Chapter 3.

Nonlinear Modeling and Methods of Analysis

Chapter 3

NONLINEAR MODELING AND METHODS OF ANALYSIS

3.1 Nonlinear Modeling

Nonlinear modeling of structural elements is the most important step in nonlinear

analysis. Hence, nonlinear modeling is discussed in the following subsequent sections.

3.1.1 Modeling of frame elements

Elements of space frame have two nodes with three DOF as illustrated in Fig. 3.1,
where Xm, Ym and Zm are member reference axes.

Fig. 3.1 Frame member

3.1.2 Modeling of shear walls

Before 1960, shear walls are modeled to account shear and flexure by analytical
methods. For practical purpose, researchers used the simplest approach. After 1960,
with the progress of advance computer technology, the use of precise modeling is in
progress. Modeling of the shear wall can be done by two approaches, i.e. macro and
micro. With the development of new software in due course of time, two approaches
have become simple and more accurate. For micro modeling finite element software
like ANSYS, ABAQUS, LS-DYNA, ATENA, ADINA utilizes a variety of elements
and material models from their libraries. In the present study, shear walls are modeled
utilizing macro modeling approach of a mid-pier frame member as recommended by
Fahjan Y.M. et al. (2010) and represented in Fig. 3.2.

24
 Chapter 3. Nonlinear Modeling and Methods of Analysis

Fig. 3.2 Mid-pier model for shear wall [Fahjan Y.M. et al. (2010)]

The stiffness matrix of a mid-pier element should consider the shear deformations. Fig.
3.3 shows a mid-pier element (wide column) with coordinates representing DOF and the
stiffness matrix for the mid-pier element is elaborated by the Eq. (3.1).

Fig. 3.3 Mid-pier element for modeling of shear wall

25
 Chapter 3. Nonlinear Modeling and Methods of Analysis

 EA 
 L 
 
 0 12 EI 
 L
3 
 
 0 6 EI EI 
4   0
1  L
2
L 
S
1      EA 0 0
EA

 (3.1)
 L L 
 12 EI 6 EI 12 EI 6 EI 
 0  3  0  2 
 
2 3
L L L L
 6 EI EI 6 EI EI 
 0 2   0  4   
 L
2 2
L L L
12 EI
where,   2
(3.2)
L GA R
The axial force value is determined as a combination of as (DL+0.25LL) for live load ≤
3kN/m2 and (DL+0.5LL) for live load > 3kN/m2.

3.1.3 Modeling of the finite size of joints

In frame-shear wall dual system, the junctions of the beam-shear wall have significant
size as presented in Fig. 3.4 a) and its effect cannot be neglected. The beam-shear wall
member can be exchanged by rigid end linear member as indicated in Fig. 3.4 b).

Fig. 3.4 a) Finite-size of the junction

Fig. 3.4 b) Modeling of the junction

26
 Chapter 3. Nonlinear Modeling and Methods of Analysis

These rigid portions have infinite stiffness of the rigid ends of an element which can be
determined as given in Eq. (3.3) and (3.4)

S  H T  S  H  (3.3)

1 dL 0 0 
0 1 0 0  (3.4)
  
H  
0 0 1 bL 
 
0 0 0 1 
This conversion is implemented in software SAP2000 by offset utility command.

3.1.4 Modeling of beam-column joints

During past earthquakes, beam-column junctions provided in RC frames were subjected

to substantial damage due to lack of design provisions to resist earthquake forces. In the
frame, performance behavior of beam-column junction subjected to lateral earthquake
forces is complex in nature, as a numerous factors influence the strength of the joints.
The shear resistance mechanism is appreciably varied for peripheral and central beam-
column junctions. Shear strength of RC beam-column joint is governed by the strength
of concrete, quantity of longitudinal rebars in beam connected to the column, joint
aspect ratio and axial force carried by the column. Fig. 3.5 indicates beam-column joint
model for effective stiffness specified by ASCE41-06 (2007). In the current study,
joints are considered as rigid.

a) ∑Mc/∑Mb = 0.8 b) 0.8 ≤ ∑Mc/∑Mb ≤ 1 c) ∑Mc/∑Mb> 1.2

Fig. 3.5 Beam-column joint model for effective stiffness

27
 Chapter 3. Nonlinear Modeling and Methods of Analysis

3.1.5 Modeling of floor

The large in-plane rigidity of floors and roof slab is simulated to vibrate as an entire
floor and act as a rigid body during deflection. In case of improper modeling, each node
vibrates independently may lead to erroneous vibration characteristics. If rigid floors are
not used in the modeling, distribution of member forces in frames cannot take place
properly and estimation of member forces will be inaccurate. The error is more
predominant in the structures having a high deviation in stiffness of columns and
asymmetric structures.

The rigidity to the floor can be assigned by using three DOF i.e. bi-directional
translation and rotation with respect to the common axis at all the nodes of the floor as
indicated in Fig. 3.6.

Fig. 3.6 Floor and DOF

The matrix [B] is a transformation matrix which relates nodal DOF u1, u2, u3 to floor
DOF U1, U2, U3 can be expressed as shown in Eq. (3.6).

 cos  sin  x sin   y cos  

 B    sin  cos  x cos   y sin  
(3.6)
 0 0 1 

In-plane rigidity of floors and roof slab can be adopted in SAP2000 by means of
diaphragm constraint utility.

3.2 Nonlinearity of Frame Elements

Two kinds of nonlinearity i.e. material and geometric are associated with frame
elements. In the case of material nonlinearity, the stress-strain behavior of the material

28
 Chapter 3. Nonlinear Modeling and Methods of Analysis

is nonlinear. While, in geometric nonlinearity, the relation between strain and

displacement is nonlinear and thus results in shape change.

