See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/342701655

Comparison Between Static Nonlinear and Time History Analysis Using

Flexibility-Based Model for An Existing Structure and Effect of Taking Into
Account Soil Using Domain Reduction...

Article  in  KSCE Journal of Civil Engineering · February 2015

DOI: 10.1007/s12205-015-0351-y

CITATIONS READS

10 91

2 authors:

Mourad Belgasmia Moussaoui Sabah

Ferhat Abbas University of Setif Ferhat Abbas University of Setif
31 PUBLICATIONS   42 CITATIONS    10 PUBLICATIONS   13 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A commun project with liege university in belgium which is about predimensioning of fluid viscous damper View project

influences of environmental parameters on the efficiency of soil mixing View project

All content following this page was uploaded by Mourad Belgasmia on 05 July 2020.

The user has requested enhancement of the downloaded file.

KSCE Journal of Civil Engineering (2015) 19(3):651-663 Structural Engineering
Copyright ⓒ2015 Korean Society of Civil Engineers
DOI 10.1007/s12205-015-0351-y pISSN 1226-7988, eISSN 1976-3808
www.springer.com/12205
TECHNICAL NOTE

Comparison Between Static Nonlinear and Time History Analysis Using

Flexibility-Based Model for An Existing Structure and Effect of Taking Into
Account Soil Using Domain Reduction Method for a Single Media
Belgasmia Mourad* and Moussaoui Sabah**
Received November 8, 2011/Revised September 14, 2012/Accepted January 29, 2014/Published Online February 5, 2015

··································································································································································································································

Abstract

This work is divided into two parts; the first one presents the nonlinear methods of analyses for seismic design of structures. The
first method is the nonlinear pushover procedure, which is based on the N2 method. The second method is the classical nonlinear
time history analysis. The objective of this paper is to make a comparative study of an existing reinforced concrete building in
Bonefro, Italy between static nonlinear analysis and time history analysis using flexibility-based finite element, and the sensitivity of
the time history analyses to the seismic parameters. The second part presents an elegant method called Domain Reduction Method,
which takes into account a small adjacent part of subsoil including structure. With this way the size of the problem to be solved is
substantially reduced. All this through Z_Soil; engineering software based on the finite-element method.
Keywords: flexibility-based finite element, seismic engineering and design, push-over, time history, nonlinear analyses, Domain
Reduced method, sensitivity to seismic parameters
··································································································································································································································

1. Introduction conducted a large number of static and dynamic simulations of

small structures with these elements with great success. The
Most studies to date concerned with the non-linear analysis of procedure is general in scope and applies to any section material
reinforced concrete frame structures are based on finite element behaviour.
models which are derived with the stiffness method. The work Modern seismic design codes allow engineers to use either
done by (Kaba et al., 1984; Zeris et al., 1988; Zeris et al., 1991) linear or nonlinear analyses to compute design forces and design
have demonstrated the advantage of flexibility-based models, but displacements. In particular, (Eurocode 8, 2003) contains four
have failed to give a clear and convincing way of determining methods of analysis: simplified static analysis, modal analysis,
the element resisting forces from the given displacements. nonlinear pushover analysis and nonlinear time-history analysis.
This difficulty arises when the flexibility-based finite element These methods refer to the design and analysis of framed
is implemented in a non-linear analysis program based on the structures, mainly buildings and bridges. The two nonlinear methods
direct stiffness method. In this case, the solution of the global require advanced models and advanced nonlinear procedures in
equilibrium equations yields the displacements of the structural order to be fully applicable by design engineers. This paper gives
degrees of freedom. During the state determination phase the a comparaison of static nonlinear pushover analysis which is
resisting forces of all elements in the structure need to be deter- based on the N2 method developed by (Fajfar et al., 2005; Fajfar,
mined. In a flexibility-based element, there are no deformation 2002; Fajfar, 1999) and classical nonlinear time history analysis
interpolation functions to relate the deformations along the of existing reinforced concrete frame structures in Italy using
element to the end displacements, therefore, the process is not flexibility-based finite element. Displacement-based and force-
straightforward and is not well developed in flexibility-based based elements are used in this study. The former is a classical
models proposed to date. This fact has led to some confusion in two-node, displacement-based, Euler-Bernoulli frame element.
the numerical implementation of previous models. To overcome The later is a two-node, force-based, Euler Bernoulli frame
this problem (Ciampi et al., 1986) proposed a consistent element. The main advantage of the second element is that it is
flexibility-based method for formulating frame member models. “exact” within the relevant frame element theory. This implies
This method was refined and applied to the development of two that one element per frame member (beam or column) is used in
types of elements in (Taucer et al., 1991; Spacone et al., 1994) preparing the frame mesh, thus leading to a reduction of the

*Lecturer, Dept. of Civil Engineering Setif University, 19000, Algeria (Corresponding Author, E-mail: [email protected])
**Ph.D. Student, Dept. of Civil Engineering Setif University, 19000, Algeria (E-mail: sabah.moussaoui @yahoo.fr)

