5.1.3.2.

1 Linear analysis methods

Linear analysis covers a broad range of methods, which use linear elastic material behavior to
determine structural dynamic properties and responses. Linear analysis can be conducted using
force-based approaches, such as linear static analysis and linear dynamic analysis (RSA); or it
can be conducted using a linear time-history approach (RHA). According to PEER/TBI (2017),
linear analysis (RSA or RHA) is appropriate for service-level earthquake (SLE) evaluation and
for design earthquake (DE) when required by ASCE 7-16 (2017a). Some considerations for each
linear approach are as follows:

Linear static analysis is a design approach where equivalent static story forces, due to
wind or earthquakes, are applied to the structure. The computation of story forces is
prescriptive, and formulations for calculating these forces are provided within the
applicable building code (Section 5.1.1). Linear static analysis is typically restricted to
use in regular structures, where dynamic behavior is dominated by the fundamental mode
of vibration, without significant higher modes and torsion effects and in regions of low
seismicity.

Since tall buildings often exhibit significant higher mode effects and the effects of torsion
are important, linear dynamic analysis, instead of linear static analysis, is typically
conducted for the seismic design of tall buildings, even in regions of low seismicity.

Linear dynamic analysis methods are based on procedures which employ the concept of
modal superposition and are often associated with seismic design. Many building codes
employ prescriptive provisions for modal RSA, where design acceleration response
spectra are used to computate the peak linear response for each mode of vibration
(Section 5.1.1). Peak modal responses are then statically combined (with a complete
quadratic combination (CQC) method (PEER/TBI, 2017)) to generate the overall
response of the structure.

For wind-governed tall buildings, equivalent static wind forces provided by the wind
consultant are often used in conjunction with linear dynamic/modal RSA for earthquakes.
The structural engineer is not directly performing linear dynamic analysis for wind
effects within their computation model, because the wind consultant creates the response
spectra upon which the static loading provided is determined. A force spectrum using
wind tunnel test data is used in conjunction with dynamic properties provided by the
structural engineer to create wind response spectra in frequency domain, which are then
used to create equivalent static loading and the associated load combination factors for x-
sway, y-sway, and torsion.

•
 Linear time-history analysis methods determine the structural response using inputs
which vary with time. Acceleration versus time input signals can be applied to above-
grade stories for determining response under wind loading, and base excitation
accelerograms (earthquake records) can be used for determining response under seismic
ground motion.

Linear time-history analysis is typically conducted using modal analysis methods;

whereas NLTHA can be conducted using fast nonlinear analysis (FNA, also called a
modal method) or using direct-integration (DI) time-history analysis (as discussed in
Section 5.1.3.2.3).

Linear methods for dynamic modification systems

When force-based approaches, such as linear static or linear dynamic analysis, are used, it is
helpful to think of results as a snapshot in time corresponding to a certain moment in the
building’s loading history. Commonly, linear analysis is used to determine the baseline building
properties provided to the wind consultant for use in motion assessment and to determine the
equivalent static forces used for the calculation of interstory drift and other serviceability criteria.

When motion-activated devices, such as TMD or tuned sloshing dampers, are employed, linear
dynamic analysis is often utilized throughout the design process, with material properties varied
when assessing the supplemental damping system’s performance. This approach is reasonable
because motion-activated devices are most commonly employed to control accelerations
resulting from dynamic wind interaction with the structure and are typically not used to improve
seismic response. In regions of low seismicity, tall structures are designed to remain essentially
elastic under service-level wind loading when compared to the much larger inelastic behavior
observed under intense ground shaking due to earthquakes (in areas of high seismicity) and,
therefore, a bounded linear dynamic analysis with appropriate assumptions/empirical parameters
could be employed. In this scenario, modal RSA or linear time-history analysis for seismic
design would generally follow code prescriptive methodology and would not account for any
supplemental damping effects created under the ground motions. Although supplemental
damping effect created by the device would not be used to improve response, its additional mass
must be considered when determining seismic forces and building frequencies.

Similarly, some displacement- or velocity-activated supplemental damping systems may be

modeled using linear methods, provided that their behavior can be adequately captured using
linear links (springs and dashpots). Whether linear or nonlinear analysis is used, damper link
formations should be based on full-scale testing conducted by the damper manufacturer and must
be carefully integrated into structural analysis models to ensure the link formulations are
validated for the range of frequencies, displacements, and temperatures being investigated. If
supplemental damping is ignored for earthquake response (i.e., wind serviceability only), the
stiffness of the links must still be considered when determining seismic forces and building
frequencies. It should be noted that when displacement- or velocity-activated distributed
damping systems are approximated using linear analysis, supplemental damping produced is
typically smeared over the entire model and equivalent modal damping is reported (for Eigen
analysis to be employed, see the different step-by-step procedures as described in the later
sections).

When linear analysis methods are used, it is critical that appropriate effective element stiffness is
considered for particular bounds, since, for example, concrete cracking, structural steel
connection fixity, and foundation–soil interaction are not varied during a given model solution.
Empirical stiffness modifiers are typically used to account for material nonlinear behavior for
each analysis. The codes (e.g., ASCE, 2017a; PEER/ATC, 2010; AISC, 2016a; PEER/TBI,
2017; CSA, 2014) usually propose property modifiers for both material strength and effective
stiffness, depending on the criteria the building needs to be designed for.

