0 - Introduction and Basics
0 - Introduction and Basics
PG 2019
Spring 2020 Semester
Fawad A. Najam
Department of Structural Engineering
NUST Institute of Civil Engineering (NICE)
National University of Sciences and Technology (NUST)
H-12 Islamabad, Pakistan
Cell: 92-334-5192533, Email: fawad@nice.nust.edu.pk
Inelastic Behaviour of Structural Components
Source:
Dr. Pramin Norachan
AIT Solutions
Thailand
• Nonstructural components can shift the location of hinging, as well as increase shear and local
deformation demands on columns significantly.
• Those effects should be eliminated by altering the details of nonstructural components, if such effects are
anticipated, their effects on member behavior should be accounted for in simulation.
• Shear failure can occur prior or subsequent to flexural yielding in frame members.
• If the estimated shear strength of a member is lower than its probable flexural strength, shear failure is expected
prior to flexural yielding at relatively low deformation levels. However, if shear strength is slightly larger than
flexural strength (less than 30% larger), then shear failure could occur after inelastic deformations in the hinge
regions where the strength of shear transfer mechanisms degrade with inelastic deformations.
• If beam and column are adequately reinforced to resist the seismic forces then joint may become a
weak link. Thus, unconfined joint may fail in pure shear with no yielding of beam or column
reinforcement.
• Joints may also fail after yielding of top and bottom reinforcements of beams, the joint experiences severe
shear cracking and subsequently joint shear failure. Compared to pure joint failure, this failure is more
ductile since it involves beam yielding.
• To prevent joint failures it is important to make sure that joint strength should be greater than the strength
of them elements framing into it.
• Joint aspect ratio has significant impact on shear strength of joints, the shear strength reduces if we have
high aspect ratio.
• Confinement of the joint either through transverse frame members or reinforcement increases the joint
shear strength.
• High axial load ratios reduce the ductility of the joints and more rapid strength degradation after initiation
of failure.
• Beam Reinforcement ratios effect the joint shear strength and determine what type of failure occurs in
joints.
Lap-Splice Failure in Columns Infill-Failure Induced Soft/Weak Story Soil Bearing Capacity Failure
Base Shear, Vb
Δ
Seismic Resistant Building
(Higher Lateral Strength + Ductile Failure)
Vb
Lateral Drift, Δ
• Buildings do not respond as linearly elastic systems during strong ground shaking
• In some cases, the moments predicted by an elastic analysis may be reasonably accurate, but the
deflections predicted by an elastic analysis may probably be too small.
• On the other hand, in some cases, the elastic analysis may overestimate the moments, e.g. if the central
support of the structure settles by 4 in (100mm) for instance, a traditional elastic analysis of this case
would overestimate the moments caused by the support settlement.
• …Hence, the decision to use nonlinear analysis should be based from a need that cannot be satisfied
by merely linear approximation.
• Stress-strain relationship must be linear and elastic. Most materials exhibit a change in stiffness or
modulus before inelastic or plastic behavior starts.
• Displacements and rotations must be small such that the assumption “plane remain plane after
deformation” is still valid. Mathematically, it is being approximated as sin(θ) = θ.
• The magnitude, orientation or direction and distribution of loads must not change.
• However, maximum stress approaching and/or exceeding yield point may be highly localized, which can
be redistributed and dissipated to less stressed geometry around it, thus nonlinear analysis may not be
necessary. It needs engineering judgment and expertise.
• Large displacement.
• Excessive displacement is usually considered a failure condition, regardless of the stress levels.
Magnitude of Changes in geometry due to displacement Displacements may not be small, hence
displacement are assumed to be small and hence an updated reference state may be
ignored, and the original (undeformed) needed.
state is always used as the reference
state.
Boundary Boundary conditions remain Boundary conditions may change, e.g. a change
Conditions unchanged throughout the analysis. in the contact area.
