Beruflich Dokumente
Kultur Dokumente
OF ADOBE DWELLINGS
Supervisors
Dr. Helen Crowley
Dr. Humberto Varum
Dr. Rui Pinho
May, 2008
ABSTRACT
Adobe buildings are highly vulnerable in earthquakes. It is painfully evident that each time a strong
earthquake strikes in areas where earthen building is common, there is widespread damage to earthen
historical monuments and housing and tragic loss of life due to the collapse of these constructions.
Recently, this has been seen in Pisco city (Peru) during the earthquake of Mw = 8.0 on August 17,
2007, where more than 500 people died and more than 37 500 dwellings collapsed, where the majority
of them were constructed in adobe.
In this research, the seismic risk of non-engineered adobe dwellings located in Cusco (Peru) is studied
based on mechanical procedures. The analysis takes into account both in-plane and out-of-plane
failure mechanisms of adobe walls. The capacity of each dwelling has been expressed as a function of
its displacement capacity and period of vibration, and has been evaluated for different limit states. A
probabilistic seismic hazard assessment for Cusco has been carried out to obtain displacement
response spectra for different return periods. Finally, from the comparison of the capacity with the
demand, probabilities of failure have been obtained for different return periods. These can be
expressed in terms of conditional or unconditional seismic risk. The first takes into account fragility
curves as function of PGA values, while the second one takes into account many possible future
earthquake scenarios that could impact upon urban areas during a given exposure time window.
For example, the conditional seismic risk and for an event with PGA= 0.18g, it is expected that 77%
of the adobe buildings can collapse or be near collapse because of insufficient in-plane capacity.
Considering the out-of-plane capacity, it is expected that around 75% of the buildings would have
wide vertical cracks at the intersections between walls, which could result in the partial collapse of
those buildings. For the unconditional seismic risk, ratios of damage of exceedance have been
calculated for different time windows, which can help to convey the message of risk to the public and
to decision makers. These results indicate the necessity of improving the seismic capacity of adobe
dwellings by providing reinforcement, which is capable of guaranteeing ductility and integrity of the
adobe walls.
Keywords: seismic risk; adobe dwellings; fragility curves; limit states; dynamic properties.
i
Acknowledgement
ACKNOWLEDGEMENTS
I would like to express my thanks to God and to all my family for being my support at every moment
wherever I am. To my parents, Pilar and Sabino, and to my sister Ximena for encouraging me to
continue in my studies, even when it took me far from home.
My gratitude goes to the supervisors: Dr. Helen Crowley, Dr. Humberto Varum and Dr. Rui Pinho for
their support and useful suggestions throughout this research work.
A special consideration to the professors Dr. Marcial Blondet and Eng. Julio Vargas from the Catholic
University of Peru, and to the professors Eng. Gustavo Flores and Eng. Genaro Delgado from the “San
Luis Gonzaga” University for being my first guides in the researching field.
Last but not least, I want to express my gratitude to Samantha and her family for their invaluable
generosity to me, and to my friends in this Master Program: Andrés, Angelo, Armando, Brankica,
Clara, Federica, Heidy, Jaime, José, Lisa, Lydell and Raffaele for their kindly help during the good
and difficult moments that we have shared.
ii
Acknowledgement
AGRADECIMIENTOS
Quiero expresar mi agradecimiento a Dios y a toda mi familia por la ayuda que recibo donde quiera
que me encuentre. A mis progenitores, Pilar y Sabino, y a mi hermana Ximena por motivarme a
continuar en mis estudios aun lejos de casa.
Mi gratitud a los supervisores: Dra. Helen Crowley, Dr. Humberto Varum y Dr. Rui Pinho por sus
ayudas y acertadas sugerencias durante el desarrollo de este trabajo de investigación.
Especial consideración a los profesores Dr. Marcial Blondet e Ing. Julio Vargas de la Pontificia
Universidad Católica del Perú, y a los profesores Ing. Gustavo Flores e Ing. Genaro Delgado de la
Universidad Nacional “San Luis Gonzaga” por ser mis primeros guías en el campo de la investigación.
Por último, pero no menos importante, a Samantha y a su familia por su invaluable generosidad
conmigo, y a mis amigos de este programa de Maestría: Andrés, Angelo, Armando, Brankica, Clara,
Federica, Heidy, Jaime, José, Lisa, Lydell y Raffaele por su gentil ayuda durante los buenos y difíciles
momentos vividos.
iii
Index
TABLE OF CONTENTS
Page
ABSTRACT ............................................................................................................................................i
ACKNOWLEDGEMENTS....................................................................................................................ii
AGRADECIMIENTOS .........................................................................................................................iii
TABLE OF CONTENTS ......................................................................................................................iv
LIST OF FIGURES ..............................................................................................................................vii
LIST OF TABLES..................................................................................................................................x
1 INTRODUCTION ...........................................................................................................................12
1.1 The state of the art....................................................................................................................12
1.2 Justification ..............................................................................................................................13
1.3 Scopes and objectives ..............................................................................................................13
1.4 Hypothesis ...............................................................................................................................14
1.5 Outline of this research ............................................................................................................14
2 MATERIAL AND DYNAMIC PROPERTIES ..............................................................................16
2.1 Material properties ...................................................................................................................16
2.1.1 Aveiro University...........................................................................................................16
2.1.2 Catholic University (PUCP)...........................................................................................16
2.1.3 Los Andes University ....................................................................................................17
2.2 Dynamic properties..................................................................................................................19
2.2.1 Yield period ...................................................................................................................19
2.2.2 Relationship Yield period-height of the structure..........................................................21
2.2.3 Analytical models ..........................................................................................................21
2.2.4 Post-yield period ............................................................................................................24
2.3 Limit states – In plane behaviour.............................................................................................24
2.4 Evaluation of the equivalent viscous damping in adobe walls ................................................29
iv
Index
v
Index
vi
Index
LIST OF FIGURES
Page
Figure 2.1. Front and plan view of the adobe wall (specimen), [Blondet et al. 2005]. ......................... 19
Figure 2.2. Hysteretic force displacement relationship from a cyclic test. ............................................ 19
Figure 2.3. Relationship between K, E and h. ....................................................................................... 20
Figure 2.4. Location of accelerometers and displacement transducers (LVDT) for the dynamic test... 21
Figure 2.5. Analytical model for 1-storey building with two rooms. .................................................... 22
Figure 2.6. Analytical model of one wall. ............................................................................................. 23
Figure 2.7. 1-storey building with 1 room. ............................................................................................ 23
Figure 2.8. 2-storey building with 2 rooms............................................................................................ 23
Figure 2.9. 2-storey building with 4 rooms............................................................................................ 23
Figure 2.10. Comparison among period of vibrations. .......................................................................... 24
Figure 2.11. Elasto-plastic force displacement relationship [Crowley et al. 2006]............................... 24
Figure 2.12. Adobe wall after tested. Colour lines show the simulated crack pattern [Blondet et al.
2005]. ............................................................................................................................................ 27
Figure 2.13. Capacity curve for an adobe wall subjected to an in-plane force. ..................................... 28
Figure 2.14. Equivalent hysteretic damping from cyclic test for masonry and adobe walls. ................ 30
Figure 2.15. Limit state 1, drift= 0.052%, damping ratio= 10%............................................................ 30
Figure 2.16. Limit state 2, drift= 0.11%, damping ratio= 10%.............................................................. 31
Figure 2.17. Limit state 3, drift= 0.26%, damping ratio= 12%.............................................................. 31
Figure 2.18. Limit state 4, drift= 0.52%, damping ratio= 16%.............................................................. 31
Figure 3.1. Political division of Peru and Cusco. .................................................................................. 33
Figure 3.2. Percentage of adobe and clay brick masonry buildings in 1993 and 2005.......................... 34
Figure 3.3. Urban growth of Cusco city [Carazas 2001]. ...................................................................... 34
Figure 3.4. Cusco city in 2007............................................................................................................... 35
Figure 3.5. Rural adobe houses [Carazas 2001]. ................................................................................... 35
Figure 3.6. Number of rooms in adobe dwellings. ................................................................................ 36
vii
Index
viii
Index
Figure 5.7. Comparison of seismic hazard curves developed with different GMPEs. Acceleration
values on rock [Alva and Escalaya 2005]. .................................................................................... 75
Figure 5.8. Input data on CRISIS2007. ................................................................................................. 76
Figure 5.9. Seismic hazard map for Cusco. Acceleration values for soil condition and for Tr= 475
years. ............................................................................................................................................. 77
Figure 5.10. Seismic hazard curve......................................................................................................... 77
Figure 5.11. Acceleration response spectra for Cusco........................................................................... 77
Figure 5.12. Distribution of accelerations evaluated on rock for 50 years as exposure period having
10% of exceedance probability [Bolaños and Monroy 2004]....................................................... 79
Figure 5.13. Distribution of accelerations evaluated on rock for 100 years as exposure period having
10% of exceedance probability [Bolaños and Monroy 2004]....................................................... 80
Figure 6.1. Example of a Cumulative Distribution Function................................................................. 83
Figure 6.2. Example of a probability of occurrence within an interval. ................................................ 83
Figure 6.3. Example of a histogram....................................................................................................... 84
Figure 6.4. Example of Probability Density Functions [Velásquez 2006]. ........................................... 84
Figure 6.5. Demand for different return periods.................................................................................... 87
Figure 6.6. Comparison of capacity vs demand for a given return period............................................. 88
Figure 6.7. Example of fragility curves [Velázquez 2006].................................................................... 88
Figure 6.8. Fragility curves for in-plane behaviour. .............................................................................. 89
Figure 6.9. Out-of-plane: cracks at upper levels [Bariola and Sozen 1990]. ......................................... 90
Figure 6.10. Vertical cracks between adobe walls................................................................................. 90
Figure 6.11. Fragility curves for LS1: top displacement= 17 mm, width of horizontal crack= 3 mm. . 91
Figure 6.12. Fragility curves for LS2: top displacement= 40 mm, width of horizontal crack= 7 mm. . 91
Figure 6.13. Fragility curves for LS3: ∆ 1 ≈ 0.12 ∆ u with a standard deviation of 0.01. ..................... 91
ix
Index
LIST OF TABLES
Page
Table 2.1. Elasticity module (E) of adobe masonry............................................................................... 17
Table 2.2. Shear module (G) of the masonry......................................................................................... 17
Table 2.3. Compressive strength of the masonry (f’m).......................................................................... 17
Table 2.4. Compressive strength of adobe blocks (fo)........................................................................... 18
Table 2.5. Maximum shear strength of the masonry (v' m ) . .................................................................. 18
x
Index
xi
Chapter 1. Introduction
1 INTRODUCTION
Collapse of adobe buildings due to earthquakes has caused considerable loss of life in third
world countries. However, earth is the least expensive construction material and almost the
only one available to world population in rural areas [Bariola and Sozen 1990].
The seismic behaviour of adobe dwellings is not good. This material is too brittle and each
time that an earthquake occurs adobe dwellings are observed to collapse. Examples of the
high vulnerability of adobe houses were seen during Huaraz (1970, Peru), El Salvador (2001),
Bam (2003, Iran) and Pisco (2008, Peru) earthquakes, amongst others.
Until now many methods have been proposed for the evaluation of the seismic vulnerability
of buildings. Those methods have been classified in a general form as empirical, hybrid and
analytical/mechanics-based methods [Crowley et al. 2006].
12
Chapter 1. Introduction
In 2004 a group of researchers of the Catholic University of Peru used a screening method
(empirical) in order to evaluate the seismic vulnerability of adobe houses [Blondet et al.
2004]. The result of this preliminary research showed the high vulnerability of existing
buildings. Another research related to earthen buildings has been developed by Lourenco and
Oliveira [2005a], where three indexes (parameters) were defined to calculate the seismic
vulnerability of churches. Those projects seem to be the only ones that have been developed
for earthen buildings.
Considering masonry buildings (bricks) D’Ayala and Speranza [2002] have recently used a
procedure to assess the seismic vulnerability of historic buildings. The approach chosen is
based on a failure analysis of the structure through the identification of the feasible collapse
mechanism and the calculation of their associated failure load factor. Restrepo-Velez [2004]
has developed a method based on mechanical procedures (collapse failure mechanics,
MeBaSe) to evaluate the seismic risk of unreinforced masonry buildings.
The common failure type in adobe houses is due to the out-of-plane behaviour of the walls.
According to Doherty et al. [2002], this kind of failure is more sensitive to displacement than
to acceleration. For this reason it is thought that a good method to assess the seismic
vulnerability of these buildings was in establishing mechanical procedures related to a
displacement-based design (limit states).
1.2 Justification
It is important to know the current seismic risk of adobe dwellings in order to take some
measures of mitigation. With that the life, the material and the economical looses can be avoid
during earthquakes.
This research (based on mechanical procedures) has been calibrated using data from Peruvian
adobe dwellings. It is believed that this procedure can easily be applied to other construction
typologies with adobe.
The following are the specific objectives that can be individuated as part of this research:
• To gather information about static and dynamic tests made on adobe buildings, and to
define the material and dynamic properties (e.g. E, G, period of vibration, etc).
• To define the limit states for walls (in-plane and out-of-plane) based on the performance
levels during earthquakes.
13
Chapter 1. Introduction
• To calibrate a methodology for the seismic risk of adobe dwellings based on the work
developed by Restrepo-Velez [2004].
• To apply the methodology to the houses studied in a previous research done by Blondet et
al. [2004].