Material strength shall be based on project-specific data. When these are not available, it is
permissible to use expected properties instead of nominal or specified properties (see Table 5.2
for an example valid for US material (PEER/TBI, 2017)). However, for evaluation and retrofit
design of existing buildings, ASCE 41-13 (ASCE, 2013) specifies lower bound properties (≈
nominal properties) in force-based checks of elements that are prone to experiencing nonductile
behavior, whereas it requires the use of “expected strength” when the element is expected to
experience ductile behavior.

Table 5.2. Expected Material Strengths (PEER/TBI, 2017)

Material Expected Yield Strength Expected Ultimate Strength

Reinforcing Steel
A615 Grade 60 482 MPa 731 MPa
A615 Grade 75 565 MPa 786 MPa
A706 Grade 60 476 MPa 655 MPa
A706 Grade 80 586 MPa 772 MPa
a
Structural Steel 1.1–1.5 fy 1.1–1.2 fua
Concrete 1.3 fc
a
For structural steel the expected strength factors are different depending on the type of
steel section utilized (i.e., hot rolled structural shapes and bars, hollow structural sections,
steep pipe, and plates).

Material stiffness shall take into account the effects of axial, flexural, and shear cracking. For
steel members the elastic stiffness is determined by the full cross-sectional properties and the
elastic modulus of steel (200,000 MPa) (PEER/TBI, 2017). However, concrete member stiffness
can be estimated with effective values as given in Table 5.3 (ASCE, 2017a), in Table 5.4
(PEER/TBI, 2017), and Table 5.5 (CSA, 2014), in lieu of detailed analysis. As it can be seen in
the tables multipliers are less than one in most of the cases since they represent the effective
linear branch of the inelastic model (PEER/TBI, 2017). In the tables, EC is the concrete elastic
modulus; Ig is the cross-sectional gross moment of inertia; Aw is the cross-sectional web
nominal area; and Ag is the cross-sectional gross area.

Table 5.3. Effective Stiffnesses of RC Structural Elements (for DE and MCER Demands)
According to ASCE 41-17 (ASCE, 2017b)
 Shear
Component Flexural Rigidity Axial Rigidity
Rigidity
Beams (nonprestressed)a 0.3EcIg 0.4EcAW –
a
Beams (prestressed) EcIg 0.4EcAW –
Columnsb with compression under gravity
0.7EcIg 0.4EcAw EcAg
load ≥0.5Agfc′
EcAg
Columnsb with compression under gravity
0.3EcIg 0.4EcAw (compression)
load ≤0.1Agfc′ or with tension
EsAs (tension)
EcAg
Walls (cracked) 0.5EcIg 0.4EcAw (compression)
EsAs (tension)
Flat slabs (nonprestressed) (4c1/l1)EcIg≥(1/3)EcIg 0.4EcAg EcAg
Flat slabs (prestressed) 0.5EcIg 0.4EcAg EcAg

Note: c1 is size of the equivalent column in the direction of the span.

l1 is the length of the slab in the direction of the seismic force.

a
For T-beams, Ig can be taken as twice the value of Ig of the web alone. For coupling
beams in coupled shear walls, the effective stiffness values given for nonprestressed
beams should be considered, unless alternative stiffness is determined by more detailed
analysis (ASCE 41-17 (ASCE, 2017b)).
b
For columns with axial compression falling between the limits provided, flexural rigidity
should be determined by linear interpolation. If interpolation is not performed, the more
conservative effective stiffness should be used.

Table 5.4. Effective Stiffnesses of RC Structural Elements According to PEER/TBI (2017)

Service-Level Linear Model MCER-Level Nonlinear Models

Component
Axial Flexural Shear Axial Flexural Shear
Structural
wallsa (in- 1.0EcAg 0.75EcIg 0.4EcAg 0.75EcAg 0.35EcIg 0.2EcAg
plane)
Structural
walls (out-of- – 0.25EcIg – – 0.25EcIg –
plane)
Basement
walls (in- 1.0EcAg 1.0EcIg 0.4EcAg 1.0EcAg 0.8EcIg 0.2EcAg
plane)
Basement
walls (out-of- – 0.25EcIg – – 0.25EcIg –
plane)
 Service-Level Linear Model MCER-Level Nonlinear Models
Component
Axial Flexural Shear Axial Flexural Shear
Coupling
beams with
0.07(lh)EcIg≤0.3E 0.07(lh)EcIg≤0.3E
conventional 1.0EcAg 0.4EcAg 1.0EcAg 0.4EcAg
cIg cIg
or diagonal
reinforcement
Composite
steel/reinforc
1.0(EA)tra 1.0EsAs 1.0(EA)tra 1.0EsAs
ed concrete 0.07(lh)(EI)trans 0.07(lh)(EI)trans
ns w ns w
coupling
beamsb
Non-post-
tension (PT)
transfer
0.5EcAg 0.5EcIg 0.4EcAg 0.25EcAg 0.25EcIg 0.1EcAg
diaphragms
(in-plane
only)
PT transfer
diaphragms
0.8EcAg 0.8EcIg 0.4EcAg 0.5EcAg 0.5EcIg 0.2EcAg
(in-plane
only)
Beams 1.0EcAg 0.5EcIg 0.4EcAg 1.0EcAg 0.3EcIg 0.4EcAg
Columns 1.0EcAg 0.7EcIg 0.4EcAg 1.0EcAg 0.7EcIg 0.4EcAg
Mat (in-
0.8EcAg 0.8EcIg 0.8EcAg 0.5EcAg 0.5EcIg 0.5EcAg
plane)
Mat (out-of-
– 0.8EcIg – – 0.5EcIg –
plane)
a
When modeled as line elements and not as fiber elements.
b
(EI)trans is the cracked trasformed sections.

