Sie sind auf Seite 1von 23

Introduction

Earthquakes are a dangerous and devastating natural phenomenon. During earthquakes, it is vital that
structural systems are engineered to withstand the forces imparted by the moving ground and hence the
performance of idealized structural systems to different type of ground motion should be analyzed in detail.

The equation of motion of a vibrating structure due to a ground motion could be expressed in a general
form as:

+ + =

Where,

M is the generalized Mass Matrix of the structure

C is the viscous damping coefficient matrix,

K is the stiffness matrix,

u is the relative displacement of the structure with respect to the ground

ug is the ground displacement and

is the influence vector.

In many structural applications, complex continuous structures can be idealized as single or multi-degree
of freedom systems, with the mass of the structure, idealized as a concentrated mass, at each floor height.

Several design codes (ASCE 7-10, IS 1893) prescribe the use of response spectrum analysis, where the
dynamic nature of the problem is simplified and an equivalent lateral force is applied to the structure
according to a response spectrum presented in the code.

For an elastic single degree of freedom structure, the equation of motion can be written as,

+ 20 + 0 2 = ,

Where,

represents the damping ratio (Ratio of damping in the structure to critical damping),

0 represents the natural frequency of the oscillator.

The Response spectrum for a given damping ratio () and a ground motion is a maximum response of the
single degree of freedom oscillator, prescribed with the same damping ratio, for a range of time periods,
for the given ground motion. The spectral displacement could be defined as, the maximum displacement
of the oscillator for the given ground motion, and damping ratio, plotted against the specified natural time
of the structure. In case of structures with no material or geometric nonlinearities, many building codes
(ASCE 7-10, IS 1893) prescribe the use of modal analysis in studying the performance of structural systems
due to earthquake loads, where the response of the multi-degree of freedom system is decoupled to form
several single degree of freedom oscillators and later combined using certain combination rules. Hence, a
careful understanding of the performance of a single degree of freedom oscillator with different hysteretic
rules, to different types of ground motion, is required to engineer structures to perform well during
earthquakes.

Part 1. Ground Motions:

The site-specific estimation of earthquake motion is an area that has obtained much interest during recent
times, as analysis of structures founded on sites, require a suite of ground motions to be selected consistent
with the hazard level of the given location. Hence several seismological characteristics i.e., source, path and
site characteristics determine the nature of the ground motion obtained at a site. Due to the uncertainty
in accurately estimating models to physically capture the rupture propagation, ground motions are treated
as an aleatory uncertainty and until more sophisticated models which capture the physics of the
phenomenon could be established. Table 1 lists the selected earthquakes used in the study, recorded on
hard or stiff soils. Figures 1.1 1.4 show characteristics of the recorded ground motion on two sites with
stiff soils (NEHRP Site Class D i.e., Stiff soils with 180 m/s < V s,30 < 360 m/s) for the 1987 Superstition hills
earthquake and the 1993 Landers earthquake.

Abbrev. EQ Event Year Mw Station Rclosest Soil Type Duration Recorded


(km) (NEHRP) (s) PGA (g)
sup1 Superstition Hills 1987 6.7 Brawley 18.2 D 22.0 0.116
sup2 Superstition Hills 1987 6.7 El Centro Imp Co. Cent. 13.9 D 40.0 0.258
sup3 Superstition Hills 1987 6.7 Plaster City 21 D 22.2 0.186
nor2 Northridge 1994 6.7 Beverley Hills 14145 Mulhol 19.6 C 30.0 0.416
nor3 Northridge 1994 6.7 Canoga Park-Topanga Can 15.8 D 25.0 0.356
nor4 Northridge 1994 6.7 Glendale-Las Palmas 25.4 D 30.0 0.357
nor5 Northridge 1994 6.7 LA-Hollywood Star FF 25.5 D 40.0 0.231
nor6 Northridge 1994 6.7 LA-N Faring Rd 23.9 D 30.0 0.273
nor9 Northridge 1994 6.7 N. Hollywood-Coldwater Can 14.6 C 21.9 0.271
nor10 Northridge 1994 6.7 Sunland-Mt Gleason Ave 17.7 C 30.0 0.157
lp1 Loma Prieta 1989 6.9 Capitola 14.5 D 40.0 0.529
lp2 Loma Prieta 1989 6.9 Gilroy Array #3 14.4 D 39.9 0.555
lp3 Loma Prieta 1989 6.9 Gilroy Array #4 16.1 D 40.0 0.417
lp4 Loma Prieta 1989 6.9 Gilroy Array #7 24.2 D 40.0 0.226
lp5 Loma Prieta 1989 6.9 Hollister Diff. Array 25.8 D 39.6 0.279
lp6 Loma Prieta 1989 6.9 Saratoga - W Valley Coll. 13.7 C 40.0 0.332
cm1 Cape Mendocino 1992 7.1 Fortuna - Fortuna Blvd 23.6 C 44.0 0.116
cm2 Cape Mendocino 1992 7.1 Rio Dell Overpass FF 18.5 C 36.0 0.385
lan1 Landers 1992 7.3 Desert Hot Springs 23.3 C 50.0 0.154
lan2 Landers 1992 7.3 Yermo Fire Station 24.9 D 44.0 0.152