1.4 Hypothesis
It is possible to know the behaviour of adobe structures subjected to earthquakes knowing
their material and dynamic properties, and possible failure mechanisms.
The buildings selected to calibrate the method are representative of the Peruvian adobe
dwellings located in Cusco.
Chapter two evaluates static tests carried out in three universities, each of them located in
Peru, Colombia and Portugal. The scope in this part is to evaluate the mean values of the
material properties such as elasticity module, Poisson module, etc. Having a cyclic test on an
adobe wall carried out at the Catholic University of Peru (PUCP), dynamic properties such as
yielding period of vibration and damping are calculated. Even analytical models have been
developed in SAP2000 to evaluate an equation that relates the period of vibration with the
height of the adobe wall. Taking advantage of the dynamic test, limit states just for the in-
plane behaviour are developed here.
In chapter three, information regarding thirty surveyed adobe buildings located in Cusco is
gathered. Then, the mean and standard deviation of the principal geometrical properties such
as length, thickness and height of the adobe walls are calculated and histograms are plotted.
Finally, it is selected the best probability density function for each of the geometrical
properties.
The fourth chapter analyzes expressions for the out-of-plane and in-plane capacity of adobe
dwellings. For the first one, collapse mechanisms are evaluated as they have an influence on
the period of vibration. The limit states are generated considering the static equilibrium of
cantilever walls. For the in-plane capacity, the limit states and expressions for the period of
vibrations are rewritten from chapter two; however, the capacity regarding displacement is
evaluated taking into account an equivalent SDOF system.
14
Chapter 1. Introduction
The fifth chapter solves the probabilistic seismic hazard assessment of Cusco considering a
soil type condition. CRISIS2007 is the software used to generate seismic hazard curves
associated with forty-six return periods from 1 to 1000 year. Taking into account a uniform
hazard spectra, acceleration response spectra and displacement response spectra are generated
for each return period.
In the sixth chapter of this report, an application of the Displacement-Based Earthquake Loss
Assessment (DBELA) methodology is presented. Using the Monte Carlo Simulation technique
and the probabilistic density function of the geometrical properties, 1000 buildings are
randomly created. The capacity of the artificial building stock is compared with the demand
for each displacement response spectrum that is associated for a given return period. The
probability of failure for each return period and for each limit states is evaluated. Finally,
fragility curves are created for each limit state in terms of Peak Ground Acceleration (PGA)
and unconditional seismic risk is evaluated for different time windows.
The conclusions and further research needed to understand better the behaviour of adobe
dwellings are given in chapter seven.
15
Chapter 2. Material and dynamic properties
Based on the in-plane behaviour of adobe walls, an equation that relates the height of the wall
has been written to evaluate the yield period of vibration. Finally, some limit states for the in-
plane behaviour have been analyzed and established for the scope of this report.
16
Chapter 2. Material and dynamic properties
Table 2.1 to Table 2.5 show the mechanical properties of the adobe. Just for Table 2.4 and
Table 2.5 it is also shown (together with the three previous references) the data obtained from
the investigation on the adobe buildings of Aliano [Basilicata, Italy – Liberatore et al. 2006].
17
Chapter 2. Material and dynamic properties
As shown in the tables, the mechanical properties of adobe blocks and masonry do not have
similar values for all the regions. This is because of the variation in properties of the soil
material. Despite this, it can be seen that the compressive strength of masonry (f’m, Table 2.3),
the compressive strength of adobe blocks (fo, Table 2.4), and the maximum shear strength of
masonry (v’m, Table 2.5) seem to have similar mean values for Aveiro and PUCP results.
18
Chapter 2. Material and dynamic properties
Figure 2.1. Front and plan view of the adobe wall (specimen), [Blondet et al. 2005].
With the force vs. displacement relationship (Figure 2.2) the elastic stiffness of the wall has
obtained. Considering the point (1.38 mm, 35 kN) as the end of the elastic part, the elastic
stiffness of the wall results in k y = F y / ∆ y = 35 × 1000 / 1.38 = 25360 kN − m .
50
40
30
20
Force (kN)
10
0
-60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60
-10
-20
-30
-40
Displacement (mm)
In this test the weight of the wall was around 135 kN. So now the yield period of vibration
can be evaluated as T y = 2π m / k → T y = 0.15 s .
19
Chapter 2. Material and dynamic properties
The elasticity module E can be evaluated directly from the elastic stiffness. The deformation
at the top of the wall can be assumed as the summation of the flexural and shear deformation
(cantilever beam, see equation 1). Besides, G= 0.4 E which means Poisson ratio = 0.20.
1
k= 3
(2.1)
h h
+
12 EI GAv
Solving Eq.(2.1) for I= 3.21 m4, Av= 0.92 m2 and h= 1.93 m it is obtained E= 135 MPa, a
value between the 16th and 50th% of the data analyzed. The explanation of why this real
elasticity module is less than those specified in Table 2.1 is due to the reduction in the
stiffness rigidity for having cracked sections. As a preliminary statement it can be said that a
reduced elasticity module has to be used for modelling sections in numerical models (see
section 2.2.3).
Figure 2.3 shows the dependency of the elastic stiffness K with the elasticity module E and
the height h. As it is seen, the K depends strongly on the E and less on the h; however, both
values are important to take into account for the evaluation of the period of vibration T, which
depends on the mass (m) and K.
50.E+3
E= 100 MPa
E= 150 MPa
45.E+3
E= 170 MPa
E= 200 MPa
40.E+3
Elastic stiffness K (kN/m)
E= 220 MPa
35.E+3
30.E+3
25.E+3
20.E+3
15.E+3
10.E+3
1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5
Height h (m)
The second test was a dynamic test [Blondet et al. 2006a] and it was carried out on a full
scaled adobe building (Figure 2.4) over a shaking table. The dimensions of the specimen were
3.21 x 3.21 m. in plan and 1.98 and 2.25 m. in elevation. The window openings had
dimensions of 1 m. x 1 m. and they were located in the two walls parallel to the excitation.
The door opening (1 m. x 1.6 m.) was located at one of the transverse walls.
20
Chapter 2. Material and dynamic properties
Figure 2.4. Location of accelerometers and displacement transducers (LVDT) for the dynamic test.
The period of vibrations were measured in each wall during a free vibration test before
imposing any load. The results are shown in Table 2.6.
Accelerometer
A1 A2 A4 A5
Period T (s) 0.167 0.151 0.121 0.167
According to the previous experimental values of yield period and the respective heights of
the specimens it is possible to find the coefficients α and β .
T y = 0.090 H 3 / 4 (2.2)
21
Chapter 2. Material and dynamic properties
Considering an average height of 2.10 m. for 1-story building and substituting this into Eq.(2.
2), obtained from experimental data, a yield period of T y = 0.16 s. is obtained.
The period of vibration of the model shown in Figure 2.5 was 0.14 s. in X-X and 0.18 s. in Y-
Y. Due to the flexibility of the roof the walls move independently of each other. For that
reason the modal participation factors (MPF) are lower; 30% for X-X and 56% for Y-Y.
Figure 2.5. Analytical model for 1-storey building with two rooms.
The model of Figure 2.6 was created in order to analyze the behaviour of only one wall. The
period of vibration in X-X was 0.18 s. with a MPF of 37%. Besides, there is a period of 0.086
s. with a MPF of 23% in the same direction. In Y-Y the period was 0.19 s. and the MPF was
55%. In Figure 2.7 another adobe building model is shown. The period of vibration in this
case was 0.142 s in X-X and 0.13 in Y-Y. The MPF was 59 and 60%, respectively.
As shown before, the yield periods obtained with SAP2000 [CSI 2005] are almost the same as
those evaluated in the equation 2.2 ( T y = 0.16 s. ).
22
Chapter 2. Material and dynamic properties
Figure 2.6. Analytical model of one wall. Figure 2.7. 1-storey building with 1 room.
To verify the validation of the Eq.(2.2) for 2-storey adobe buildings two models in SAP2000
were analyzed. In Figure 2.8 a model with 3.21 x 3.21 m. in plan and a total height of 4 m. is
shown The period of vibration in X-X was 0.25 s. and in Y-Y it was 0.265 s; with a MPF of
68 and 71%, respectively.
The model in Figure 2.9 has dimensions in plan of 3.21 x 6.0 m. and a total height of 4 m. The
period of vibration in X-X was 0.21 s. and in Y-Y it was 0.31 s. In both cases the MPF was
almost 70%. According to the Eq.(2.2), the elastic period of vibration in both cases has to be
T y = 0.26 s As shown here, the equation fits more or less well the experimental and analytical
yield periods of vibration (Figure 2.10).
Figure 2.8. 2-storey building with 2 rooms. Figure 2.9. 2-storey building with 4 rooms.
23
Chapter 2. Material and dynamic properties
0.30
0.25
0.20
Period (s)
0.15
Equation
0.10
Experimental
SAP2000
0.05
0.00
1.60 1.93 2.10
2.10 2.25 2.60 3.10 3.60 4.00 4.10
Height (m)
The effective period ( TLSi ) of the inelastic structure can be represented as a function of the
elastic period ( T y ) and ductility ( µ LSi ). This relationship is shown in Eq.(2.3), [Crowley et al.
2006].
TLSi = T y µ LSi
(2.3)
24
Chapter 2. Material and dynamic properties
The test carried out by Blondet et al. [2005] consisted of an unreinforced wall (with a vertical
load simulating the weight of the roof) subjected to cyclic load to eight different levels of
deformation. The height of this wall was 1.93m. The sequence of damage was as follows (see
Table 2.7 and Table 2.8):
Nine small diagonal cracks (thickness around 0.25 mm.) started to appear up to 0.05% of drift
( ∆ = 1 mm.). Then, there was a slight reduction of the wall strength, almost constant until
0.1% of drift ( ∆ = 2.12 mm.). After that, more diagonal cracks started to appear and the
previous ones became thicker than before.
From 0.1% to 0.26% of drift ( ∆ = 2.12 mm. and 5 mm, respectively) there was no increment
of the lateral force (plateau, yielding zone), having a maximum value of 37.5 kN. At the end
of this deformation new diagonal cracks appeared around the opening. The thickness of the
cracks ranged from 0.15 to 1.6 mm, but two of them reached 4 mm.
After the drift of 0.26% a clear loss of resistance was registered, which means the beginning
of instability on the wall. The curve force vs. displacement plotted a descending branch.
Thicker diagonal cracks formed and some horizontal cracks started to appear.
Cracks of 10 mm. and 50 mm. of thickness appeared at a drift of 0.52% and 1%, which mean
the beginning or even the collapse of the wall. Figure 2.12 shows the final state of the adobe
wall.
Table 2.7. Summary of the behaviour of the wall during the test.
25
Chapter 2. Material and dynamic properties
Table 2.8. Sequence and thick of cracks (during phase 1 no appeared cracks).
26
Chapter 2. Material and dynamic properties
Figure 2.12. Adobe wall after tested. Colour lines show the simulated crack pattern [Blondet et al. 2005].
27
Chapter 2. Material and dynamic properties
Figure 2.13 shows the normalized capacity curve of the adobe wall and its respective drift
limits. Until 0.05% drift the structure can be considered as elastic (LS1), which means fully
operational. After that the structure can have some cracks but is still functional (LS2) until
0.1% of drift. Then the life-safety performance (LS3) is reached at 0.26% of drift and finally
the structure is considered near to collapse or collapsed at 0.52% of drift.
0.045
0.040
LS1 LS4
0.025 0.52 %
0.052%
0.020
0.015
0.010
0.005
0.000
0.00 0.25 0.50 0.75 1.00
-02
Drift (x10 )
Figure 2.13. Capacity curve for an adobe wall subjected to an in-plane force.
According to the previous analysis the four limit states have been adapted for 1-story adobe
masonry buildings and are described as follows:
LS1: Only very minor damage has occurred. The building retains its original stiffness and
strength. The behaviour of the building is essentially elastic and stable. The risk of life-
threatening injury during the earthquake is negligible.
LS2: Minor structural damage can be seen as slightly diagonal cracks. The structure retains
nearly all its original stiffness and strength. Repairs may be instituted at the convenience of
the building users. The risk of life-threatening injury during the earthquake is very low.
LS3: Significant structural damage is reported. Beginning of horizontal cracks. Some adobe
bricks have been cut. The building has lost a significant amount of its original stiffness, but
retains some lateral strength and margin against collapse. The building cannot be used after
the earthquake without significant repair. The risk of life-threatening injury during the
earthquake is low.
LS4: Near collapse or collapse of the building is expected. Repairing the building is neither
possible nor economically reasonable. The structure will have to be demolished after the
earthquake. Beyond these LS global collapse with danger for human life has to be expected.
Similar to Calvi [1999], five situations can result from the comparison of an expected demand
and the capacities discussed above.
28
Chapter 2. Material and dynamic properties
Ah
ξ hyst =
4 ⋅ π ⋅ Ae (2.4)
where Ah is the area within a complete cycle of stabilized force-displacement response, and Ae
is the elastic area. Typically, the dissipated energy in each cycle evolves with the increase of
damage and with the increase of displacement demand [Magenes and Calvi 1997].
In the work developed by Magenes and Calvi [1997], equivalent damping ratios for masonry
walls were evaluated from cyclic and pseudo-dynamic tests and considering flexural response,
diagonal shear cracking response and shear sliding response. For the shear cracking response
it was found that greater values of equivalent damping ratios are obtained if a cyclic test is
considered as opposed to a pseudo-dynamic test, the ratio of the two values is found to be
around 1.34. Furthermore, it was seen now a larger amount of hysteretic energy is dissipated
in the first cycle, while the remaining cycles tend to show lower dissipation.