− 651 −
 Belgasmia Mourad and Moussaoui Sabah

global number of degrees of freedom. The complete theory for assumed that the section constitutive relation
the force-based element can be found in (Aydinoglu, 2004;
f ( x ) = KD ( x ) (3)
Antoniou et al., 2004; Conte et al., 2004). In this paper four
applications are presented. The first , second and the third one is
With K = EA 0
the study of a SDOF problem under step axial loading, a cantilever 0 EI
beam with bending loading and the earthquake response of 2D
where EA is the axial and EI the flexural rigidity
model for an existing reinforced concrete frame structures in
linear and nonlinear case using flexibility-based and displacement
2.1 Flexibility Method
based formulation model in order to compare between the two.
If equilibrium is considered in the deformed element configuration
The forth one is to make a comparative analysis of 3D model of
in Fig. 1, the relation between nodal forces Q in the system
an existing reinforced concrete building between static nonlinear
without rigid body modes and internal forces f(x) is
analysis and time history analysis using flexibility-based finite
element according to (Eurocode 8, 2003), and the sensitivity f ( x ) = b [x, w ( x ) ]Q (4)
study of the time history analyses to the seismic parameters.
where
2. Flexibility Based Formulation 1 0 0 , ξ = --x- ,
b[ x, w ( x ) ] = (5)
–w ( ξ ) ξ ξ – 1 L
The proposed formulation is flexibility based and uses force
interpolation functions for the bending moment variation that Is the matrix of displacement-dependent force interpolation
depend on the transverse displacements and strictly satisfy functions. Since shear deformations are neglected, the shear force
equilibrium in the deformed configuration. The derivation of the does not appear in Eq. (4), but it can be determined a posteriori
governing equations is substantially more involved than for from the equilibrium condition (see Fig. 1)
stiffness-based elements (Neuenhofer et al., 1998). V = −Hw + T (6)
Nonetheless, the element offers significant advantages over
existing stiffness-based approaches, since no discretization error The weak form of the compatibility condition in Eq (1),
occurs and all governing equations are satisfied exactly. Conse-
∫0 δF ( x )D( x )dx
L
(7)
quently, fewer elements are needed to yield results of com-
parable accuracy. This is demonstrated with the analysis of Leads to three compatibility equations for the frame element
several simple example structures by comparing the results from without rigid body modes. one for the axial displacement q1 and
flexibility and stiffness-based elements. two for the end rotations q2 and q3 in Fig. 1, the latter are
The use of flexibility instead of stiffness is motivated by the identical with the linear case. The former becomes
fact that the dynamically measured flexibility matrix is dominated
1
∫0 δN u′( x ) + --2- [w′( x) ] – ε( x) dx = 0
L 2
by the lowest modes of a structure, which can be easily measured, (8)
while the dynamically measured stiffness matrix is dominated by
the highest modes of the structure, which are hard if not After integrating the preceding expression by parts and accounting
impossible to measure (Reynders et al., 2010). for the boundary terms. The compatibility condition for q1 reduces
The plane frame finite element models are based on the Euler- to
Bernoulli beam theory (Hjelmstad et al., 2002) for geometrically
L 1
∫0 ε( x)dx− ∫0 --2- χ ( x )w ( x )dx
L
nonlinear behavior (Valipour et al., 2010). In this case, the q1 = (9)
governing variables are the axial and transverse displacement
fields u(x) and w(x), respectively, of the element reference axis Thus the complete set of governing equations, the flexibility-
that give rise to deformation fields based is

∫0 b [x, w( x) ]D( x)dx

L *t
1 2 t q= (10)
D ( x ) = [ ε( x ) χ ( x ) ] = u′( x ) + --- w′ ( x ) – w″ ( x ) (1)
2
1 0 0
Where ε( x ) is the axial strain at the reference axis and χ ( x ) is With b* [x, w ( x ) ] = w( ξ ) (11)
– ----------- ξ ξ – 1
the curvature, with the prime denoting differentiation with 2
respect to x. Displacements and strains are assumed to be small.
The nonlinear axial strain-displacement relation in Eq. (1) forms
the basis for the proposed geometrically nonlinear formulation.
The corresponding stress resultants or internal force fields are
f ( x ) = [N ( x ) M ( x ) ]
t
(2)
where, N(x) is the axial force and M(x) the bending moment. It is Fig. 1. Equilibrium in Deformed Configuration

− 652 − KSCE Journal of Civil Engineering

 Comparison Between Static Nonlinear and Time History Analysis Using Flexibility-Based Model for An Existing Structure

In geometrically nonlinear flexibility-based analysis the evaluation length L = 1 m), mass M = 1 kg, force F0 = 1N.
of the flexibility matrix F requires special attention. This is because
both b(x) and b*(x) depend on the transverse displacements w(x). 3.1.1 Axial Loading (Biggs, 1964)
Starting from the governing equation in Eq. (10) we obtain Linear case
*t
Equation of motion:
∂q- = L b*t( x ) ∂D ( x ) + ---------------
∂b ( x -D ) ( x ) dx
F = ------
∂Q 0
∫ --------------
∂Q ∂Q Mu·· + Ku = F0
∂q ∂D ( x ) ∂f ( x ) ∂b ( x )
*t
∂w ( x ) F
*t
F = ------- = ∫L0 b ( x ) -------------- ------------ + ----------------D ( x ) -------------- dx The exact solution is u = ust(1−cosωt) with ust = -----0 . For this case
∂f ( x ) ∂Q ∂w K
∂Q ∂Q the maximum displacement is umax = 2 m.
∂q- = L b*t( x )fb ( x ) + b*t( x )f -------------
∂b ( x )-Q ------
∂w- dx Figure 3 compares the results obtained using displacement-
F = ------
∂Q 0
∫ ∂w ( x ) ∂Q and force-based (flexibility based) element. As expected the
*t
∂b ( x )-D ( x ) ∂w ( x -) dx exact solution is obtained using a single element in both cases.
+ ∫L0 --------------- ------------- (12) Newmark’s algorithm with γ = 0.5 and β = 0.25 is used.
∂w ( x ) ∂Q
where, f = K −1 is the section flexibility matrix. Since Eq. (12) is 3.1.2 Nonlinear Case
derived from Eq. (10), which is based on a variational principle, Nonliniarity is characterized by a yield stress σyield = 1.5 N/m2.
it consists of three contributions. Elastic-perfectly plastic behavior is assumed.
The first term represents the “direct” change in the element end The equation of motion is:
displacements q caused by a change in the element end forces
Q for fixed displacements w(x). Mu·· + F ( u ) = F0
The second and third terms are of indirect nature accounting There are 3 cases to consider in the response
for the change in q caused by the change in w(x), which in turn case 1) u ≤ uyield
originates from a change in Q Since Fyield = 1.5 N, uyield = 1.5 m. The time corresponding to
yield can be computed by setting :
3. Nonlinear and linear Comparison between Flex-
σyield
ibility and Displacement Base Formulation - l = 1.5 = ust ( 1 – cosωtyield )
uyield = ----------
E
Advanced models for frame analysis are used. In particular, with ust = ω = 1 We obtain tyield = arccos(−0.5) = 2.0944 s
nonlinear beams with fiber section models are available. The case 2) uyield ≤ u ≤ um
cross section is divided into fibers and the constitutive law of After yielding the internal force is σyield A, thus the equation of
each fiber is assigned from the constitutive law library available motion is
in software. Both displacement-based and force-based formulations
Mu·· + F yield = F0 ⇒ Mu·· = F 0 – Fyield = –0.5
are available. Force-based elements (Spacone, 1996) are exact
Mu·· + F yield = F0 ⇒ Mu·· = F 0 – Fyield = –0.5
within the classical Euler-Bernoulli beam theory. As for geometric
linearities, these are considered in the general framework of the Mu·· = F0 – Fyield ⇒ Mu· = ( F 0 – Fyield )t′+c1 with t′ = t – tyield
program and thus follow a corotational approach (Zimmermann,
Mu = ⎛⎝ ---------------------
F0 – F yield⎞ 2
⎠ t′ + c1t′ +c2 with t′ = t – tyield
2008). 2
The initial conditions are used to determine c1 and c2.
3.1 SDOF Test Problem
We consider a step load (F = 0 for t < 0 and F = F0 for t ≥ 0)
applied to SDOF nonlinear oscillator (see Fig. 2). The oscillator,
modeled as a bar, has the following characteristics: stiffness K =
1 N/m (modulus of elasticity E = 1 N/m2, section A = 1 m2,