Table 5.5. Effective Stiffnesses of RC Structural Elements According to Canadian Standard

Association (CSA) A23.3 (CSA, 2004)

Ranges for Linear Elastic Analysis

Load Type Component
Serviceability (SLS) Strength (ULS)
Wind 0.50Ig 0.40Ig
Nondiagonally reinforced coupling beams
Ave=0.40Ag Ave=0.25Ag
0.45Ig 0.35Ig
Diagonally reinforced coupling beams
Ave=0.45Ag Ave=0.40Ag
Shear walls 0.95Ig 0.75Ig
Ave=0.95Ag Ave=0.75Ag
 Ranges for Linear Elastic Analysis
Load Type Component
Serviceability (SLS) Strength (ULS)
Shear walls in net tension Refined calculation required
Slabs with mild reinforcement 0.35Ig 0.20Ig
Posttensioned slabs 0.60Ig 0.45Ig
0.50Ig 0.40Ig
Beams (excluding coupling beams)
Ave=0.75Ag Ave=0.50Ag
Columns 1.0Ig 0.70Ig
Ave=1.0Ag Ave=0.70Ag
Columns in net tension Refined calculation required
Beams (excluding coupling beams) 0.40Ig 0.40Ig
Columns acIg acIg
0.40Ig 0.40Ig
(Nondiagonally reinforced) coupling beams
Ave=0.15Ag Ave=0.15Ag
Seismic 0.25Ig 0.25Ig
(Diagonally reinforced) coupling beams
Ave=0.45Ag Ave=0.45Ag
Slab frame element 0.20Ig 0.20Ig
awIg awIg
Wall
Axe=awAg Axe=awAg

Note: From CSA A23.3 (CSA, 2014) Table N9.2.1.2 and Table 21.1.

Stability analysis of natural circulation systems

Pallippattu Krishnan Vijayan, ... Naveen Kumar, in Single-Phase, Two-Phase and Supercritical
Natural Circulation Systems, 2019

8.2.2 Nonlinear analysis

Linear analysis is well suited to generate a stability map. However, linear analysis tells us
whether a particular steady state is stable or unstable for infinitesimal disturbances. It does not
tell us how the steady state can be approached. Stability thresholds of an NCS depend on the way
we approach (i.e., the operating procedure) the steady state which involves finite disturbances. In
addition, due to the hysteresis effects (Chen et al., 2001) or the existence of a conditionally stable
(Vijayan and Date, 1992; Vijayan et al., 2001, 2007) region, the stability threshold depends on
the operating procedure. Achard et al. (1985) showed that finite amplitude perturbation can cause
instability on the stable side of the linear stability boundary. Hence, nonlinear analysis is
required to establish the stable operational domain and start-up procedure for NCSs.

Nonlinear analysis is carried out by directly solving the governing transient partial differential
equations numerically. The nonlinear analysis can be carried out with the transient formulation
described for single-phase NCSs in Chapter 4 which is based on the dimensional equations. The
nonlinear analysis with dimensionless equations, however, is much simpler and is independent of
the fluid used. The analysis is carried out as a normal transient analysis with the steady-state
initial conditions for which an analytical solution exists. The nonlinear analysis is usually carried
out by perturbing one of the parameters (say 5% increase) in Grm (dimensionless power) or Stm
(modified Stanton number which represents the overall heat transfer coefficient) or ω
(dimensionless flow rate) for a short duration (say 10 seconds) and observing the behavior of the
solution. If the perturbation grows in time, then it is considered as unstable. On the other hand, if
the perturbation dies out with time, then it is stable. If the perturbations continue with the same
amplitude, then it is neutrally stable. The threshold of neutrally stable conditions is arrived at in
this way for different operating conditions to arrive at the stability map. Strictly speaking, the
perturbation is not required as this kind of perturbation exists in numerical computations because
of the truncation error.

Before the calculations can commence the loop is divided into a number of nonoverlapping small
segments. A node separates two such segments. The energy equation for the various segments of
the loop (i.e., Eqs. 8.6a–8.6c) is discretized to obtain the following equations for the temperature
of the ith node at the new time step (i.e., n + 1) as a function of the old (i.e., nth) nodal
temperatures at the ith and (i−1)th nodes.

(8.42)θin+1=θin[1−ϕωnΔτΔS]+θi−1n[ϕωnΔτΔS]+LtΔτLh1<N≤Nh(heater)
(8.43)θin+1=θin[1−ϕωnΔτΔS]+θi−1n[ϕωnΔτΔS]Nh+1<N≤NhlandNc+1<N≤Nt(pipes)
(8.44)θin+1=θin[1−ϕωnΔτΔS−StmΔτ]+θi−1n[ϕωnΔτΔS]Nhl+1<N≤Nc(cooler)

Fig. 8.6A and B can be referred for the values of Nh, Nhl, Nc and Nt used above. Similarly, the
momentum equation can be discretized implicitly as

Sign in to download full-size image

Figure 8.6. Nodalization for nonlinear stability analysis: (A) clockwise flow and (B)
anticlockwise flow.

