Title: Seismic vulnerability of non-ductile RC buildings in Pakistan

through stochastic analysis of design/construction deficiencies

Publisher: Springer (2018)

Introduction:
This paper is about the assessment of seismic vulnerability of reinforced concrete framed
structures, in terms of probabilistic analysis of parameters having an effect on structure, as they
represent a huge part of structure accumulation in Pakistan. The results of the parameters
(strength, serviceability and seismic assessment) are greatly being disrupted by faulty
construction, geometric uncertainty of structure and material uncertainty.

Most of the buildings in the region built without appropriate structural and seismic design as
only the gravity loads were taken into consideration with practically no or little protection
against lateral forces. After the Kashmir earthquake of 2005, early research showed structures
were failed with brittle modes being principal. Because of absence of guidelines or unawareness
in construction industry, territories that have not yet encountered an earthquake but lie in the
seismic regions have a great number of buildings designed against gravity loads only.

Earthquakes coming from time to time especially Kashmir earthquake uncovered the poor local
design and construction practices in Pakistan. Subsequently after the Kashmir earthquake of
2005, a new building code named as Building Code of Pakistan (BCP-SP 2007) with Seismic
Provision for RC structures was developed which is in practice now. This effort to get better the
construction industry needs more research in seismology. It is not wrong to say that actually
Kashmir earthquake justifies the problem statement.

This investigation intends to deal with a part of this issue by assessing seismic vulnerability of
various buildings with fluctuating stories while including material and geometric uncertainties in
the building models to make a model that is closest in its properties to the real structures. The
coefficient of variation in terms of geometry, material & alignment is taken in to account to be
better in simulation process. The response of buildings is assessed using static pushover
analysis. Seismic vulnerability curves for various structural configurations are derived using the
framework developed for deficient RC structures. Structure’s behavior is also assessed in
different seismic zones as described in BCP-SP 2007.

Page 1
Objectives & Hypothesis:
The objective of this study is to develop awareness in construction industry & to make them
compatible with upcoming challenges. It’s just like the social service to provide knowledge
about the deficiencies that greatly affect the seismic potential of structures. And to develop a
method previously used to determine the level of vulnerability and how the buildings are going
to perform in various earthquake zones.

About 10–15% of the absolute building stock is of Reinforced Concrete. It is assumed that in
homogeneity of concrete, poor construction & unskilled workmanship in Pakistan causing great
uncertainties in various parameters of predicting the structure’s behavior. The present guidelines
are also very crude and in an early stage of development. As the construction industry is not
being regulated therefore substandard design practices are yet common in it and the building
stock even in a single locality exhibited different response to earthquakes. This calls for further
refinement of these guidelines.

Research Methodology:
Description of RC building stock and characteristics:

One of the most widely observed deficiency of buildings subjected to earthquakes was
insufficient or no joint reinforcement to account for shear force produced in joints. Formation of
plastic hinges in columns was seen due to the strong beam and weak column connections. Strong
column weak ductile beam connection should be made for low transfer of forces to column from
beam. In Soft storey mechanism, stiffness of stories varies from one another so is the inter storey
drift and this phenomenon has been reported as the reason of failure of many buildings in
Kashmir earthquake. Inadequate transverse reinforcement, lap splices and development lengths
cause diagonal cracks and bar pull-out mechanisms respectively. Lack of structural concrete
strength and reinforcement also contributed in it.

Numerical Modeling:

Typical residential & commercial buildings of 2, 3, 4 and 7 storeys with 2 to 5 bays are under
consideration in this study. The design of these RC building frame structures is done in ETABS
which is general software of design and finite element methods. Then, modeling of RC frames is
carried out in DRAIN 3DX which is commonly known as an analytical tool for static pushover
analysis due to its capability of modeling all the deficiencies related to design and detailing. To
get precise results fiber modeling was introduced in this study. Precision of flexural capacity
analysis can be increased by increasing the number of fibers but the computations will become
tough.

Page 2
The committee anticipated the following relations for shear strength of various varieties of beam
column connections.
Vn = 1.67√ fc′ * Aj
Vn = 1.25√ fc′ *Aj
Vn = 1.00√ fc′ *Aj

Equation 1 is for calculation of shear strength of interior connections confined on all four faces,
Eq. 2 for interior connections confined on two or three faces and Eq. 3 for exterior connections.
Where, Aj is corresponding area of joint and fc′ is representing 28 days compressive strength of
concrete.

Material Uncertainty:

Uncertainty in earthquake demand and structural capacity are the predominant parameters
effecting the vulnerability assessment of structures. According to many researchers, variability in
material properties is one of the main sources of uncertainty in vulnerability assessment studies.