Fig. 3. Comparison Between Displacement-Based and Force-Based

Formulation in The Evolution of Displacement in Time, ∆t =
Fig. 2. Applied Step Load 0.1

Vol. 19, No. 3 / March 2015 − 653 −

 Belgasmia Mourad and Moussaoui Sabah

⎧ u ( tyield ) = ust( 1 – cosωt ) = 1.5

⎨· ⇒ c1 = 0.866 and c2 = 1.5
⎩ u ( tyield ) = ωsin ( ωtyield ) = 0.866
this case is true till u· = 0 . Based on the previous equations:
u· = Fo – Fyield ( t – tyield ) + 0.866 = 0
⇒ tm ′ = t – tyield = 1.732
⇒ tm = 3.8265s Fig. 6. Model of a Cantilever Beam Discretised With One Element
(Force Based Element)
thus

um = ⎛ ---------------------
F0 – F yield⎞ 2 0.5 2
t ′ +c1 tm ′+c1= – ------- ( 1.732 ) +0.866 ( 1.732 )+1.5
⎝ 2 ⎠m 2

um = 2.25m

case 3) t > 3.8265s The displacement is

F yield – F 0⎞ Fyield – F 0
u = ⎛ um – --------------------- + ---------------------cosωt″ with t″ = t – tm
⎝ K ⎠ K
Fig. 7. Model of a Cantilever Beam Discretised with one 20 Ele-
u = ( 2.25 – 0.5 ) + 0.5cos ( t – 3.8265 ) for t = 7s ⇒ t″ = 3.1735s ments (Displacement Based Element)
and u = 1.25m
and force-based element and agree to the theory. A single
In this last case the amplitude of vibration will remain below element was used in both cases.
um.
All values of the nonlinear three cases are checked in Figs. 4 3.2 Bending Loading
and 5. We can see that the results are identical for both displacement- A cantilever beam is used in this second example. The beam
has the following characteristics: L = 1 m, rectangular section
b = h = 0, 3 m, E = 493.83 kN/m2, n = 0.16, ft = fc = 3000 kN/m2.
The cantilever has a 750 kg lumped mass at the free end and is
loaded at the free end with a transverse step load of 15 N applied
at time t = 0 (Figs. 6 and 7). The cantilever response in terms of
tip displacement (in the transverse direction) and shear force at
the fixed end is shown in Figs. 8 and 9, respectively.
We can deduce after studying these two examples, that there is
a clear convergence between the values given by the theory
either in linear or nonlinear cases with those given by the software
used and especially with the flexibility-based finite element.

3.3 Response of 2D Model of Bonefro Building to Ground

Fig. 4. Displacement Response for Nonlinear Biggs’s Example, ∆t Acceleration
= 0.1 The response of a single 2D frame is studies. The side frame
analyzed is shown in Fig. 10, the response of the frame of to the

Fig. 8. Response in Terms of Displacement of Cantilever Beam,

Fig. 5. Force Response for Nonlinear Biggs’s Example, ∆t = 0.1 ∆t = 0.1

− 654 − KSCE Journal of Civil Engineering

 Comparison Between Static Nonlinear and Time History Analysis Using Flexibility-Based Model for An Existing Structure

Fig. 9. Response in Terms of Shear Force at Fixed End of Cantile- Fig. 13. Top Floor Response to Fruili Accelerogram
ver Beam, ∆t = 0.1
an existing reinforced concrete building between static nonlinear
analysis and time history analysis using flexibility-based finite
element.

4. Equivalent SDOF Model and Push-Over Ap-

proach

The starting point is the equation of a planar MDOF subjected

to base motion model explicitly includes only lateral translational
degrees of freedom. The equation of motion can be written:
Fig. 10. 2D Frame Studied
Mu·· + F = −M1a (13)
u and F vector representing displacements and internal forces, 1
is a unit vector, and a is a earthquake ground acceleration as
function of time. The displacement vector u is defined as:
u = φd (14)
Where d is time-dependent top displacement. φ is the displacement
shape normalized in such a way that the component at the top is
equal to 1.
In the pushover analysis the vector of the lateral loads p used is
Fig. 11. Top Floor Response to El Centro Accelerogram determined as
P = pMφ (15)
El Centro accelerogram is shown in the following.
Figure 11 shows the 2D frame top (third) floor response to the Where M is the diagonal mass matrix. The magnitude of the
El Centro earthquake using displacement based and flexibility– lateral loads is controlled by p and φ is the assumed displacement
base element. The same displacement is plotted for the Hollister shape. From Eq. (16) it follows that the lateral force in the ith
and Friuli accelerograms in Figs. 12 and 13. level is proportional to the component φi of the assumed
After satisfactory comparison of flexibility and displacement displacement φ, weighted by the story mass mi
based formulation, only flexibility-based was used in the following
Pi = pmiφ (16)
study which is a comparative study of a fully 3D frame model of
From statics
P=F (17)
By introducing Eqs. (15), (14) and (17) into Eq. (13), and by
multiplying from the left side with φ t, we obtain
··
φ Mφd + φ Mφp = –φ M1a
t t t
(18)

