19] S. O. Sajedi, S. R. Mirghaderi, F. Keshavarzi, "Nonlinear evaluation and optimization of frame‐type load bearing system for steel cantilevers with genetic
algorithm." Proc., 2nd International ISSS Conf., Tehran, Iran. 2015
[20] MATLAB [Computer software], MathWorks, Natick, MA.
[21] OpenSees [Computer Software]. Univ. of California, Berkeley, CA.
[22] K. Khandelwal, S. El‐Tawil, Eng. Struct. 2011, 33(9), 2653.
[23] M. Liu, A. Pirmoz, Eng. Struct. 2016, 123, 372.
[24] B. I. Song, H. Sezen, Eng. Struct. 2013, 56, 664.
[25] Federal emergency management agency. "Prestandard and commentary for the seismic rehabilitation of buildings", FEMA‐356, Washington D.C. 2000.
[26] D. Vamvatsikos, C. Cornell. "Seismic performance, capacity and reliability of structures as seen through incremental dynamic analysis." Stanford
University. 2002
[27] Federal emergency management agency. "Quantification of building seismic performance factors", FEMA‐P695, Washington D.C. 2009,
[28] Perfrom‐3D V6.0.0 [Computer software], Computers and Structures, CA.
[29] Pacific Earthquake Engineering Research Center (Peer). "NGA‐West2 ground motion database."< http://ngawest2.berkeley.edu> (Apr.1.2017) 2014
[30] D. Vamvatsikos, C. A. Cornell, Earthquake Eng. Struct. Dynam. 2002b, 31(3), 491.
[31] Federal emergency management agency. "NEHRP recommended provisions for seismic regulations for new buildings and other structures", FEMA‐450,
Washington D.C. 2003
[32] International code council. "International Building code", IBC‐2012. 2012

Seyed Omid Sajedi is a PhD student at University at Buffalo SUNY (USA) in earthquake and structural engineering. He received his BSc in civil
engineering in 2013 and his MSc in structural engineering in 2017 from University of Tehran (Iran). He is currently employed as a graduate
teaching assistant at University at Buffalo. His research interests include seismic analysis and design of steel and composite building structures,
computer‐aided design, and structural optimization.

Seyed Rasoul Mirghaderi is an associate professor at the University of Tehran. He received his BSc in civil engineering at Sharif University (Iran)
in 1977 and his MSc and PhD in structural engineering at Purdue University (USA) in 1981 and 1984, respectively. He has authored several
books on the subject of structural seismic design and has been the head of Iranian national code committee for steel construction, Iranian code
of practice for seismic‐resistant design of buildings. Dr Mirghaderi is currently the director of Iranian Society of Steel Structures and Depart-
ment of Structural Engineering Studies at BHRC. His research focuses on seismic analysis and design, innovative earthquake resistant systems
in composite and steel structures, structural retrofitting, and seismic evaluation of buildings.

Farhad Keshavarzi is a senior structural engineer. He received his B.Sc. in civil engineering at the University of KNTU in 2004 and his M.Sc. in
earthquake engineering at University of Tehran in 2008. He has been the head of structural design team for several high and mid‐rise buildings
in Iran. His areas of expertise include nonlinear structural evaluation, steel and composite structures, nonlinear analysis of steel connections
and structural seismic evaluation.

How to cite this article: Sajedi SO, Mirghaderi SR, Keshavarzi F. Frame‐type load bearing system for long‐span cantilevers. Struct Design
Tall Spec Build. 2018;e1469. https://doi.org/10.1002/tal.1469
SAJEDI ET AL. 19 of 20