Table 1. Lists details of the other ground motion records used in the study, recorded on hard/stiff soils.
Figure 1.1-1.4. Recorded Ground Motion records of sup2 and lan2

The frequency domain representation of both the acceleration records may be found in figure 1.1a and
1.2a, respectively.
Figure 1.1a-1.2a. Frequency domain representation of sup2 and lan2

It could be noticed that sup2 has a PGA (Peak Ground Acceleration) of 0.26g, PGV (Peak Ground Velocity)
of 41cm/s and lan2 has a PGA of 0.15g and a PGV of 30m/s, and both records possess a broadband range
of frequencies.

The site effects are a near earth effect, which account for the modification of seismic waves, when it passes
through a deposit are of importance to structural engineers as evident with historic earthquakes (Mexico
City (1985), Christchurch (2011) and the Nepal Gorkha (2014) earthquakes. Soil conditions are known to
modify earthquake waves, acting as a filter, and hence must be accounted for in the analysis.

Figure 1.5 and 1.5a show the time and frequency domain representation of the SCT-long signal recorded
due to the 1985 8.0 Mw Mexico City earthquake. The city was founded on an ancient lacustrine deposit
and significant amplification took place around the natural time period of the structure due to the
amplification due to the soil. Records from hard rock deposits at nearby mountains did not possess the
same effect as the ground motions in the valley. The record SCT-Long was obtained from a NEHRP Site
Category F, which requires that site response analysis to be conducted, due to the nature of the soft soil.
Figure 1.5. Time Domain Representation of SCT-Long

Figure 1.5a. Frequency Domain Representation of SCT-Long

While the energy content is distributed evenly through a broad band set of frequencies in both sup2 and
lan2, the SCT-Long record possessed significant energy content concentrated a single frequency of around
0.5Hz. It could be observed that due to the energy density of the signal around 0.5 Hz, many structures
with similar natural frequencies experienced collapse due the prolonged duration of the record and the
resonance. Table 2 summarizes the characteristics of the both the ground motions.

Record Name
sup2 lan2 SCT-Long
Intensity Measure
PGA (g) 0.26 0.15 0.165
Duration (s) 40 44 183.5
PGV(cm/s) 41 30 57.7
Predominant Period (s) 0.46 0.22 2.04
Bracketed Duration(s) 35.86 39.62 64.9

Table 2. General Characteristics of the ground motion records sup2, lan2 and SCT-Long
Part 2. Linear Oscillator:

The performance of a SDOF oscillator with several hysteretic relationships including linear, Bi-Linear (Elastic
Perfectly Plastic), A Clough Degrading stiffness oscillator and an oscillator using a combination of these
springs in parallel.

The idealized hysteretic response of the oscillators may be found in figure 2(a-d)

Figure A: Basic Material Models Used

Description of Material Models:

a) Linear: This Model is used to study the response of a Linear Oscillator. The yield Force is set to high
value so as to not exhibit any sort of nonlinear behavior. This is used by the code ASCE 7-10 to
derive the response spectra.
b) Nonlinear Elastic: This model although nonlinear exhibits elasticity post yield. This model is used in
combination with other material models described later to obtain the desired Re-centering Model
commonly used in modelling the behavior of pre-stressed systems.
c) Bilinear Model with Kinematic Hardening: This Model is mainly used to simulate a well-detailed
Steel Structure. Strain Hardening can be captured too in a simple way, describing the behavior of
the material post yield, but an elastic-perfectly plastic relationship is used for the analysis.
d) Clough Stiffness Degrading Model: This Model is used to simulate a well detailed Reinforced
Concrete Structure. Significant softening is observed in concrete structures due to cracking and
stiffness degradation and hence is used in concrete structures.