According to Calvi [1999], 2% of damping can be assumed for LS1 and LS2 for masonry
buildings. However, Priesley et al. [2007] suggest a value of 5%. For LS3 and LS4, damping
ratios of 10% and 15% are assigned. The drift limits associated to each one of the limit sates
for masonry are 0.1%, 0.1%, 0.3% and 0.5%, respectively.
For the adobe wall two values of equivalent damping have been computed evaluating the
energy dissipated by the first and second cycle of each hysteretic curve (see Figure 2.15). The
resulting values related to each of the drift limits are shown in Table 2.9.
ξ (%)
Limit state drift
1st cycle 2nd cycle average Assumed
LS1 0.052% 16.5 11.0 13.8 10.0
LS2 0.110% 18.0 11.2 14.6 10.0
LS3 0.260% 18.5 12.5 15.5 12.0
LS4 0.520% 20.0 12.2 16.1 16.0
29
Chapter 2. Material and dynamic properties
Figure 2.14 shows the damping ratios obtained from masonry wall tests [Magenes and Calvi
1999]. As it can be seen the equivalent damping ratios suggested for LS1 and LS2 (0.1%
drift), and LS3 (0.3% drift) are less than those obtained from the masonry hysteretic curves,
while for the LS4 of masonry, the damping ratio is related to the first cycle of the test.
Looking into these aspects considering the damping ratios for masonry and knowing the
limitations to infer in damping values from just one cyclic test, the average equivalent
damping ratios for adobe walls has been adjusted to 10%, 10%, 12% and 16% for LS1, LS2,
LS3 and LS4, respectively (Table 2.9).
25.0
20.0
Equivalent damping
15.0
10.0
Adobe-1st cycle
Adobe-2nd cycle
Adobe-damping
5.0
Masonry-1st cycle
Masonry-2nd cycle
Masonry-damping
0.0
0 0.001 0.002 0.003 0.004 0.005 0.006
Drift
Figure 2.14. Equivalent hysteretic damping from cyclic test for masonry and adobe walls.
Figure 2.15 to Figure 2.18 show the hysteretic curves obtained for the different limit states for
an adobe wall.
50 50
40 40
30 30
20 20
10 10
Force (kN)
Force (kN)
0 0
‐25 ‐20 ‐15 ‐10 ‐5 0 5 10 15 20 25 ‐1.5 ‐1 ‐0.5 0 0.5 1 1.5
‐10 ‐10
‐20 ‐20
‐30 ‐30
‐40 ‐40
Displacement (mm) Displacement (mm)
30
Chapter 2. Material and dynamic properties
50 50
40 40
30 30
20 20
10 10
Force (kN)
Force (kN)
0 0
‐25 ‐20 ‐15 ‐10 ‐5 0 5 10 15 20 25 ‐2.5 ‐2 ‐1.5 ‐1 ‐0.5 0 0.5 1 1.5 2 2.5
‐10 ‐10
‐20 ‐20
‐30 ‐30
‐40 ‐40
Displacement (mm) Displacement (mm)
50 50
40 40
30 30
20 20
10 10
Force (kN)
Force (kN)
0 0
‐25 ‐20 ‐15 ‐10 ‐5 0 5 10 15 20 25 ‐5.5 ‐4.5 ‐3.5 ‐2.5 ‐1.5 ‐0.5 0.5 1.5 2.5 3.5 4.5 5.5
‐10 ‐10
‐20 ‐20
‐30 ‐30
‐40 ‐40
Displacement (mm) Displacement (mm)
50 50
40 40
30 30
20 20
10 10
Force (kN)
Force (kN)
0 0
‐25 ‐20 ‐15 ‐10 ‐5 0 5 10 15 20 25 ‐15 ‐10 ‐5 0 5 10 15
‐10 ‐10
‐20 ‐20
‐30 ‐30
‐40 ‐40
Displacement (mm) Displacement (mm)
An elastic behaviour until the LS1 has been assumed. However, in the Figure 2.15 a kind of
hysteretic behaviour is seen, which means that the Eq.(2.4) can be used to evaluate even the
equivalent viscous damping ratio for this limit state.
Table 2.10 summarizes the drifts, equivalent damping and ductility values for the adobe limit
states.
31
Chapter 2. Material and dynamic properties
32
Chapter 3. Typology of adobe dwellings in Cusco
A data base containing the principal geometrical properties (such as height and thickness of
walls, dimensions of openings, etc) of the dwellings has been created. With the previous
information it is possible to have a representation of the geometrical data of adobe dwellings
(mean and standard deviation) through probabilistic density functions, which will be useful as
input data to generate artificial data. Finally, pictures of the surveyed dwellings are shown to
have an idea of the typology of adobe dwellings.
33
Chapter 3. Typology of adobe dwellings in Cusco
According to the last census [INEI 2005], adobe buildings have decreased from 43% to 37%
at a national level (Figure 3.2a) from 1993 to 2005. However, the region of Cusco (with 1 171
503 inhabitants and 348 493 inhabitants in the province) maintains almost 80% of adobe
houses (Figure 3.2b). This describes with certainty than in Cusco people build with adobe as a
principal material and with clay bricks as a second material (Figure 3.2c).
50% 90%
Clay brick or cement block Adobe block or tapial Clay brick or cement block Adobe block or tapial
45% 80%
40%
70%
% of total dwellings
% of total dwellings
35%
60%
30%
50%
25%
40%
20%
15% 30%
10% 20%
5% 10%
0% 0%
1993 2005 1993 2005
Year Year
60%
50%
40%
30%
20%
10%
0%
1993 2005
Year
c) Province of Cusco
Figure 3.2. Percentage of adobe and clay brick masonry buildings in 1993 and 2005.
The Figure 3.3 shows the urban growth of Cusco city from 1924 to 1995, while Figure 3.4
shows the actual situation of Cusco.
34
Chapter 3. Typology of adobe dwellings in Cusco
Adobe has good thermal properties. It retains the warmth of the environment during the day
and releases it at night. With that the house always is warm even in seasons with low
temperature. For that reason the construction in adobe is common in the highland.
However, adobe dwellings have high seismic vulnerability due to an undesirable combination
of mechanical properties of dry earth: 1) earthen structures are massive and thus attract large
inertia forces, 2) they are weak and cannot resist these forces, and 3) they are brittle and break
without warning [Blondet et al. 2006a].
According to Carazas [2001] the rural zone in Cusco (periphery of the city) has a huge
influence of pre-Hispanic characteristics. This means adobe dwellings of only one level and
with two rectangular rooms, one for social activities such as cooking or eating and the other
one for sleeping. The entrance is defined with an opening between the two rooms (Figure
3.5).
There is no statistical data about the quantity of adobe houses of one or more stories in Cusco.
However, it can be seen from Figure 3.6 that the majority of adobe houses has 1 or 2 rooms.
Indirectly it can be concluded that those houses, or at least the majority of them, have only
one level. According with that in the department of Cusco there are more than 50% of 1-
35
Chapter 3. Typology of adobe dwellings in Cusco
storey adobe dwellings. However, this number increases referring to the province (with
348493 inhabitants), where the value reaches almost 60%.
45% 45%
1
40% 40%
35% 35%
% of total dwellings
% of total dwellings
1 2 1 2 1
30% 30%
25% 25%
2 2
20% 3 3 20%
4 3 4 3
15% 4 15% 4
10% 10%
5% 5%
0% 0%
1993 2005 1993 2005
Year Year
In the reference Blondet et al. [2004] 30 adobe houses located in the province of Cusco has
been surveyed. These dwellings are shown in Figure 3.7.
ID 1 ID 2
ID 3 ID 4
36
Chapter 3. Typology of adobe dwellings in Cusco
ID 5 ID 6
ID 7 ID 8
ID 9 ID 10
ID 11 ID 12
37
Chapter 3. Typology of adobe dwellings in Cusco
ID 13 ID 14
ID 15 ID 16
ID 17 ID 18
ID 19 ID 20
38
Chapter 3. Typology of adobe dwellings in Cusco
ID 21 ID 22
ID 23 ID 24
ID 25 ID 26
ID 27 ID 28
39
Chapter 3. Typology of adobe dwellings in Cusco
ID 29 ID 30
Figure 3.7. Photos of the surveyed dwellings.
A summary of the principal characteristics such as dimensions of those houses are shown in
Table 3.1 and Table 3.2. With that data it is possible to define the mean values and the
standard deviations of dimensions such as thickness, height, length of walls and openings
(Table 3.3 and Table 3.4). In the case of the 2-storey building it has been assumed that the
height of each floor is half of the total height.
40
Chapter 3. Typology of adobe dwellings in Cusco
Table 3.2. Characteristics of surveyed dwelling – gable, opening and room dimensions.
41
Chapter 3. Typology of adobe dwellings in Cusco
42
Chapter 3. Typology of adobe dwellings in Cusco
According to the previous data, it is concluded that the mean value of the wall thickness is
0.44 m. and the wall height is 2.45 m. and 4.88 m. for buildings of one and two levels,
respectively (Table 3.3). These mean heights have been calculated without considering the
height of the gable. The thickness of the wall seems to be uniform due to the low standard
deviation (0.04 m). Besides, Table 3.4 shows the statistics obtained from the adobe blocks,
where it is seen that the mean values are 0.44, 0.21 and 0.15 m. for the length, width and
thickness, respectively.
43
Chapter 3. Typology of adobe dwellings in Cusco
Table 3.5 shows the mean and standard deviation values for the gables and openings
dimensions. Also, the average values of the length and width measurements for a typical
adobe room are shown. With these values it is concluded that the mean gable height is around
1.33 m, the dimensions of doors are 1.08 m. (width) and 1.80 m. (height) and the dimensions
of windows are 1.07 m. (width) and 1 m. (height). Typically the rooms have dimensions in
plan of 4.53 x 5.38 m.
Openings Room
Gable
Door Window Length Width
Width Height Width Height
(m) (m) (m) (m) (m) (m) (m)
16th% 1.20 0.90 1.80 0.80 1.00 4.79 4.00
50th% 1.40 1.00 1.80 1.00 1.00 5.00 4.43
84th% 1.50 1.20 1.90 1.40 1.00 6.00 5.00
Mean 1.33 1.08 1.80 1.07 1.00 5.38 4.53
SD 0.18 0.18 0.15 0.28 0.13 1.10 0.59
The variability of each of the parameters in this study (i.e. length of walls, width of adobe
bricks, etc) represented in histograms plots can also be adapted to a probability density
function (PDF, see section 6.2.2). The loss assessment method DBELA [Crowley et al. 2006]
-see section 6.4- allows the variation in the geometrical, material and limit state properties
(used in the displacement-based definition of buildings class vulnerability) to be directly
modelled using probabilistic/statistical distributions [e.g. Bal et al. 2007].
44
Chapter 3. Typology of adobe dwellings in Cusco
5% and 1%). It is noted that different histograms shape may be obtained if the data are
grouped considering different interval sizes.
3
# of Buildings
3 # of Data : 26
2 # of Buildings : 30
2 Mean : 2.44
1 COV (%) : 8
0
2.10
2.20
2.30
2.40
2.50
2.60
2.70
2.80
Storey height (m)
Figure 3.8. The histogram and PDF of total storey height for 1-storey buildings.
5
Observed
4
LgNorm. Dist.
4
3
# of Buildings
3 # of Data : 22
2 # of Buildings : 30
2 Mean : 4.85
1 COV (%) : 7
1 Chi-Squ. Test : 10%, satisfied
0
4.20
4.40
4.60
4.80
5.00
5.20
5.40
5.60
Figure 3.9. The histogram and PDF of total storey height for 2-storey buildings.
16
Observed
14
LgNorm. Dist.
12
10
# of Data : 30
# of walls
8
# of Buildings : 30
6
Mean : 5.38
4 COV (%) : 20
0
3.70
4.30
4.90
5.50
6.10
6.70
7.30
7.90
8.50
9.10
Figure 3.10. The histogram and PDF of wall length for walls located perpendicular to the façade.
45
Chapter 3. Typology of adobe dwellings in Cusco
16
Observed
14
LgNorm. Dist.
12
10
# of Data : 30
# of walls
8
# of Buildings : 30
6
Mean : 4.53
4 COV (%) : 13
2 Chi-Squ. Test : Not Satisfied
3.50
4.70
5.90
Length of walls (m)
Figure 3.11. The histogram and PDF of wall length for walls located parallel to the façade.
14
Observed
12
Exp Dist.
10
# adobe bricks
8 # of Data : 30
6 # of Buildings : 30
Mean : 0.44
4
COV (%) : 9
2 Chi-Squ. Test : Not Satisfied
0
0.40
14
Observed
12
Exp Dist.
10
# adobe bricks
8 # of Data : 31
6 # of Buildings : 30
Mean : 0.22
4
COV (%) : 9
2 Chi-Squ. Test : Not Satisfied
0
0.20
0.25
12
Observed
10 Norm. Dist.
8
# of Data : 29
# of gables
6
# of Buildings : 30
4 Mean : 1.33
COV (%) : 13
2
Chi-Squ. Test : 10%, satisfied
0
1.00
1.20
1.40
46
Chapter 3. Typology of adobe dwellings in Cusco
25
Observed
20 LgNorm. Dist.
15
# of Data : 66
# of doors
# of Buildings : 30
10
Mean : 1.08
COV (%) : 17
5
Chi-Squ. Test : 10%, satisfied
0.80
1.00
1.20
1.40
1.60
Width of doors (m)
60
Observed
50 Norm. Dist.
40
# of Data : 66
# of doors
30
# of Buildings : 30
20 Mean : 1.80
COV (%) : 9
10
Chi-Squ. Test : Not Satisfied
0
1.50
1.70
1.90
25
Observed
20 Norm. Dist.
15
# of windows
# of Data : 84
# of Buildings : 30
10
Mean : 1.07
COV (%) : 26
5
Chi-Squ. Test : Not Satisfied
0
0.60
1.00
1.40
1.80
80
Observed
70
LgNorm. Dist.
