You are on page 1of 6

1.

Distinct types of Collapse simulation models


Pujol, Ramirez and Sozen, 1999
Kato and Ohnishi, 2002
Sezen and Moehle, 2004

1.1 Lumped plasticity model

The Lumped plasticity or concentrated plasticity model is an efficient model for


beam-column elements with inelastic springs located at member ends. The model was
developed by Ibarra, Medina and Krawinkler (2003, 2005). Usually the springs have
zero lengths and for that reason they are called point hinges. It is used to predict the
collapse simulation of a beam-column element. It captures the strength and stiffness
deterioration to assess global collapse. The model captures the negative stiffness of the
post-peak response, enabling modelling of the strain softening behaviour of reinforced
concrete members. It is incompetent to capture shear flexibility due to lumping and no
degrees of freedom along the section. It captures deterioration of reinforcement due to
buckling and low cycle fatigues.

1.2 Fiber elastic model

The Fiber elastic model is a better model than lumped plasticity model in terms of
shear detection along the length of the member. It considers shear flexibility by
modelling shear degrees of freedom along the section. It is more where cracking ad
tension stiffening behaviour governs. It is unable to capturedeterioration of
reinforcement due to buckling and low cycle fatigues. It is not able to capture strength
and stiffness deterioration too.

There are several types of collapse simulation models to idealise the collapse of
any Non-ductile concrete structure. Modern seismic codes discuss the ductile detailing
of reinforcement and impose strong-column weak-beam mechanism. But the non-code
compliant buildings were vulnerable to seismic forces. In order to understand the
mechanism of failure of members due to gravity and seismic loads, simulation models
are developed. The collapses of most non-ductile concrete buildings are controlled by
loss of support due to gravity loads prior to gravity loads (Shoraka et al. 2013). Loss of
gravity load support is one of the primary cause of failure in Non-ductile concrete
buildings. Probabilistic assessment of these concrete buildings is performed to identify
the collapse risk. Among the members, column is a critical, which if fails, will cause
major damage to the building. Hence several collapse mechanisms have been proposed
on the simulation models of columns and beam-column joint. The simulation models
are based on axial, shear, flexure and longitudinal slip failure. Pincheira et al. (1999)
proposed a two-dimensional model based on flexural slip and shear deformation
including degradation of lateral stiffness ad strength. Lee and Elnashai (2001) evaluated
inelastic flexure and shear response of bridge columns using lumped hysteresis. They
developed the model by combining flexure-shear-axial interaction (Lee and Elnashai,
2002). Another analytical model based on flexural yielding followed by degrading
shear was proposed by Elwood (2004) which was incorporated in OpenSees as Limit
State material model. Later, Mostafaei et al. (2009) had proposed a model based on
displacement to predict the point of shear failure which includes axial-shear-flexural
interaction (ASFI).It considered the effect of shear deformations in sectional analysis.