• Material Nonlinearity
• Due to inelastic behavior of constituent materials such as concrete and steel when strained beyond proportional
limit resulting to cracking, crushing, sliding, yielding, fracture, etc.
• Geometric Nonlinearity
• Due to Due to change in shape of the structure.
• Includes P-Δ and large displacement/rotation effects.
• Nonlinear boundary conditions
• Due to contact such as constraints and restraints
• In many cases, if material nonlinearity is encountered, one or both of the other types will be required as
well.
• There are two causes of geometric nonlinearity, the first based on equilibrium and the second on
compatibility (continuity).
• Geometric nonlinearity occurs when the displacements of a structure are large enough to affect one
or both of the following.
• (1) The equilibrium relationships. Equilibrium in the deformed position of the structure may be
significantly different from that in the undeformed position.
• (2) The compatibility relationships. The relationships between element deformations and
element end displacements may be significantly nonlinear.
• For analysis, the effects of large displacements on the equilibrium and compatibility relationships can be
treated separately.
• Consequently, there are three different types of analysis that can be carried out, as follows.
• (1) Small displacements analysis. This is one extreme. Equilibrium is considered in the undeformed position,
and for compatibility the displacements are assumed to be vanishingly small.
• (2) True large displacements analysis. This is the other extreme. Equilibrium is considered in the deformed
position, and for compatibility the displacements are assumed to be finite. The compatibility relationships are
nonlinear. In this case, geometric nonlinearity is considered with no approximations.
• (3) P-∆ analysis. This is in the middle. Equilibrium is considered in the deformed position (with some minor
approximations), and the compatibility relationships are assumed to be linear. In this case, geometric
nonlinearity is considered approximately.
• There is a fourth type (deformed position for equilibrium, small displacements for compatibility), but this is
never used.
Assume that the bar is stiff axially, so that it has negligible axial deformation.
• The differences among the three methods depend on the relative values of the loads 𝑃 and 𝐻, and on the
displacement ∆. Consider two example cases as follows.
• (1) 𝑃 = 0, and ∆/ℎ = 0.1 (i.e., 10% drift ratio, which is a very large drift for most structures). For all three methods
the force in the spring is 𝐻 and the force in the bar is zero. The only difference is that the vertical displacement is
negligible for small displacements and 𝑃-∆. analysis, and equal to a small value (0.005h) for large displacements
analysis.
• 2) 𝑃/𝐻 = 5, ∆/ℎ = 0.10. For the small displacements case the forces in the spring and bar are respectively 𝐻 and
𝑃. For the 𝑃-∆ case the forces are 1.5𝐻 and 0.995𝑃. For the large displacements case the forces are 1.503𝐻 and
0.995𝑃. The vertical displacements are essentially the same as for 𝑃 = 0.
• These examples show that small displacements analysis can be in error when there are substantial gravity loads and
large drifts, but only in the force in the spring (in the second example above there is an error of 50% in the spring
force).
• The large-stress and large-displacement effects are both termed geometric (or kinematic) nonlinearity, as
distinguished from material nonlinearity. Kinematic nonlinearity may also be referred to as second-order
geometric effects.
• For each nonlinear static and nonlinear direct integration time-history analysis, you may choose to consider:
• No geometric nonlinearity
• P-delta effects only
• Large displacement and P-delta effects
• Material nonlinearity has a wide range of • Geometrical nonlinearity has clear causes and is
causes, many of which are poorly understood, governed by a well-defined mathematical theory.
and it is not governed by any single theory.
• Geometrical nonlinearity has two well-defined
• Our knowledge of material nonlinearity
causes (equilibrium and compatibility), both of
depends almost entirely on what we observe in
which are governed by clear mathematical rules
experiments on actual structures and structural
• Geometrical nonlinearity is not subject to judgment
components.
and interpretation.
• Material nonlinearity is subject to judgment and
interpretation. • This does not mean, however, that geometrical
nonlinearity is easy to account for in an analysis.