(8.45)ωn+1+pLtΔτ2Ressbωn+12−b=ωn+GrmΔτRess3θI

where
(8.46)θI=∮θin+1dZ=∫ZN1ZN2θin+1dZ−∫ZN3ZN4θin+1dZ

The limits ZN1, ZN2, ZN3, and ZN4 correspond to the dimensionless elevation at nodes N1, N2, N3 and
N4, respectively (see Fig. 8.6A and B). The explicit scheme has been used for the energy
equation for which the stability criterion satisfying Eq. (8.44) is

(8.47)Δτ≤1ϕωnΔS+Stm

To ensure stability, the calculated time step is usually multiplied by a number less than unity. A
uniform segment length of 1 cm is selected for the present calculations. For clockwise or
anticlockwise flow, the marching calculations started with node 1 (first node in the heater) and
ended with node Nt. Eq. (8.42) was used to calculate the nodal temperatures in the heater (i.e., for
nodes 1 < N ≤ Nh). From Nh + 1 < N ≤ Nhl Eq. (8.43) was used. Similarly, Eq. (8.44) was used for
Nhl + 1 < N ≤ Nc. For the cold leg Eq. (8.43) was used for Nc + 1 < N ≤ Nt. Once all the nodal
temperatures are calculated ωn+1 is calculated using Eq. (8.45) after evaluating the temperature
integral numerically by Simpson’s rule. For the horizontal heater horizontal cooler (HHHC)
orientation, flow direction could be clockwise or anticlockwise. For flow in the anticlockwise
direction the marching direction was reversed, keeping the same sign for ω in Eqs. (8.42)–(8.44).
The flow rate so calculated will always be positive.

The adequacy of the nonlinear formulation was checked by comparing with the analytical
steady-state equations. Also, the stability threshold predicted by the code can be compared with
the corresponding threshold obtained by the linear method. For this comparison, the initial
conditions shall correspond to the steady-state value for the orientation considered.

Tire Characteristics and Vehicle Handling and Stability

Hans B. Pacejka, in Tire and Vehicle Dynamics (Third Edition), 2012

Effective Axle Cornering Stiffness

Linear analysis, valid for relatively small levels of lateral accelerations, allows the use of
approximate tire characteristics represented by just the slopes at zero slip. We will first derive the
effective axle cornering stiffness that may be used under these conditions. The effects of load
transfer, body roll, steer compliance, side force steer, and initial camber and toe angles will be
included in the ultimate expression for the effective axle cornering stiffness.

The linear expressions for the side force and the aligning torque acting on a tire have been given
by Eqns (1.5). The coefficients appearing in these expressions are functions of the vertical load.
For small variations with respect to the average value (designated with subscript o), we write for
the cornering and camber force stiffnesses the linearized expressions:

(1.7)CFα=CFαo+ζαΔFzCFγ=CFγo+ζγΔFz

where the increment of the wheel vertical load is denoted by ΔFz and the slopes of the coefficient
versus load curves at Fz = Fzo are represented by ζα,γ.
When the vehicle moves steadily around a circular path, a centripetal acceleration ay occurs and a
centrifugal force K = may can be said to act on the vehicle body at the center of gravity in the
opposite direction. The body roll angle φ that is assumed to be small is calculated by dividing the
moment about the roll axis by the apparent roll stiffness which is reduced with the term mgh′ due
to the additional moment mgh′φ:

(1.8)φ=−mayh'cφ1+cφ2−mgh'

The total moment about the roll axis is distributed over the front and rear axles in proportion to
the front and rear roll stiffnesses. The load transfer ΔFzi from the inner to the outer wheels that
occurs at axle i (= 1 or 2) in a steady-state cornering motion with centripetal acceleration ay
follows from the formula:

(1.9)ΔFzi=σimay

with the load transfer coefficient of axle i:

(1.10)σi=12si(cφicφ1+cφ2−mgh'h'+l−ailh')

The attitude angle of the roll axis with respect to the horizontal is considered small. In the
formula, si denotes half the track width, h′ is the distance from the center of gravity to the roll
axis, and a1 = a and a2 = b. The resulting vertical loads at axle i for the left (L) and right (R)
wheels become, after considering the left and right increments in load,

(1.11)ΔFziL=ΔFzi,ΔFziR=−ΔFziFziL=12Fzi+ΔFzi,FziR=12Fzi−ΔFzi

The wheels at the front axle are steered about the king-pins with the angle δ. This angle relates
directly to the imposed steering wheel angle δstw through the steering ratio nst:

(1.12)δ=δstwnst

In addition to this imposed steer angle, the wheels may show a steer angle and a camber angle
induced by body roll through suspension kinematics. The functional relationships with the roll
angle may be linearized. For axle i, we define

(1.13)ψri=ɛiφγri=τiφ

Steer compliance gives rise to an additional steer angle due to the external torque that acts about
the king-pin (steering axis). For the pair of front wheels, this torque results from the side force
(and, of course, also from here not considered driving or braking forces) that exerts a moment
about the king-pin through the moment arm which is composed of the caster length e and the
pneumatic trail t1. With the total steering stiffness cψ1 felt about the king-pins with the steering
wheel held fixed, the additional steer angle becomes when for simplicity the influence of camber
on the pneumatic trail is disregarded:

(1.14)ψc1=−Fy1(e+t1)cψ1
In addition, the side force (but also the fore-and-aft force) may induce a steer angle due to
suspension compliance. The so-called side force steer reads

(1.15)ψsfi=csfiFyi

For the front axle, we should separate the influences of moment steer and side force steer. For
this reason, side force steer at the front is defined to occur as a result of the side force acting in a
point on the king-pin axis.