APPENDIX A
SECTION ASSIGNMENTS FOR THE SIX FRAMES STUDIED FOR INCREMENTAL DYNAMIC ANALYSES

IPE550 IPE550 IPE550

IPE600

IPE600

IPE600

IPE600
IPE550 IPE550 IPE550

IPE600

IPE600

IPE600

IPE600
IPE500 IPE500 IPE500
IPE600

IPE600

IPE600

IPE600
Hanging Structure

Hanging Structure
IPE500 IPE500 IPE500
HE500B

HE500B

HE500B

HE500B
IPE600 IPE600 IPE600
HE600B

HE600B

HE600B

HE600B
IPE600 IPE600 IPE600
HE600B

HE600B

HE600B

HE600B
IPE600 IPE550 IPE600
HE600B

HE600B

HE600B

HE600B
IPE500 IPE500 IPE500
HE800B

HE800B

HE800B

HE800B

FIGURE A1.A Main frame section assignments

HE600B IPE550 IPE400 IPE450 IPE600 IPE500


HE100B
IPE500

IPE450

IPE330

IPE550

HE600B IPE360 IPE450 HE240B IPE360 IPE500


HE100B

HE120B
IPE360

IPE400

HE550B IPE400 IPE500 IPE600 IPE400 IPE450


IPE400

IPE180

IPE600

IPE550

IPE550

HE550B IPE360 HE240B IPE600 IPE450 IPE500


HE100B

HE120B
IPE400

HE550B IPE330 IPE600 IPE450 IPE450 IPE550


HE260B

HE260B
IPE600

IPE180

IPE400

HE500B IPE400 HE500B IPE600 IPE500 IPE600


HE140B

IPE160

IPE550

IPE500

HE500B IPE550 HE280B IPE600 IPE600 IPE450


HE260B
IPE550

IPE500

IPE100

IPE550

HE500B IPE550 HE550B IPE450 HE500B IPE500

Frame A Frame B Frame C Frame D Frame E Frame F


FIGURE A1.B FCS section assignments
20 of 20 SAJEDI ET AL.

APPENDIX B
G RO U N D M O TI O N DA T A BA S E
Table of ground motion record data

ID no. Event name RSNa V/H Record type

1 Northridge 953 0.73 Far field


2 Northridge 960 0.75 Far field
3 Hector Mine 1787 0.56 Far field
4 Imperial Valley 169 0.60 Far field
5 Imperial Valley 174 0.39 Far field
6 Kobe 1,111 0.80 Far field
7 Kobe 1,116 0.28 Far field
8 Landers 900 0.56 Far field
9 Landers 848 0.62 Far field
10 Loma Prieta 752 1.09 Far field
11 Manjil 1,633 1.05 Far field
12 Superstition Hills 721 0.36 Far field
13 Chi‐Chi 1,244 0.49 Far field
14 Chi‐Chi 1,485 0.75 Far field
15 San Fernando 68 0.73 Far field
16 Gazli 126 2.42 Near‐field NPRSb
17 Denali 2,114 0.72 Near‐field NPRS
18 Chi‐Chi 1,517 0.32 Near‐field NPRS
19 Chi‐Chi 1,504 0.47 Near‐field NPRS
20 Imperial Valley 160 0.89 Near‐field NPRS
21 Imperial Valley 165 0.80 Near‐field NPRS
22 Imperial Valley 181 4.24 Near‐field PRSc
23 Imperial Valley 182 1.70 Near‐field PRS
24 Irpinia 292 1.04 Near‐field PRS
25 Loma Prieta 802 0.77 Near‐field PRS
26 Erzican 821 0.61 Near‐field PRS
27 Landers 879 1.14 Near‐field PRS
28 Northridge 1,063 1.10 Near‐field PRS
29 Northridge 1,086 0.89 Near‐field PRS
30 Kocaeli 1,165 0.63 Near‐field PRS
31 Chi‐Chi 1,503 0.33 Near‐field PRS
32 Chi‐Chi 1,529 0.58 Near‐field PRS
33 Duzce 1,605 0.86 Near‐field PRS
34 Tabas 143 0.75 Near‐field PRS
35 Bam 4,040 1.20 Near‐field PRS
a
Record sequence number defined by Pacific Earthquake Engineering Research Center.
b
No pulse record subset.
c
Pulse record subset.

Potrebbero piacerti anche