Its effects can be complex and subtle, and they
can be difficult to capture in an analysis model.
Link:
https://www.youtube.com/playlist?list=
PL48SKuieCUq9WzNWSgbv44KoAA
SXukGXe
46
Boundary Nonlinearity
• Contact
• Contact conditions such as constraints and restraints which allow parts or portions of the same part to touch or lift
off each other.
• Model the interactions of certain systems.
• Forces
• Represent loads that can be defined as displacement or velocity based, such as earthquakes and soil conditions
Static Equilibrium
𝐹
𝐾𝑢 = 𝐹𝐿
𝐹𝑁𝐿
𝐾𝑢 − 𝐹𝑁𝐿 = 𝐹
𝑀𝑢(𝑡) + 𝐶 𝑢(𝑡) + 𝐾𝑢(𝑡) + 𝐹𝑁𝐿 (𝑡) = 𝐹(𝑡)
𝑢
Non-linear Equilibrium
𝑲𝒖 − 𝑭𝑁𝐿 = 𝑭 3 𝐾𝑢 − 𝐹𝑁𝐿 = 𝐹
• Nonlinear-Dynamic 𝑀𝑢 + 𝐶 𝑢 + 𝐾𝑢 + 𝐹𝑁𝐿 = 𝐹
𝑴𝒖(𝑡) + 𝑪𝒖(𝑡) + 𝑲𝒖(𝑡) + 𝑭(𝑡)𝑁𝐿 = 𝑭(𝑡) 4 𝑢
• Cover all Static, Dynamic, Elastic, Non-elastic, Damped, Un-damped, Linear, Non-Linear cases
and their combinations
• Handles response for
• Basic Dead and Live Loads
• Seismic, Wind, Vibration and Fire analysis
Damping-Velocity Nonlinearity
• The overall resistance of the structures to overall loads, called the Global Structure Stiffness.
• This is derived from the sum of stiffness of its members, their connectivity and the boundary or
the restraining conditions.
• The resistance of each member to local actions called the Member Stiffness is derived from the
cross-section stiffness and the geometry of the member.
• The resistance of the cross-section to overall strains (Cross-section Stiffness). This is derived from
the cross-section geometry and the stiffness of the materials from which it is made.
• The resistance of the material to strain derived from the stiffness of the material particles (Material
Stiffness).
2 5
𝐾 = 𝑓(𝐸, 𝐴, 𝐼, 𝐿)
A 2D frame element with 3
3 degrees of freedom per node
6
1 4 Material Cross-section Member
Property Properties Property
𝐸𝐴/𝐿 0 0 −𝐸𝐴/𝐿 0 0
Structure Stiffness
Material Stiffness
Cross-Section Geometry
Member Geometry
Non-Linear
Member Stiffness
Structure Geometry
Structure Stiffness
• The Action-Deformation curves show the entire response of the structure, member, cross-
section or material
Point ‘A’ corresponds to the serviceability design considerations and working strength or
allowable strength design concepts.
Point ‘B’ is the point up to which the relationship between load and deformation can be
considered nearly linear and the deformations are relatively small.
Point ‘C’ roughly corresponds to the ultimate strength considerations or the design capacity
consideration.
Point ‘D’ is the point at which the load value starts to drop with increasing deformations
Point ‘E’ is the point at which the load value is reduced to just a fraction of ultimate load
(residual strength)
• Probably the most important action-deformation curve for beams, columns, shear walls and consequently
for building structures
• Significant information can be obtained from Moment Curvature Curve to compute:
• Yield Point
• Failure Point
• Ductility
• Stiffness
• Crack Width
• Rotation
• Deflection
• Strain
Load
deformation and a given stage to the maximum
deformation capacity” Yield/ Design
• Normally ductility is measured from the Strength
Deformation
Ductility = Du / Dy
• The most important factor effecting ductility of reinforced concrete cross-section is the
confinement of concrete.