After multiplying and dividing the left hand side with φ M1 , the
t

equation of motion of the SDOF system can be written as

* ··* * *
Fig. 12. Top Floor Response to El Hollister Accelerogram m d + F = –m a (19)

Vol. 19, No. 3 / March 2015 − 655 −

 Belgasmia Mourad and Moussaoui Sabah

Where m* is the equivalent mass of the SDOF system spectrum. The superposition of both the capacity and demand
*
spectra leads to obtaining target displacement wd for the single
m = φ M1 = ∑ mi φi
* t
(20)
degree of freedom oscillator, and finally to target displacement
wd for the real structure (wd = Γ w d ) see Fig. 15, this target dis-
*
d* and F* are the displacement and force of the equivalent SDOF
system placement represents the maximal horizontal displacement
which will be experienced by the structure during an earthquake
* d * V
d = --- and F = -----b (21) corresponding to the given Acceleration Displacement Response
Γ Γ
Spectrum (ADRS spectrum).
Vb is the total base shear of the MDOF model
5. Nonlinear Frame Models in Z_Soil
Vb = ∑ Pi = –φ M1p = p ∑ mi φi = pm
t *
(22)
The constant Γ controls the transformation from the MDOF to Two fully 3D frame models are implemented in the β-version
the SDOF model and vice-versa. It is defined as of the new Z_Soil program, (Urbaski et al., 2007) and (Belgasmia
et al., 2007), which includes full dynamic capabilities. The first
∑ miφ 1 - Where φi is the i-th component of the first mode
i
Γ = ----------------------- is a classical two-node, displacement-based, Euler-Bernoulli frame
i 2
1
∑ mi( φ 1 ) (23) element. The second is a two-node, force-based, Euler Bernoulli
frame element. The main advantage of the second element is that
The push-over approach is a nonlinear static method described
it is “exact” within the relevant frame element theory. This
in (Eurocode 8, 2003). Its implementation in Z_SOIL.PC is
implies that one element per frame member (beam or column see
described in detail in (Urbaski et al., 2007). In short, force
Fig. 17) is used in preparing the frame mesh, thus leading to a
distribution (unitary or modal) is applied to the structure and
reduction of the global number of degrees of freedom. The
monotonously increased. A capacity curve is obtained, drawing
complete theory on the force-based element is found in (Spacone
the total shear force at the base of the structure with respect to the
et al., 1996). For describing the section response, both elements
top displacement (in our case, the bonefro building). This curve
use a fibre discretization. Fibre sections automatically account
is then expressed for an equivalent single degree of freedom
for moment-axial load interaction. In the present implementation,
oscillator and it is bi-linearized, giving birth to the so-called
simple uniaxial constitutive laws have been implemented for
capacity spectrum (see Fig. 14).
concrete and steel. Enhancements to these laws are planned for
The seismic action depends on the type of the structure, soil
future developments. Both elements include both material and
conditions and the zone of application. It is expressed as an
geometric nonlinearities. Material nonlinearities derive from the
Acceleration-Displacement Response Spectrum (ADRS), or demand

Fig. 16. 3D Frame Model Using Displacement Based Formulation

(Four Elements per Frame Member)

Fig. 14. Capacity Spectrum

Fig. 17. 3D Frame Model Using Flexibility Based Formulation (One

Fig. 15. Demand Spectrum and Target Displacement Element per Frame Member)

− 656 − KSCE Journal of Civil Engineering

 Comparison Between Static Nonlinear and Time History Analysis Using Flexibility-Based Model for An Existing Structure

Table 1. Computational Cost between Flexibility and Displacement

Based Formulation
Flexibility based Displacement based
model for pushover model for pushover
analysis analysis
Number of node 280 1369
Number of beam 363 1452
Number of equations 1542 8076
Total execution time 184s 430s

fibre nonlinear constitutive laws. Geometric nonlinearities are Fig. 19. 3D Model
included in the framework of the co-rotational formulation.
Geometric nonlinearities are important for analyses carried out Table 2. The Reinforcement Details of Elements
up to the collapse limit state. elements &
b(m) h(m) Reinforcement
caracteristics
5.1 Computational Cost Comparison between Flexibility element 1 0.3 0.3 2Φ16 top; 2Φ16 bottom; 2Φ14 middle
and Displacement Based Formulation element 2 0.3 0.3 2Φ16 top ; 2Φ16 bottom
The comparison of total execution time of 3D frame model element 3 0.3 0.5 2Φ14 top ; 4Φ14 bottom
using flexibility based formulation in which only one finite element 4 0.8 0.2 6Φ14 top ; 6Φ14 bottom
element per frame member is used with displacement based element 5 0.5 0.2 2Φ14 top ; 2Φ14 bottom
formulation in which 4 finite elements per frame member is used element 6 0.3 0.2 2Φ14 top ; 2Φ14 bottom
as shown in Figs. 16 and 17, the results of computational cost are
summery in Table 1.

6. Application

The nonlinear response of the 3D model of an existing building

is presented. The building is a residential two-storey reinforced
concrete building in Bonefro, Italy. It is representative of typical
residential building construction in Italy in the 1970’s and 1980’s. Fig. 20. Rectangular Section for Floor Beams
The building is shown in Fig. 18.
The following reinforcement details are used:
6.1 Bonefro Building Modeling and Material Properties For the floor beams geometry, an equivalent rectangular beam
A general 3D model is presented in Fig. 19, the following is used. The beam dimensions are found imposing the same
material properties are used for the fiber section (see Table 2). inertia of the T-beam (see Fig. 20). The results are:
Concrete:E = 2.7e7 kN/m2 Steel E = 2.1e8 kN/m2
In our case b/d = 6.67 m & h/t = 2.5 m thus b'/b = 0.336 m so b'
υ= 0.16 υ = 0.2
= 0.672 m.
ft = 300 kN/m2 ft = fc = 500000 kN/m2
fc = 3000 kN/m2
6.2 Pushover Response and Target Displacement of 3D
Where E is Young modulus, υ Pisson ratio, ft and fc are tensile Model (Belgasmia, 2013)
and compressive strenght respectively. The design spectrum for the building was obtained from
(Eurocode 8, 2003) using the local soil properties and the peak
ground acceleration given by the new Italian seismic map. The
building is regular in height but is irregular in plan because of the
eccentric position of the staircase. The pushover analysis of the
building in the x direction is shown in Fig. 21, the determination
of the target displacement is shown in
Figure 22 the target displacement of Single Degree Of
Freedom (SDOF) is 0.06 m. in order to have the displacement of
Multi Degree Of Freedom (MDOF) we must multiply the results
of SDOF by the mass participation factor Γ which is equal for
3D modal to 1.28 thus the MDOF results 0.077 m.
Fig. 18. Two-Storey Bonefro Building Used for Analyses In order to perform the time-history analyses, and to compare

