Seismic Strengthening of Buildings and Monuments by Creating Large and Stiff Foundations

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Seismic Strengthening of Buildings and Monuments by Creating

Large and Stiff Foundations


P.G. Carydis
Professor of Earthquake Engineering, Director of Laboratory for
Earthquake Engineering, National Technical University of Athens, Greece

ABSTRACT: The size of the foundation as a two dimensional elastic beam and the velocity of the
propagation of the ground motion are examined as basic parameters for the scope of the present
paper. Two strong ground motions have been used, each one with quite different characteristics
compared to the other one. An artificial time history of rather high frequency, fitting to EC8, ground
class A and a natural ground motion of the Edessa, Greece 1990, M = 5.9, earthquake. The Edessa
earthquake is characterized by much longer predominant periods of vibration compared to the
artificial one. Various lengths of the foundation beam has been examined in combination with the
velocity of the propagation of the ground motion along the longitudinal dimension of the beam. The
achieved motions at the center of gravity of the beam as well as the respective response spectra are
calculated. These spectra are compared to the free field ones. The results of the investigation are
demonstrated in practical case studies.
KEYWORDS: Response spectrum, stiff foundation, wave propagation, near field ground motion,
strengthening of buildings and monuments.

1. INTRODUCTION – HISTORICAL REVIEW

The influence of the size of the foundation to the earthquake response of its structure has been
attracted, since almost 50 years ago, the interest of various researchers. Housner (1957) observed and
explained the reduction of up to 50% of the free field motion along the longitudinal direction of the
foundation during the Arvin – Tehachape earthquake of 1952. Towards the transversal direction of
the foundation he found not any change in the motion compared to the free field one. According to his
investigation the obvious difference is due to the length of the foundation compared to the wave
length of the seismic motion. Along the longitudinal direction the length of the foundation is
comparable to the wave length, while along the perpendicular direction the foundation length is much
smaller. This was an indication that the longer the foundation, the smaller are the seismic excitations
introduced to the building.
Nevertheless, in the code for earthquake resistant design of the Ministry of Construction of
U.S.S.R (1957) a limit of the maximum allowable length of the foundation was introduced according
to the seismicity of the region. The higher the seismicity, the smaller the length and stiffer should be
each statically independent unit. This was set in order to avoid the evaluation of high tensile and
shear stresses in the foundation as well as torsions around vertical axis. In the book of Poliakov
(1973) the increase of stresses in one type of large panel building was demonstrated by presenting the
results of calculations by E.S. Medvedeva. According to these calculations using the maximum
allowable by the code lengths in the building dimensions, considerable stresses, could be created
along the longitudinal direction due to earthquake as high as 1.35MPa for the tensile and 0.74MPa for
the shear one. In this book it was stated that “increasing the length of a building usually will adversaly
affect its earthquake resistance”.
In the same book it was also demonstrated in a figure the reason for the creation of these stresses.
It was not written explicitly but one could infer that this is due to the phasing phenomenon of the

1
seismic waves that propagate along the longitudinal direction of the building foundation.
Nevertheless, later Poliakov (1985) explicitly wrote that for some buildings erected on strip
foundations of continuous footings considerable decrease of the total seismic loads may be developed
compared to the values that result under the assumption that the ground particles during the
earthquake vibrate “synchronously throughout the foundation”. He mentioned that in this case the
averaging principle may yield under which the mean acceleration of the foundation

1 x =l (1)
&&b (t ) =
a ∫ a && (x, t )dx
l 0

in which &&b (t )
a is the mean seismic acceleration in the body of the foundation as a function of time (t)
is the seismic acceleration at the free field along the foundation as a function of
&&(x, t )
a
position (x) and time (t)
l is the total length of the foundation
It is obvious that the max a&& (t ) is smaller than the max a&&(x, t ).
b
In the same book Poliakov (1985) presented the results of the analytical work by G.L. Nikiporets,
who calculated the response spectra of the longitudinal component of the Jurika, 1954, earthquake for
various lengths of the foundation of a building. The ratio:

τ = l / v (sec) (2)

in which v (m / sec) is the velocity of the propagation of the seismic waves, was used as a basic
parameter in the said study. He found that for building periods of 0.1 (sec) up to 0.8 (sec) the
response spectra are reduced, compared to the one in the free field, with the increase of the ratio (τ).
Namely, for values of τ = 0.05 (sec), 0.1 (sec), 0.2 (sec) and 0.4 (sec) the maximum values of the
obtained response spectra on the foundation body were the 95%, 83%, 65% and 33% respectively of
the one corresponding to the free field motion. This means that keeping the foundation length the
same, the lower the seismic velocity is, the greater is the reduction of the response spectra. The same
results may be achieved with the longer foundation while keeping the seismic velocity unchanged.
Anastassiadis (1989) in his book mentioned the fact of the reduction of the earthquake response
of large size foundations due to local mutual refutations of the input motions at the various points
along the foundation at both horizontal directions as well as the vertical one. He mentioned also, that
internal forces in the foundation body due to these refutations may occur.
All these analytical and field observations are in good agreement among themselves. Actually,
due to the phasing phenomenon of the input seismic motion at the various points of the foundation as
this is shown in Figure (1), internal forces in the foundation (axial, bending and shear) will be
developed. These forces are increased with a mutual increase of the length of the foundation. The
foundation body should be strong enough in order to be able to withstand these forces. Therefore, it is
logic to observe damages in inadequately reinforced cases according to the length of the foundation
and therefore it is logical to introduce limitations concerning the maximum permissible lengths of the
foundations. Once the extra forces in the foundation body are adequately confronted, the seismic
response of the whole structure is reduced compared to the case in which the input excitations at the
various points of the foundation were in phase.
It is obvious that the whole subject under consideration is based on the variation of the input
motion along the foundation body. This variation is due to amplitude and phase difference of the
seismic waves that travel along the foundation of a structure. A good measure of this variation is the
coherency between two motions at points i and j. According to Bold (1991) this coherency in the
frequency domain is expressed by

sij(ω) (3)
γij(ω) =
sii(ω)sjj(ω)

2
in which Sij is the cross spectral function and Sii or Sjj are the auto spectral ones. According to Bold
(1991) the coherence is a normalized value between zero and one. Coherence equal to 1 means
complete coherence. The coherence is identified as the square of the modulus of the coherency i.e. the
2
γ ij (ω)
A great help in the subject under consideration was given by the analyses of the strong motion
records obtained from the various strong motion arrays as this is given in Abrahamson and Bold
(1985), in Niazi (1986), in Bold (1991) and many other researchers. Besides this, the practical interest
is for the creation of strong motion time histories for the seismic analysis of structures with multiple
supports, long structures e.t.c, as for example in Bold (1991) and in Petronijevic (2002) is described.
According to Bold (1991) the recorded motions by the SMART 1 array at frequencies above 2Hz
are dominated by incohevent energy when are averaged over a large distance.
It is interesting to mention here the conclusions by Niazi (1986) according to which the peak
cross corrilation values of the accelerogramms, of the El Centro Differential Array may be
approximated by the expression

0.00035×d
− v
P=e λ (4)

where d (m) is the distance between the two stations or the foundation length
λ (m) is the wave length and
v (msec-1) is the velocity of the wave propagation along the foundation.
Since the period of the particle motion is T = λ / v = f the frequency of the motion, expression (4)
could be written as:

P = e − 0.00035 ×d× f (5)

From this equation Niazi (1986) concludes that the frequency and the length of the foundation as far
as the reduction of the input seismic motions is concerned are quantities interchangeable.
Luco and Wong (1986) contributed considerably towards the better understanding of the
earthquake response of rigid foundations to spatially varying ground motion excitations by analytical
approach producing closed mathematical forms, including all basic parameters. In that paper the
authors used a massless rigid foundation bonded to a viscoelastic half space excited by a wave
passage and they found reduction of the translational component of the motion of the beam at high
frequencies (>1Hz), while torsional and rocking components of the motion were created. They noted
that these effects highly depend an the degree of spatial incoherence of the free field motion along the
foundation.