3.2.1 Material nonlinearity

Material nonlinearity may be incorporated in SAP2000 by means of

1. Assigning nonlinear properties through link/support to an element,
2. Defining limits in frame elements for tension and compression and
3. Assigning plastic hinges to frame members.

In present work, material nonlinearity is considered by assigning plastic hinges to the

structural elements. Nonlinear stress-strain behavior of materials are included by using
Mander, Park and Simple model for confined concrete, rebar and structural steel as
indicated in Fig. 3.7 a), b), c) respectively. These stress-strain curves are readily
available in SAP2000 software library.

a) Mander‟s model for confined concrete b) Park‟s model for rebar

[Mander J.B. et al. (1988)] [Park and Paulay (1975)]

c) Simple curve for structural steel

[Yun and Gardner (2017)]
Fig. 3.7 Stress-strain curves

29
 Chapter 3. Nonlinear Modeling and Methods of Analysis

3.2.1.1 Hysteretic behavior

The RC structures, when exposed to the series of cyclic loads due to seismic excitation,
acquire a large increase in strain after elastic limit. Such behavior of the material is
known as the hysteretic response. The area enclosed by the hysteretic loop is the strain
energy dissipated during the loading and unloading cycle.

Nonlinear analysis exhibit load reversal under cyclic loadings may affect the hysteretic
behavior of materials. To account stiffness degradation, Kinematic model is used for
steel and Takeda model is used for concrete as shown in Fig. 3.8.

a) Kinematic model for steel b) Takeda model for concrete

Fig. 3.8 Hysteresis curves

Kinematic hysteresis model

This model is dependent on the kinematic hardening phenomenon which is normally

experienced in all types of metals, and it is the in-built hysteresis model in the software
program. The model dissipates a considerable amount of energy and also suitable for
ductile materials.

According to laws of kinematic hardening, plastic distortion occurs in one direction

extends the loop in another direction along with it. The identical sets of two ordinates
are aligned. No other parameters are necessary to define this model.

After unloading and during reverse cyclic loading, the loop pursues a direction followed
by sections parallel to as well as of a similar span as obtained by earlier loaded sections
and their reverse path counterparts till it rejoins to the backbone curve while loading in
the reverse path. This behavior is illustrated in Fig. 3.8 a) for cycles of rising
deformation.

30
 Chapter 3. Nonlinear Modeling and Methods of Analysis

While defining the ordinates on the multi-linear curve, it is to be remembered that

symmetrical sets of ordinates will be aligned, regardless of whether the curve isn't
symmetrical. This provides a certain restriction about the state of the hysteretic curve.

The kinematic model forms the source for numerous other models, such as Takeda,
degrading and BRB hardening.

Takeda hysteresis model

This model is fundamentally same as the kinematic model, however, utilizes a

degrading hysteretic loop dependent on the Takeda model, as specified by Takeda et al.
(1970). This basic model requires no other parameters and is more efficient for RC in
comparison to other metals. The energy dissipation of this model is less than the
kinematic model.

Unloading of the elastic portions of the Takeda model is identical to the kinematic
model. While reloading, the loop pursues a secant line up to the backbone diagram for
reverse direction loading. The ultimate deformation that takes place along the path of
preceding load cycles is considered as a target point for this secant. This causes
diminishing in the quantity of energy dissipation with higher deformations. Along the
elastic sections, unloading is observed. This relation is displayed in Fig. 3.8 b).

3.2.1.2 Modeling of Material Nonlinearity

Material nonlinearity can be adopted using either fiber-based structural modeling or

plastic hinge dependant structural modeling procedure.

Fiber-based structural modeling

Fiber-based frame element modeling is a numerical procedure in which structural

elements are separated into two end different frame elements, and then connecting
every extremity to a discrete cross-section (C/S) by means of the grid of fibers. As per
Casarotti C. and Pinho R., (2006), a fiber-based model is capable to capture nonlinear
effects across C/S of the structural member and also along an extended portion of the
member shown in Fig. 3.9.

31
 Chapter 3. Nonlinear Modeling and Methods of Analysis

Fig. 3.9 Fiber-based structural modeling

Plastic hinges

Plastic hinge modeling assumes either concentrated or distributed plasticity regions for
the structural members, with subsequent formation of plastic hinges. Here, concentrated
plasticity is assumed for prismatic sections as suggested by FEMA 356 (2000).
Distributed plasticity is defined for the non-prismatic section like columns of a flat slab.
Here, lp denotes a length of plastic hinge. The plastic hinge data is prerequisite in the
format of moment-rotation relation for nonlinear analysis. The length of the plastic
hinge may be estimated according to several formulae available in past literature and
guidelines of design codes. It is noticed that no significant differences were observed
from the use of any of these formulae [Monteiro R. et al. (2008)] indicated in Table 3.1.

Table 3.1 Length of plastic hinge

Plastic Hinge Length (Lp) Researchers Guidelines

Lp = 0.08Lv + 0.022fydb ≥ 0.044fydb Paulay and Priestley (1992), ATC32 (1996),
NZSEE(2006)
Lp = 0.05h (less conservative) Priestley and Park (1987), Paulay and Priestley
(1992), ATC32(1996), NZSEE(2006)
Lp = 0.06Lv + 0.035h + 0.036dbfy Eurocode 8-Part 3(2004)
Lp = 0.5h + 0.025dbfy
Lp = 0.03h + 0.016dbfy
Lp = 0.2Lvh + 0.036dbfy
Lp = 0.025Lv + 0.125h + 0.1dbfy
Lp = klv + 0.022fyØAslwith k = 0.2[(fu/fy)-1]≤0.04 Kappos (1991)

Note: Lv is shear length constant for symmetrical elements, fy is the yield strength of the longitudinal
reinforcement [MPa], fu is reinforcement ultimate strength, db is longitudinal bars mean diameter in the
affected section, h is the total height

In the present study, plastic hinge length is computed by ATC 32 (1996) prescribed
formula.