60
50
# of windows
# of Data : 84
40
# of Buildings : 30
30
Mean : 1.00
20 COV (%) : 13
10 Chi-Squ. Test : Not Satisfied
0
0.80
1.00
1.20
1.40
47
Chapter 3. Typology of adobe dwellings in Cusco
Looking into Figure 3.12 and Figure 3.13 a uniform distribution can be assumed instead the
exponential one. For Figure 3.16 and Figure 3.18 a unique value for the height door and
height window can be considered for analysis.
48
Chapter 4. Capacity of adobe dwellings
In this chapter the out-of-plane and in-plane capacity of adobe walls (expressed by the
displacement capacity and period of vibration) have been evaluated. For the former a collapse
multiplier based on a collapse mechanism was defined. Furthermore, it was possible to relate
the collapse multiplier to the resistant force and then to the stiffness and period of vibration.
Limit states based on displacements for the out-of-plane resistant were established as function
of the walls thickness. For the in-plane behaviour the period of vibration was evaluated in
chapter 2, where an expression considering different limit states based on drift limits were
created.
According to the Adobe Peruvian Code [NTE E.080 2000], slenderness values less than 6 can
result in stable walls with an addition of a collar ring-beam. While, slenderness greater than 6
results in an unstable wall and the addition of vertical and horizontal reinforcement is
compulsory.
On 15th August 2007, an earthquake of Mw= 8.0 [USGS 2007] occurred in Pisco (south
central part of Peru). Around 80% of the adobe dwellings collapsed and the majority of them
due to the lack of proper connection between adobe walls (Figure 4.2). Other factors that
49
Chapter 4. Capacity of adobe dwellings
could have influenced the dwellings collapse are the soil conditions, the low quality of
materials, thinner adobe walls, inadequate locations of openings, etc [Tarque et al. 2007].
One important fact to remark is that in Pisco the adobe walls are thinner (~ 0.25 m) and the
mean height can be assumed as 2.50 m, which results in an slenderness value of 10.
Buildings located at the corner of the square suffered partial collapse of the two façades. In
this case the cracks are diagonal starting at the top of the building and ending at the base as a
V shape (Figure 4.3).
50
Chapter 4. Capacity of adobe dwellings
Figure 4.4. Unreinforced masonry walls support configurations [Doherty et al. 2002].
It is important to remark that out-of-plane walls tend to behave as rigid bodies subjected to
rocking and are more sensitive to displacement than acceleration [Restrepo 2004].
The capacity of the URM walls (cantilever or simple supported walls) for an ultimate limit
state is evaluated taking into account the secant stiffness ( K 2 ) of the wall and the ultimate
displacement ( ∆ u ≈ t ), measured at the top or at the mid-height of walls, for parapets or
simply supported walls, respectively. This capacity can be directly compared to the
Displacement Response Spectrum (DRS) considering a 5% damping for maximum
displacements greater than 0.5∆ u [Griffith et al. 2003]. For maximum displacements less than
0.5∆ u the stiffness can be represented as function of ∆ 1 (see section 4.1.3). The maximum
displacement is referred to the ordinates of the DRS.
It can be assumed that displacement demand can be estimated via a simplified approach
which makes use of elastic displacement response spectra [Doherty et al. 2002].
4.1.2 Demand
For the out-of-plane behaviour the ultimate displacement is measured at the top of the wall
because we are considering cantilever walls without any collar ring-beam over the walls.
51
Chapter 4. Capacity of adobe dwellings
If the wall is located above the first level, it is logical to think that the input demand at the
ground floor should be amplified by the effect of the height (ground-floor acceleration). To
evaluate this amplification some equations have been written in national codes, where not
necessarily it is indicated that those are applied for walls above the first floor, even those can
be applied to walls located on the ground floor. For example the Euro-Code 8 [CEN 2003]
gives the following expression, Eq.(4.1):
a g S ⎡ 3(1 + Z / H ) ⎤ ag S
Sa = ⎢ − 0.5⎥ ≥
g ⎣⎢1 + (1 − Ta / T1 ) 2
⎦⎥ g (4.1)
where a g is the peak ground acceleration, g is the gravity acceleration, S is a soil factor, Z is
the height from the foundation to the centroid of the weight forces applied on rigid elements,
H is the height of the structure, Ta the period of vibration of the wall and T1 the period of
vibration of the structure.
The Italian Code [OPCM 3431:2005] gives directly a Displacement Response Spectrum for
design as in Eq.(4.2):
⎛ ⎞
⎜ 3⎛1 + Z ⎞ ⎟
⎜ ⎟
TS2 ⎜ ⎝ H ⎠ ⎟
TS < 1.5T1 ∆ d (Ts ) = a g S ⎜ − 0. 5 ⎟
4π 2 ⎜ ⎛ T ⎞ 2 ⎟
⎜ 1 + ⎜⎜1 − ⎟⎟
s
⎟
⎝ ⎝ T1 ⎠ ⎠
1.5T1 TS ⎛ Z⎞
1.5T1 ≤ TS < TD ∆ d (Ts ) = a g S ⎜ 1 .9 + 2 .4 ⎟
4π 2 ⎝ H⎠ (4.2)
1.5T1 TD ⎛ Z⎞
TD ≤ TS ∆ d (Ts ) = a g S ⎜ 1 .9 + 2 .4 ⎟
4π 2 ⎝ H⎠
where a g is the peak ground acceleration, S is a soil factor, Z is the height from the
foundation to the centroid of the weight forces applied on the rigid bodies, H is the total
height of the building form the foundation, TS the period of vibration of the wall and T1 the
first period of vibration of the structure along the analysis direction, and TD is the corner
period.
The two expressions do not seem to be calibrated. They produce inconsistencies when applied
to the estimation of displacement demand on isolated blocks situated at ground levels. For
example, evaluating the corresponding Acceleration Response Spectrum from EC-8 and from
the Italian code, multiplying it by T 2 / 4π 2 to get the corresponding Displacement Response
Spectrum for design (DRSd) and then this DRSd is compared with those calculated from
Eq.(4.1) and (4.2), a de-amplification instead an amplification is calculated.
In the updated Italian Code [NTC 2008] Eq.(4.2) has been deleted and instead of that there are
some Explanatory Notes (unofficial version until the submission of this dissertation) where it
52
Chapter 4. Capacity of adobe dwellings
is mentioned that for walls located at the ground floor, the demand for evaluating the out-of-
plane should be the same as the one for evaluating the in-plane behaviour with 5% viscous
damping.
Figure 4.5. Trilinear idealization of the static force-displacement relationship [Griffith et al. 2003].
∆ 1 is related to the end of the initial stiffness and ∆ 2 is related to the secant stiffness. ∆ u is the
ultimate displacement, which means the point of static instability (ultimate limit state). From
static equilibrium ∆ u ≈ t for cantilever or simple supported walls.
Displacements greater than ∆ u mean that the wall will collapse. Fo = λW is the force at
incipient rocking and is also called the “Rigid Threshold Resistance”, λ is the collapse
multiplier factor (see section 4.1.5).
From simple static equilibrium of the parapets and simple supported walls, the ultimate
displacement at the top and at the mid-height of the walls can be obtained, respectively
(Figure 4.6). In both cases the ultimate displacement is equal to the wall thickness, ∆ u = t . At
the equivalent height, the equivalent ultimate displacement is represented as (2 / 3) ⋅ t .
The lateral static strength ( F ) and the ultimate displacement ( ∆ u ) are not affected by
uncertainties in properties such as the elasticity module, whereas geometry, boundary
conditions and applied vertical forces are the essential parameters [Griffith et al. 2003].
The ∆ 1 and ∆ 2 parameters can be related to the material properties and the state of
degradation of the mortar at the pivot points as a proportion of ∆ u (Table 4.1). For adobe
dwellings in Cusco a moderate degradation can be assumed.
53
Chapter 4. Capacity of adobe dwellings
Figure 4.6. Inertia forces and reactions on rigid URM walls [Doherty et al. 2002].
Table 4.1. Displacement ratios for the tri-linear model [Griffith et al. 2003].
The ultimate limit state is related to the complete stability or the collapse of adobe walls,
which means displacement at the top of walls less or greater than the ultimate displacement.
Since we are considering collapse mechanisms A, C and D -where walls are rotating at the
base- a conservative value of ∆ u ≈ 0.8 ⋅ t can be assumed, where some of the reasons for the
reduction are the consideration of dynamic effects and degradation in walls [Restrepo 2004].
In this case the secant stiffness K 2 is considered for the calculation of the period as suggested
by [Griffith et al. 2003].
Knowing that adobe walls will have cracks at the base before they collapse, another
intermediate limit state can be created. For this, the initial stiffness K 1 should be considered
when we are dealing with maximum displacements less than 0.5 ∆ u [Griffith et al. 2003]. The
following limit states described in Table 4.2 have been assumed for the out-of-plane
behaviour. The top displacements and crack width have been calculated considering mean
values of thickness and height of the adobe walls.
Table 4.2. Limit states for adobe walls subjected to out-of-plane forces.
Limit state Top displacement Crack width at the base, Eq.(4.16) ζ (%)
LS-1 ≈ 17 mm. ≈ 3 mm. 5
LS-2 ≈ 40 mm. ≈ 7 mm. 5
LS-3 ∆ 1 ≈ 0.12 ∆ u + σ SD ≈ 45 mm, σ SD =0.01 ≈ 8 mm. 5
54
Chapter 4. Capacity of adobe dwellings
The relationship between top displacement and crack width is described further in the next
section. The LS1, LS2 and LS3 indicate the beginning and increment of vertical cracks at the
edges of perpendicular walls, which can lead to the separation of them. The ultimate limit
state indicates the loss of static stability for the walls.
4.1.4 Capacity
The scope in this step will be the definition of the period of vibration for a given limit state.
Then, with the displacement known as described in the previous section and with the period
of vibration, it can be possible to compare the capacity with the demand for each limit state.
In this case is not necessary to go from a MDOF system to a SDOF one because we are going
to analyze the displacement at the top of the wall. The tri-linear representation of the non-
linear response of the wall can be given in terms of ultimate displacement at the top
and Fo = λW , where λ is the collapse multiplier factor (see section 4.1.5).
Following the work of Griffith et al. [2003], the lateral static strength F can be evaluated
using Eq.(4.3) and the secant stiffness K 2 by Eq.(4.4), where Fo = λW is the force necessary
to trigger overturning.
⎛ ∆ ⎞
F = Fo ⎜⎜1 − 2 ⎟⎟
⎝ ∆u ⎠ (4.3)
F
K ∆2 =
∆2 (4.4)
The lateral static strength F and the ultimate displacement ∆ u of a wall subjected to out-of-
plane action are not affected significantly by uncertainties in the material properties as the
elasticity module or the masonry compressive strength, whereas geometry, boundary
conditions and applied vertical forces (including self weight) are the essential parameters
[Griffith et al. 2003].
For the ultimate displacement is used the secant stiffness K 2 because it is a valid parameter in
order to determine if the wall will collapse or not [Griffith et al. 2003]: “…the use of the
effective stiffness K 2 and of the effective period T2 combined with an elastic, 5% damped
displacement response spectrum seems to work rather well in the prediction of the
displacement demand in the large amplitude displacement region ( ∆ max > 0.7 ∆ u ), and can be
regarded as suitable for predicting whether a wall will collapse or not” . Even Doherty et al.
[2002] says that the peak response of the tri-linear oscillator can be estimated via an
equivalent linear system with secant stiffness K 2 .
55
Chapter 4. Capacity of adobe dwellings
where ∆ LSu = φ ⋅ ∆ u , with φ a factor that can be assumed from 0.8 to 1 just to reduce the
ultimate limit state, and ρ 2 = ∆ 2 / ∆ u (Table 4.2).
For intermediate limit states (where displacement limits are less or equal to ∆ 1 ) a value of
0.12 for ∆ 1 / ∆ u is assumed with a standard deviation of 0.01 (Table 4.1). From static
equilibrium (Figure 4.7) a relation between the crack width ( ω ) and the displacement at the
top can be obtained, Eq.(4.7). According to this it is seen that the greater the crack width, the
greater the displacement.
t ⋅ ∆i ω⋅h
ω= or ∆i =
h t (4.7)
∆i
ω
Figure 4.7. Horizontal cracks at the base due to rocking.