Besides the wheel angles indicated above, the wheels may have been given initial angles that
already exist at straight ahead running. These are the toe angle ψo (positive pointing outward)
and the initial camber angle γo (positive: leaning outward). For the left and right wheels, we have
the initial angles:

(1.16)ψiLo=−ψio,ψiRo=ψioγiLo=−γio,γiRo=γio

Adding all relevant contributions (1.12) to (1.16) together yields the total steer angle for each of
the wheels.

The effective cornering stiffness of an axle Ceff,i is now defined as the ratio of the axle side force
and the virtual slip angle. This angle is defined as the angle between the direction of motion of
the center of the axle i (actually at road level) when the vehicle velocity would be very low and
approaches zero (then also Fyi → 0) and the direction of motion at the actual speed considered.
The virtual slip angle of the front axle has been indicated in Figure 1.4 and is designated as αa1.
We have, in general,

(1.17)Ceff,i=Fyiαai

The axle side forces in the steady-state turn can be derived by considering the lateral force and
moment equilibrium of the vehicle:

(1.18)Fyi=l−ailmay

The axle side force is the sum of the left and right individual tire side forces. We have

(1.19)FyiL=(12CFαi+ζαiΔFzi)(αi−ψio)+(12CFγi+ζγiΔFzi)(γi−γio)FyiR=(12CFαi−ζαiΔFzi)
(αi+ψio)+(12CFγi−ζγiΔFzi)(γi+γio)

where the average wheel slip angle αi indicated in the figure is

(1.20)αi=αai+ψi

and the average additional steer angle and the average camber angle are

(1.21)ψi=ψri+ψci+ψsfiγi=γri
The unknown quantity is the virtual slip angle αai which can be determined for a given lateral
acceleration ay. Next, we use Eqns (1.8, 1.9, 1.13, 1.18, 1.14, 1.15), substitute the resulting
expressions (1.21) and (1.20) in (1.19), and add up the two equations (1.19). The result is a
relationship between the axle slip angle αai and the axle side force Fyi. We obtain, for the slip
angle of axle i,

(1.22)αai=FyiCeff,i=FyiCFαi[1+l(ɛiCFαi+τiCFγi)h'(l−ai)(cφ1+cφ2−mgh')
+CFαi(ei+ti)cψi−CFαicsfi+2lσil−ai(ζαiψio+ζγiγio)]

The coefficient of Fyi constitutes the effective axle cornering compliance, which is the inverse of
the effective axle cornering stiffness (1.17). The quantitative effect of each of the suspension,
steering, and tire factors included can be easily assessed. The subscript i refers to the complete
axle. Consequently, the cornering and camber stiffnesses appearing in this expression are the
sum of the stiffnesses of the left and right tire:

(1.23)CFαi=CFαiL+CFαiR=CFαiLo+CFαiRoCFγi=CFγiL+CFγiR=CFγiLo+CFγiRo

in which (1.7) and (1.11) have been taken into account. The load transfer coefficient σi follows
from Eqn (1.10). Expression (1.22) shows that the influence of lateral load transfer only occurs if
initially, at straight ahead running, side forces are already present through the introduction of
e.g., opposite steer and camber angles. If these angles are absent, the influence of load transfer is
purely nonlinear and is only felt at higher levels of lateral accelerations. In the next subsection,
this nonlinear effect will be incorporated in the effective axle characteristic.

An Overview of Part and Assembly Stress Analysis

Wasim Younis, in Up and Running with Autodesk Inventor Simulation 2010, 2009

LINEAR ANALYSIS
Sign in to download full-size image

Some definitions are as follows:

Stress=ForceArea, Strain=Change in LengthOriginal Length, Young's Modulus=StressStrain

TheYoung’s Modulus provides the stiffness of the material; for example, a higher Young’s
modulus will produce a stronger material and a lower Young’s modulus will produce a weaker
material.

Sign in to download full-size image

Sign in to download full-size image

Note: For linear analysis, it is assumed that the change in length is very small compared to the
original length.

Factor of Safety=Calculated StressYield Limit Stress
Assumptions normally made when conducting a linear analysis
 The material properties of the component remain linear after the yield limit. Hence, results
1
beyond this limit are not valid using Autodesk Inventor Simulation Suite.
2 The deflections of components are small compared to the overall component size.
3 The components are rigid and ductile; for example, metal components (not rubber).
The components deform equally in all three directions; for example, material properties are
4
isotropic.

The Stress Analysis Environment

Wasim Younis, in Up and Running with Autodesk Inventor Simulation 2011 (Second Edition),
2010

Linear analysis

Sign in to download full-size image

Young's modulus provides the stiffness of the material; for example a higher Young's modulus
will produce a stronger material and a lower Young's modulus will produce a weaker material.

Sign in to download full-size image

Sign in to download full-size image
Strain=Changeinlengthoriginallengthand=ForceArea

(Note: for linear analysis it is assumed that the change in length is very small compared to the
original length.)