The moment-rotation (M-θ) relationship can also be transferred into a force-deflection

relationship by substituting Force (F = M/Lv) and Displacement (∆ = θ.Lv), where Lv is
the shear length which is constant for the symmetrical element (1/2 of the total length L)

32
 Chapter 3. Nonlinear Modeling and Methods of Analysis

as indicated in Fig. 3.10. The reinforcing steel is mainly responsible for the value of
yield rotation θy and concrete is accountable for the value of ultimate rotation θu.

Fig. 3.10 M-θ relationship

For assigning non-linear hinge properties to structural elements, moment-rotation data is

required. To develop moment-rotation relation, moment-curvature analysis is required
to be carried out. It can be performed analytically by using stress-strain curves of
materials or by using software like EXTRACT, Response 2D, SAP2000, etc.

Fig. 3.11 shows actual moment-curvature curve and approximated idealized curve.
Yield and ultimate moment and corresponding curvature are considered according to
idealized moment-curvature curve for the development of moment-rotation relation.

Fig. 3.11 Moment-curvature curve and approximated idealized curve

33
 Chapter 3. Nonlinear Modeling and Methods of Analysis

Beams are assigned with the only moment and shear hinges as they can yield in flexure
and shear. Columns and shear walls are assigned with axial force-moment interaction
hinges and shear hinges as suggested by Raju R.K. et al. (2012), Cinitha A. et al. (2012)
and as displayed in Fig. 3.12.

Fig. 3.12 Hinge assignment for beam and column

Flexural hinges

For each frame element, moment-curvature (M-ø) analyses are carried out considering
axial forces. The moment-rotation (M-θ) relationship is essential in terms of input for
SAP2000. Hence, moment-rotation (M-θ) values are obtained from moment-curvature
(M-ø) values by using formulae specified in Eurocode 8-Part 3 (CEN 2004).

Yield rotation θy can be determined according to the Eurocode 8-Part 3 (CEN 2004) as
stated in Eq. (3.7).

𝐿𝑣 +𝑎 𝑣 𝑧 𝐻 𝑑 𝑏 𝑓𝑦
𝜃𝑦 = ∅𝑦 + 0.00135 1 + 1.5 𝐿 + 0.13∅𝑦 (3.7)
3 𝑣 𝑓𝑐

with:

y = yield curvature,
Lv = the shear length which is constant for symmetrical members,
(1/2 of the total member length L),
H = total element height,
db = longitudinal bar mean diameter in the affected portion,
(e.g. ø10 = 10mm)
fy = yield strength corresponding to longitudinal rebar steel [MPa],

34
 Chapter 3. Nonlinear Modeling and Methods of Analysis

fc = compressive concrete strength [MPa],

av = av = 1, if cracking due to shear is probable to lead flexural yielding at
the edge of cross-section; otherwise, av = 0.
z = internal lever arm length, used equivalent to dr-d'‫ ׳‬in beams, columns
and for walls provided for barbell/T-section, or equal to 0.8h with
rectangular walls cross-section [mm].

Ultimate rotation θu can be determined according to the Eurocode 8-Part 3 (CEN 2004)
as stated in Eq. (3.8).
0.5𝐿𝑝
𝜃𝑢 = 𝜃𝑦 + ∅𝑢 − ∅𝑦 𝐿𝑝 1 − (3.8)
𝐿𝑣

y = yield curvature,
u = ultimate curvature,
Lv = shear length [mm],
Lp = plastic hinge length [mm], where, 𝐿𝑝 = 0.08𝐿𝑣 + 0.022𝑓𝑦 𝑑𝑏 (3.9)

The major output of the pushover analysis is a load-displacement curve, called as

pushover (capacity) curve. Five points are designated as A, B, C, D and E to explain the
force-deformation performance of the plastic hinge as shown in Fig. 3.13 a).

a) Deformation controlled action b) Force controlled action

Fig. 3.13 Force-deformation behavior of a plastic hinge [FEMA356 (2000)]

As per FEMA356 (2000) and ATC40 (1996) guidelines, plastic rotation θp is the
difference between ultimate rotation and yield rotation. According to these codes, at
point B, the moment is equal to yield moment and rotation is equal to yield rotation. At
point C, the moment is approximately equal to 1.1 times yield moment and rotation is
equal to the plastic rotation. At point D, the moment is approximately equal to 0.2 times
yield moment and rotation is equal to the plastic rotation. At point E, the moment is

35
 Chapter 3. Nonlinear Modeling and Methods of Analysis

approximately equal to 0.2 times yield moment and rotation is approximately equal to
1.5 to 2 times plastic rotation as indicated in Fig. 3.14.

CP
LS
IO

Fig. 3.14 Bending hinge

According to Fig. 3.14 before attaining yield point „B‟, deformation is linear. It takes
place within frame members, and not in the hinge. When hinge unloads elastically, it
yields exclusive of plastic deformation, i.e. parallel to linear slope A-B. The plastic
deformation after point „B‟ takes place within the hinge additional to elastic
deformation which may take place in the element. Consequently, hinge properties are
allocated to the frame members are rigid-plastic in nature according to CSI Analysis
Reference Manual, (2009).