In this case the period of vibration for all the intermediate limit states will be related to the
given ∆ 1 (initial stiffness) as follows, Eq.(4.8):
F F ⎛ ∆ ⎞
K1 = = o ⎜⎜1 − 2 ⎟⎟
∆1 ∆1 ⎝ ∆ u ⎠ (4.8)
1/ 2
⎛ M⎞
Replacing Eq.(4.8) into T1 = ⎜⎜ 4π 2 ⎟ Eq. (4.16) is obtained which is a fixed period of
⎝ K 1 ⎟⎠
56
Chapter 4. Capacity of adobe dwellings
When buildings do not have a horizontal restriction such as a collar ring-beam, the following
mechanism can be seen: Mechanism A assumes that no connection is present at the edges of
the wall, or this is insufficient to generate restraint by the party wall. Mechanism B1 and B2
will occur instead of mechanism A when the level of connection is sufficient to involve,
beyond the façade wall, respectively, one or both party walls into overturning, due to
sufficient length of overlapping between elements common to both walls. Mechanism C refers
to the overturning of the corner and it will occur when at least one of the corners of the
building is free, which means without adjacent structures. Mechanism D occurs when only a
portion of the façade is subjected to overturning and the party walls are not involved directly
in the mechanism. Mechanism E is considered when due to the window layout there might be
solution of integrity within the façade plane leading to partial failures (Figure 4.8), [D’Ayala
and Speranza 2003].
Restrepo [2004] has modified the equations for the aforementioned mechanisms in order to fit
experimental data and added a new model of collapse (Mechanism H, Figure 4.9). The base of
the new equations is the consideration of a pure rigid body motion plus a friction term (just in
those cases where friction has been identified as an important source of lateral strength).
57
Chapter 4. Capacity of adobe dwellings
These new equations seem to have more accuracy than other previous expressions, and for
that reason those are going to be applied to 1-storey buildings in this report.
According to the survey damage done in Pisco [Tarque et al. 2007] it has been seen that for 1-
storey buildings the most typical failure can be due to Collapse Mechanisms A, C and D. It
can be assumed that those mechanisms can be extended to Cusco dwellings; however, as it
was mentioned before, the adobe buildings from Pisco have thinner walls, while in Cusco the
thickness of adobe dwellings is around 0.44 m, which results in a slenderness value close to 6
and according to the Adobe Peruvian Code [NTE E.080 2000] those walls can be considered
as stable.
A description of each of the equations (modified by Restrepo 2004) for the collapse
mechanisms is described below -see Eq.(4.10) to (4.16).
Mechanism A
T 2L h
+ β ⋅ Ω pef s µ ⋅ s ⋅ b
(r + 1) + K r LT
λ= 2 3 2 2
⎛ TL ⎞ (4.10)
hs ⎜ + Kr L⎟
⎝ 2 ⎠
Qr
Kr =
γ m hs (4.11)
where T and L are the thickness and length of the front walls, β is the number of edge and
internal perpendicular walls, Ω pef is a partial efficiency factor to account for the limited
effect of the friction, hs is the height of the failing portion of the wall, µ is the friction
coefficient, s is the staggering length, b is the thickness of the brick units, r is the number of
courses within the failing portion (assuming courses in the rocking portion). K r is the
overburden load, Qr is the load per unit length on top of the front wall and γ m is the unit
weight of the masonry (18 kN/m3).
L
Ω pef = 1.0 − 0.185 ≥0
hs (4.12)
58
Chapter 4. Capacity of adobe dwellings
Even Eq.(4.1) results in a collapse multiplier that represents a collapse mechanism between A
and B2 . The friction coefficient for adobe blocks varies from tan 30° ≈0.6 [Corazao and
Blondet 1974] to µ = 1.09 [Tejada 2001]. In this report a value of 0.8 will be assumed.
Mechanism C
⎡ ⎤
1 ⎢ 3T L − L2 T L2 ⎥
λ= ⎢ ⋅ + ⋅ ⎥
π ⎢ ⎛ Lh ⎞ L rh L ⎥ (4.13)
cos 2 min⎜ rh, ⎟
4⎣⎢ ⎝ s ⎠ ⎥
⎦
( )
L1 = min r , nrhs ⋅ s
(4.14)
⎧ 0; r < nrhs
L2 = ⎨
⎩ L − L1 ; otherwise (4.15)
It is important to remark that when the height of the mechanism is less than the total height of
the façade wall, then L2 is equal to zero. rhs is the number of courses within the storey height
and n is the number of storeys.
Mechanism D
3T L − L2 T L2
λ= ⋅ + ⋅
⎛ Lh ⎞ L rh L (4.16)
2 min⎜ rh, ⎟
⎝ s ⎠
Consider the adobe building shown in Figure 4.10 and designed in Tejada [2001]. The wall
located at axis A is bearing the weight of the inter-story and of the roof.
W-A1 and W-A2 are the walls located at the first and second level respectively along axis A.
WA-1 receives the weight of the inter-story and the roof and the self-weight of WA-2.
However, WA-2 receives just the weight of the roof.
The wall perpendicular to WA-1 will be named WB-1 and WB-2 for the first and second level
respectively. If the inter-storey and the roof weight are lying just on walls parallel to axis A,
so the perpendicular walls just receives its self-weight (see Figure 4.10).
59
Chapter 4. Capacity of adobe dwellings
0.025
0.025
1
1
A A
0.2
0.2
0.2
0.2
4
2.775
0.05
B B
1.3725
0.17
4 4
4
4.4 4.4
5.6 5.6
3.425
2.2
4.4
5.6
For the influenced area of WA-1 (similar to the one of WA-2) the weight of the inter-storey is
2.6 kN, of the roof is 12 kN and the self weight of WA-2 is 35.6 kN. So the Qr is
(2.6 + 12 + 35.6) / 1.8 ⇒ Qr = 28 kN / m .
For WA-2 the Qr takes into account just the weight of the roof: ⇒ Qr = 6.7 kN / m
For WB-1 the Qr = 19.8 kN / m due just to the self-weight of WB-2, and for WB-
2 Qr = 0 kN / m .
As it has been shown, the load per unit length can vary from 0 to 28 kN, where the walls at
the first level have a larger demand due to the self-weight of walls located at the second floor.
60
Chapter 4. Capacity of adobe dwellings
4.2.2 Demand
From a probabilistic analysis the acceleration response spectrum (ARS) is obtained and this
can be transformed to have the displacement response spectrum (DRS). Since those spectra
are usually evaluated for a 5% damping, it is necessary to multiply them by a coefficient that
takes into account different values of damping for different limit states, Eq.(4.17), [Priestley
2007].
7
η=
2+ξ (4.17)
Table 4.3. Limit states for adobe walls subjected to in-plane forces.
These drift values of the limit states are quite closed to those obtained by Calvi [1999] for
brick masonry buildings (Table 4.4).
61
Chapter 4. Capacity of adobe dwellings
Table 4.4. Limit states for brick masonry buildings [Calvi 1999].
Limit state Median drift (%) Coefficient of variation (%) ζ (%) Ductility
LS-1 & LS-2 0.1 --- 2 1
LS-3 0.3 --- 5 1+3/n
LS-4 0.5 10.9 10 1+6/n
4.2.4 Capacity
As in the previous section, the scope in this step will be to recall the expression for the period
of vibration at a given limit state and to produce an expression to calculate the displacement
for a given limit state.
A multi degree of freedom system (MDOF) can be represented as a SDOF system having as
principal parameters the effective mass ( meff ), the effective stiffness ( k eff ) and the effective
height ( heff ).
The maximum displacement for a given limit state ( ∆ LS ) can be represented as the summation
of the yield displacement ∆ y and the plastic displacement ∆ p (Eq. 4.18, 4.19 and 4.20),
[Restrepo 2004]. The coefficients k1 and k 2 takes into account the conversion from MDOF to
SDOF system.
∆ y = k1 ⋅ hT ⋅ δ y
(4.18)
∆ p = k 2 ⋅ (δ LS − δ y )⋅ hsp
(4.19)
∆ LS = ∆ y + ∆ p
(4.20)
where k1 and k 2 are coefficients that depends on the mass distribution and on the
hsp (effective height of the piers going to the inelastic range). The effective displacement is
computed then with Eq.(4.21), (4.22) and (4.23), which assumes lumped masses m f at each
floor and the masonry is assumed to have a distributed mass mm per unit length (Figure 4.12).
n n n
δ y2 ∑ hi2 m f + 2δ y δ p hsp ∑ hi m f + δ p2 hsp2 ∑ m f + M m
i i i
∆e = i =1
n
i =1
n
i =1
(4.21)
δ y ∑ hi m f + δ p hsp ∑ m f + N m
i i
i =1 i =1
h1 hT
h1 hT
62
Chapter 4. Capacity of adobe dwellings
(a) Evaluation of k1
The coefficients k1 can be evaluated in an explicit way equalling the effective displacement
∆ e (having µ = 1 ) to ∆ LS . For example, assuming that for 1-storey building h1 is measured at
the mid-height and µ = 1 ( δ p = 0 ), then Eq.(4.24), (4.25) and (4.26) are found, where mmT is
the total mass of the wall ( mmT = mm ⋅ hT ).
δ y2 (hT2 ⋅ m f ) + M m
∆e =
δ y (hT ⋅ m f ) + N m
1
(4.24)
1
mmT δ y2 hT2
Mm =
3 (4.25)
mmT δ y hT
Nm =
2 (4.26)
mmT
mf +
k1 = 3
mmT (4.27)
mf +
2
(b) Evaluation of k 2
Considering µ = 2 , k1 = 0.80 and the effective height of the piers hsp = hT , the value of
k 2 can be evaluated analyzing again with Eq. (4.21), (4.22) and (4.23).
4δ y2 hT2 m f + M m
∆e =
2δ y hT m f + N m (4.28)
63
Chapter 4. Capacity of adobe dwellings
19mmT δ y2 hT2
Mm =
12 (4.29)
N m = δ y hT mmT (4.30)
Replacing the last expressions into Eq(4.21) where ∆ LS = ∆ e it is obtained the expression for
k 2 , Eq.(4.31):
19
4m f + mm
k2 = 12 T − k
1
2m f + mmT (4.31)
Evaluating Eq.(4.31) for the mass values explained before, it is obtained k 2 = 0.95 . It is
important to mention that variation in ductility does not affect greatly k 2 values [Restrepo
2004].
(c)Period of vibration
The limit state period of vibration of adobe walls is rewritten for convenience:
This period is assumed equal to the period of the SDOF system. This is because the
fundamental period of a MDOF is related more or less to 80% of the total mass, which can be
a similar value to the effective mass meff in a SDOF system.
64
Chapter 5. Probabilistic seismic hazard assessment
The seismic hazard can be measured by two ways: by a deterministic and by a probabilistic
analysis. The first one just takes into account of the greatest earthquake or a few relevant ones
than can occur at a specific site; while the second one takes into account all possible
earthquakes affecting the site of interest together with their probabilities of being exceedance.
There are two kind of uncertainty related to a probabilistic seismic hazard assessment
(PSHA): aleatory and epistemic uncertainties. Aleatory (intrinsic randomness) is inherent to
the unpredictable nature of future earthquake events as randomness in rupture mechanism and
heterogeneity of sources. Epistemic (scientific uncertainty) is related to the lack of knowledge
and data about the physics of earthquakes phenomenon [Cotton et al. 2006].
Aleatory uncertainties are directly incorporated into the hazard calculations in terms of the
standard deviation of the median from the Ground Motion Prediction Equations (GMPEs).
However, epistemic uncertainty has to be handed within the framework of a logic tree
approach in which the ground motion models occupy different branches related to subjective
weights (Figure 5.1). At each branch tip, a hazard estimate is obtained along with an
associated relative weight that is calculated by multiplying the weights of the branches in that
particular calculation. Procedures for assigning weights to the branches of a logic tree can be
even seen in Bommer et al. [2005].
The limited nature of the datasets from which empirical models are derived gives rise to
epistemic uncertainty in both the median estimates and the associated aleatory variability of
65
Chapter 5. Probabilistic seismic hazard assessment
those predictive equations [Cotton et al. 2006]. In addition to the aleatory variability in each
model there is uncertainty, usually referred to as epistemic (knowledge related), on the
median and the sigma value of the model itself. Factors that influence the epistemic
uncertainty are the limited database, models imported from other parts of the world
(differences in terms of seismic sources, wave propagation or site response between the target
and the host region).
The PSHA follows the procedure established by Cornell [1968] and summarized in four steps
by Reiter [1990], (Figure 5.2):
First step: identification and characterization of earthquakes sources and definition of the
probability distribution of potential rupture locations within each source. Uniform probability
distributions are usually assigned to each source zone, implying that earthquakes are equally
likely to occur at any point within the source zone.
log( N m ) = a − b ⋅ M (5.1)
where N m is the mean annual rate of exceedance of a magnitude M , 10 a is the mean yearly
rate of earthquakes of magnitude greater than or equal to zero, and b express the relative
relation between large and small earthquakes.
66
Chapter 5. Probabilistic seismic hazard assessment
Third step: definition of the Ground Motion Prediction Equation (GMPE) which gives the
ground motion value (e.g. acceleration) due to a given magnitude M for a given distance.
The GMPEs used to consider standard deviations that are taken into account in the PSHA.
Fourth step: the uncertainties in earthquake location, earthquake size and ground motion
parameter prediction are combined to obtain the probability that the ground motion parameter
will be exceeded during a particular time period.
Taking into consideration the map distribution of epicentres and the tectonic characteristics of
Peru, Castillo and Alva [1993] gathered earthquakes due to the subduction process and crustal
events; with that information 20 sources have been identified in Peru. The subduction events
are due to the interaction of the Nazca and South-American plates, where the former pass
beneath the latter. The crustal events are related to the shallow earthquakes due to the
compression stresses originated on the Peruvian highlands that results in the orogenic process
of the Andes Cordillera (Mountain).
