Nonlinear

Nonlinear structural behavior may be associated with either geometric or material response, each
described as follows:

Geometric nonlinearity concerns the P-Delta effects which are associated with the
application of external loading upon the displaced configuration of a structure.

Material nonlinearity concerns inelastic structural response in which the behavior of a

component, system, or connection deviates from the initial stiffness tangent characteristic
of linear-elastic behavior.

Linear vs. nonlinear analysis

Nonlinear analysis methods are best applied when either geometric or material nonlinearity is
considered during structural modeling and analysis. If only elastic material behavior is
considered, linear analysis methods should suffice, though P-Delta formulation may still be
applied. Linear and nonlinear methods may be static or dynamic. A few of the traditional analysis
methods, and the relations between their attributes, are presented in Figure 1:

Figure 1 - Analysis methods

Each of these analysis methods has benefits and limitations. An overview of each method is as
follows:

Strength-based analysis is a static-linear procedure in which structural components are

specified such that their elastic capacities exceed the demands of loading conditions.
Strength-based demand-capacity (D-C) ratio indicate the adequacy of each component.
Since only the elastic stiffness properties are applied to the analytical model, strength-
based analysis is the most simplified and least time-consuming analysis method.

Static-pushover analysis is a static-nonlinear procedure in which a structural system is

subjected to a monotonic load which increases iteratively, through an ultimate condition,
to indicate a range of elastic and inelastic performance. As a function of both strength and
deformation, the resultant nonlinear force-deformation (F-D) relationship provides insight
into ductility and limit-state behavior. Deformation parameters may be translational or
rotational. Pushover is most suitable for systems in which the fundamental mode
dominates behavior. When higher-order modes contribute, as with taller buildings,
dynamic analysis is most effective.
 Response-spectrum analysis is a dynamic-linear method in which maximum structural
response is plotted as a function of structural period for a given time-history record and
level of damping. For a set of structural mode shapes and corresponding natural
frequencies, the linear superposition of SDOF systems represents response. Response
measures may be in terms of peak acceleration, velocity, or displacement relative to the
ground or the structure. Structures must remain essentially elastic since response-
spectrum analysis is dependent upon the superposition of gravity and lateral effects.
Results may be enveloped to form a smooth design spectrum.

Time-history analysis is a dynamic-nonlinear technique which may involve either the

FNA or the direct-integration method. FNA is a modal application, whereas with direct
integration, the equations of motion are integrated at a series of time steps to characterize
dynamic response and inelastic behavior. Loading is time-dependent, and therefore
suitable for the application of a ground-motion record. Time-history analysis may account
for both material nonlinearity and P-Delta effects.

Analysis objective

Engineers may use any of these analysis methods to:

Characterize and gain insight into structural behavior.

Generate information useful to the design decision-making process.

Capacity Design

Nonlinear modeling and analysis is fundamental to Capacity Design.

P-Delta effect

P-Delta effect, also known as geometric

nonlinearity, involves the equilibrium and
compatibility relationships of a structural

system loaded about its deflected

configuration. Of particular concern is the
application of gravity load on laterally
displaced multi-story building structures.
This condition magnifies story drift and
certain mechanical behaviors while
reducing deformation capacity.

P-Delta effect typically involves large

external forces upon relatively small
displacements. If deformations become sufficiently large as to break from linear compatibility
relationships, then Large-Displacement and Large-Deformation analyses become necessary. The
two sources of P-Delta effect are illustrated in Figure 1, and described as follows:
 P- effect, or P-"small-delta", is associated with local deformation relative to the element
chord between end nodes. Typically, P- only becomes significant at unreasonably large
displacement values, or in especially slender columns. So long as a structure adheres to
the slenderness requirements pertinent to earthquake engineering, it is not advisable to
model P-, since it may significantly increase computational time without providing the
benefit of useful information. An easier way to capture this behavior is to subdivide
critical elements into multiple segments, transferring behavior into P- effect (Powell
2006).

P- effect, or P-"big-delta", is associated with displacements relative to member ends.