Vol. 19, No. 3 / March 2015 − 657 −

 Belgasmia Mourad and Moussaoui Sabah

Fig. 21. Pushover Response of 3D Model with Modal Load Distri-

bution in x Direction

Fig. 22. Target Displacement for Pushover Response

Fig. 24. Verification that Generated Earthquake Ground Motion

Spectra are Close to Mean Spectrum: (a) Earthquake1
Spectrum, (b) Earthquake2 Spectrum, (c) Earthquake3
Spectrum

Fig. 23. Generated Accelerogram Spectrum vs EC8 Spectrum

the results with those obtained with the pushover analysis, the
first step is to select a set of spectrum-compatible ground motions.
In this application, three artificially generated ground motions
are used. The ground motions are generated using a computer
program based on the theory presented in (Sabetta, 1996). The
most important input parameters are the epicentral distance, the
magnitude and the type of soil. In this case, the Magnitude was Fig. 25. Response of Bonefro Building to Earthquake Ground Motion
set to 6.02, the epicentral distance to 22.8 km and the soil type to 1 Applied in the x Direction
shallow. The results are three ground motions whose response
spectra are shown in Figs. 23, Fig. 24 compares the mean 6.3 Nonlinear Time History of 3D Model
spectrum to the three separate spectra. The time history responses A time history analysis of the 3D model is presented with the
in terms of top floor center of mass displacement are shown in three accelerograms of the previous section. The results for input
Figs. 25, 26 and 27. ground motion applied in one direction only, are compared for

− 658 − KSCE Journal of Civil Engineering

 Comparison Between Static Nonlinear and Time History Analysis Using Flexibility-Based Model for An Existing Structure

Fig. 26. Response of Bonefro Building to Earthquake Ground

Motion 2 Applied in the x Direction

Fig. 28. Base Shear at Selected Top-Floor Displacements from TH

Analyses: Comparison with Pushover Curve

are 0.0708, 0.0316, 0.0507 meters. Because only three ground

motions are used, the design displacement is 0.0708 m. Fig. 28
shows on the same plot the pushover curve and the based shear
measured during the time history analysis at given displacements,
0.01, 0.02 and 0.04 m. Because during a cyclic analysis the top
displacement reaches at several instances a given value, several
points correspond to a given displacement.
Several points correspond to a given displacement, the dis-
Fig. 27. Response of Bonefro Building to Earthquake Ground placement of 0.04 m is attained at several instances. It is interesting
Motion 3 Applied in the x Direction to note from Fig. 28 that in the time history analysis the maximum
based shear is reached even for small displacements due to the
pushover and time-history analysis. Stiffness proportional Rayleigh cylcic nature of the response. This figure points out how the
damping is prescribed, with 5% damping at 2 Hertz. The resulting pushover curve gives different information compare to the three
values for Rayleigh damping are α = 0 and β = 0.008. The top- time history.
floor response to the three accelerograms is shown in Figs. 25, 26
and 27. 7. Sensitivity to Seismic Parameters
The three responses show a residual displacement at the end of
the time histories, indicating a nonlinear response in parts of the The sensitivity of the time history analyses to the seismic
building. The residual displacement is larger under ground motion parameters used in generating the accelerograms with the program
1. The maximum displacements due to the three ground motions by (Sabetta., 1996) is presented here. The three parameters

Table 3. (a) Sensitivity to Magnitude Seismic Parameter

Magnitude = 5.7 Magnitude = 6.0 Magnitude = 6.3
Maximum response for static pushover analysis 1.28 × 0.06 m = 0.077 m
Maximum response under earthquakes in × direction for Time
0.06 m 0.0708 m 0.106 m
history analysis

(b) Sensitivity to Epicentre Seismic Parameter

Epicenter distance = 11.4 km Epicenter distance = 22.8 km Epicenter distance = 45.6 km
Maximum response for static pushover analysis 1.28 × 0.06 m = 0.077 m
Maximum response under earthquakes in × direction
0.065 m 0.0708 m 0.095 m
for Time history analysis

(c) Sensitivity to Soil Type Seismic Parameter

Soil type shallow Soil type deep
Maximum response for static pushover analysis 1.28 × 0.06 m = 0.077 m
Maximum response under earthquakes in × direction for Time history analysis 0.0708 m 0.065 m

Vol. 19, No. 3 / March 2015 − 659 −

 Belgasmia Mourad and Moussaoui Sabah

+
one by Ω̂ , Γ is the boundary that separates interior and exterior
+
domains Γ̂ is a boundary where viscous damping are to be put
to cancel wave reflexions.

8.1.1 Remarks
Kinematic quantities at any point in interior domain will be
Fig. 29. Effect of the Inclusion of the Soil in the Behaviour of the denoted with the lower index ( )i, at boundary Γ with index ( )b
Superstructure and in the exterior domain with index ( )e
+
After partitioning of the whole domain into Ω and Ω̂ one
magnitude, epicenter distance and soil type were changed. The may write equation of motion, neglecting viscous damping terms,
+
results are shown in the following tables. The reference values in Ω and Ω̂ respectively
are magnitude = 6.02, epicenter distance = 22.8 km, shallow soil
··
Mii Mib ⎧ Ui ⎫ + Kii Kib ⎧ Ui ⎫ = ⎧ 0 ⎫
Ω Ω Ω Ω
type. These are the values used for generating the three
⎨ ·· ⎬ ⎨ ⎬ ⎨ ⎬ (24)
earthquakes used for the time history analyses (see Table 3).
M bi Mbb ⎩ Ub ⎭ Kbi Kbb ⎩ Ub ⎭ ⎩ P b ⎭
Ω Ω Ω Ω