Assumptions normally made when conducting a linear analysis

The material properties of the component remain linear after the yield limit. Hence, results
1
beyond this limit are not valid using Autodesk Inventor Simulation Suite
2 The deflections of components are small compared to the overall component size
3 The components are rigid and ductile like metal (not rubber)
The components deform equally in all three directions; that is, the material properties are
4
isotropic

Risk-based inspection technique

Mohamed Abdallah El-Reedy Ph.D., in Offshore Structures (Second Edition), 2020

Design-level method

The DLM involves linear analysis using the standard API RP2A approach for new platform
design by checking the platform on a component basis. The same factors of safety used for a new
design are used for the DLM. The key difference for assessment is the use of special assessment
criteria. The platform must be shown to perform linearly at loads equal to or greater than the
assessment criteria.

The DLM is typically the first level of direct analysis of the platform. The DLM is a simpler and
more conservative check than the USM, which is more complex and less conservative. It is
generally more efficient to begin with the DLM because it is usually simpler to implement.

Computer assistance in electronics

Ian Sinclair, in Passive Components for Circuit Design, 2001

Linear circuit analysis

The analysis of linear circuits is based on the principles of the effect of resistive and reactive
components on the amplitude and phase of a sine wave. For very simple circuits, this can be
done either by drawing phasor diagrams to scale or by the use of algebra to express the circuit
impedance as R + jX where X is the reactive component and j is the square root of minus one.
These methods are comparatively simple but very tedious, and when the circuit is one that is
only slightly more complicated than the most basic filter, the amount of work is enormously
increased.

For many standard circuits, the formulae can be obtained from reference books (such as the
splendid ITT Reference Data for Radio Engineers, now published by Howard Sams & Co. Inc.
with the ISBN of 0-672-21218-2), but the amount of manipulation that is required becomes
much greater, and in many cases you still have a lot of work to do after working out the results of
each formula. The repetitive nature of the work, unless a programmable calculator is used, means
that it is very easy to make mistakes.

When the circuit is not a standard one that can be looked up in a reference book, the analysis
becomes very much more difficult. It amounts then to combining components, resistive or
reactive, in series or in parallel, working out the first two, then combining with the next, and so
on until the whole circuit has been covered. The aim is to express the effect of the whole circuit
in the R + jX format so that the impedance magnitude (Z) is the square root of (R2 + X2) and the
phase angle (ϕ) is the angle whose tangent is X/R. This analysis is long, tedious, and very liable
to errors. It requires a good grasp of working with complex numbers (numbers including j) and
for a large circuit the effort is very considerable. The working for a simple parallel resistor and
capacitor will convince you, if you do not already know, that there is quite a lot of work
involved.

Another dimension is added when a circuit contains one or more active components. The gain of
an active component converts a passive circuit, whose power gain is always less than unity, into
an active circuit which can have a power gain of more than unity over a considerable frequency
range (the bandwidth). The gain–bandwidth product of the active device needs to be considered,
however, as does the effect of the impedances of the active device.

None of this need be a problem for anyone with access to a PC-compatible computer because
there are now several programs which ensure that even for very complex linear circuits, the
effort of calculating frequency response can be performed by the computer. This brings linear
circuit analysis, once possible only on very large and expensive computers, within the reach of
any user, amateur or professional.

There are many linear analysis programs available at the time of writing, and if you use the
Schematics portion of the PSPICE package the logical answer is to use PSPICE itself for
analysis, and the companion PROBE for producing a graphical output. When you have a circuit
diagram, with input and load, you can then produce an analysis, after saving the circuit as a file.
The steps are:

1.

Run Electrical Rule Check

2.
 Create Netlist

3.

Run PSPICE

4.

Run PROBE

5.

Use Trace to get output graph

You may find that the PSPICE step refuses to run, either because there is an error in the circuit
(like no input) or because you have not specified the correct form of analysis (like AC sweep), or
because you have not correctly specified the range of frequencies (PSPICE uses k for 1000 and
MEG for 1 000 000 – a common fault is to use M, this means 10−3). Figure 10.4 shows a PROBE
output graph for the circuit shown in Figure 10.3.

Sign in to download full-size image

Figure 10.4. The graphical output for the circuit of Figure 10.3 taken at the output.

•
 Note that you can obtain an output at several designated points in a circuit. If your graph
shows no changes, it may be because your designated point is the grounded end of a
component.

The PSPICE package is ideal if you are using US conventions for circuit diagrams, but another
analysis program that is particularly well suited for the small-scale UK user, professional or
amateur, is called ACIRAN, and information is available from the website:

http://www.aciran.co.uk/about.htm

ACIRAN has been written by an engineer in the UK and is available in shareware form. For
anyone not familiar with this way of distributing software, shareware programs are not
obtainable in the usual way from computer shops but by downloading over the Internet and also
from specialists in Public Domain and Shareware disks. Although most shareware programs
originate in the USA, the ACIRAN program and one other linear analysis program have been
written in the UK, so upgrades and support are more readily available.

In addition to its obvious applications to printing tables and graphs of response for a circuit, the
use of the ACIRAN program makes it possible to allow for the effect of input and output
impedances and of component tolerances, something that is particularly time consuming if done
in the traditional ways. This is, however, the type of information that is particularly needed for
small-scale production circuit design, so the use of methods based on the computer is a valuable
aid to anyone involved in such work. In addition to the conventional R, C and L passive
components, transmission lines can be dealt with, a considerable help to anyone involved in UHF
design. As well as the built-in systems for dealing with transistors, FETs and op-amps, there is a
current-generator component which can be used to simulate the action of any active components
much more closely at high frequencies.