Unlike P-, this type of P-Delta effect is critical to nonlinear modeling and analysis. As
indicated intuitively by Figure 2, gravity loading will influence structural response under
significant lateral displacement. P- may contribute to loss of lateral resistance, ratcheting
of residual deformations, and dynamic instability (Deierlein et al. 2010). As shown in
Figure 3, effective lateral stiffness decreases, reducing strength capacity in all phases of
the force-deformation relationship (PEER/ATC 2010). To consider P- effect directly,
gravity load should be present during nonlinear analysis. Application will cause minimal
increase to computational time, and will remain accurate for drift levels up to 10%
(Powell 2006).

The Difference between P-Delta article further explains the difference between P- and P-
.

Material nonlinearity

Material nonlinearity is associated with the inelastic behavior of a component or system.

Inelastic behavior may be characterized by a force-deformation (F-D) relationship, also known as
a backbone curve, which measures strength against translational or rotational deformation. The
general F-D relationship shown to the right indicates that once a structure achieves its yield
strength, additional loading will cause response to deviate from the initial tangent stiffness
(elastic behavior). Nonlinear response may then increase (hardening) to an ultimate point before
degrading (softening) to a residual strength value.
A variety of F-D relationships may characterize material nonlinearity, including the following:

Monotonic curve

Hysteretic cycle

Interaction surface

Monotonic curve

A monotonic curve is produced when a load pattern is progressively applied to a component or

system such that the deformation parameter (independent variable) continuously increases from
zero to an ultimate condition. The corresponding force-based parameter (dependent variable) is
then plotted across this range, indicating the pattern of material nonlinearity.

Static-pushover analysis is a nonlinear method which generates a monotonic response curve. The
P-M2-M3 hinge is best suited for modeling a condition of static pushover. Some examples of
monotonic F-D relationships (and their associated physical mechanism) include stress-strain
(axial), moment-curvature (flexure), and plastic-hinging (rotation).

To simplify the expression of a monotonic F-D relationship, and to provide for numerically-
efficient formulation, the nonlinear curve may be idealized as a series of linear segments. Figure
2 presents one such model. When the general curve is compared with the idealized, it is evident
that an exact formulation may be simplified with minimal compromise to accuracy.

Figure 2 - Idealized monotonic backbone curve

Serviceability parameters may then be superimposed onto the nonlinear F-D relationship to
provide insight into structural performance. Property owners and the general public are very
much interested in performance measures which relate to daily use. Therefore it may be useful to
introduce such limit states as immediate-occupancy (IO), life-safety (LS), and collapse-
prevention (CP), which indicate the correlation between material nonlinearity and deterministic
projections for structural damage sustained. Figure 3 depicts the serviceability limit states of a F-
D relationship.

Figure 3 - Serviceability limit states

Limit states may also be specific to inelastic behavioral thresholds. For example, under static
pushover, a confined reinforced-concrete column may experience 1). yielding of longitudinal
steel; 2). spalling of cover concrete; 3). crushing of core concrete; 4). fracture of transverse
reinforcement; and 5). fracture of longitudinal steel.

Hysteretic cycle

Another relationship type which indicates material nonlinearity is the hysteretic cycle. When the
F-D relationship is developed for a component or system subjected to cyclic loading, hysteretic
loops are produced. When modeling hysteretic dynamics, the fiber hinge is best applied.

Figure 4 illustrates hysteretic behavior. Again, translational or rotational deformation is the

independent variable. As the orientation of loading continually reverses, a strength-based
parameter is plotted against the physical oscillation of the system. Hysteresis is useful for
characterizing dynamic response under application of a time-history record.

Figure 4 - Hysteresis loop

As seen in Figure 4, both stiffness and strength deviate from their initial relationship once
yielding occurs. This behavior advances with additional hysteretic cycles, and becomes more
pronounced with greater inelastic deformations. Initially, strength may increase through
hardening behavior, though ultimately, stiffness and strength will both degrade through softening
behavior. Whereas strength gain or loss is indicated by the strength level achieved, the decrease
in slope upon load reversal indicates degradation of stiffness. Ductility describes the ability of a
system to maintain post-peak strength levels during hysteretic behavior and increasing levels of
deformation.