Scaling of Hinge curve in SAP2000

While characterizing the force-deformation (or moment-rotation) diagram, it is possible

to assign values of force-deformation straightforwardly or normalized values may be
allocated along with determined scale factors (SF) to normalize the hinge curve.

Generally, the hinge curve will be normalized with the value of yield force (moment)
and yield displacement (rotation). It should be noted that any deformation from point A
to B is discarded.

Acceptance criteria for plastic deformations are as recommended by FEMA356 (2000)

guidelines as Immediate Occupancy (IO), Life Safety (LS) and Collapse Prevention
(CP) as indicated in Fig. 3.15 defined as 10%, 60% and 90% of summation of yield
rotation and plastic rotation respectively.

36
 Chapter 3. Nonlinear Modeling and Methods of Analysis

Fig. 3.15 FEMA 356 (2000) performance levels

At the Operational level, no damage has observed in structural and non-structural

elements. The building is habitable for functional use. Immediate occupancy causes
hairline cracking, partial yielding probable at the few locations, but no crushing (strain
below 0.003). The building is habitable, but repairs are required to be done. Life Safety
causes widespread damage to beams, splitting of cover concrete and propagation of
shear cracks in ductile columns, severe damage of non-ductile columns and
development of joint cracks. The building is not habitable until repairs are done.
Collapse Prevention leads to development hinge pattern in ductile members, partial
cracks and few ductile columns undergo splice collapse, extensive damage observed in
short columns. The building is not habitable and the hazard of falling debris is
increased. Collapse causes complete failure of the structure.

Shear hinges

The shear failure has been modeled for frame elements as force-controlled action
(brittle) as per ASCE/SEI 41-06 (2007). Force deformation relation of the shear hinge is
as represented in Fig. 3.13 b).

As per Weng C.C. et al. (2001) shear capability of the composite member with encased
steel section in concrete can be evaluated with the help of Eq. (3.10) to Eq. (3.14)

Total shear strength =Vcm +Vs + Vrc (3.10)

Vcm = 0.6fysAws (3.11)

where,

Vcm = steel section shear capacity,

Vrc = RC portion shear capacity,
Vs = reinforcement shear capacity,

37
 Chapter 3. Nonlinear Modeling and Methods of Analysis

Aws = area of a steel web of encased steel section,

fys = yield strength of steel section.

Axial compressive force, bending and shear applied to elements according to ACI 318-
08 (2008), Diagonal shear capacity of the concrete is evaluated using Eq. (3.12),

Vrc = Vc + Vrs (3.12)

Shear capacity of the concrete is given by Eq. (3.13),

𝑁
𝑉𝑐 = 0.166 1 + 0.073 𝐴𝑢 𝑓𝑐′ 𝑏𝑠 𝑑𝑠 (3.13)
𝑔

Shear capacity of reinforcement is denoted in Eq. (3.14),

𝐴𝑠 𝑓𝑠 𝑑 𝑠
𝑉𝑟𝑠 = (3.14)
𝑆

Where,
bs = section width,
ds = effective depth,
’
fc = unconfined compressive strength,
Nu = the axial load carried by the member,
Ag = concrete area,
As = area of confined reinforcement,
fs = yield strength of transverse confined hoops,
S = the spacing of transverse confined hoops.

Here, effective stiffness suggested by Eurocode 8 (2004) has been adopted for frame
elements.

3.2.1.3 Effect of concrete cracking

Effect of material nonlinearity due to concrete cracking on pushover response of the

frame elements has been included. The flexural stiffness reduction in structures, due to
concrete cracking amplifies the lateral deflection.

The main parameters affecting the stiffness of the cracked section are elasticity modulus
and moment of inertia (M.I.) denoted as I. The effective M.I., Ie, concept to reflect the
concrete cracking which is more than or equivalent to the M.I. of a cracked element and
is lesser than or equivalent to the M.I. of the total gross section.

Ahmed et al. (2008) provided modification factors recommended in different country

standards to evaluate effective M.I. for fractured section. To estimate it with certain

38
 Chapter 3. Nonlinear Modeling and Methods of Analysis

accuracy is very difficult. The values of effective stiffness differ significantly from each
other as specified by different design standards as represented in Table 3.2.

Table 3.2 Effective stiffness of frame members

RC Element Eurocode 8 ACI 318 FEMA356 ASCE 41

(2004) (2005, 2008), (2000), Supplement 1
IS 1893 (2016) ASCE41-06 (2007)
(2007)
Non-prestressed beam 0.35EcIg 0.5EcIg 0.3EcIg
Column with design gravity
0.7EcIg 0.7EcIg
loads ≥ 0.5Agfc’
Column with design gravity 0.5EcIg Linear
0.7EcIg 0.5EcIg
loads ≤ 0.3Agfc’ interpolation
Column with design gravity
- 0.3EcIg
loads ≤ 0.1Agfc’ or tension
Note: Ec = Elasticity modulus of concrete, Ig= M.I. of C/S, Ag = Gross C/S area, fc’= concrete compressive
strength

3.2.2 Geometric Nonlinearity

Geometric nonlinearity is related to change in the shape of the element and strain-
displacement relation becomes nonlinear due to the application of the lateral load. For
higher values of displacements and strains, this effect is more predominant. Fig. 3.16 a)
denote column subjected to load in vertical and horizontal directions and bending
moment diagram. It is experienced that, extra moment observed at the foundation level
of the column as the secondary moment because of gravity load and recognized as P-Δ
moment.

a) Column undeformed, deformed geometryand bending moment

39
 Chapter 3. Nonlinear Modeling and Methods of Analysis

b) Frame subjected to P-Δ c) Force-deformation relation due to the P-Δ effect

Fig. 3.16 P-Δ effect

Fig. 3.16 b) describes the building frame subjected to the P-Δ effect. Fig. 3.16 c)
describes the capacity curve considering the P-Δ influence and without considering the
P-Δ influence. The inclusion of P-Δ consequence during analysis results into decrease in
lateral stiffness.