ACIRAN cannot be used until it has been notified of the circuit components and connections,
and since ACIRAN cannot read a circuit diagram (although if circuit diagrams were prepared
with standardized drafting programs such as ORCAD this could be done) it is necessary to use a
system for entering the component positions and values. This system depends on identifying and
numbering the circuit nodes, as you would when laying the circuit out for construction on PCB
or stripboard.

Figure 10.5 shows a typical simple (in terms of components) passive circuit, with input and
output and a common earth line, with nodes marked by numbers. A node, in this sense, means a
point where components join, and this will normally include the input and the output as well as
the earth (ground) connection. Each node can be numbered, and the convention followed by
ACIRAN is that the earth (ground) node is always 0 (zero) and the input node is always 1. Other
nodes can be numbered as you please, but it is better to take a logical arrangement of node
numbers, moving from input to output.
Sign in to download full-size image

Figure 10.5. A simple circuit with nodes marked.

The general rule about nodes is that no node can have more than one number, and nodes are
always separated by components. If there is no component between two nodes, one of the nodes
must be redundant. If you need to work with nodes connected, use a small resistor value such as
0R33 between the nodes. You can use whatever number you like for the output node, because
this will be notified to ACIRAN, but you must use the numbers 0 and 1 for the earth and input
nodes respectively. Other linear analysis programs allow you to specify node numbers for all
points, but this simply takes longer to set up; since every circuit will have an input and an output
there is no reason why these should not be pre-allocated with node numbers.

The total number of nodes in this circuit example is five (numbered 0 to 4). There are also five
components, but this is purely coincidental because the number of nodes does not depend in any
simple way on the number of components. The point to watch in this illustration is the node
numbered 4, because it is easy to overlook this one. Nodes where three or more components join
are easy to spot on a well-drawn diagram because of the blob at the junctions, but this type of
two-component junction is often less easy to see.

The form of the data can be seen in part in Figure 10.6, a view of the ACIRAN screen when the
Data option of the menu has been selected. Small circuits can be checked from this display, but
for a complex circuit it is better to print the data out and check against the circuit diagram, which
should be numbered to show node positions. Once this set of data items has been entered,
component values can be changed, but you cannot change the connections of a component. If
you find, for example, that you have mistakenly entered R3 as lying between nodes 3 and 0 and
wish to change this to nodes 4 and 0, you have to do this by deleting the component and re-
entering R3 with its correct value between the correct nodes. The list will always show that a
component has been deleted as a reminder of what you have done.
Sign in to download full-size image

Figure 10.6. The components list, showing values and node numbers.

The form of entry of values follows the standard methods of specifying value, so that a figure by
itself is taken to be in fundamental units of ohms, farads or henries. The suffix values of k, M, m,
n, p and so on (lower case or capitals are taken as identical except for M and m) are all
recognized, and you can enter values in the format 3K3 if you wish.

The analysis of this circuit is started with a wide frequency range and logarithmic response
selected. This is always advisable so that the overall picture of the response can be seen,
allowing you subsequently to change the frequency limits and sweep type if you want to see
more detail. By specifying a large range initially, you ensure that the circuit has no surprises
lurking outside the frequency range for which it is intended. This is unlikely in such a simple
circuit, but the point about using ACIRAN is that it allows you to analyse circuits that are far
from simple.

The table that appears following analysis takes the form shown in Figure 10.7 – this is only a
part of the table as displayed on the screen. For each frequency in the range, the values of
magnitude (amplitude), phase angle and time delay are printed, using units of decibels for
amplitude, degrees for phase angle and seconds for time delay. The graph for amplitude is
obtained by selecting Graph from the main menu, and appears, Figure 10.8, as the first of a set of
at least three.
Sign in to download full-size image

Figure 10.7. The table of values of magnitude, phase and time delay.

Sign in to download full-size image

Figure 10.8. The graph of gain plotted against frequency.

The other graphs are phase, Figure 10.9, and time delay, of which the time delay is usually of
least interest for linear circuit work other than for specialized purposes. When options such as
return loss and impedance values have been selected, there will be several more graphs of
magnitude and phase angle for each of the other quantities covered.

Sign in to download full-size image

Figure 10.9. The graph of phase plotted against frequency.

Circuit Fundamentals
Martin Plonus, in Electronics and Communications for Scientists and Engineers (Second
Edition), 2020

1.6.2 Superposition

Circuit theory is a linear analysis; i.e., the voltage-current relationships for R, L, and C are linear
relationships, as R, L, and C are considered to be constants over a large range of voltage and
currents. Linearity gives rise to the principle of superposition, which states that in a circuit with
more than one source present, the voltage or current anywhere in the circuit can be obtained by
first finding the response due to one source acting alone, then the second source acting alone, and
so forth. The response due to all sources present in the circuit is then the sum of the individual
responses. This is a powerful theorem which is useful since a circuit with only one source
present can be much easier to solve. Now, how do we shut off all sources in the circuit but one?15
Recall what happens to an ideal voltage source when the amplitude is cranked to zero? One is
left with a short circuit (see Fig. 1.9). Similarly, when a current source is cranked down to zero,
one is left with an open circuit (see Fig. 1.12). Therefore, in a circuit with multiple sources, all
sources except one are replaced with their respective short or open circuits. One can then proceed
to solve for the desired circuit response with only one source present. The following example
illustrates the technique.