As hysteresis loops develop, the profile of peak values forms the cyclic envelope. The backbone
curve produced by the cyclic envelope will be less than the monotonic curve which would result
from the same structure being subjected to monotonic loading. This may be attributed to strength
and stiffness degradation. An important provision of nonlinear modeling is the accurate
characterization of strength and stiffness relationships as a structure progresses through hysteretic
behavior. PERFORM-3D is a computational tool which offers this capability.

Depending on structural geometry and materials, a hysteretic cycle may follow one of many
different possible patterns. Four possible hysteretic-behavior types are illustrated in Figure 5:

Figure 5 - Hysteresis loop types

Information on plotting hysteresis loops is available in the Plotting link hysteresis article.

Interaction surface

An interaction surface is developed for a structural element when the combined relationship
between various strength parameters is plotted. Von Mises, Mroz, or another such plasticity
theory may be used to develop a 2D or 3D surface which represents a performance envelope for a
given limit state. Behavior exceeds the limit state when the performance measure is outside the
envelope. An example may be a 3D P-M2-M3 interaction surface describing the yielding of a
column under combined axial, strong-axis, and weak-axis bending. These three performance
measures interact in a way which may be plotted to create a 3D ellipse. A response measure
outside of the P-M-M envelope would indicate that the column has yielded.
Capacity Design

Capacity Design is a design process in which it is decided which objects within a structural
system will be permitted to yield (ductile components) and which objects will remain elastic
(brittle components).

Once ductile and brittle systems are decided upon, design proceeds according to the following
guidelines:

Ductile components are designed with sufficient deformation capacity such that they
may satisfy displacement-based demand-capacity ratio.

Brittle components are designed to achieve sufficient strength levels such that they may
satisfy strength-based demand-capacity ratio.

It is best to implement Capacity Design because structural performance is then a deliberate

intention of the designer, and not revealed in a secondary manner by computational tools.
Further, because of the many sources of uncertainty inherent to structural modeling and analysis,
unless ductile systems are predetermined, a computational tool may not accurately indicate which
systems will achieve inelastic response. In summary, Capacity Design enables the creation of a
more reliable computational model, which should lead to better structural design.

Capacity Design also comes to the relief of computational time. When an engineer knows which
objects will behave elastically, and which will be permitted to yield, material nonlinearity need
only be modeled for ductile components, while components which will not yield need only
consider elastic stiffness properties. These relationships are linear, which provides for a more
simple formulation of less computational demand. Brittle components are redesigned such that
strength capacity exceeds that demanded. A level of complexity comes with the redesign of
ductile components, however, in that ductile components may satisfy nonlinear demand-capacity
criteria through a balance of both strength and deformation capacity.

While Capacity Design should lead to more reliable modeling and more accurate results,
engineers should note that computational models only represent a mathematical simulation of
physical phenomena, and cannot exactly predict structural behavior. Too many sources of
uncertainty exist, and it is up to the designer to best characterize as many behavioral parameters
as is practical.

Response-spectrum analysis

Response-spectrum analysis (RSA) is a linear-dynamic statistical analysis method which

measures the contribution from each natural mode of vibration to indicate the likely maximum
seismic response of an essentially elastic structure. Response-spectrum analysis provides insight
into dynamic behavior by measuring pseudo-spectral acceleration, velocity, or displacement as a
function of structural period for a given time history and level of damping. It is practical to
envelope response spectra such that a smooth curve represents the peak response for each
realization of structural period.

Response-spectrum analysis is useful for design decision-making because it relates structural

type-selection to dynamic performance. Structures of shorter period experience greater
acceleration, whereas those of longer period experience greater displacement. Structural
performance objectives should be taken into account during preliminary design and response-
spectrum analysis.

Time-history analysis

Time-history analysis provides for linear or nonlinear evaluation of dynamic structural response
under loading which may vary according to the specified time function. Dynamic equilibrium
equations, given by K u(t) + C d/dt u(t) + M d2/dt u(t) = r(t), are solved using either modal or
direct-integration methods. Initial conditions may be set by continuing the structural state from
the end of the previous analysis. Additional notes include:

Step Size Direct-integration methods are sensitive to time-step size, which should be
decreased until results are not affected.