3.3 Methods of Analysis

3.3.1 Pushover analysis technique

Pushover analysis is performed by considering material and geometric nonlinearities to

assess the capacity of structure which essentially consists of subjecting the structure to
gravity loads and incremental load or displacement.

In a force-controlled pushover analysis, the forces are uniformly increased until the
entire force achieves a target value or the structure develops a hinge mechanism. While
in a displacement-controlled pushover analysis, the displacements are uniformly
increased till either the displacement of a pre-defined node reaches a target value or the
structure has built a hinge mechanism. In the present study, displacement-based
pushover analysis is executed. During this approach, the structures will start yielding
and finally develop the collapse mechanism. SAP2000 provides default FEMA356
(2000) hinges and provision to define the user-defined hinges, used for structural
elements. Plastic hinges can be allocated to a structural element through the software at
the relative distance of both ends. It is possible to assign flexure and shear hinges at the
identical location. SAP2000 automatically displays the hinges on the frame with
different color code depending upon the moment capacity of the section.
40
 Chapter 3. Nonlinear Modeling and Methods of Analysis

Three different lateral load patterns utilized in pushover analysis are as follows:
i. Uniform load distribution where forces are equivalent to multiplication of
lumped floor mass m and corresponding acceleration a, as indicated in Eq.
(3.15).
F = m×a (3.15)

ii. Modal load pattern is directly equivalent to the multiplication of amplitude of

elastic first mode and each storey mass m as described in Eq. (3.16).
𝑚 𝑖 ∅𝑖
𝐹𝑖 = (3.16)
𝑚 𝑖 ∅𝑖

where,
mi = mass of ith storey,
Øi = the amplitude of the first elastic mode of the storey.

iii. Codal lateral load pattern where the lateral load is distributed along with the
building height dependent on equation prescribed by the IS: 1893-Part 1 (2016)
are used for the analysis represented in Eq. (3.17) to Eq. (3.19).

𝑉𝐵 = 𝐴𝑕 𝑊 (3.17)

𝑊𝑖 𝑕 𝑖2
𝑄𝑖 = 𝑉𝐵 𝑛 𝑊 𝑕2 (3.18)
𝑖=1 𝑖 𝑖

where,
VB = design base shear

𝑍 𝑆𝑎 (3.19)
2 𝑔
𝑉𝐵 = 𝐴𝑕 𝑊 = 𝑊
𝑅
𝐼

Z = seismic zone factor,

Sa/g = design acceleration coefficient,
R = response reduction factor,
I = importance factor,
W = seismic weight,
Qi = lateral force at storey i,
Wi = seismic weight at storey i,
hi = floor i height from ground level.

41
 Chapter 3. Nonlinear Modeling and Methods of Analysis

The equation of motion to be solved during pushover analysis is given by Eq. (3.20).

MΔẍ(t) + CΔẋ(t) + ΔKx(t) = -M{I}Δẍg(t) (3.20)

Where,

M = diagonal mass matrix,

C = damping matrix,
K = stiffness matrix,
Δẍ(t), Δẋ(t) and Δx(t) = incremental acceleration, velocity, displacement
vector respectively of the structure w.r.t. time,
{I} = an influence vector characterizing the
displacements of the corresponding masses
subjected to statically applied unit ground
displacement,
Δẍg(t) = incremental load vector (ground acceleration).

The outcome of the pushover diagram is established from the relationship between the
total base shear and top storey displacement. Pushover analysis is simple, easy to
perform and requires lesser computational time to complete analysis. A cautious
analysis by pushover method will offer insight on structural aspects to obtain enhanced
performance during severe seismic excitation. Such analysis will provide excellent
estimates of not only global but also local nonlinear deformation demands for buildings
which primarily respond according to the first fundamental mode. With an increasing
predominance of higher modes, the accuracy of pushover analysis starts decreasing. To
select load patterns and to interpret the results, good judgement is required for the
incorporation of all deformation modes that are originated and propagated through
ground motions and nonlinear response of the building structures.

3.3.2 Capacity spectrum method (CSM)

The capacity curve can be rebuilt into a capacity spectrum to exploit CSM specified by
ATC 40 (1996).The storey displacement and the total base shear are transformed to the
spectral displacement Sd and spectral acceleration Sa by the application of mode
participation factor and effective modal mass for the first fundamental mode. The
performance of a building to an earthquake can be evaluated by superimposing the
capacity diagram and seismic demand diagram.

42
 Chapter 3. Nonlinear Modeling and Methods of Analysis

Capacity spectrum (CS) intersects reduced demand spectrum (DS) to give the
performance point (inelastic seismic demand). The Eq. (3.20) to Eq. (3.24) are used to
make the transformation.

𝑁 𝑊 𝑖 𝜑 𝑖1
𝑖=1 𝑔
𝛤1 = 2 (3.21)
𝑁 𝑊 𝑖 𝜑 𝑖1
𝑖=1 𝑔

𝑊 𝑖 𝜑 𝑖1 2
𝑔
𝛼1 = 2 (3.22)
𝑁 𝑊𝑖 𝑁 𝑊 𝑖 𝜑 𝑖1
𝑖=1 𝑔 𝑖=1 𝑔

𝑉/𝑊
𝑆𝑎 = (3.23)
𝛼1

∆
𝑆𝑑 = 𝛤 𝜑𝑟𝑜𝑜𝑓 (3.24)
1 𝑟𝑜𝑜𝑓 ,1

Where,
Γ1 = modal participation factor for 1st fundamental mode,
α1 = modal mass coefficient intended for1st fundamental mode,
φi = the amplitude of the first mode at level i,
N = level N; the uppermost height of the structure,
V = base shear,
W = seismic weight of the building (dead load + likely live loads),
Wi/g = mass lumped at floor height,
∆roof = storey top displacement,
Sa = spectral acceleration,
Sd = spectral displacement

Conversion of capacity curve into the capacity spectrum is displayed in Fig. 3.17.