Numerous other models existed which simulated the maximum shear strength or
limiting deformation of a reinforced concrete column under lateral loads. But most of
them did not provide proper simulating behaviour at the critical limits. Among all the
models proposed, three models were selected to represent ideal behaviours.They were
based on regression and mechanics base equations. First, a model based on lumped
plasticity framework with zero length rotational spring and elastic line (beam-column
element) was proposed by Haselton et al. (2008). The springs use non-linear hysteretic
model to include damage accumulation features. 255 rectangular columns (which
include 220 for flexure and 35 for flexure-shear failure) were tested and the test results
were combined to get an average model. The model was not calibrated for shear failures
prior to flexural yielding. The model provides regression based equations to estimate
linear and non-linear parameters of column and loading. The material model
incorporated cyclic and in-cyclic degradation behaviours. The boundary conditions and
the model building parameters were fixed and cannot be altered during analysis. The
model estimates the stiffness of elastic line elements and include the effects of bar-slip
induced deformations. It is compatible with line-based joint models. The model utilizes
a non-linear material model that is implemented in OpenSees as Pinching4 using
hysteretic model. The model does not distinguish between flexure and flexure-shear
failures; hence it is unsuitable to combine it with axial failure model (Elwood 2004).
The second model was proposed by Leborgne and Ghannoum in 2012 which based on
plastic rotation at two ends rather than drift ratio. It includes a zero length shear springs
at the joints with 3 non-linear line elements. The model can be used with either a fiber
section or lumped plasticity column element but calibrated using fiber sections.The
model is calibrated using 32 rectangular column tests in which flexural yielding
occurred prior to shear induced strength degradations.It has the property to incorporate
the shear failure mode and flexure-shear failure mode by monitoring deformations
between two nodes through the shear and flexure springs. If the model reaches its
limiting shear force or limiting rotation then the degrading behaviour is triggered. The
model can transit from the shear failure to flexure-shear failure (only flexure failure is
not captured). Like the first model, it incorporates the cyclic and in-cyclic degradation
behaviour. The material model has several damage functions which include strength
and stiffness degradation. The damage parameters are defined based on regression
analysis and column properties. It is also compatible with line based joint models. In
this model, the user has the choice to input fixed values for rotation and shear force
limits or use the calibrated values obtained from ASCE41 shear strength equation and
regression based plastic rotation equation. The model includes pinching characteristics
and is implemented in OpenSees as PinchingLimitSate Material. As it is applicable for
shear and flexure-shear failure models, so it can be coupled with axial failure models.
Thirdly, based on the previous concepts of column modelling by Elwood and ASFI
method, Baradaran and Elwood (2013), proposed a model for columns which can
represent flexure, shear, bar slip failures. A model which simulates the lateral
displacements due to flexure, bar slip and shear independently has already been
developed (Sezen and Chowdhury, 2009). But, this model combines all types of failures
giving priority to shear mode failure bya mechanical-based model.The model consists
of a non-linear line element along with zero length springs. The shear response ranges
from pre-peak to post-peak point. The Modified Compression Field theory (MCFT)
(Shoraka and Elwood, 2013) is used to simulate the pre-peak shear behaviour. The
ASFI method is used to predict the point of shear failure (peak point shear). The post
peak behaviour is represented by shear-friction model. The model is validated with 20
rectangular column tests. The model detects shear or flexure-shear failure based on
shear strains in the plastic hinge zone of the column element. It can simulate full
degrading behaviour (cyclic and in-cyclic degradation). It is already implemented in
OpenSees as LimitState Material.

The flexural response is estimated by the moment-curvature relationship based on


fiber beam-column elements with five sections. The five sections are located at Gaus-
Lobatto integration points along the length of the element for optimum integration of
section deformations. These section deformations (curvatures and axial strains)
integrated over the length determine the element end deformations(rotations and axial
elongation). The column section is discretized into uniaxial fibers for both concrete and
reinforcement. The modified Kent and Park model (Kent and Park, 1971) is used for
modelling concrete while Giuffre-Menegotto-Pinto model is used for reinforcement.
The bar slip model is based on the Sezen and Moehle model (Sezen and Moehle, 2006).
The moment vs bar slip rotation relationship is based on column geometry, amount and
location of longitudinal reinforcement. The relationship shows is a trilinear curve based
on equal area of calculated and approximated response (Shoraka and Elwood, 2013).