HHT Value A slightly negative Hilber-Hughes-Taylor alpha value is also advised to

damp out higher frequency modes, and to encourage convergence of nonlinear direct-
integration solutions.

Nonlinearity Material and geometric nonlinearity, including P-delta and large-

displacement effects, may be simulated during nonlinear direct-integration time-history
analysis.

Links Link objects capture nonlinear behavior during modal (FNA) applications.

Damping

Damping is a property of material which influences dynamic response. A certain type of

damping is available for each type of load case. Within all load cases of a certain type, damping
is applied consistently, though additional damping may be added to individual load cases.

Modal damping is used for response-spectrum and modal time-history analyses. Material
modal damping, also known as composite modal damping, is weighted according to
element and modal stiffness. For each material, users specify a material modal damping
ratio r, where 0 <= r < 1, which relates to the damping ratio of each mode.

Viscous-proportional damping is used for direct-integration time-history analysis. This

property is proportional to mass and stiffness.

Hysteretic-proportional damping, also mass- and stiffness-proportional, is used for

steady-state and power-spectral-density analyses. Additional information on each of these
damping types may be found in the CSI Analysis Reference Manual (Material Damping,
page 78).
 Damping devices may also be modeled as a structural subsystem, as described in the
Tuned-mass damper tutorial.

Pushover

Pushover is a static-nonlinear analysis method where a structure is subjected to gravity loading

and a monotonic displacement-controlled lateral load pattern which continuously increases
through elastic and inelastic behavior until an ultimate condition is reached. Lateral load may
represent the range of base shear induced by earthquake loading, and its configuration may be
proportional to the distribution of mass along building height, mode shapes, or another practical
means.

Output generates a static-pushover curve which plots a strength-based parameter against

deflection. For example, performance may relate the strength level achieved in certain members
to the lateral displacement at the top of the structure, or bending moment may be plotted against
plastic rotation. Results provide insight into the ductile capacity of the structural system, and
indicate the mechanism, load level, and deflection at which failure occurs.

When analyzing frame objects, material nonlinearity is assigned to discrete hinge locations where
plastic rotation occurs according to FEMA-356 or another set of code-based or user-defined
criteria. Strength drop, displacement control, and all other nonlinear software features, including
link assignment, P-Delta effect, and staged construction, are available during static-pushover
analysis.

Modal analysis

Modal analysis, or the mode-superposition method, is a linear dynamic-response procedure

which evaluates and superimposes free-vibration mode shapes to characterize displacement
patterns. Mode shapes describe the configurations into which a structure will naturally displace.
Typically, lateral displacement patterns are of primary concern. Mode shapes of low-order
mathematical expression tend to provide the greatest contribution to structural response. As
orders increase, mode shapes contribute less, and are predicted less reliably. It is reasonable to
truncate analysis when the number of mode shapes is sufficient.

A structure with NDOF will have N corresponding mode shapes. Each mode shape is an
independent and normalized displacement pattern which may be amplified and superimposed to
create a resultant displacement pattern, as shown in Figure 1:
 Figure 1 - Resultant displacement and modal components

Numerical evaluation proceeds by reducing the equations of motion (N simultaneous differential

equations coupled by full mass and stiffness matrices) to a much smaller set of uncoupled second
order differential equations (N independent normal-coordinate equations). The orthogonality of
mode-shape relations enables this reduction.

Buckling

Buckling occurs physically when a structure becomes unstable under a given loading
configuration, and mathematically when a bifurcation occurs in the solution to equations of static
equilibrium. The two primary means for performing buckling analysis include Eigenvalue and
Nonlinear buckling analyses. Buckling must be explicitly evaluated for each set of loads
considered because, unlike natural frequencies, buckling modes are dependent upon a given load
pattern. When evaluating buckling, any number of load cases may be defined, each of which
should specify loading, convergence tolerance, and the number of modes to be found. Since the
first few buckling modes may have similar factors, we recommend finding a minimum of six
modes.