67
Chapter 5. Probabilistic seismic hazard assessment
Figure 5.3. Example of seismic sources geometries. (a) Small source that can be modelled as punctual; (b)
Shallow source that can be modelled as a lineal source; (c) Volumetric source. [Bolaños and
Monroy 2004, adapted from Kramer 1996].
A 3D scheme of the subduction process in Peru is shown in Figure 5.4 where two types of
subduction are present: flat subduction at the northern and central Peru and normal subduction
in the southern Peru. The flat subduction in northern and central Peru has a maximum depth
of 100–150 km. Normal subduction has a constant dip angle of 30° till 350 km depth [Tavera
and Buforn 2001].
Figure 5.4. 3D scheme for subduction process in Peru. Arrows show tensional axis dipping parallel the
subduction plate [Tavera and Buforn 2001].
68
Chapter 5. Probabilistic seismic hazard assessment
The area representation of the seismic sources is due to the lack of sufficient data to model the
active faults as lineal sources [Castillo and Alva 1993]. The shallow subduction zone (Benioff
zone, 0-70 km) is represented by the sources 1 to 5 located along the Peruvian coast. The
intermediate seismicity (71-300 km) is represented by sources 13 to 19 and the deep
seismicity (500-700 km) is represented by source 20. The crustal events are represented by
sources 6 to 12 (Figure 5.5).
Figure 5.5. Peruvian seismic sources: subduction and crustal sources [Castillo and Alva 1993].
Table 5.1 shows the geographical coordinates of the shallow subduction and crustal seismic
sources.
Table 5.1. Geographical coordinates of shallow subduction sources and crustal sources.
69
Chapter 5. Probabilistic seismic hazard assessment
Table 5.2 shows the geographical coordinates for the intermediate and deep subduction
seismic sources [Castillo and Alva 1993].
70
Chapter 5. Probabilistic seismic hazard assessment
The a and b values from Eq.(5.1) can be obtained through regression analysis of a historical
earthquake catalogue which must be corrected for completeness in magnitude and for removal
of fore- and aftershocks. Alva and Escalaya [2005] used the seismic catalogue of the SISRA
[1985] project (dated from 1963 to 1992) and the updated seismological data till 2003 given
by the Peruvian Geophysical Institute [IGP 2003] in order to evaluate the seismic parameters
(Table 5.3).
The M max is the probable maximum magnitude that can be released as seismic energy. Alva
and Escalaya [2005] established as M max a value similar to the greater event that has occurred
in the past in the source. In the same work, the evaluation of the representative hypocentral
depth at each seismic source was obtained by a calculation of frequency of earthquakes versus
depth.
71
Chapter 5. Probabilistic seismic hazard assessment
The general form of an attenuation relationship takes into account the decay of the magnitude,
the geometrical spreading, the anelastic attenuation and the site effect by the following
general expression, Eq.(5.2), [Kramer 1996]:
where M is the magnitude of the earthquake, R is a distance from source to site (there are
different kind of distance: closest distance, epicentre distance, etc) and a1 and b1 are
parameters developed for each zone.
Regarding to Cotton et al. [2006], the flow chart shown in Figure 5.6 should be considered for
pre-selecting of ground motion models.
Then, the next steps are criteria that could be considered for rejecting equations from the
complete list of available candidate models, arranged in order to descending hierarchy [Cotton
et al. 2006]:
72
Chapter 5. Probabilistic seismic hazard assessment
Figure 5.6. Overview of procedure for the selection of ground motion models for PSHA [Cotton et al.
2006].
According to Bolaños and Monroy [2004], the attenuation model that represents Peruvian
earthquakes due to subduction and fits the criteria of selection is the given by Youngs et al.
[1997]. While the one that can represents crustal events is given by Sadigh et al. [1997].
The data base used to developed the attenuation equation has records from Alaska, Chile,
Cascadia, Japan, Mexico, Peru (14 records) and Salomon Island [Bolaños and Monroe 2004].
There are two expressions of ground motion models, one for general rock and another one for
general soil condition, Eq.(5.3) and (5.4). The standard errors are dependent on magnitude.
Rock:
ln( y ) = 0.2418 + 1.414M + C1 + C 2 (10 − M ) 3 + C 3 ln(rrup + 1.7818e 0.554 M ) ) + 0.00607 H + 0.3846Z T (5.3)
Soil:
ln( y) = −0.6687 + 1.438M + C1 + C 2 (10 − M ) 3 + C 3 ln(rrup + 1.097e 0.617 M ) ) + 0.00648H + 0.3643Z T (5.4)
Standard deviation: C 4 + C 5 M
73
Chapter 5. Probabilistic seismic hazard assessment
where y is the spectral acceleration in terms of g , M is the moment magnitude, rrup is the
closest distance, H is the hypocentral depth, Z indicates the source type (interface or
intraslab) and Ci are coefficients.
The attenuation models indicate that for large events at large distances, one should expect that
the peak motions from subduction zone earthquakes will be larger than those predicted using
attenuation relationships for shallow crustal earthquakes. At small source-to-site distances, the
empirical models predict that the peak motions from interface earthquakes are lower than
those for shallow crustal earthquakes.
There are two expressions of ground motion models, one for rock sites and another one for
deep soil sites, Eq.(5.5) and (5.6). The standard errors are dependent on magnitude.
Rock sites
where y is the spectral acceleration in terms of g , M is the moment magnitude, rrup is the
closest distance and Ci are the regression coefficients.
Alva and Escalaya [ 2005] have evaluated seismic hazard curves considering different
attenuation relationships for the coordinate -71.33° Latitude and -14.90° Longitude. The
results are shown in Figure 5.7. Even they have concluded that some GMPE such as Idriss
[1993], Sadigh et al. [1993] and Boore et al. [1994] are not applicable to Peruvian seismic
sources, while the others show less dispersion.
74
Chapter 5. Probabilistic seismic hazard assessment
Figure 5.7. Comparison of seismic hazard curves developed with different GMPEs. Acceleration values on
rock [Alva and Escalaya 2005].
The software CRISIS2007 (program for computing seismic hazard) has been used due to the
facility of inputting attenuation relationships and visualization of results (Figure 5.8). Here it
is calculated the Seismic Hazard Curve and the Acceleration Response Spectrum (5%
damping) for difference return periods. Just to be sure of the results, attenuation relationships
for Youngs et al. [1997] model were generated out of the program for different magnitudes
and hypocentral distances, and entering as input data to CRISIS2007, the resulting Seismic
Hazard Curve was similar to the one generated using the default attenuation relationship for
Youngs et al. [1997] that CRISIS2007 has as an option.
Shallow subduction sources S3, S4 and S5, intermediate subduction sources S15, S16, S17
and S19, the deep subduction source S20, and the crustal sources S11, S12, S8 and S9 have
been considered as active seismic sources than can influence the seismic hazard at Cusco. A
soil type condition has been estimated for the site of interest. Due to the limits of the
attenuation relationships, the maximum distance site-to-source was 500 km.
75
Chapter 5. Probabilistic seismic hazard assessment
As can be seen in Figure 5.9, the PGA related to a return period of 475 years at -71.97°
Latitude and -13.52° Longitude is around 0.29g, a greater value of PGA compared to the
0.20g evaluated by Bolaños and Monroy [2004] for the same return period but for rock type
condition. An amplification of the PGA value due to soil type condition was expected.
Figure 5.10 shows the Seismic Hazard Curve for -71.97° Latitude and -13.52° Longitude and
Figure 5.11 the corresponding Acceleration Response Spectra developed for different return
periods (Tr). As can be seen, the spectral ordinates use to increase according to the increment
in Tr, which means that larger ground motions have lower annual rates of exceedance.
76
Chapter 5. Probabilistic seismic hazard assessment
0.291g
0.287g
0.285g
Figure 5.9. Seismic hazard map for Cusco. Acceleration values for soil condition and for Tr= 475 years.
1.00E+01
1.00E+00
Annual rate of exceedance
1.00E-01
1.00E-02
1.00E-03
1.00E-04
1.00E-05
0.0 0.1 1.0
Acceleration (g)
0.80
0.70
0.20
0.10
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Period T (s)
77
Chapter 5. Probabilistic seismic hazard assessment
Another model has been generated in CRISIS2007 considering a soil type: general rock, and
with the same attenuation relationships. The objective in this case was to evaluate the PGA
for a return period of 475 and 949 years and to compare the results to those given in the
distribution of acceleration for Peru [Bolaños and Monroy 2004], (Figure 5.12 and Figure
5.13). The values obtained were closed to the reference, just with a difference of 0.015g that
can be explained due to the different criteria for selecting the depth of sources and seismic
parameters (Table 5.4).
78
Chapter 5. Probabilistic seismic hazard assessment
Figure 5.12. Distribution of accelerations evaluated on rock for 50 years as exposure period having 10% of
exceedance probability [Bolaños and Monroy 2004].
79
Chapter 5. Probabilistic seismic hazard assessment
Figure 5.13. Distribution of accelerations evaluated on rock for 100 years as exposure period having 10%
of exceedance probability [Bolaños and Monroy 2004].
80
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
The fragility curves for each limit state can be obtained from the probability of failure and the
PGA related to the return period. Fragility curves for in-plane and out-of-plane behaviour are
shown in this chapter and the results are compared with previous studies carried out for Cusco
adobe dwellings.
As can be understood, probability theory and statistics are not the same but those complement
each other to study aleatory phenomena. The statistics give information that helps the
probabilistic models to represent real cases. Therefore, probability theory would be useless
without the parameters given by the statistics.
81
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
An aleatory event is defined as the result of a non casual test, which means an experiment
where a lot of results are expected. The final result of the experiment can or can not be equal
to another result given by another trial. Due to that and after a large repetition of the
experiment (random experiment) a sample space (population) that contains all possible
sample points and describes the aleatory phenomena is obtained. The sample space may be
continuous, in which case the number of sample points is infinite, or it may be discrete, as
when the number of sample points are finite and countable [Kramer 1996].
The concept of probability of the event can be used to define the inherent aleatory to the
sample space. The events can be defined as commands: “values greater than x”, “values
between a and b”, etc.
Given an event A , with N as size of the population (sample space) and n A as size of the
sample points that fulfil that event (outcome space), the probability of the event A can be
represented by Eq.(6.1):
nA
P[A] = (6.1)
N
FX ( x ) = P ( X ≤ x ) (6.2)
82
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
The probability that X is within the interval [a, b] is given by Eq.(6.3), Figure 6.2:
P (a ≤ X ≤ b ) = FX (b ) − FX (a ) (6.3)
The CDF of X can be defined in terms of the probability density function as follows, Eq.(6.4):
x
FX ( x ) = ∫ f (x )dx
X (6.4)
−∞
b) lim FX ( x ) = 0
x → −∞
c) lim FX (x ) = 1
x →∞
83
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
As in Eq.(6.4), the PDF can be represented as the derivative of the CDF, Eq.(6.5):
dFX (x )
f X (x ) = (6.5)
dx
There are many models representative of the PDFs developed by the probability theory (e.g.
normal distribution, log-normal distribution, exponential distribution, etc). The objective of
the statistic is to adjust the best CDF model to the available information, while the Monte
Carlo method use the PDF models with the aleatory variables (standard deviation) to generate
artificial data in order to use as input data within a group of a deterministic process
[Velásquez 2006].
84
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
Let’s define a function G (called limit-state function) of the design parameters x (e.g.
geometrical properties), such that:
When some of the parameters of the function G are considered as random (vector x ), a
statement on whether the structure is safe or not can only be given in terms of probability. The
probability that the random parameters lie in the failure domain F is called probability of
failure Pf , and it is given by Eq.(6.7):
Pf = Pr (G ≤ 0) = FG (0 ) (6.7)
Where FG (0 ) is the CDF of the variable G . Eq.(6.8), which is not so easy to evaluate, can be
rewritten as Eq(6.8) where it is shows the dependence of Pf on a Joint Probability Density
Function (JPDF).
Pf = Pr (x ∈ F ) = ∫ f X (x )dx (6.8)
F
⎧1, if x ∈ F
I f (x ) = ⎨
⎩0, if x ∉ F
Eq(6.8) can be expressed as follows, Eq.(6.9):
Pf = ∫ I (x ) f (x )dx = E[I (x )]
F ∪S
f X f (6.9)
85
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
1 N Nf
Pf ≈ Pˆ f (N ) = ∑ I (x ) =
f i (6.10)
N i =1 N
For the limit states of in-plane behaviour a 10% coefficient of variation is assumed. The
friction coefficient is fixed to 0.8, φ = 0.8 (reduction of the ultimate limit state), and the
ultimate displacement for out-of-plane analysis is taken as 0.85 ⋅ t with a standard deviation
of 0.05 ⋅ t . The relationship between ∆ 1 / ∆ u and ∆ 2 / ∆ u is taken from Table 4.1 as 0.12 (with
a standard deviation of 0.01) and 0.4 (fix value), respectively.
Vector X µx σx
T (m) 0.44 0.04 Width of the wall
L (m) 4.53 0.59 Length of the wall
hT (m) 2.45 0.21 Height of 1-story building
hs(m) 2.45 0.21 Failing portion (height of the wall)
s (m) 0.103 0.008 Staggering length
b (m) 0.44 0.04 Thickness of the brick units
h(m) 0.152 0.01 Height of the brick units
Qr (kN/m) 6.70 0.50 Overburden load
γ (kN/m3) 18 --- Specific weight
β
Number Total Cumulative
2 24 0.40
3 22 0.77
4 12 0.97
5 2 1.00
86
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
Table 6.3. Number of courses within the storey height = number of courses within the failing portion.
r=rhs
Number Total Cumulative
12 1 0.04
13 1 0.08
14 7 0.35
15 11 0.77
16 2 0.85
17 4 1.00
The equations for the in-plane and the out-of-plane behaviour have been generated in order to
calculate the capacity, which means the displacement and the respective period of vibrations
for different limit states (see section 4). The corresponding capacity can thus be evaluated
even with DBELA for all the 1000 building stock.