2. CALCULATION OF SOME RESPONSE SPECTRA AS A FUNCTION OF THE LENGTH OF


A MASSLESS FOUNDATION
2.1 Basic considerations

With the advancement of computers and computer codes in Earthquake Engineering it became
feasible to carry out quickly and reliably series of parametric analyses under various assumptions in
order to examine complex phenomena as the one under consideration.
The motion that is recorded by an accelerograph within the epicentral region is the projection of
the particle motion of the ground along the respective three axes. Except for the tectonic and creep
deformations within the epicentral region, various motions that are due to P, S, R and L waves as
well as the reflected and refracted waves take place, as this is shown in Figure (2). These motions
may create even torsional motions around vertical axis in the epicentrical region.
The particles of the ground in the Rayleigh waves perform elliptical motions. The longer axis of
the ellipse is along the vertical direction and the shorter one is along the horizontal direction. The
closer to the epicentre, the greater is the ratio between the vertical over the horizontal axis of the
ellipse.

3
In this way the seismic motion that excites the foundation of a structure is a motion that results
from the convolution of the above mentioned five types of basic motions, plus the initial tectonic and
due to creep motions. Any convoluted motion propagates with some velocity along the ground where
the foundation lies. This velocity some relation may have with the velocities of any of the P, S, R and
L waves. On the other hand the frequency of the alteration of the position of the ground particles
from one site to other is in general higher than the basic ones from which those motions result. This
alternation determines the wave propagation of the motion. As it is shown in figure (2c) the resulted
horizontal and vertical motions are a superposition of the horizontal projections of the Rayleigh
(WR,h) and Shear (WS), and vertical projections of the Rayleigh (WR,V) and P (WP) waves
respectively. Due to the vicinity of the R and S wave velocities is the form of the resulted wave in
Figure (2e). It must be taken under consideration that within the epicentral region the tectonic
deformations should be superimposed, since they affect the region almost simultaneously with the
other types of waves. Also, there are indications that the “wave” lengths in epicentral regions are
relatively very small, and there is an inconsistency between wave length (λ), wave velocity (v) and
frequency (f).
One may understand after the above mentioned that the phenomenon of the motion at a point in
the epicentral region is quite complicated and accelerographs record the final motion of the
respective point. They do not record the motion of a surface or a body (rigid) with certain
dimensions. Nevertheless, one may easily understand that one single point at the surface of the free
field performs the maximum possible motions during the passage (transient) of the seismic waves,
while on the contrary, a linear body with its longitudinal axis oriented along the direction of the
propagation of the waves receives the various seismic motions with phase lag. This, in general is a
function of the length of the body, the propagational velocity and the coherence of the motion along
the body. Due to the mentioned facts in chapter (1), the anticipated motion on the body is certainly
less than the corresponding to the free field excitation.

2.2 Calculation of the response spectra in rigid and massless foundations.

For the parametric study a massless beam out of reinforced concrete with various lengths of
l1=25(m), l2=50(m), l3=100(m) and l4= 150(m) as it is shown in Figure 3 has been used. Its cross
section is 0.5(m)x2.0(m). The dimension of 2.0(m) is vertical. The beam rests on elastic springs with
a stiffness of

Kh = 4.0 (MN/m) and Kv = 5.0 (MN/m) (6)

its modulus of elasticity is E = 2.9 . 107 (KPa) and the poisson ratio is ν = 0.2 .
The transient dynamic analysis of the computer code ABACUS has been used throughout the
present study, as well as other programs of every day use for minor calculations (response spectra
e.t.c)
The input motion is defined as follows. The time history of an accelerogramm has been selected.
This ground motion acts horizontally along the longitudinal direction of the above mentioned beam.
The whole accelerograph is propagated with a certain velocity along the beam and the excitation is
applied through the above mentioned horizontal beams. The velocity of the propagation is

v1= 100 (m/sec), v2 = 200 (m/sec), v3 = 400 (m/sec) and v4 = 600 (m/sec) (7)