43
 Chapter 3. Nonlinear Modeling and Methods of Analysis

a) Capacity curve b) Capacity spectrum

Fig. 3.17 Capacity spectrum transformation from capacity curve [ATC40 (1996)]

To convert a standard response spectrum Sa Vs T to ADRS style, it is required to

evaluate the magnitude of Sdi for all coordinates on the diagram Sai-Ti. Spectral
displacement is associated with spectral acceleration and time period as given in Eq.
(3.25).
𝑇𝑖2
𝑆𝑑𝑖 = 𝑆 𝑔 (3.25)
4𝜋 2 𝑎𝑖

Where,
Sdi = spectral displacement at ith coordinate,
Ti = the fundamental natural period at ith coordinate,
Sai = spectral acceleration at ith coordinate,
g = gravitational acceleration

The representation of the response spectrum in usual and ADRS format is as indicated
in Fig. 3.18.

a) Usual spectrum b) ADRS spectrum

Fig. 3.18 Response spectra in usual and ADRS format [ATC 40]

44
 Chapter 3. Nonlinear Modeling and Methods of Analysis

The performance of the building structures is assessed on the basis of the performance
point, with reference to these performance levels. If the damage condition at this
performance point is satisfactory, afterwards the building is assumed to be safe for the
design basis earthquake (DBE). Consequently, this kind of design is also recognized as
PBD.

3.3.3 Development of fragility curve from the capacity spectrum

Past earthquakes in India like Koyana (1964), Uttarkashi (1991), Killari (1993),
Chamoli (1999), Bhuj (2001), Off Northern Sumatra (2004), Gangtok (2011), Nepal
(2015) and Bihar (2015) have shown that research in seismic analysis has to be intended
on the assessment of the vulnerability of the structure by probabilistic approach.
Fragility diagrams are vulnerability diagrams expressing the probability to overcome
damage state for specified seismic intensity. These curves account the unpredictability
and ambiguity related to capacity spectrum properties, damage states and characteristics
of ground motion.

The procedure of developing fragility curve is explained in HAZUS-MH (2003)

technical manual. The HAZUS (2003) suggested capacity curves and fragility curves
are applicable to buildings in USA only. These curves required to be reassessed for
other regions for existing stock and also extended to very high-rise buildings.

3.3.3.1 Building type and classification (Table 5.1 HAZUS)

Buildings are also classified depending on the structural system as concrete MRF, shear
walls, steel MRF, etc.

Further, buildings are classified depending on structural properties such as the number
of stories as:
i. Low rise (one to three stories)
ii. Mid-rise (four to seven stories)
iii. High-Rise (more than eight stories)

3.3.3.2 Seismic design states and class of construction

[Table 5.3, 5.19 and 5.20 HAZUS]

Buildings are designed using different seismic design standards, have variation in the
quality of construction and also the different age of occupancy. Therefore, they are
expected to deform differently during an earthquake. By considering all these factors,

45
 Chapter 3. Nonlinear Modeling and Methods of Analysis

seismic design levels are provided based on UBC seismic zones and NEHRAP MAP
area provisions, age of the structure and seismic performance levels as High-code
(newer construction, after 1975), Moderate code (built between 1940-1975), Low code
(built before 1940) and pre-code (older construction, not designed as per seismic
standards). Detailed guidelines regarding selection criteria of seismic design levels are
provided in section 5.7 of HAZUS technical manual.

3.3.3.3 Damage states [Section 5.3.1 HAZUS]

Damage states are stated individually for structural and nonstructural components of the
building. It is illustrated with the help of Slight, Moderate, Extensive and Complete
damage states. Loss functions associate the physical situation of structure to a variety of
loss variables (i.e. direct financial loss, disaster and loss function). The four damage
states are explained as follows:

Slight Structural Damage: It deals with the development of diagonal fine cracks on the
majority surfaces of the concrete shear wall; a small amount of concrete falling at some
locations.

Moderate Structural Damage: The majority surfaces of the shear walls how evidence of
the formation of diagonal cracks; few shear walls have overtaken elastic capability
demonstrated by the establishment of massive diagonal cracks and falling of concrete at
edges of the wall.

Extensive Structural Damage: Mainly RCSW have overtaken their yield capabilities;
some walls have crossed over their maximum capabilities confirmed by outsized,
throughout diagonal cracks in the wall, severe spalling near about the cracks and
noticeably buckled rebars in walls or overturning of thin walls with insufficient
foundations. Limited disintegration may take place due to damage of non-ductile
columns which are not provided with reinforcement for ductile detailing to withstand
lateral loads.

Complete Structural Damage: Structure has failed or is in forthcoming hazard of failure

because of the disintegration of most of the shear walls and collapse of the few critical
beams or columns.