Yielding strength of steel (fy) and compressive strength of concrete (fc′) are taken into account
for demonstrating material related uncertainty. Data related to testing of both of steel and
concrete is gathered from various testing research labs & construction sites from the whole
country. Concrete samples with 20 MPa target strength are taken under consideration which are
assumed to be generally used in Pakistan. A normal distribution for compressive strength of
concrete with a coefficient of variation (CV) of 23% was observed. Samples of concrete having
5MPa strength are also observed in residential areas with little or no supervision, which is not
acceptable at all. Field data collection reveals grade 40 steel having yield strength of 275 MPa
was widely used instead of grade 60 steel which should be used as per ACI. Grade 40 steel
normally providing yield strength value slightly greater than 275 MPa with a mean value of 319
MPa and a coefficient of variation of 10.23%.

Geometric Uncertainty:

Because of faulty construction & unskilled workmanship, huge variations is been observed in
geometry of structural members which affect the strength parameters of respective members and
finally increasing the vulnerability of integral structure. To account for this effect, probability
density functions of concrete cover and column size are defined. Un-skilled labour and
inappropriate formwork are reasons behind the variation in column sizes. Irregular alignment of
formwork has an effect on the concrete cover too. Data related these functions gathered across
the different areas of Pakistan. Data depicted a normal distribution for variation in column size
having a coefficient of variation of 5.30%. In addition, the data suggests a coefficient of variation
of 7.8% for concrete cover distribution. For considering these random variables, random
sampling is needed from their probability density functions.

Page 3
Evaluation Methodology:

For generating the seismic vulnerability curves of various structures, the framework of
probabilistic vulnerability assessment developed by Kyriakides was used. It involves improved
form of Capacity Spectrum Method known as Modified Acceleration Displacement Response
Spectrum method (MADRS). Major steps involved in this framework are modeling of structures,
how the structure is going to perform in case of different loading scenarios & damage assessment
of structures.

When the representative structures had been extracted out then the design of them was carried
out by application of gravity loads. Then response of RC frame structures obtained in terms of
capacity curves using static pushover analysis. In static cyclic analysis, to get the targeted
displacement value lateral displacement was cyclically increased. Applied lateral load was
according to the Building Code of Pakistan (BCP-SP 2007). Damage experienced by structures is
evaluated using damage index which depends upon time periods.

DI = (Tsec − Tinitial)/(T100 − Tinitial)

Where

Tsec = 2π√SD ÷ √SA

SD and SA are representing the spectral displacement and spectral acceleration respectively.

Tinitial is the time period of structure at zero damage & T100 is time period of structure at
collapse.

DI is correlated with Mean Damage Ratio (MDR) which is more meaningful parameter by this
relation:

MDR = f(DI)

Where value of f is 1.

A seismic vulnerability curve is shown which is a plot of PGA vs the mean damping ratio for 2
bays of 2 storey building. Initially, the damage is increased linearly with hazard. Because of
yielding of compression steel the curve becomes steep. At 50% of damage compression concrete
yields in joints curve becomes steeper. Then at 90% of damage pulling out of bars occurred. The
behavior of taller buildings up to 7 storeys is slightly different than this building.

Page 4
 Figure1: Seismic Vulnerability curve of 2-storey building with 2 bays

Discussion on Results:
From seismic vulnerability curves it is clearly shown that failure was in brittle manner (shear and
pull out failure modes). This brittleness is due to inappropriate design, detailing & construction
practices. All these deficiencies are incorporated during the design process for better results of
simulation. Faulty design practice of strong beam and weak column forms plastic hinges in
column. This thing will lead to the abrupt brittle failure of columns & of ultimately building.
Another reason for this brittle behavior is lack of ductility in the structure because ductility acts
as a damper to dissipate the energy generated by earthquake vibrations. The total ductility of
structures comprises of material level ductility, cross level ductility & member ductility
respectively. If material is not that much ductile just like concrete then its ductility can be
enhanced by the provision of transverse reinforcement & good proper detailing. The pull and
push forces generated by earthquake are taken by concrete and steel which in turn depends on
steel and concrete strengths, development and anchorage length. Shear failure is represented by
diagonal cracks because of geometric distortion at joints, which is most common after
earthquake. That’s why column size must be sufficiently large for the prevention of shear failure.
Unfortunately, all these things are not currently being considered and not in practice in Pakistan
at different construction sites commonly.

Building performance in different seismic zones of Pakistan:

If a certain building is subjected to various seismic loadings its response would be different
against them. It was seen that structure with different storeys performed acceptably in Zone 1
and Zone 2, only 2-storey building was slightly damaged. In zone 2B, while 3, 4 and 7 stories are
slight damaged but the 2-Storey structure is about to collapse. None of the building would stand
a chance against PGA value greater than 0.353 g.