8. Soil-Structure Interaction Problems +

·· + + +

M bb M be ⎧ Ub ⎫ + Kbb Kbe ⎧ Ub ⎫ = ⎧ –P b ⎫
Ω̂ Ω̂ Ω̂ Ω̂
⎨ ·· ⎬ ⎨ ⎬ ⎨ ⎬ (25)
M eb M ee ⎩ Ue ⎭ Keb Kee ⎩ e ⎭ ⎩ e ⎭
When we take soil into account in calculation, we can apprehend Ω̂
+
Ω̂
+
Ω̂
+
Ω̂
+
U P
the actual behaviour of the structure in a good manner, as shown
in Fig. 27 (Gazeta et al., 2006). The two above equation can be written in global form as follows:
We can notice from Fig. 29(b) the rocking is allowed, there is
0 ⎧ U·· i ⎫ Kii
Ω Ω Ω Ω
less damage, no plastic hinging in the superstructure and design M ii Mib Kib 0 ⎧ Ui ⎫ ⎧ 0 ⎫
+ + ⎪ ·· ⎪ + + ⎪ ⎪ ⎪ ⎪
Mbe ⎨ Ub ⎬ + Kbi Kbb +Mbb Mbe ⎨ Ub ⎬ = ⎨ 0 ⎬ (26)
with soil is more favourable that with structure only , because we Ω Ω Ω̂ Ω̂ Ω Ω Ω̂ Ω̂
M bi M bb+M bb
have a strongly nonlinear foundation response. ⎪ ·· ⎪ ⎪ ⎪ ⎪ ⎪
Pe
Mee ⎩ Ue ⎭ Mee ⎩ ⎭ ⎩ ⎭
Ω̂
+
Ω̂
+
Ω̂
+
Ω̂
+
Ue
0 M eb 0 Meb
8.1 Basic Assumption and Theory of Domain Reduction Let us decompose displacement vector in the exterior domain Ue
Method for a Single Media 0
into free field displacement Ue and residual one Ûe
The main goal of Domain Reduction Method (DRM) is to
0
analyze the computational model that concerns the structure and Ue = Ue + Ûe (27)
a small adjacent part of subsoil. With this way the size of the
Substituting Eq. (27) into Eq. (26) modifies this later to the
problem to be solved is substantially reduced. The basic
following form
assumptions and theory are shown in the following explanations.
The complex model with a large subsoil zone and source of Ω Ω ·· Ω Ω
load represented in Fig. 30 is split into reduce model as shown in Mii Mib 0 ⎧ Ui ⎫ Kii Kib 0 ⎧ Ui ⎫
+ + ⎪ ·· ⎪ + + ⎪ ⎪
Mbe ⎨ Ub ⎬ + Kbi Kbb +Mbb M be ⎨ Ub ⎬
Fig. 31. Ω Ω Ω̂ Ω̂ Ω Ω Ω̂ Ω̂
M bi Mbb +M bb
In Fig. 31 the interior domain is denoted by Ω while exterior ⎪ ·· ⎪ ⎪ˆ ⎪
ˆ
Mee ⎩ Ue ⎭ M ee ⎩ e ⎭
+ + + +
Ω̂ Ω̂ Ω̂ Ω̂ U
0 M eb 0 Meb

⎧ 0 ⎫
⎪ +
Ω̂ ·· 0
+ ⎪
= ⎨ – M be Ue − KΩ̂be U0e ⎬ (28)
⎪ + + ⎪
⎩ P e – M Ω̂ee U·· 0e – KΩ̂ee U0e ⎭

The Pe term can now be derived from Eq. (25)

Fig. 30. Full Model of Subsoil, Structure and Source of the Loading
Pe (t) + + + +

Pe = M eb U·· b + M ee U·· e + Keb Ub + Kee Ue

Ω̂ 0 Ω̂ 0 Ω̂ 0 Ω̂ 0
(29)
By substituting the above Pe term to the Eq. (28) the following
form of the right hand side term of Eq. (28) is obtained:

⎧ 0 ⎫
eff ⎪ +
Ω̂ ·· 0 Ω̂
+
0⎪
P = ⎨ – M be Ue − Kbe Ue ⎬ (30)
⎪ + ⎪
⎩ MΩ̂eb U ·· 0 – KΩ̂+ U0 ⎭
b eb b
Fig. 31. Reduce Model

− 660 − KSCE Journal of Civil Engineering

 Comparison Between Static Nonlinear and Time History Analysis Using Flexibility-Based Model for An Existing Structure

+ +

Remark: for the lumped mass matrix terms M Ω̂be U·· e , M Ω̂eb U·· b
0 0

disappear.

8.2 Seismic Input in the Relative Format

Seismic input is given as an imposed acceleration ag of the
ground, common to the whole structure. In this case the external
force vector can be written as follows:
F ( t ) = –Mag ( t )−Fext ( t ) (31)
In this approach the displacements are relative ones with respect Fig. 33. Displacement Time History Analysis Results (Command,
2013)
to the fixed nodes and are referred to non inertial coordinate
system. n the Domain Reduction Method the inertia force term
Table 4. Pushover Analysis Report for Structure Only
−Mag(t) on the right hand side must be modified for nodal points
+ Pushover analysis report
which are in the exterior domain Ω̂ (this forces are not present
Item UnitP SH 1/Default
in the exterior domain). By partitioning it into interior, boundary
MDOF Free vibr. period........T [s] 0.445066
and exterior nodes it takes the following form:
SDOF Free vibr. period.......T* [s] 0.837848363
Ω Ω SDOF equivalent mass.........M* [kg] 12369.6
⎧ –Mii ag – M ibag ⎫
⎪ ⎪ Mass participation factor Gamma - 1.20401
–Mag( t ) = ⎨ – M a – M a – M a – M a ⎬
Ω Ω Ω̂
+
Ω̂
+
(32)
⎪ bi g bb g bb g be g
⎪ Bilinear yield force value..Fy* [kN] 48.9047294
⎩ 0 ⎭ Bilinear displ. at yield....Dy* [m] 0.07030178
Target displacement.........Dm* [m] 0.166111577
8.3 Application to Two-storey RC Frame (Command, 2013) SDOF displacement demand....Dt* [m] 0.053940741
The nonlinear time history and pushover response for two Energy......................Em* [kN*m] 6.404596979
cases without soil and with taking into account soil of the 2D Reduction factor.............qu - 1
model of the two-storey reinforced concrete building see Fig. 32, Demand ductility factor......mi - 3.079519728
using flexibility-based model for superstructure and Domain Capacity ductility factor...miC - 2.362836024
Reduction Method DRM for soil. MDOF displacement demand.....D t[m] 0.064945192
The results using time history analysis and pushover analysis
without taking into account soil are represented in Fig. 33 and
Table 4 respectively. The multi degree of freedom displacement
demand in this case is 6.5 cm. For the value of the Bending
moment envelope for both time history analysis and pushover
one are +106/−103 kNm, 93 kNm respectively.
Figure 34 represents the 2D model of the structure taking into
account the soil using the reduced model. The results using time
history analysis and pushover analysis with taking into account
soil are represented in Fig. 35 and Table 5 respectively. The multi