Figure B: Re-centering Model (Combination of Nonlinear Elastic and Bilinear models)

e) Re-centering Model: This model is obtained as a combination of the Nonlinear Elastic Model and
Bilinear Model with varying contribution factors as desired. This model is used to simulate a well
detailed Pre-stressed Concrete Structure. It captures all the details of both the Nonlinear Elastic
Model and Bilieal Model with very little to no residual displacement at the end of the earthquake
which is usually the case in a pre-stressed concrete structure.
Figures 2.1-2.5: Displacement Time History of the Linear and Nonlinear Structure to lan2 ground motion

The behavior of the various idealized structures are identified in figures 2.1-2.5 using a single degree of
freedom oscillator. Insight into the structural behavior could be identified using the obtained displacement
time histories. The linear oscillator generates the maximum number of cycles, during the free vibration
period. Though as observed in figure 1.2, the ground motion lan2 has very little amplitude, hence
generating the free vibration response of the system, where the oscillator vibrates at its natural time period
of 2s. Due to the prescribed damping ratio of 2.5%, the time window (0-45 s) is not sufficient to decay the
motion of the oscillator and hence at the end of the record, though a linear oscillator is used, there exists
a residual displacement.

Figures 2.1a shows the response of various linear oscillators of time period 2s and different damping ratios.
It could be observed that with higher damping, the response dies out faster and within the observed time
window.

Figure 2.1a. Effect of damping on the free vibration response

In case of the bilinear oscillators as shown in figure 2.2 and 2.3 for different response reduction factors,
post yield during free vibration, the oscillators will vibrate at its natural time period of 2s, but due to
yielding, will change the equilibrium position where it oscillates, as noted in the figures 2.2 and 2.3, where
residual displacements are observed. Though greater plastic deformation is expected in the nonlinear
oscillator with the high response reduction factor during the entire time window (As observed in figure 2.2
and figure 2.3), the residual displacement cannot be evaluated using the response reduction factor as the
two different systems behave using different hysteretic behaviors and is strongly dependent on the pulses
in the ground motion. Though different material models show similar maximum displacement response (as
shown in figures 2.1-2.5), if the residual displacement is considered as an important factor in the design,
the selection of material models play an important role in its estimation.

An important conclusion to studying this behavior is that while residual displacements are important in
structural systems, to retrofit and rehabilitate structures, the estimation of residual displacements (in case
of elastic-perfectly plastic systems, as there is significant plastic deformation) require accurate modelling
of behavior of the structural system and hence, as noted by Priestley (2003).
Figures 2.6-2.8 show the hysteretic response (restoring force vs. deformation) of the different material
models (Bilinear, Clough Stiffness degrading and re-centering oscillators)

On observation of the residual displacement, it could be seen that at the end of the observed time window,
as was the case with the elastic-perfectly plastic oscillator, the estimation of the residual displacement
varies heavily on the ground motion record considered. It could though be stated that on increasing the
time window, it could be seen that the re-centering oscillator would have the least residual displacement
as for small amplitudes of shaking at the end of the input record, the material behavior forces least plastic
deformation. As the response of the oscillator is heavily dependent on its history, and due to the random
nature of the earthquake motion, it is very difficult to estimate the residual displacement and perform a
comparison with different hysteretic models. Table 3 shows estimates of the residual displacement for
various types of oscillators.

Displacement at
Material Model R Value end of EQ
(mm)
Linear 1 42.5
Bi-Linear 4 97.13
Bi-Linear 8 11.2
Clough 8 18.8
Re-Centering 8 10.4

Table 3. Summary of residual displacement with response reduction factor for different oscillators
Figure 2.(9-10). Variation of restoring force for different oscillators.