• Amount of confinement steel
• Shape of confinement steel
• Other factors include:
• Presence of Axial Load
• Stress-strain curve of rebars
• Amount of rebars in tension
• Amount of rebars in compression
• The shape of cross-section
1. By actual measurements
• Apply load, measure deflection
• Apply load, measure stress and strain
2. By computations
• Use material models, cross-section dimensions to get Moment-Curvature Curves
Moment
Moment
𝜃
Rotation, 𝜃
Ductility Degradation
Damping 5%
𝑘 𝑐
Assumed
80
Results from
Nonlinear
Analysis
81
Nonlinear Static and Dynamic
Analysis Results
Software Demonstration
PERFORM 3D
• The nonlinear analysis aims to simulate all significant modes of deformation and deterioration in the
structure from the onset of damage to complete collapse.
• Therefore, unlike a linear elastic model, the nonlinear model of an RC structure should be able to
capture all local inelastic phenomenon including concrete cracking, crushing, steel yielding,
buckling, fracture and bond slip between steel and concrete, etc.
• The nonlinear models can generally be classified based on the degree of idealization used in the model.
A comparison of three idealized model types for simulating the nonlinear response of a reinforced concrete
beam-column is shown in Figure (taken from ATC 72 [4]).
• Continuum Models explicitly model the nonlinear behavior of the materials and elements that comprise
the component.
• A continuum model might include finite elements representing the concrete, longitudinal reinforcement,
and shear reinforcement, in which associated constitutive models (e.g. the nonlinear stress-strain curves
of concrete and steel) would represent various nonlinear phenomenon.
• Continuum models generally do not enforce any predefined behavioral modes and, instead, seek to
model the underlying physics of the materials and elements.
• They do not require definitions of member stiffness, strength or deformation capacity, as these effects are
inherently captured in the model through the material properties.
• At the other extreme are lumped plasticity (concentrated hinge) models in which the nonlinear action is
lumped at certain points of the structure and the nonlinear functions between various actions and
corresponding deformations are assigned at those points. In this way, these models are defined entirely
by the phenomenological description of the overall force-deformation response of the component.
• For example, a “concentrated hinge element” assigned at both ends of a beam or column might represent
a lumped nonlinear flexural behavior (cracking, yielding and post-yielding behavior) defined by a
nonlinear function between end moment and resulting curvature (or rotation) of the member. This
nonlinear function should correspond to the observed force-deformation behavior and hysteretic test data
of similar beam or column components.
• In between the two extremes are distributed inelasticity (fiber) models, which can explicitly capture some
aspects of nonlinear behavior while some effects are captured implicitly.
• For example, the complete nonlinear stress-strain curves of materials can be defined to capture important
aspects of material nonlinearity. However, the integration of flexural stresses and strains through the
cross section and along the member is considered implicitly.
• These models typically enforce some behavior assumptions (e.g., plane sections remain plane) in
combination with explicit modeling of uniaxial material response.
2 5
𝐾 = 𝑓(𝐸, 𝐴, 𝐼, 𝐿)
A 2D frame element with 3
3 degrees of freedom per node
6
1 4 Material Cross-section Member
Property Properties Property
𝐸𝐴/𝐿 0 0 −𝐸𝐴/𝐿 0 0
Cross-section Properties
Linear
Cross-section Stiffness
Member Properties
Nonlinear
Member Stiffness
Structure Properties
Structure Stiffness
Defining Inelastic
Behavior at
Nonlinear Modeling
Material Level Fiber Modeling Approach
of Materials
Nonlinear Modeling
Cross-section Level
of Cross-sections Plastic Hinge Modeling
Nonlinear Modeling Approach
Member Level
of Members
Shear Hinges
Defining Inelastic Axial Hinges
Behavior at
Torsion Hinges
Nonlinear Modeling
Cross-section Level
of Cross-sections
Shear Hinges
Axial Hinges
Torsion Hinges
Moment-Curvature
Flexural Hinges Plastic Hinges
Torsion Hinges
• For beam-column elements, models for nonlinear analysis can range from uniaxial spring or hinge models, to more fundamental
fiber-type models, to detailed continuum finite element models.