Fig. 34. 2D Model of the Structure Taking Into Account the Soil
Using the Reduced Model (Command, 2013)

Fig. 32. Geometry and Member Properties of the Typical Frame Fig. 35. Displacement Time History Analysis Results Taking into
Structure Analysed: (a) Plan View, (b) Cross Section A-A' Account Soil (Command, 2013)

Vol. 19, No. 3 / March 2015 − 661 −

 Belgasmia Mourad and Moussaoui Sabah

Table 5. Pushover Analysis Report for Structure with Taking into Table 3(a)).
Account Soil − The maximum base shear indicated by the pushover analysis
Pushover analysis report is systematically reached for almost any maximum top dis-
Item Unit PSH 1/Default placement, this is probably indicative of significant influence
MDOF Free vibr. period........T [s] 0.529679 of 2nd 3rd modes.
SDOF Free vibr. period.......T* [s] 1.1692452 − Only one finite element per frame member is used, thus
SDOF equivalent mass.........M* [kg] 14036 introducing a significant decrease in the global number of
Mass participation factor Gamma - 1.28476 degrees of freedom, number of equation and total execution
Bilinear yield force value..Fy* [kN] 19.58028002 time (see Table 1).
Bilinear displ. at yield....Dy* [m] 0.048308883 − Taking soil into account leads to more important value of
Target displacement.........Dm* [m] 0.075135711 displacement (about 50% and 32%) than taking structure
SDOF displacement demand....Dt* [m] 0.075181918 only, for both time history and pushover analysis respec-
Energy......................Em* [kN*m] 0.998227533 tively. For the designed moment the results are very interest-
Reduction factor.............qu - 1.556275235 ing so when we take soil into account leads to more lower
Demand ductility factor......mi - 1.556275235 values of moment (about 59% and 61%) than taking struc-
Capacity ductility factor...miC - 1.555318745
ture only, for both time history and pushover analysis respec-
MDOF displacement demand.....Dt [m] 0.096590721
tively.

Table 6. Summary of Results of Section 8 9. Conclusions

Time history (with DRM) Pushover analysis
dtop min/max M min/max dt |M|max (dt) The element formulation is based on force interpolation functions
(cm) (kNm) (cm) (kNm) strictly satisfy element equilibrium and, thus, belongs to the
Structure only +4.0 -103/+106 6.5 94 category of flexibility-based elements.
Structure with soil -8.0 -43/+47 9.65 37 The use of exact force interpolation functions in the element
requires fewer elements for the representation of the non-linear
behaviour of a structure, and gives good numerical results without
degree of freedom displacement demand in this second case is difficulties. The element presents two major advantages over
9.65 cm. For the value of the Bending moment envelope for both classical displacement base element, firstly a numerical accuracy
time history analysis and pushover one, are +47/−43 kNm, 37 of the element with only one finite element per frame member is
kNm respectively. used, thus introducing a significant decrease in the global number
of degrees of freedom, secondly no discretization error occurs
C = Cohesion
and all governing equations are satisfied exactly.
E50 = Is a secant Young modulus at 50% of failure
We can now compare the pushover and the time history analysis;
deviatoric stress.
in the case of time history analysis the Eurocode propose to take
Eo = Young modulus at very small strains.
the maximum response, if we apply to structure less than 7
Eur = Is the unloading/reloading Young modulus.
earthquakes, and compare it to the target displacement of multi
γ = Specific weight
degree of freedom pushover model.
Φ = Friction angle
The maximum response to the three earthquakes is 0.0708 m.
Ψ = Dilatation angle
In pushover analysis study the target displacement of a multi
σh;ref = Reference vertical stress.
degree of freedom is 0.077 m. We can say that the result of time
Let us do a summary of All results of section 8 see Table 6. history and pushover analysis are very similar.
There are no doubt advantages in using nonlinear analyses vs
8.3.1 Discussion of the Results using linear methods. Most importantly, nonlinear analyses allow
− The comparison between pushover and dynamics gives dif- designers to follow more closely the nonlinear response of
ference of 8% . buildings and bridges to the design earthquakes corresponding to
− The sensitivity to Magnitude, in dynamics a (+−5%) varia- the ultimate and collapse limit states.
tion in reference magnitude value (6.02) yields a variation in The displacement of 0.04 m is attained at several instances. It
max displacement of( +33% and –15%) (see Table 3a). is interesting to note from Fig. 28 that in the time history analysis
− The sensitivity to epicenter distance, in dynamics (via Sabetta the maximum based shear is reached even for small displace-
program) a (+100%, −50%) variation in reference magni- ments due to the cylcic nature of the response. Fig. 28 points out
tude value (22.8 Km) yields a variation in max displacement how the pushover curve gives different information compare to
of (+25%, −8%) (see Table 3(b)). the three time history. Pushover analysis provides the maximum
− The sensitivity to soil, in dynamics (via Sabetta program): base shear one can expect for a given maximum target displace-
shallow and deep soil type the difference is about 8% (see ment (corresponding to a given earthquake intensity). The time