The shear response is governed by three models namely, the pre-peak response
model, the point of shear failure model and the post peak response model. All the three
models are combined to produce a single model based on shear friction. The MCFT
method which is based on the modified Kent and Park model is used to represent the
pre-peak shear response as it relates the average stress to average strain in a cracked
reinforced concrete element (Vecchio, 1990). Based on MCFT method and sectional
analysis, the shear strain is evaluated. The shear failure in columns occurs due to
diagonal tension or diagonal compression. The diagonal tension failure occurs due to
insufficient transverse reinforcement while the diagonal compression failure occurs due
to excessive axial loads in the compression region. The plastic hinge region detects the
shear failure based on ASFI method. As per ASFI method, the uniaxial averaged
compression strain and longitudinal strain are determined.Based on MCFT method,
principal strain, transverse strain, crack angle, compression stress and aggregate
interlock shear stress are computed. Finally, the diagonal tension shear failure and
diagonal compression shear failure is evaluated. The degrading post peak response is
based on the shear friction model. Hence, combining all the three models the final shear
response can be evaluated. Combining the shear response with the flexural and bar slip
response, the combined response is evaluated.
The work done by Ibarra, Medina and Krawinkler includes the empirical functions
based on seven calibrated model parameters to the physical properties of the beam-
column (i.e. confinement, axial load, concrete strength, etc.). The model composed of
a tri-linear monotonic backbone curve. It captures the four basic modes of cyclic
deterioration strength deterioration of the inelastic strain-hardening branch, strength
deterioration of the post-peak strain-softening branch, accelerated reloading stiffness
deterioration, and unloading stiffness deterioration. The cyclic deterioration is based on
two factors - one is the normalized energy dissipation capacity () and second one is
the exponent term to define the rate of cyclic deterioration changes (c) with
accumulation of damage. Hence, the seven parameters to determine the empirical
function were - My, y, Ks, cap, and Kc, , and c. Among the total 255 tests, 220 were
for flexure mode tests and 35 were for flexure-shear mode test. A range of column
parameters selected to check their individual behavior. Some of them include the axial
load ratio (), compressive strength of unconfined concrete (fc), yield stress of
longitudinal reinforcement (fy), area ratio of transverse reinforcement (sh), etc. During
calibration, it is seen that the stiffness of the end springs in a column model is very high
than the line element. Hence it is accordingly adjusted so that it behaves correctly
during lateral loads. In the Ibarra model, the monotonic and cyclic models are
interdependent to each other as they are calibrated together. It is an approximation so;
a standardized procedure is adopted for calibration of test data. The P-delta effects,
proper estimation of yield shear force, yield displacement of reinforced concrete at the
yield point (as the concrete becomes non-linear before reinforcement), displacement at
40% of yield force, post yield stiffness, finalizing the values of normalized cyclic
energy-dissipation capacity , and exponent term, c. The cyclic accelerated
deterioration and cyclic unloading deterioration modes have zero deterioration. The
value of was assumed to be same for the full displacement (matching more for more
damaging cycles). The value of c is approximately assumed to be 1.0 for flexure and
flexure-shear failure modes (Ibarra, 2003).

After the yield point, the capping point for peak behaviour and post peak behaviour
were taken into account by cyclic and in-cyclic deterioration. The cyclic deterioration
is the one where strength deterioration happens after one complete cycle of loading
while in-cycle deterioration happens within the same cycle by a negative post capping
stiffness. The in-cyclic deterioration is more dangerous than the cyclic deterioration due
to the negative slope of the curve which defines negative stiffness and finally strength
loss too. It has been seen during cyclic loading, the displacement is not so larger for the
capping point to reach. Then, the Lower Bound (LB=1) value, which means that the
value is lesser than the value of capping point is used in plots. During earthquakes it is
said that the number of cycles of load is few till collapse. Hence, the tests where the
numbers of cycles were more were neglected as it did not idealize an earthquake. The
pinching effects were prominent in 9 out of 35 flexure-shear tests. The major effect of
pinching is increasing the displacement of the building, so, it is not considered in the
Ibarra model. The calibration of stiffness deterioration is critical, hence, correct
calibration of impacts hugely on the overall building collapse simulation. If the stiffness
degradation is cyclic (i.e. drift did not reach the capping point) then the stiffness
remains positive which is less important for buildings than in-cyclic (i.e. drift reached
capping point or more) where the stiffness becomes negative (PEER Report 2007-03;
Ibarra et al. 2003, 2005; Haselton et al. 2006).

You might also like