46
 Chapter 3. Nonlinear Modeling and Methods of Analysis

3.3.3.4 Standard Deviation [Table 5.9 a) HAZUS]

The standard deviations (SD) are presented in Table 5.9 a) titled as “Structural Fragility
Curve Parameters” of HAZUS technical reference manual (2003). This table is related
to four different classes, defining four kinds of threshold magnitude (for Slight,
Moderate, Extensive and Complete) of variability as per the seismic performance levels
as High-code (newer construction, after 1975), Moderate code (built between the year
1940-1975), Low code (built before 1940) and Pre-code (older construction, not
designed as per seismic standards).

3.3.3.5 Evaluation of cumulative damage probabilities of particular damage states

Due to the haphazard nature of earthquakes, the performance-based earthquake

engineering (PBEE) is more suitable than deterministic to evaluate the performance of
the structure at specified seismic loading. The measure utilized to express the intensity
of ground motion is recognized as Intensity Measure (IM). IMs of earthquakes are the
Richter scale or Modified Mercalli scale so as to describe as the base shear, PGA,
spectral acceleration Sa, spectral displacement Sd. Fragility curves can be developed
utilizing spectral acceleration as an IM and relating with Engineering Demand
Parameter (EDP) at particular damage state.

The conditional probability of attaining or overtaking a specific damage state, ds, is

defined by the fragility function as specified in [HAZUS, (2003)] Eq. (3.26).

𝑑𝑠 1 𝑆𝑑
𝑃 𝑆𝑑
=∅ 𝛽𝑑𝑠
𝑙𝑛 𝑆𝑑 ,𝑑𝑠
(3.26)

Sd = given spectral displacement,

Sd,ds = given median displacement,
βds = the standard deviation for specific damage state and
Ø = the standard normal distribution function

The total variability of a particular damage state, βds is obtained by combining three
components of structural standard deviation, βC, βD and βM as prescribed in Eq. (3.27).

2 2
𝛽𝑑𝑠 = 𝐶𝑂𝑁𝑉 𝛽𝐶 , 𝛽𝐷 , 𝑆𝑑 ,𝑑𝑠 + 𝛽𝑀 𝑑𝑠 (3.27)

Where,
βC = lognormal SD which expresses the unpredictability of capacity

47
 Chapter 3. Nonlinear Modeling and Methods of Analysis

curve,
βD , = lognormal SD which expresses the unpredictability demand
spectrum,
βM = lognormal SD which expresses the unpredictability the ambiguity
in the estimation of the medium magnitude of the margin of
structural damage state, ds,
CONV = convolving probability distributions of the demand spectrum and
capacity spectrum

Probabilities of specified damage levels are defined as:

PR[SD|Sd] = probability of attaining or overtaking Slight damage state,

PR[MD|Sd] = probability of attaining or overtaking Moderate damage state,
PR[ED|Sd] = probability of attaining or overtaking Extensive damage state,
PR[CD|Sd] = probability of attaining or overtaking Complete damage state,

3.3.3.6Determination of discrete damage probabilities of damage states

Discrete damage probabilities may be computed as stated in Eq. (3.28)

Complete damage probability,

PR[CD] = PR[CD|Sd]
Extensive damage probability,
PR[ED] = PR[ED|Sd] - PR[CD|Sd]
Moderate damage probability,
PR[MD] = PR[MD|Sd] - PR[ED|Sd]
Slight damage probability,
PR[SD] = PR[SD|Sd] - PR[MD|Sd]
No damage probability,
PR[ND] = 1 - PR[SD|Sd] (3.28)

The seismic fragility analysis of building is performed and damage probability matrix is
developed. Every fragility curve is expressed by a median magnitude of the EDP (e.g.
Sd, PGA, PGV, etc.) that relates to the threshold of the damage level and by the
unpredictability linked with that damage level.

According to Barbat A.H. et al. (2008), four states of damage are defined: Slight,
Moderate, Extensive and Complete. They have defined median spectral displacement

48
 Chapter 3. Nonlinear Modeling and Methods of Analysis

for each damage state where, Sd,y is yield state spectral displacement and Sd,u is the
spectral displacement at ultimate state obtained from capacity spectrum curve. It is
displayed in Table 3.3.

Table 3.3Damage state thresholds [Barbat A.H. et al. (2008)]

Damage State Median spectral displacement (Sd,ds)
Slight Sd,S = 0.7 Sd,y
Moderate Sd,M = Sd,y
Extensive Sd,E = Sd,y + 0.25 (Sd,u-Sd,y)
Complete Sd,C = Sd,u

As per HAZUS (2003), four states of damage are stated in Table 3.4.
Table 3.4Damage level thresholds [HAZUS (2003)]

Damage State Median spectral displacement (Sd,ds)

Slight Sd,S = Sd,y (First yield)
Moderate Sd,M = 1.5Sd,y
Extensive Sd,E = 0.5(Sd,M + Sd,C)
Complete Sd,C = Average median value from the capacity spectrum of the building near
collapse state

Fragility curves can be developed by choosing spectral acceleration Sa as IM, the

median magnitude of spectral displacement Sd,ds is transformed in the spectral
acceleration Sa,ds in ADRS format as shown in Fig. 3.19.

Fig. 3.19 Development of equivalent PGA-median damage values [HAZUS (2003)]

Here, a log-normal distribution, the conditional probability of achieving or defeating a

special damage state, ds, for specified spectral acceleration Sa, at given median
acceleration Sa,ds and standard variation βds for particular damage state, Ø is standard
normal distribution function is defined by the fragility function as in Eq. (3.29).