− 662 − KSCE Journal of Civil Engineering

 Comparison Between Static Nonlinear and Time History Analysis Using Flexibility-Based Model for An Existing Structure

histories provide not only the maximum values, but the entire 28, No. 9, pp. 979-993.
history. For the example at hand, tha maximum displacement of Fajfar, P. (2002). “Structural analysis in earthquake engineering – A
the time history is 0.0708 m, and the maximum base shear is breakthrough of simplified non-linear methods.” Proceedings of 12th
European Conference on Earthquake Engineering.
approximately equal to 1500 kN, but during the time history this
Fajfar, P., Kilar, V., Marusic, D., and Perus, I. (2005). “The extension of
value of the base shear can be reached at several instances and the n2 method to asymmetric buildings.” Proceedings of the 4th
for different values of the top displacement. On the pushover European Workshop on the Seismic Behaviour of Irregular and
curve this translates into a single point that provides maximum Complex Structures, No. 41, Thessaloniki, Greece.
displacement and maximum base shear that can be expected for Gazetas, G. (2006). Seismic design of foundation and soil-structure
that given earthquake. interaction, First European Conference on Earthquake Engineering
Taking soil into account leads to more important displacement and Seismology, Geneva, Swiszerland.
because of the actual behaviour of the structure, this show what Hjelmstad, K. D. and Taciroglu, E. (2002). “Mixed methods and
flexibility approaches for nonlinear frame analysis.” Journal of
is neglected when we don’t take into account the soil in a
Constructional Steel Research, Vol. 58, Nos. 5-8, pp. 967-993.
numerical model, in the same time modeling soil leads to more Kaba, S. and Mahin, S. A. (1984). “Refined modeling of reinforced
lower value of moment which is gain. concrete columns for seismic analysis.” EERC Report 84/03, Earthquake
Engineering Research Center, University of California, Berkeley.
References Neuenhofer, A. and Filippou, F. (1998). “Geometrically nonlinear flexibility-
based frame finite element.” Journal of Structure Engineering. Vol.
Antoniou, S. and Pinho, R. (2004). “Advantages and limitations of 124, No. 6, pp. 704-711.
adaptive and non adaptive force-based pushover procedures.” Reynders, E. and Roeck, G. (2010). “A local Flexibility Method for
Journal of Earthquake Engineering, Vol. 8, No. 4, pp. 497-522. Vibration-Based Damage Localization and Quantification.” Journal
Aydinoglu, M. N. (2004). An improved pushover procedure for of Sound and Vibration, Vol. 329, No. 12, pp. 2367-2383.
engineering practice: Incremental Response Spectrum Analysis Sabetta, F. and Pugliese, A. (1996). “Estimation of response spectra and
(IRSA), International Workshop on Performance-based seismic simulation of non stationary earthquake ground motions.” Bulletin
design. Concepts and Implementation, Bled, Slovenia, June 28-July of the Seismological Society of America, Vol. 86, No. 2, pp. 337-352.
1, In Press by Pacific Earthquake Research Center, PEER. Spacone, E. (1994). Flexibility-based finite element models for the
Belgasmia, M. and Moussaoui, S. (2013). “Comparaison of static nonlinear static and dynamic analysis of concrete frame structures,
pushover analysis in the case of small and large deformation with PhD Thesis, Department of Civil Engineering, University of
time history analysis using flexibility-based model for an existing California, Berkeley.
structure.” International Journal of Current Engineering and Spacone, E., Filippou, F. C., and Taucer, F. F. (1996). “Fiber beam-
Technologie, Vol. 3, No. 2, pp. 655-665. column model for nonlinear analysis of R/C frames. I: Formulation, II:
Belgasmia, M., Spacone, E., Urbanski, A., and Zimmermann, Th. Applications.” Earthquake Engineering and Structural Dynamics,
(2007). Seismic evaluation of constructions: Static pushover Vol. 25, No. 7, pp. 711-742.
procedure and time history analysis for nonlinear frames, LSC Taucer, F. F., Spacone, E., and Filippou, F. C. (1991). A fiber beam-
Internal Report, Swiss Federal Institute of Technology Lausanne column element for seismic response analysis of reinforced concrete
Switzerland. structures, EERC Report 91/17, Earthquake Engineering Research
Biggs, J. M. (1964). Introduction in structural dynamic, McGraw-Hill, Center, University of California, Berkeley.
New York. Urbaski, A., Spacone, E., Belgasmia, M., Sarf, J. L., and Zimmermann,
Ciampi, V. and Carlesimo, L. (1986). “A nonlinear beam element for Th. (2007). Static pushover analysis in Z_Soil.PC & related
seismic analysis of structures.” Proceedings of 8th European. EUROCODE 8 regulations, Report 070202 Zace Services.
Conference on Earthquake Engineering, Lisbon. Valipour, H. R. and Foster, S. J. (2010). “A total secant flexibility-based
Command, S. (2013). “Pushover analysis with zsoil taking soil into formulation for frame elements with physical and geometrical
account.” Numerics in Geotechnics and Structures Symposium, 28 nonlinearities.” Finite Elements in Analysis and Design, Vol. 46, No.
Years of Zsoil and Structure PC, Swiszerland. 3, pp. 288-297.
Conte, J. P., Barbato, M., and Spacone, E. (2004). “Finite element Zeris, C. A. and Mahin, S. A. (1988). “Analysis of reinforced concrete
response sensitivity analysis using force-based frame models,” beam-columns under uniaxial excitation.” Journal of Structure
International Journal for Numerical Methods in Engineering, Vol. Engineering, Vol. 114, No. 4, pp. 804-820.
59, Issue 13, pp. 1781-1820. Zeris, C. A. and Mahin, S. A. (1991). “Behavior of reinforced concrete
Eurocode 8 (2003). Design of structures for earthquake resistance, structures subjected to biaxial excitation.” Journal of Structure
european committee for standardization, Part 1: General Rules, Engineering, Vol. 117, No. 9, pp. 2657-2673.
Seismic Actions and Rules for Buildings. Zimmermann, Th., Truty, A., Urbanski, A., and Podles, K. (2008). Z-
Fajfar, P. (1999). “Capacity spectrum method based on inelastic demand soil user manual, Zace Services, Switzerland.
spectra.” Earthquake Engineering and Structural Dynamics, Vol.

Vol. 19, No. 3 / March 2015 − 663 −

View publication stats