For example, for the case of the beam l3 = 100 (m) and the velocity u3 = 400 (m/sec) its last support
starts to be excited at a time lag equal to ∆t = 0.25 (sec) after the excitation of the first support of the
beam. The coherency of the two input motions in the case under consideration depends only on the
dominating frequency of the motion.
The following two input motions have been selected. The first one is an artificial accelerogram
the response spectrum of which fitts to FC8, ground class A, corresponding to a design acceleration
value of ag=0.24(g), of a total duration of 2.4 (sec), as it is shown in Figure (4). The second one is the
Edessa N –S component due to Griva northern Greece, 21 Dec. 1990, M = 5.9 R, earthquake,
recorded by ITSAK. The epicentral distance is 31 (Km) and the depth of the hypocentre is h < 15

4
(Km). The record is on rather soft soil materials. Therefore it is logical to measure predominant
periods between 0.5(sec) and 0.7(sec). The component used is shown in Figure (5).
For each type of excitation 4x4=16 combinations of beam lengths and wave velocities have been
analysed. The total number of final analyses were 32. The obtained horizontal motion at the center of
gravity of each beam was used in order to calculate its response spectrum for 5% damping ratio. Here,
for the sake of brevity, the most representative response spectra are presented. It was proved, as this
has been already indicated by G.L. Nikiporets in Poliakov (1985) that the determinating parameter
among the various response spectra is the ratio τ = l / v (sec). Therefore, in Figure (6) these response
spectra for τ1 = 0 (sec), (is the corresponding to the free field motion), τ2 = 25/200 =0.125 (sec),
τ3=25/100 = 0.25 (sec), τ4= 100/100 = 1.0 (sec) and τ5 = 150/100 = 1.5 (sec) are presented for the
artificial ground motion.
In Figure (7) the response spectra for τ1 = 0 (sec), τ2 = 25/100 =0.25 (sec), τ3=150/400 = 0.375
(sec), τ4= 100/100 = 1.0 (sec) and τ5 = 150/100 = 1.5 (sec) are presented for the Edessa ground
motion.
Although the earthquake excitation records are quite different in shape, duration, and frequency
content in engineering point of view, the resulted response spectra look to have almost the same
reduction as that derived also by G.L. Nikiporets in Poliakov (1985). Namely, for the two earthquakes
and the various τ = l / v values, the following ratios of the obtained maximum response spectra to the
free field one are approximately as follows:

τ =l / v (sec) 0.0 0.125 0.25 0.375 1.0 1.5


Artificial 1.0 0.84 0.76 - 0.66 0.50
Edessa 1.0 - 0.84 0.60 0.50 0.50
Table 1. Comparison of the ratio of the obtained at the center of gravity of the beam maximum response
spectra to the free field one

As it may be seen from Table (1) the reduction of the obtained response spectra for the artificial
motion is quicker than that for the Edessa earthquake, with the increase of the τ(sec). The final and
lowest value of 0.50 was achieved quicker for the Edessa earthquake compared to the artificial one. In
this investigation it is worth to discuss what is the meaning of the used velocity of the wave
propagation. This must not be confused with the well known wave velocities of P, S, R and L waves.
It might be better the respective “velocity” to be determined on the base of the ratio v = λ /Τ, where λ
is the apparent wave length and T is the dominating period of vibration of the ground. In epicentral
regions λ might be found relatively very small (50 – 100 (m)).
The maximum calculated stresses in the beam were the following: Axial stress : about 1.0 MPa.
(This can not be tensile, since the soils can not transmit any tensile motion). Axial stress due to
bending 1.5 MPa and shear stress 0.6 MPa. Under the foundation body it must be a zone of soil that
may create plastic deformations and absorb energy. This is a transition zone from the moving soil to
the less moving or out of phase moving foundation beam, as it is shown in Figure (8).