• In general, all models are phenomenological in that they rely on empirical calibration to observed behavior at some level of
idealization.
• The choice of model type for a given application involves a balance between reliability, practicality, and computational efficiency,
subject to the capabilities of available software and computational resources. The optimal model type depends on many factors,
including the structural system and materials, governing modes of behavior, the expected amount of nonlinearity, and the level of
detail available for the input and output data. The reliability of the model comes from its ability to capture the critical types of
deformation that are of interest to the modeler and control the response.
• Hinge and spring models have the advantage of being computationally efficient by modeling highly nonlinear effects in localized
regions of the structure with few degrees of freedom. The models generally employ pre-defined functions to define the nonlinear
response of the component. Concentrated spring models are typically implemented to capture single degree of freedom response (e.g.,
M-θ), but they may include multiaxial response through yield surfaces (e.g., Px -My -Mz interaction) or other means. By capturing
complicated behavior with highly idealized models, concentrated hinge models are very versatile, but they are also empirical and limited
to modeling phenomena over the range of components and behavior modes for which they have been calibrated.
• For beam-columns, fiber-type models provide the capability to numerically integrate material response through the member cross
sections at a more fundamental material level. The fiber-type integration through the cross section can be used either in conjunction with
a finite length hinge zone or with model formulations that simulate distributed inelasticity along the member length. Fiber-type models for
beams and beam-columns generally invoke kinematic assumptions, such as the Euler-Bernoulli (plane sections remain plane)
assumption, to relate uniaxial stresses and strains through the member cross section to stress resultants and generalized strains for the
cross section.
• Continuum finite element models represent the behavior at the most fundamental level and provide the ability to model the
complete interaction of three-dimensional behavior, including complex geometries and multi-axial stress and strain states.
• However, three-dimensional (3D) continuum models are the most computationally intensive, particularly where the numerical
mesh refinement is controlled by the smallest dimension of the member (e.g., where finite element meshing through the
thickness or depth of a member will dictate the mesh size required along the member length).
• For this reason, 3D continuum models are typically only used to simulate portions of overall systems. In addition, while continuum
models offer the potential for capturing response at very fundamental levels, their practical application is limited by both
computational resources and data to calibrate certain localized behavioral effects. For example, 2D shell or 3D continuum finite
element models can capture the response of isotropic steel materials fairly well, whereas many unresolved challenges remain for
simulating the detailed behavior of reinforced concrete members, considering concrete cracking/dilation and interactions between
steel reinforcing bars and concrete (e.g., bond and anchorage).
Performance-based Seismic Design of Buildings – Semester: Spring 2020 (Fawad A. Najam) 100
Flowchart aid for selecting appropriate nonlinear element and component models
• The first decision point outlined in Figure 2-15 is the level of detail of the model, with the primary choice being between a
continuum model and a discrete model (i.e., fiber or concentrated hinge). For discrete models (which is typical for design), the
next decision is how each component in the model is classified and modeled, either as a force- or deformation-controlled
component, as well as which components to model (i.e., decision of if the gravity framing components are included in the
model).
• After this, the model types for each component and the level of detail of the modeling can be decided. All of these modeling
decisions are centered on how simplified versus complex the structural model should be and the needed level of complexity of
the model depends on aspects such as the expected failure modes and their consequences, the anticipated locations of
damage, and the anticipated level of nonlinearity.
Performance-based Seismic Design of Buildings – Semester: Spring 2020 (Fawad A. Najam) 101
Nonlinear Models for Building Components
Foundations
Performance-based Seismic Design of Buildings – Semester: Spring 2020 (Fawad A. Najam) 102
Nonlinear Models for Building Components
Performance-based Seismic Design of Buildings – Semester: Spring 2020 (Fawad A. Najam) 103
Thank you