On observation of the force deformation relationship, several conclusions could be made, the linear
oscillator, post 35s (end of the earthquake) oscillates in free vibration with the natural period of the
oscillator (2s), whereas softening has taken place in the re-centering oscillator. Due to the yielding of the
bilinear oscillator, and the very small strain hardening ratio, the response history of the re-centering
oscillator appears capped. If an elastic perfectly plastic oscillator was used along with the nonlinear elastic
oscillator, we could observe flat peaks in the restoring force, when the force reaches the yield strength.
Part 3. Nonlinear Response:

Figures 3.1-3.3: Acceleration and Pseudo Acceleration Response Spectra for Linear Oscillator

Figures 3.1-3.3, show the absolute acceleration response spectra and the Pseudo-Acceleration response
spectra. Sa and PSa are very close to each other, which is very typical of structures with low damping ratios
(usually <20%). As damping increases, PSa (obtained as 2Sd) starts to deviate from the actual absolute
acceleration Sa. This is also evident from the figure where for the values are practically equal for = 2.5%,
very slightly differ for = 5% and visible different (even though can be considered equal for practical cases)
for = 15%.

Also, as the time period of the structure (T) , tends to zero,( 0), the structure acts like a stiff rigid body
attached to the ground and moves along with it. Thus, at T=0, both S a and PSa are the Peak Ground
Acceleration PGA. As , both PSa and Sa tend to zero, no matter what the damping. This is because an
infinitely flexible structure behaves like isolated from the ground motion, leading to zero acceleration. At
large time periods, for larger damping ratios, PSa degrades faster than Sa because it is inversely dependent
on the square of the T.

Figure 3.4 Variation of Spectral Displacement Ratio

It can be said that in general, higher damping leads to lower spectral displacements for linear systems.
However, Sd converges to zero as 0 because the structure acts as a rigid body attached to the ground
with no relative displacement and as , it converges to Peak Ground Displacement because the
(,)
structure would be infinitely flexible. This means that at large time periods, ( = 5%) =
(,=5%)
tends to 1. Also, as can be seen above, the average B values for both the damping ratios of 2.5% and 15 %
of the given 20 stiff soil ground motion records agree well with the value of B obtained from the code (Table
17.5.1, ASCE 7-10) as 1 . So, when using the code, critical damping ratio is used to obtain the

value of BD or BM and the response spectrum given in the code for =5% is modified as needed.

An observation can be made that, though for each ground motion, B values might vary, for a large enough
ensemble, the observed B values are consistent, for the average for 20 ground motions used in the analysis,
the B values approach the code specified values. If a big enough ensemble is considered, the B factor for
different damping ratios could be established. Identifying B is useful as design codes specify response
spectral ordinates for a given value of damping (usually 5%), and using the obtained B values, response
spectral ordinates for other damping ratios could be easily evaluated.

Part 4. CR Spectra:

The displacement coefficient spectra is a useful tool in estimating the inelastic deformation (Spectral
Deformation) of structures under a particular force deformation relationship. It was seen in figures 1.2-1.5,
the maximum response of the nonlinear oscillators were similar. CR is a useful tool in estimating the inelastic
deformation, given the elastic response. Figures 4.1-4.3 show the variation of the CR averaged over each
ground motion plotted against the time period of the oscillator.
Figures 4.1-4.3: Average Displacement Coefficient CR Response for Nonlinear Models
There is a clear influence of R on the CR spectra. At longer time periods, the equal displacement rule applies
for inelastic systems and hence both systems have almost equal displacement irrespective of the R value.
Increasing R, we introduce ductility in the system, and hence should expect larger maximum displacements,
hence as observed in the figure 4.1-4.3, increasing R increases the inelastic response and by definition CR.
Miranda (2003) clearly notes this issue using a suite of 72 ground motions, where for larger R values, the
inelastic deformation increases. Miranda (2003) also notes that the site category where the ground motion
is recorded heavily nature of the CR spectra identifying that for various site classes, as the soil gets softer,
the time period where the equal displacement principle starts to work, elongates.