49
 Chapter 3. Nonlinear Modeling and Methods of Analysis

𝑑𝑠 1 𝑆𝑎
𝑃 =∅ 𝑙𝑛 (3.29)
𝑆𝑎 𝛽𝑑𝑠 𝑆𝑎 ,𝑑𝑠

Discrete damage probabilities may be computed as given in Eq. (3.30)

Complete damage probability,

PR[CD] = PR[CD|Sa]
Extensive damage probability,
PR[ED] = PR[ED|Sa] - PR[CD|Sa]
Moderate damage probability,
PR[MD] = PR[MD|Sa] - PR[ED|Sa]
Slight damage probability,
PR[SD] = PR[SD|Sa] - PR[MD|Sa]
No damage probability,
PR[ND] = 1 - PR[SD|Sa] (3.30)

3.3.4 Nonlinear time history analysis (NLTHA)

The analysis will be depended on the interpretation of the seismic vulnerability in the
pattern of time history database. Under the NLTHA, design seismic forces and their
distribution along with the height of the structure and the subsequent inside forces and
displacements are estimated accurately. NLTHA is gradual time increment analysis of
the dynamic performance of a structure exposed to particular loading which differs with
time. The equation of motion for dynamic equilibrium to be comprehended is specified
in Eq. (3.31).

MΔẍ(t) + cΔẋ(t) + KΔx(t) = -M{I}Δẍg(t) (3.31)

The load incorporates acceleration of ground, the displacement, velocity and

acceleration which are associated with ground motion.

Time history may be transient or periodic. There are two distinct techniques, i.e. modal
and direct-integration for NLTHA. Direct-integration solves wave propagation
problems more efficiently which might excite a large number of modes. It is also
possible to include all types of nonlinearity i.e. material and geometric nonlinearities.
Generally, NLTHA continues from the previous case of nonlinear gravity analysis. It
will internally find a solution for the equation of motion at every output time increment
and at every load function time increment. Solutions are obtained utilizing techniques
like Newmark‟s procedure, Wilson θ procedure, central difference procedure, Hibler

50
 Chapter 3. Nonlinear Modeling and Methods of Analysis

Hughes Taylor procedure (HHT), Newton-Raphson balance iterative procedure and

Chung-Hulbert procedure. In the current study, NLTHA performed by Direct-
integration method. The general solution is obtained for second-order differential Eq.
(3.31) by using HHT [Hibler H.M. et al. (1977)] method. Newmark‟s method utilizes
the identical finite difference expression with parameters illustrated in Eq. (3.32) to
(3.34).

α=0 (3.32)
1−2𝛼
𝛾= (3.33)
2

1−𝛼 2
𝛽= (3.34)
4

While HHT (Hibler-Hughes-Taylor) method uses parameter α in the range to acquire

converge solution as shown in Eq. (3.35) and γ, β parameters are the same as that of
Newmark's method.

α = [-1/3, 0] (3.35)

Mass and stiffness relative damping, generally recognized as Rayleigh damping is

normally utilized in NLTHA. In the estimation of the damping matrix, the damping
coefficient is considered to be corresponding to the mass and stiffness matrices are as
stated in Eq. (3.36).

[c] = ɳ[M] + δ[K] (3.36)

Where,
ɳ = damping coefficient proportional to mass,
δ = damping coefficient proportional to stiffness

The modal and orthogonality conditions relationship permits in form of Eq. (3.37).
1 𝜔𝑛
𝜉𝑛 = ɳ+ 𝛿 (3.37)
2𝜔 𝑛 2

Where,
ξn = critical damping ratio,
ωn = natural frequency = 2πfn

If damping ratios (ξi and ξj) allied with two particular frequencies (ωi and ωj) or modes
are identified, the two Rayleigh damping factors (ɳ and δ) are able to determine by the
solution of the set of two simultaneous equations, expressed numerically by Eq. (3.38).

51
 Chapter 3. Nonlinear Modeling and Methods of Analysis
1
ξ𝑖 𝜔𝑖
1 𝜔𝑖 ɳ
ξ𝑗 = (3.38)
2 1 𝜔𝑗 𝛿
𝜔𝑗

SAP2000 allows user to either determine coefficient ɳ and δ straightforwardly or by

means of the critical damping proportion either at two distinct frequencies, f (Hz) or at
two unique periods, T (sec).

When ξi = ξj, coefficient ɳ and δ are determined by Eq. (3.39).

2ξ
𝛿= and ɳ = 𝜔𝑖 𝜔𝑗 𝛿 (3.39)
𝜔 𝑖 +𝜔 𝑗

But, the utilization of NLTHA is restricted in light of the fact that dynamic response is
extremely susceptible to nonlinear modeling and earth-shaking attributes. It is necessary
to model deformation properties appropriately subjected to cyclic load taking into
account deterioration characteristics of all significant elements. With the NLTHA, the
design displacements aren‟t determined utilizing target displacements but rather
utilizing real-time histories. Established behavior can be extremely sensitive in qualities
of every time history. Along these lines, it is prescribed to do the investigation with
more than one real time history record. Since the analytical model considers
straightforwardly the influence of material nonlinearity, computed internal forces will
be sensible approximations those predicted during the design earthquake.

3.4 Closure

Nonlinear modeling of frame elements is explained in detail. Material and geometric

nonlinearities are also illustrated. Pushover analysis procedure implemented in
SAP2000 is briefly described. Different load patterns used during analysis are
illustrated with the help of equations. Output interpretation of pushover analysis is
discussed. Capacity spectrum formulation from the capacity curve as prescribed by
ATC40 is mentioned. In order to accommodate variability and uncertainty related to
capacity spectrum fragility curves are essential. The method used to derive fragility
curves and damage probabilities are narrated. NLTHA is proposed for structure when
the higher modes become predominant. Hence, the procedure to perform NLTHA is
also specified. Pushover analysis, NLTHA and fragility analysis are used in subsequent
chapters to understand the seismic performance of SCCMSW systems.

52