3. APPLICATION OF THE RESULTS IN SOME CASE STUDIS

During the Egion (northern Peloponnesus) earthquake of 1995 a church erected during the decade of
1890 was severely damaged. The church is a Greek national monument. The damages were more
along the perimetric walls which support arched roof construction. Their damage was
unsymmetrical. The four central main pillars suffered almost symmetrical compression damages at
their base. Based on the findings already mentioned in the previous chapters as a first remedy for the
diminuation of the first strong input motion, the creation of the special foundation was selected, for
the case of the Egion church. The foundation beams create a grillage, which is in contact to the
foundation of the church and is quite strong, stiff and as long as the available site permit, as it is
shown in Figure (9). Similar system has been applied during the construction of the new earthquake
resistant city of A. Liossia, after the destructive earthquake of 1999, Carydis (2001).

5
This method could be considered as a strengthening method, since the ratio L/R
(Load/Resistance) is decreasing. In general, by conventional methods, the earthquake resistance of a
structure is usually increased, and the resulting ratio (L/R) is decreased.

CONCLUSIONS

The increase of the size and especially of the horizontal dimensions of the foundation of a structure
creates internal forces in the foundation body. Also, the obtained motion at the center of gravity of
the foundation gives spectra which are smaller compared to that of the free field motion. This
reduction may be up to 50%. The dominating parameters for this reduction are the coherency
between the various ground motions along the foundation body, their frequency content, the length
of the foundation, the wave length and wave velocity. A way for strengthening of any existing or
new structure may be achieved by creating in its foundation a grillage with strong beams as long as
possible (up to 150m might be enough). This method that may decrease the seismic loads up to 50%
compared to that of the free field, can be applied for the strengthening of monuments without
touching the above ground structure.
The concept for constructing long and stiff foundations could be applied under various types,
according to the particular site conditions. In order to create the placticity zone indicated in Figure
(8) a band of well compacted coarse gravel of a thickness of about 30cm (depending on the
anticipated difference of motions) could be interfere between the physical ground and the foundation
body.

REFERENCES

1. Abrahamson, N. A., Bolt, B. A., (1985): The spatial variation of the phasing of seismic strong ground
motion, Bulletin of tee Seismological Society of America, Vol. 75, No. 5, p.p. 1247-1264.
2. Anastassiadis, K., (1989): Earthquake resistant structures, Ziti publishers, Vol. 1, p.p. 400, Thessaloniki (in
greak).
3. Bolt, B. A., (1991): Development of the phased strong motion time- histories for the structures with
multiple supports, 6th Canadian Conf. Earthquake engineering, Toronto.
4. Carydis, P.G., (2001): A system for earthquake proofing of structures through a special foundation.
Application to the new earthquake resistant city of A. Liossia in Attica. Society of Civil Engineers of
Cyprus, Nicosia, June, p.p. 10
5. Housnev, G. W., (1957): Interaction of building and ground during an earthquake. Bull.seism. soc. Am. 47,
p.p. 179-186.
6. Luco, J. E., Wong, H. L., (1986): Response of a rigid foundation to a spatially random ground motion,
Earthquake Engineering and Structural Dynamics. Vol. 14, p.p. 891-908.
7. Ministry of Construction of U.S.S.R (1957): Earthquake Resistant Regulations, A world List – 1963, IAEE,
p.p. 225 - 306 (in russian).
8. Niazi, M., (1986): Inferred displacements, velocities and rotations of a long rigid foundation located at El
Centro differential array site during the 1979 Imperial Valley California earthquake, Earthquake
Enggineering and Structural Dynamics, Vol. 14, p.p. 531-542.
9. Petronijevic, M., Nefovska, M., Brcic, S., (2002): Multiple – support seismic analysis of bridges including
soil – structure interaction, 12th European Conference on Earthquake Engineering, Elsevier Science Ltd.
London, paper ref. 180.
10. Poliakov, S. V., (1973): Earthquake resistant construction, Translated from Russian, Izdatel’ stvo
‘VYSSHAIA SHKOLA’, 1969, p.p. 282, Moscow.
11. Poliakov, S.V., (1985): Design of earthquake resistant structures, Mir. Publ., Moscow
12. Sali, T., (2002): Earthquake engineering problems and proposals for their solutions in retrofitting of
masonry monuments. Proceedings of the International Conference Hazards and Modern Heritage, Rhodes,
Greece, April 2-5.