In earthquake engineering design philosophy, a larger hysteretic energy response is believed to help reduce
the system response through hysteretic damping and stiffness degradation. Bilinear oscillators have the
largest area under the hysteresis curve and hence through this philosophy should have least response and
the re-centering oscillators have the least area under the hysteresis curve and hence have the maximum
response. The philosophy is proven to be false as in this specific case, the re-centering oscillator has the
minimal response. Priestley (2003) notes this fallacy and identifies that the response quantity considered
is important in using this design philosophy. Though larger hysteretic curves lead to higher equivalent
damping, the nature of the stress-strain curve itself is important, while designing structures. The dispersion
of B is shown in figure 4.4
Figure 4.4: Distribution of B for Linear Oscillators

It could be observed from figure 4.4 that though for individual records, B values maybe scattered, for an
ensemble of records, the average B values remain consistent Since a linear oscillator is considered, the
distribution of B is fairly narrow. Each ground motion record is a random process and hence since the
output obtained for individual records might be erratic but on considering a large enough number of
ground motion records from a similar soil category, it could be observed that the distribution of B values is
centered around a mean and is fairly consistent with the values observed in the design codes as observed
in figure 3.4.

(WHY IS THE RESPONSE <1)

Figures 4.5 and 4.6 plots the distribution of CR obtained for T = 2s using the Clough Oscillator with R = 4 and
R = 8 respectively.
Figure 4.5. Distribution of CR for Clough Oscillator, R=4
Figure 4.6. Distribution of CR for Clough Oscillator, R=8

It could be seen from the distribution that a larger R values induces an earlier nonlinear response
in the system and hence as the behavior of the Clough model at any time instant is related to the
history of the deformation, the results are more scattered. If R = 1, in case of a linear system CR
would tend to 1. On Increasing R, we subjected the system to experience greater ductility and
higher nonlinear behavior is expected. As ground motions are a random process, in itself,
increasing R, increases the coefficient of variation in CR. This can be confirmed using Miranda
(2003) where for a suite of 216 ground motions, the increase in R, leads to an increase in the
coefficient of variation of CR for a system with natural time period of 2 s.
Displacement Coefficient Spectrum for Soil Soil Record:

For this response analysis, far field record on soft soil (Soil Type F) for 1985 Mw 8.1 Mexico Earthquake is
used.

Figure 5.1: Displacement Response Spectra Comparison of Linear and Clough Oscillator (r=0) for Far Field
Ground Motion

The Spectral Displacement for Linear Oscillator with a time period of 2 seconds shows a peak because of
resonance with the period of the ground movement. Since the structure is linear, there is no change in its
stiffness and time period unlike for a nonlinear structure where upon yielding it undergoes period
elongation and out of the region of T close to 2 seconds as can be seen in Figure 1.5(a). This shows that
displacement demands of linear oscillator are not always less than those of nonlinear ones.
Figure 5.2: Displacement Coefficient Spectra for Soft Soil Record (SCT-Long)
using Clough Oscillator (r=0)
Figure 5.2(a): Displacement Coefficient Spectra for Soft Soil Record (SCT-Long)
using Clough Oscillator (r=0)
Figure 5.3: Comparison of Displacement Coefficient Spectra for Near Field and Far Field Ground Motion
Records using Clough Model (r=0)


As can be seen in Figure 5.2, the value of = , where Sdi is the Spectral displacement of the Inelastic

structure and Sd is that of the Linear Elastic Structure, at low time periods, the Spectral displacements of
inelastic system is very large as compared to that of linear system. This is because upon period elongation,
the time period of the structure falls in the range of the predominant period of the ground motion for soft
soil, leading to amplification. Also, at low time periods, the initial stiffness of the structure is high and for a
linear system, the spectral displacement is low. Whereas after yielding, the structure undergoes significant
deformation and thus, at low periods, equal energy rule is more relevant. The very high CR values for very
small time periods is because in case of linear system, the spectral displacement is almost zero and for
nonlinear system, upon yielding undergoes very large deformation due to perfectly plastic model used. At
larger time periods, the nonlinear system undergoes period elongation and it moves further away from the
predominant range of time period of the ground motion. Also, since the period is high, the time period of
the system in the linear range is low. Upon yielding, the deformation it undergoes is comparable to the
deformation of the nonlinear system and hence, equal displacement rule can be used. However, for a
better understanding of the response analysis, more ground motion records are to be analyzed.

Das könnte Ihnen auch gefallen