Page 5
Increment in time period means structure is subjected to lesser lateral forces. A stiffer building
early collapses because it attracts more load. Figure showed a trend that Up to 4 storeys, stiffness
phenomenon becomes dominates mass, leading to decrement in time period & structure fails at
early stages. On the other side, by adding another bay mass dominates the stiffness leading to
increment in time period and making the structure to behave in ductile manner.

Figure 1: Effect of variation in building configuration

Comparison with past earthquake damage data:

The structure’s performance was also observed with respect to the previous earthquakes like
Kashmir whose damaged data was used as a reference. This data comprises of RC structures of
various storeys and different ages. The RC frame structures were having in-fills of block and
brick work masonry but with deficient joints. Their mean damage ratio is obtained from repair
and replacement cost of these structures and their respective PGA is obtained from a recent
hazard framework. In Kashmir earthquake, collapse situation of medium-rise buildings is in
accord with collapse condition predicted by this study.

Conclusions:
The investigation of material & geometric uncertainties depicted huge variation in concrete
compressive strength fc′ with coefficient of variation of 23%. Average fc′ value is 16.31 MPa
which is 18% less than that of ACI 318-02 (2002) prescription for design purposes. And the
coefficient of variation for yield strength of steel is observed to be 10%. Also, average yielding
strength value is 319 MPa which is 22% less than that of ACI 318-02 (2002) prescription for
design purposes in seismic areas. Similarly, concrete cover is 4–37% less than that of ACI 318-
02 (2002) specifications for the design. Above observation strengthens the idea of need of
vulnerability assessment of structures with material and geometric uncertainties.

From the development of vulnerability curves for structures taking into account the varying
parameters then evaluating the building performance in various seismic zones of Pakistan
showed the brittle failure of structures as they were designed for gravity loads only. The selected

Page 6
representative buildings perform satisfactorily in zone 1 and zone 2A with a slightly damage in
2-storey building. In the zone 2B, 2-storey building will collapse while other would be damaged
accordingly. All of the selected representative buildings will become very vulnerable in zone 4
and collapse will occur.

Structure’s performance is also evaluated by exposing it to previous earthquake hazard level. For
low rise buildings, the damage is underestimated by 6–15% value from the mean curve before
occurrence of slight damage then damage is overestimated by 20% amount at collapse state. In
case of medium rise buildings, the damage seems to be underestimated by 10%, 10% and 3% at
less, moderate and extreme damage conditions from the mean curve respectively. This
underestimation is because of difference in height and bay width of actual and theoretical data
and also attributed to the irregularities in structure.

Recommendations:
Seismic vulnerability of the structures ought to be reassessed while taking into account more
building bays & storeys to be more accurate in results.

Irregularities in the structures ought to be modeled to assess its effect on the seismic performance
of structures under consideration.

Praise:
Seismic vulnerability assessment of non-ductile buildings was carried out incorporating the in
homogeneity of concrete, coefficient of variation of faulty construction practices, material and
geometric uncertainties. Structural response was evaluated using the static pushover analysis and
its performance in various hazards according to zones was observed.

Critique:
In this paper already developed framework for seismic parameters is used which itself has
variation. Geological conditions of the structure were not taken into account i.e either the soil
will amplify the ground shaking or lessen its effect. The paper didn’t tell about the any upcoming
earthquake & effect of epi-central and hypo-central distance is ignored.

Suggestions:
Probabilistic seismic hazard analysis of the whole country is needed to be done in more
refinement and must be the part of building code as there is a lot variation in the current
proposed parameters. Research must be done how the retrofitting of existing building can be
done to make them safe against seismic events. Seismic fragility assessment of high rise
buildings must be the area of interest in Pakistan. Research should be done on non linear
dynamic analysis for seismic assessment of structures.

Page 7
Intellectual Nugget:
In the design process of a structure incorporation of poor workmanship, material and geometric
uncertainty, poor joint connections and predicting the performance of structure in various
seismic zones was Pakistan.

Remarks:
The seismic vulnerability of a building (foundations, columns, beams, and floor slabs) is its
vulnerability to damage by ground shaking. The vulnerability of a structure exposed to an
earthquake depends upon seismic deficiency of that structure. And the seismic deficiency is
attributed as a state that will prevent a structure from meeting its intended purpose.

Static pushover analysis is a non linear cyclic analysis method where a structure is subjected to
gravity loading and lateral load pattern is applied on it to get repeated displacements until the
specific value of displacement is reached.

Page 8