From the Probabilistic Seismic Hazard Assessment (PSHA) the Acceleration Response
Spectra (ARS, Figure 6.5a) for 5% damping can be obtained for different return periods (Tr);
these response spectra are thus uniform hazard spectra. Multiplying the ARS by T 2 / 4π 2 the
Displacement Response Spectra (DRS, Figure 6.5b) is obtained, which is the demand. The
next figures show the ARS and DRS evaluated for different return periods; as can be seen, the
demand increases with return period. The DRS seem to have a plateau zone after 3 s.
0.7 0.14
Tr= 5
0.6 0.12
Tr= 10
Tr= 50
0.5 0.1
Displacement (m)
Tr= 100
Acceleration (g)
0.1 0.02
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Using DBELA, the capacity of buildings for each limit state is compared with each DRS
(Figure 6.6). For the cases of in-plane behaviour, a reduction in the demand is considered
because the damping for adobe walls is different from 5% (see section 2.4). The probability of
failure is obtained with Eq.(6.10), which it is the ratio between the number of dwellings below
the demand over the total number of dwellings. This evaluation is repeated for all the return
periods and for each limit state.
87
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
0.05 0.02
N: total number
of buildings
0.04
Displacement (m)
Displacement (m)
0.03
0.01
0.02
Pfailure: Nf/N
0.01
Demand for a Nf: # of building having
given Tr capacity less than the
demand
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8
Period (s) Period (s)
The demand evaluated for each return period can be related to a Peak Ground Acceleration
value (PGA), which is the acceleration response ordinate at zero seconds for each spectrum.
Since the probability of exceedance (failure) has been calculated for each return period, it is
possible to plot the PGA vs the probability of exceedance for the building stock and for a
given limit state, to obtain fragility or vulnerability curves.
A fragility curve represents the relationship between the intensity of a seismic event and the
vulnerability of the structure. Specifically, a fragility curve describes the probability that the
structure will meet or exceed a specified limit state during an earthquake. In probabilistic
terms it is a cumulative density function (CDF) which represents the susceptibility of a
structure to failure [Jozefiak 2005].
For example, Figure 6.7 shows another example related to fragility curves [Velásquez 2006].
The seismic intensity has also been defined by the PGA. It is observed that for a PGA= 0.30g
the probability that the group of buildings can collapse is 37%, 35% of the building stock will
have severe damage, while just 20% and 8% will have moderate and minor damage,
respectively.
88
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
The Adobe Peruvian Code [NTE E.080 2000] specifies that the maximum acceleration at
Cusco city is around 0.15g times a soil factor equal to 1.2. This calculation results in 0.18g.
According to Figure 6.8a it is expected that 100% of the building stock will exceed the LS3
and 82% exceeds the LS4 for a 0.18g event.
Figure 6.8b shows fragility curves considering adobe damping ratios. Regarding this figure it
is seen that LS1 and LS2 will be exceeded completely after an event of 0.15g. Considering an
event of 0.18g, it can be said that 98% of the building stock will exceed the LS3, while
around 77% will be nearly to collapse or already collapsed (LS4). Since equivalent viscous
damping ratios for masonry are lower than for adobe, greater values of probability of
exceedance are expected to be obtained considering the former. This can also be seen in
Figure 6.8.
100% 100%
90% 90%
80% 80%
Probability of exceedance
Probability of exceedance
70% 70%
60% 60%
50% 50%
40% LS1: very minor damage 40% LS1: very minor damage
30% LS2: minor structural damage 30% LS2: minor structural damage
20% LS3: significant structural damage 20% LS3: significant structural damage
LS4: near to collapse or collapse LS4: near to collapse or collapse
10% 10%
0% 0%
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
PGA (g) PGA (g)
89
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
An important parameter to take into account before any analysis is the consideration of the
slenderness value, which is the ratio between the height and the wall thickness. According to
the Adobe Peruvian Code a slenderness value 6 means stable walls. In the case of the building
stock, the slenderness is around 2.43/0.44 ≈ 5.5.
As in masonry and concrete material, it is assumed that adobe does not have any resistance to
tension. Since adobe dwellings do not have confinement elements that permit a similar
displacement for all the walls, it is possible to think that for a certain grade of movement
vertical cracks will appear at the joint of perpendicular walls with the wall subjected to
overturning forces (Figure 6.10). Looking at this aspect, limit states 1 and 2 -for which the top
displacements are 17 and 40 mm, respectively- have been introduced into the DBELA
calculations. Furthermore a limit state 3, where the maximum displacement is ∆ 1 (see section
4.1.3), has been added.
Figure 6.11 to Figure 6.13 show the fragility curves considering failure mechanism A, C and
D. It is seen that the controlling mechanism is D, followed by the mechanism A in all the limit
states. Looking at the LS1 it is said that more than the 90% of adobe dwellings will tend to
have slightly vertical cracks at the edges of perpendicular walls when an event of PGA= 0.1g
occurs (Figure 6.11). The vertical crack (top displacement) will continue increasing according
90
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
to the increment of the PGA. For example, considering an event of PGA= 0.18g for the LS2
(Figure 6.12) it is seen that around 75% of the buildings will exceed the 40 mm. of
displacement at the top.
100%
90%
80%
Probability of exceedance
70%
60% Mechanism. A
Mechanism C
50%
Mechanism D
40%
30%
20%
10%
0%
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
PGA (g)
Figure 6.11. Fragility curves for LS1: top displacement= 17 mm, width of horizontal crack= 3 mm.
100%
90%
80%
Probability of exceedance
70%
60% Mechanism. A
Mechanism C
50%
Mechanism D
40%
30%
20%
10%
0%
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
PGA (g)
Figure 6.12. Fragility curves for LS2: top displacement= 40 mm, width of horizontal crack= 7 mm.
For the LS3 (top displacement equal to ∆ 1 ≈ 0.12 ∆ u + σ SD ≈ 45 mm), it is seen that for a
PGA= 0.18g almost 25% of the building stock will exceed this limit if mechanism A or D is
expected (Figure 6.13). In these three limit states a collapse of walls is not expected because
those will maintain their static equilibrium due to rocking behaviour; however, it is expected
that vertical cracks will appear that produce partial or total separation between the
perpendicular walls.
100%
90%
80%
Probability of exceedance
70%
60%
50%
40%
30%
Mechanism A
20% Mechanism C
10% Mechanism D
0%
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
PGA (g)
Figure 6.13. Fragility curves for LS3: ∆ 1 ≈ 0.12 ∆ u with a standard deviation of 0.01.
91
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
Figure 6.14 shows the fragility curves considering just the influence of mechanism A and D
for all the intermediate limit states. The LS2 indicates the separation of adobe walls at the top
of perpendicular walls due to vertical cracks at the edges, while the LS3 indicates the growing
in width and depth of the vertical cracks.
100% 100%
90% 90%
80% 80%
Probability of exceedance
Probability of exceedance
70% 70%
60% 60%
LS1= 17 mm.
50% 50%
LS2= 40 mm.
40% 40% LS1= 17 mm.
LS3= 45 mm.
LS2= 40 mm.
30% 30%
LS3= 45 mm.
20% 20%
10% 10%
0% 0%
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
PGA (g) PGA (g)
a) Mechanism A b) Mechanism D
Figure 6.14. Fragility curves for out-of-plane behaviour.
A similar risk assessment for non-engineering masonry houses located at the Peruvian Coast
[Tarque and Mosqueira 2005] has even developed, where special survey formats considering
architectonic, structural and constructive characteristics were generated. Those formats were
then modified for adobe dwellings in the case of buildings for Cusco. All the information
obtained on the field was summarized in report formats.
The evaluation of the seismic hazard was considering the expected PGA of 0.15g over rock.
Since the type of soil in Cusco is not rock, a soil factor of 1.2 was considered in order to
amplify the PGA. Furthermore, a force-based analysis for the in-plane capacity of walls was
performed to know the required wall density. For the out-of-plane behaviour a force-based
analysis was carried out to know if the adobe walls were stable or not. The seismic
vulnerability was evaluated placing some weights given by experts to each of the input
parameters such as quality in the construction, deterioration on the walls, density of walls,
stability of walls due to out-of-plane forces, amongst others. It is important to mention that
more weight was given to the in-plane behaviour than to the out-of-plane one.
The results given by Blondet et al. [2004] show that for Cusco the principal problem of adobe
buildings is the lack of an adequate wall density, which can infer in in-plane failures. 74% of
the surveyed dwellings have high seismic vulnerability and 17% have moderate seismic
vulnerability. Combining the results from vulnerability and hazard, it was concluded that
92
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
almost 75% of adobe dwellings in Cusco has high seismic risk (related to LS4) and 25% has
moderate seismic risk considering an earthquake with a PGA of 0.18g.
Regarding in-plane capacity (see section 6.5 ) it has calculated herein that for a PGA= 0.18g
almost 77% of dwellings will collapse, while 23% of them will have significant structural
damage. However, for the out-of-plane capacity (see section 6.6) it is seen that for the same
PGA it is expected that around 75% of the buildings show vertical cracks at the edges of walls
(LS2) and probably separation among perpendicular walls.
Comparing the results from the conditional seismic risk and the project carried by Blondet et
al. [2004], it is concluded that almost similar values for probabilities of exceedance for LS3
and LS4 are reached by both methodologies considering an event of 0.18g. In this dissertation
the evaluation of the seismic risk considering another PGA values is possible, while in
Blondet et al. [2004] all the calculations were based on a PGA= 0.18g.
The equation to relate the annual frequency of exceedance (AFE, reciprocal to the return
period) to probability of exceedance (PE) of a given ground motion level for a given exposure
time (L) is given by the Poisson process, Eq.(6.11).
PE = 1 − e − AFE×L (6.11)
Figure 6.15 to Figure 6.17 show the annual probability of exceedance of a given ground
motion (exposure period L= 1 year) related to the probability of exceedance for each limit
state. These plots may be termed Damage State Exceedance Curves. For the in-plane
behaviour equivalent damping ratios for adobe walls have been selected rather than the values
for masonry walls.
93
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
1.000
LS1: very minor damage
0.010
0.001
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Probability of exceedance for a given limit state
Figure 6.15. Damage state exceedance curve for in-plane behaviour, 1 year time window.
0.100
Annual probability of exceedance
Mechanism A
Mechanism C
Mechanism D
0.010
0.001
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Probability of exceedance for a given limit state
Figure 6.16. Damage state exceedance curve for LS3, out-of-plane behaviour, 1 year time window.
1.000 1.000
LS2= 40 mm.
LS3= 45 mm. LS3= 45 mm.
0.100 0.100
0.010 0.010
0.001 0.001
0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100%
Probability of exceedance for a given limit state Probability of exceedance for a given limit state
a) Mechanism A b) Mechanism D
Figure 6.17. Damage state exceedance curves, out-of-plane behaviour, 1 year time window.
94
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
of-plane behaviour of adobe dwellings. For the in-plane case, damping ratios from the cyclic
test of the adobe wall and those recommended for masonry buildings have both been used
(see section 2.4).
The Mean Annual Probability of Damage Exceedance obtained in this manner can be
multiplied by the number of buildings in each building class, and the sum of this product for
all building classes is calculated to give the absolute risk, which can then be divided by the
total number of buildings to obtain the percentage risk. For example, if just 1000 adobe
buildings of 1-storey are considered, it is expected that around 14 buildings (1.41% of 1000)
exceed the limit state 4 (in-plane) annually (evaluated considering equivalent damping values
for adobe walls). It is noted that a Poisson model does not have memory in time and different
results may be obtained with the use of renewal models.
Table 6.4. MAPDE for in-plane behaviour. Table 6.5. MAPDE for out-of-plane behaviour.
When presenting seismic risk results to public, the use of different exposure periods can help
to convey the message of risk. Table 6.6 and Table 6.7 show the Mean Probability of Damage
Exceedance (MPDE) for three different time windows. Just to clarify that in each time
window all the possible ground motions that can occur are considered with their respective
return periods.
95
Chapter 6. Fragility curves for in-plane and out-of-plane behaviour
Looking into Table 6.4 and Table 6.6 it is seen that values evaluated with the masonry
damping ratios are greater than those evaluated with the adobe ones. This difference is
especially greater for LS1 and LS2 for lower time windows and it is due to the lower damping
ratios that are specified for masonry buildings, which results in more conservative values.
However the damping is seen to have a much lower influence on the results for the higher
limit states, thought this is also due to the smaller difference between the damping ratios for
the two materials at these limit states.
96
Chapter 7. Conclusions
7 CONCLUSIONS
7.1 Summary
This dissertation presented a study of the seismic risk of non-engineered adobe structures. The
first section showed the adobe material properties based on experiments. Then, dynamic
properties like inherent damping and period of vibration of adobe walls were obtained from a
cyclic and dynamic test. It is important to note that the yielding period of vibration, Ty, is
strongly dependent on the elastic stiffness K. It was also shown that, in order to obtain
accurate predictions from analytical models, it is necessary to reduce the elastic stiffness to a
fraction of EIgross. A reduction factor of 0.6 gave periods of vibration similar to those obtained
from the cyclic and dynamic tests. It was also possible to develop an expression for the value
of Ty associated with the in-plane capacity of the wall as a function of its height.