Example 1.6

Use superposition to find the current i in Fig. 1.19a. For linear responses only (voltage and
current, but not power), the circuit of Fig. 1.19a is a superposition of the two circuits of Figs.
1.19b and c. Hence the current is the superposition of two currents: the first, due to the current
source acting alone, is flowing to the left, and the second, due to the voltage source acting alone,
is flowing to the right. Hence

i=iυs=0+iis=0=3105+10−65+10=2−0.4=1.6A
and a current of 1.6 A is flowing to the left in the circuit of Fig. 1.19a. Superposition, by
breaking up a problem into a set of simpler ones, often leads to a quick solution and provides us
with insight as to which sources contributes more.

Sign in to download full-size image

Figure 1.19. (a) A circuit with a voltage and current source present. (b) The voltage source is
removed. (c) The current source is removed.

It is tempting to calculate i2R power dissipated in, for example, the 5 Ω resistor by adding
powers. We would obtain (2)2·5 + (0.4)2·5 = 20.8 W, when in fact the actual power dissipated in
the 5 Ω resistor is only (1.6)2·5 = 12.8 W. This demonstrates succinctly that superposition applies
to linear responses only, and power is a nonlinear response, not subject to superposition.

Types of existing buildings: detailed introduction and

seismic rehabilitation
Reza Mokarram Aydenlou, in Seismic Rehabilitation Methods for Existing Buildings, 2020

3.3.4.4.2 Linear analysis method (static and dynamic) for CBF brace

3.3.4.4.2.1 Steel concentric braced frame—CBF brace

Frames that the earthquake force is depreciated by the concentric bracing are called CBF. In
which the seismic load capacity is evaluated in terms of the axial force capacity tolerated by the
braces. Bracketed frames with axes (CBF) are bracketed frame systems in which the member
axes are connected each at one point, unless the distance between the farthest intersection point
of the intersection of the beam and column axes to the cross-section are smaller than or equal to
the smallest member connecting at the intersection point.

3.3.4.4.2.1.1 Determining of stiffness of components for CBF brace frame

The stiffness determination is performed in linear analysis as described in the stiffness modeling
of FR moment frame and PR moment frame. Also to derive more realistic results, it is advisable
to consider the stiffness of the panel zone in the model. Braces shall be modeled as columns
[1,2].

3.3.4.4.2.1.2 Determination of strength of CBF brace

Determining the strength of these components includes the following.

3.3.4.4.2.1.2.1 Expected compression strength of CBF brace

1.

The expected compression strength in the brace must be one of the two minimum values
corresponding to the local buckling or the overall buckling of the member. The effective
design strength, PCE, shall be calculated in accordance with AISC (1993) LRFD
Specifications, taking φ=1.0 and using the expected yield strength Fye for yield strength.

1.

In CBF brace the effective length of each bracing member in the frame in-plane behavior
is 0.5 for length inboard plane and 0.7 for the outboard.

2.

In CHEVRON braces, the buckle length of the bracket is equal to the length of the off-
plate and is 0.8 for length.
3.3.4.4.2.1.2.2 Expected tensile strength for brace for CBF brace

The expected tensile strength QCE of steel bracing should be calculated similar to the behavior
of the tensile columns (Fig. 3.3.35).

Sign in to download full-size image

Figure 3.3.35. Types of concentric braces frame.

inear analysis

All the analyses discussed so far have been linear: there is a linear relationship between
the applied loads and the response of the system. For example, if a linear spring extends
statically by 1 m under a load of 10 N, it will extend by 2 m when a load of 20 N is
applied. This means that in a linear Abaqus/Standard analysis the flexibility of the
structure need only be calculated once (by assembling the stiffness matrix and inverting
it). The linear response of the structure to other load cases can be found by multiplying
the new vector of loads by the inverted stiffness matrix. Furthermore, the structure's
response to various load cases can be scaled by constants and/or superimposed on one
another to determine its response to a completely new load case, provided that the new
load case is the sum (or multiple) of previous ones. This principle of superposition of
load cases assumes that the same boundary conditions are used for all the load cases.

Abaqus/Standard uses the principle of superposition of load cases in linear dynamics

simulations, which are discussed in Linear Dynamics.

Nonlinear analysis

A nonlinear structural problem is one in which the structure's stiffness changes as it

deforms. All physical structures exhibit nonlinear behavior. Linear analysis is a
convenient approximation that is often adequate for design purposes. It is obviously
inadequate for many structural simulations including manufacturing processes, such as
forging or stamping; crash analyses; and analyses of rubber components, such as tires or
engine mounts. A simple example is a spring with a nonlinear stiffening response (see
Figure 1).

Figure 1. Linear and nonlinear spring characteristics.

Since the stiffness is now dependent on the displacement, the initial flexibility can no
longer be multiplied by the applied load to calculate the spring's displacement for any
load. In a nonlinear implicit analysis the stiffness matrix of the structure has to be
assembled and inverted many times during the course of the analysis, making it much
more expensive to solve than a linear implicit analysis. In an explicit analysis the
increased cost of a nonlinear analysis is due to reductions in the stable time increment.
The stable time increment is discussed further in Nonlinear Explicit Dynamics.

Since the response of a nonlinear system is not a linear function of the magnitude of the
applied load, it is not possible to create solutions for different load cases by
superposition. Each load case must be defined and solved as a separate analysis.