6
ACKNOWLEDGEMENTS

The author expresse his gratitude to his collaborators for their contribution to the present study.
Especially to the lectures Dr. H. Mouzakis, M. Sc. St. Antoniou, M. Sc. L. Karapitta, M.Sc. G.
Karydis and Mr. B. Axiotis.

Figure 1. The actual excitation of a structure. (a): The earthquake motion – coming from the left hand, stricks
first the less parts of the building foundation. The rest of the building is not moving, until (b) The excitation
penetrates into the building by the various (A, B, C and D) foundations. The seismic motions at the various
input points present a phase lag.

EPICENTRE

’’P’’ wave
’’P’’ wave

’’S’’ wave Focus


(a)
Surface wave
RAYLEIGH
WR,v

Distance WR,h Distance

Elliptical orbit EPICENTRE


of the ground
(b) particles

Time

(c) WR,h + Ws (VR = 0,9Vs)

Time

(d) WR,v + WP (VR = 1/3 VP)

Figure 2. The final motion on the surface within the epicentral region results from the combination of the
reflected and refracted body and surface waves, as well as of the source mechanism, path scattering and seismic
energy emission peculiarities.

7
Beam length = 25m

Beam length = 50m

Beam length = 100m

Beam length = 150m

Figure 3. The various beams and points with spring supports – input motions used in the parametric analyses

Artificial Earthquak
Artificial Earthquake
Edessa Earthquak
Edessa Earthquake
4 1.2

0.8

0.4
A cce leration (m /sec 2 )
Acceleration (m/sec 2 )

0 0

-0.4

-2

-0.8

-1.2
-4
0 1000 2000 3000
0 50 100 150 200 250
Time x100 (sec) Time (sec)

Figure 4. Artificial accelerogramm fitting to EC8 Figure 5. The N – S component of the Edessa, 21
response spectrum, ground class A, design acceleration Dec. 1990, M = 5.9, Northern Greece, Griva
0.24g. earthquake recorded by ITSAK

8
Response Spectra of an Artificial Motion Fitting EC8 Response Spectra of the Edessa Earthquake
for various beam length/wave velocity ratios for various beam length/wave velocity ratios
and for the damping ratio 5% and for the damping ratio 5%
8 5

-----FREE FIELD l/v=0 sec

4
6

---l/v=25/100=0.25 sec
-------FREE FIELD l/v=0 sec
Acceleration (m/sec2)

Acceleration (m/sec 2)
3
--------l/v = 25/200 = 0.125 sec --------l/v=150/400=0.375 sec
4

-----------l/v=100/100=1.0 sec
2
----------------l/v=150/100= 1.5 sec

-----------l/v = 25/100 = 0.25 sec


2

-----l/v = 100/100 = 1.0 sec 1


---------l/v = 150/100 = 1.5 sec

0
0
0 0.4 0.8 1.2 1.6 2
Period (sec) 0 0.4 0.8 1.2 1.6 2
Period (sec)

Figure 6. Response spectra of the artificial motion Figure 7. Response spectra of the Edessa
fitting to EC8, for various τ (sec), (=l/v=beam length / earthquake, for various τ (sec), (l/v=beam
wave velocity) and for 5% damping ratio. length/wave velocity) and 5% damping ratio.

Figure 8. A transition plastic zone between the soil and the rigid beam may be developed under and around the
foundation body

Figure 9. Creation of as long, as grillage network of foundation beams as possible, after Sali (2002)

You might also like