The seismic risk assessment of adobe dwellings done in this research has been based on
mechanical procedures, which means the analysis of different limit states for the in-plane and
out-of-plane failure mechanisms of adobe walls. The capacity of each building has been
expressed in terms of displacement capacity and period of vibration for each limit state. The
demand has been evaluated using the displacement response spectrum obtained from a
Probabilistic Seismic Hazard Assessment (PSHA). Then, the capacity of the building stock
was compared with the demand to obtain the probability of failure for each return period.
Taking into consideration a conditional seismic risk, fragility curves were plotted. These
curves relate each peak ground acceleration (PGA) and its associated return period to the
probability of reaching each limit state for the in-plane and out-of-plane behaviour.
The analysis of the fragility curves were based on a PGA of 0.18g (Tr ~ 120 years) because
this is the value specified in the Adobe Peruvian Code for the design of adobe buildings
located in Cusco, 0.15g, multiplied by 1.2 due to the soil conditions. From the PSHA, the
expected PGA for a return period of 475 years was calculated as 0.28g, including the effects
of the soil type condition.
97
Chapter 7. Conclusions
Regarding the fragility curves for the in-plane capacity, it was seen that for an event with a
PGA of 0.18g and considering a soil type condition representative of Cusco, around 77% of
the adobe dwellings would collapse or be near collapse. In this case, four limit states that were
obtained from a cyclic test carried out on an adobe wall were considered.
Some intermediate limit states were defined for the out-of-plane capacity. These limit states
are related to the degree of degradation of the edges, that is, the progress of forming vertical
cracks at the intersections between perpendicular walls. For a PGA of 0.18g, it was seen that
around 75% of the dwellings will have vertical cracks that are wide enough to imply
separation of perpendicular walls. None of the walls are expected to reach the ultimate
collapse limit state; however, more research should be done to assure this result.
For an unconditional seismic risk, all the ground motions with their respective return periods
are taken into account for a given time window. According to the Mean Annual Probability of
Damage Exceedance (MAPDE, time window = 1 year), it was found that more buildings are
likely to be affected by in-plane failure than by out-of-plane failure.
As has been shown in this report, a simple methodology based on mechanical procedures
which suggest limit states for adobe walls can be applied for the evaluation of the seismic risk
of a group of building stock. The results from the fragility curves and from the unconditional
seismic risk assessment indicate the necessity of mitigating the seismic risk of adobe
dwellings. One way of doing this is by using reinforcement that can work as confining
elements. Recent researches are dealing with this aspect, where polymer mesh is placed on the
wall surfaces in order to increase the in-plane ductility and to maintain the out-of-plane
integrity of the walls [Blondet et al. 2006a, b].
• In order to account for the epistemic uncertainties for the probabilistic seismic hazard
assessment, the use of more Ground Motion Prediction Equations should be considered
within the logic tree concept.
• The analysis of 2-storey buildings is important in order to better assess the seismic risk in
Cusco. For this, it is necessary to evaluate the in-plane and out-of-plane capacity of the
walls, considering more complex expressions for the failure mechanisms.
• Equivalent damping ratios for adobe walls loaded in-plane have been derived from just
one cyclic test. It is advisable to consider more cyclic and dynamic tests in order to better
estimate the equivalent damping values for adobe walls.
98
Chapter 7. Conclusions
• The evaluation of the limit states for out-of-plane loading has been carried out assuming a
simple rocking behaviour of cantilever walls. In this case, the ultimate displacement at the
top is approximately equal to the wall thickness, in order to satisfy static equilibrium.
However, walls are not built alone and they interact with perpendicular walls and with the
roof. Therefore, in order to develop more accurate limit states for the out-of-plane wall
behaviour, more research is needed.
• Analytical models should be developed to understand the common limit states that include
interaction between in-plane and out-of-plane behaviour of adobe walls. For this would be
important to consider even the grade of degradation within the connection of two
perpendicular walls.
• The cost of repairing and retrofitting adobe buildings at different limit states should be
explored such that the economic cost of future earthquakes can be estimated using the
Mean Annual Probability of Damage Exceedance results presented here.
• Additional data in the population living in adobe buildings in Cusco should also be
collected such that social losses in terms of fatalities can also be estimated.
99
References
8 REFERENCES
Alva J., Escalaya M. [2005] “Actualization of the seismological parameters for the evaluation of the
seismic hazard in Peru”, in Spanish, CISMID/FIC/UNI, Lima, Peru.
Bal I., Crowley H., Pinho R., Gulay F. [2007] Structural Characteristics of Turkish RC buildings stock
in Northen Marmara Region for Loss Assessmment Aplications, IUSS Press, Pavia, Italy.
Bariola J., Sozen M. [1990] “Seismic Tests of Adobe Walls”, Earthquake Spectra, Vol. 6, No 1, pp.
37-56.
Blondet M., Tarque N., Acero J. [2004] “Study of the seismic vulnerability of non-engineering
buildings located at the Peruvian Highlands”, in Spanish, Joint Project SENCICO-PUCP, Catholic
University of Peru, Lima, Peru.
Blondet M., Madueño I., Torrealva D., Villa-García G., Ginocchio F. [2005] “Using industrial
materials for the construction of safe adobe houses in seismic areas”, Proceedings of Earth Build
2005, Sydney, Australia.
Blondet M., Vargas J., Torrealva D., Tarque N., Velázquez J. [2006a] “Seismic reinforcement of
adobe houses using external polymer mesh”, Proceeding of the First European Conference on
Earthquake Engineering and Seismology, Switzerland.
Blondet M., Vargas J., Tarque N., Velázquez J. [2006b] “Dynamic tests on adobe dwellings”, in
Spanish, internal report, Division of Civil Engineering, Catholic University of Peru, Lima, Peru.
Bolaños A., Monroy M. [2004] “Uniform Seismic Hazard Spectrum”, in Spanish, Master thesis,
Graduate School, Catholic University of Peru, Lima, Peru.
Bommer J., Scherbaum F., Bungum H., Cotton F., Sabetta F., Abrahamson N.A. [2005] “On the use of
logic trees for ground-motion prediction equations in seismic-hazard analysis”, Bulletin of the
Seismological Society of America, Vol. 95, pp. 377–389.
Boore D., Joyner W. [1993] “Estimation of response spectra and peak accelerations from western
North American earthquakes: An Interim Report”, U.S. Geological Survey, Open File Report 93-
509.
CSI. [2005] “Structural Analysis Program: SAP v10.01”, Computer and Structures Inc. California,
USA.
Calvi G.M. [1999] “A displacement-based approach for vulnerability evaluation of classes of
buildings”, Journal of Earthquake Engineering, Vol. 3, No. 3, pp. 411-438.
100
References
Carazas W. [2001] Vivienda Urbana Popular de adobe en el Cusco, in Spanish, UNESCO, Grenoble,
France.
Castillo J., Alva J. [1993] “Seismic Hazard in Peru”, in Spanish, Proceedings of the VII National
Congress of Soil Mechanism and Engineering of Foundations, Lima, Peru.
CEN – Comité Européen de Normalisation. [2003] “Eurocode 8, Design of Structures for Earthquake
Resistance – Part 1: General rules, seismic actions and rules for buildings, Pr-EN 1998-1”. Final
Draft. December 2003.
Corazao M., Blondet M. [1973] “Experimental study of the structural behaviour of adobe buildings
due to seismic solicitations”. Division of Civil Engineering, Catholic University of Peru, Lima,
Peru.
Cotton F., Scherbaum F., Bommer J., Bungum H. [2006] “Criteria for selecting and adjusting ground-
motion models for specific target regions: Application to Central Europe and rock sites’, Journal of
Seismology, Vol. 10, pp. 137-156.
Cornell C. [1968] “Engineering Seismic Risk Analysis”, Bulletin of the Seismological Society of
America, Vol. 58, pp. 1583-1606.
CRISIS2007 v1.1. [2007] “Program for Computing Seismic Hazard”, Engineering Institute of the
UNAM. Mexico.
Crowley H., Pinho R., Bommer J., Bird J. [2006] Development of a Displacement Based Method for
Earthquake Loss Assessment, European School for Advanced Studies in Reduction of Seismic Risk
(ROSE School), IUSS Press, Pavia, Italy.
D’Ayala D., Speranza E. [2002] “An integrated procedure for the assessment of seismic vulnerability
of historic buildings”, Proceedings of the 12th European Conference on Earthquake Engineering,
New Zealand.
D’Ayala D., Speranza E. [2003] “Definition of Collapse Mechanisms and Seismic Vulnerability of
Historic Masonry Buildings”, Earthquake Spectra, Vol. 19, No. 3, pp. 479–509.
Doherty K., Griffith M.C., Lam N., Wilson J. [2002] “Displacement-based seismic analysis for out-of-
plane bending of unreinforced masonry wall”, Journal of Earthquake Engineering and Structural
Dynamic, Vol. 31, pp. 833-850.
Goel R., Chopra A. [1997] “Period formulas for moment-resisting frame buildings”, Journal of
Structural Engineering ASCE, Vol. 123, No 11, pp. 1454-1461.
Griffith M.C., Magenes G.M., Melis G., Picchi L. [2003] “Evaluation of Out-of-Plane Stability of
Unreinforced Masonry Walls Subjected to Seismic Excitation”, Journal of Earthquake
Engineering, Vol. 7, special Issue No. 1, pp. 141-169.
Gutenberg B., Richter C. [1944] “Frequency of earthquakes in California”, Bulletin of the
Seismological Society of America, Vol..34, No 4, pp. 1985-1988.
Gutenberg B., Richter C. [1954] Seismicity of the Earth and related phenomena, Princeton University
Press, New Jersey, USA.
Houben H., Guillaud H. [1994] Earth Construction – A Comprehensive Guide, ITDG Publishing,
London, UK.
101
References
Idriss I. [1993] “Procedures for Selecting Earthquake Ground Motions at rock sites”, NIST GCR 93-
625.
IGP. [2003] “Seismic activity in Peru: period 28th August 2001 to 3rd June 2003”, Peruvian
Geophysical Institute, Lima, Peru.
INEI. [2005] “Census 2005”, National Institute of Statistic and Informatics, Lima, Peru. Available at
July 2007 from www.inei.gob.pe
Josefiak S. [2005] “Fragility Curves for Simple Retrofitted Structures - CM-4 Consequence
Minimization through Structure Retrofit Strategies”, University of Michigan, USA.
Kramer S. [1996] Geotechnical Earthquake Engineering, Pretince Hall International Series.
Liberatore D., Spera G., Mucciarelli M., Gallipoli M., Santarsiero D., Tancredi C., Masini N., Racina
V., Capriuoli A., Cividini A., Tedeschi C. [2006] “Typological and Experimental Investigation on
the Adobe Buildings of Aliano (Basilicata, Italy)”, Proceedings of the Structural Analysis of
Historical Constructions, New Delhi, India.
Lourenço P., Oliveira D. [2005a] “Métodos simplificados para análise da vulnerabilidade sísmica de
construções de terra”, in Portuguese, Division of Civil Engineering, Minho University, Portugal.
Lourenço P., Oliveira D. [2005b] “Seismic Vulnerability Overview of Historical Masonry Churches in
Europe”, Proceedings of the 1st Canadian Conference on Effective Design of Structures, McMaster
University Hamilton, Ontario, Canada.
Magenes G., Calci G.M. [1997] “In-plane seismic response of brick masonry walls”, Journal of
Earthquake Engineering and Structural Dynamics, Vol. 26, pp. 1091-1112.
Melis G. [2002] “Displacement-based Seismic Analysis for Out-of-Plane Bending of Unreinforced
Masonry Walls”, Master Dissertation, European School for Advanced Studies in Reduction of
Seismic Risk (ROSE School), University of Pavia, Pavia, Italy.
NTC – Norme Tecniche per le Costruzione. [2008]. “Norme Tecniche per le Costruzione”, Italy,
January 2008.
NTE E.080 – Norma Técnica de Edificación. [2000] “Adobe Peruvian Code”, in Spanish,
MTC/SENCICO. Peru. March 2000.
OPCM 3431:2005 – Ordinanze del Presidente del Consiglio dei Ministri. [2005] “Masonry sections,
Chapter 11: Existing Masonry Buildings”. Italy. May 2005.
Pinto P., Giannini R., Franchin P. [2004] Seismic Reliability Analysis of Structures, Istituto
Universitario Superiori di Pavia. IUSS Press. Pavia, Italy.
Priestley M.J.N. [2003] Myths and fallacies in earthquake engineering, IUSS Press. Pavia, Italy.
Priestley M.J.N., Calvi G.M., Kowalsky M.J. [2007] Displacement-Based Seismic Design of
Strcutures, IUSS Press. Pavia, Italy.
Reiter L. [1990] Earthquake Hazard Analysis: Issues and Insights, Columbia University Press, New
York, USA.
102
References
103
References
Yamin L., Rodriguez A., Fonseca L., Reyes J., Phillips C. [2005] “Seismic behaviour and retrofit
alternatives for adobe buildings and tapia with bases on reduced models tested over a shaking
table”, in Spanish, Division of Civil Engineering, Los Andes University. Bogota, Colombia.
Youngs R.R., Chiou S., Silva W., Humphrey J. [1997] “Strong Ground Motions Attenuation
Relationships for Subduction Zone Earthquakes”, Seismological Research Letters, Vol. 68, No. 1,
pp. 58-73.
104