Designing for Tornadoes:

New Requirements in ASCE 7-22

Second Order Effects (P-d, P-D, P-Q) and Buckling Analysis

Week 9 Handout
 DF P DF P DF P
V V V

P=0.01 Pcr P=0.2 Pcr P=0.5 Pcr

2
 Week 9: Second-Order Effects

Topics presented this week include:

◼ Second-order effects
◼ P-Delta (P-D)
◼ p-delta (p-d)
◼ P-Theta (P-Q)

◼ Including second-order effects in structural analysis

◼ Buckling calculations

◼ Conforming to requirements of ACI 318 and AISC 360

◼ SAP 2000 Procedures for second-order and buckling analyses

3
 Week 9: Learning Outcomes

◼ Upon completion of this week’s content you will be able to:

◼ Explain the underlying principles of second-order analysis
◼ Differentiate between P-Delta, p-delta, and P-Theta effects
◼ Appropriately include second-order effects in structural analysis
◼ Use second-order modeling principles to determine system-level buckling loads
◼ Apply the second-order analysis requirements of ACI 318 and AISC 360

24
 Week 9: Significance on the Job

◼ Building codes and standards are evolving to require that second-order effects be evaluated
explicitly rather than by use of effective length factors, moment magnifiers, and other approximate
techniques.

◼ For example, the Direct Analysis Method provided for design of steel structures in AISC 360
requires the use of second-order analysis. For this reason, it is essential that the engineer be
familiar with the underlying concepts.

◼ Understanding the theoretical basis of second-order analysis is also necessary for successful use
of modern software that provides an assortment of techniques for incorporating second-order
effects.

25
Next Scene: Basic Concepts of Second Order
Analysis

6
 DF P DF P DF P
V V V

P=0.01 Pcr P=0.2 Pcr P=0.5 Pcr

7
Overview of Stability Issues
 D0 P DF P

V V

D0 =Vh3/3EI DF =lD0

First Order Analysis Second Order Analysis

Overturning moment Overturning moment PDF is included.
PD0 is ignored DF is determined by iteration.

9
 Second-Order Effects: Basic Concepts

◼ In the vast majority of structural analysis, it is assumed that materials remain elastic and that
displacements are so small that their influence on equilibrium is negligible.

◼ Where displacements become large, their influence on equilibrium can become important. The
degree to which the effect becomes important depends on the structural configuration, the
magnitude, direction, and sequence of loading, the stiffness of the elements, the slenderness of
the elements, whether the elements are in tension or compression, and other factors.

◼ The influence of the displacements on structural response are generally referred to as second-
order effects. This type of behavior is nonlinear because the displacements which are being
determined are dependent on the displacements in the system.

◼ The nonlinearity is referred to as a geometric nonlinearity because it is due to the change in

geometry of the structure as it is loaded.

10
Solution Methods for Capturing Second Order Effects
There are two methods for capturing second-order effects in structural analysis.

▪ Large Displacement Methods where the equilibrium is formed in the deformed configuration. In this
method the loading must be applied gradually in several steps. At each step, the system stiffness
is updated based on the geometry determined at the end of the previous step and iteration is
required to establish equilibrium. This method is required for some systems (e.g., cable net structures)
but is rarely needed for typical building structures. The method is theoretically exact.
▪ Geometric Stiffness based approaches. In this method equilibrium is satisfied in the original configuration
and the influence of changes in displacement are accounted for by use of geometric stiffness. Iteration
is required in most cases but can be avoided in certain circumstances. The method is approximate
but of sufficient accuracy for building type structures (e.g., moment frames and braced frames) when
the structural material remains elastic.

(K + K g ,i )D i = Fi i is load step number

Elastic stiffness Geometric Stiffness

(Constant for linear materials) (Depends on current element
forces and hence current
displacement)
11
 Large Displacement Analysis vs Geometric Stiffness Analysis
DF DF P
P
V V

Note that the top node

moves in the X and Y direction Note that the top node
due to change in geometry moves only in the X direction
Y
due to change in geometry

Where large displacement analysis is used Where only geometric stiffness is considered
equilibrium is satisfied in the deformed equilibrium is satisfied in the undeformed
configuration configuration

These differences are not important for framed structures when

displacements are less than about 5% of the element length
12
In Some Cases, Large Displacement Analysis is Essential

qU

P1 For cable type systems the

change in geometry produces
Y P2 significant changes in the
X stiffness of the structure.
Here, large displacement analysis
is essential.
qU
qD

Note: Geometric stiffness is often used in

large displacement analysis. The main benefit is
P1 to speed up convergence (fewer load steps
P2 and iterations)
13
The P-Delta Effect
◼ For building structures, the second order effect that is most important to capture is the
P-Delta (P-D) effect. D0 and DF are displacements without and with P-Delta effects, respectively.

P D0 P DF P
V V V

h, E, I D0 =Vh3/3EI DF =lD0

System to be analyzed First Order Analysis: Second Order Analysis:

Overturning moment Overturning moment PDF is included.
PD0 is ignored 14
The P-Delta Effect
◼ The magnitude of the P-D effect depends on how close P is to the elastic buckling load pcr for the column.
The lateral displacement DF approaches infinity as P approaches pcr (but the column would yield before that
happens)

DF P DF P DF P
V V V

P=0.01 pcr P=0.2 pcr P=0.5 pcr

15
Simplified Approximate Approach For Forming equilibrium Forming equilibrium
by summing moments by summing moments
Fixed-Base Column in the undeformed in the deformed
configuration configuration

P P D0 P DF P

V V V V
ks=3EI/h3

h, E, I Assumed linear
deflected shape

System to be analyzed Equivalent System Vh-ksD0h=0 Vh+PDF -ksDFh=0

The elastic stiffness of the column is replaced by ksD0=V (ks-P/h)DF=V
a spring at the top of the column D0=V/ks DF=V/(ks-P/h)
16
 Clarification of Sign Convention for P
Forming equilibrium Forming equilibrium
by summing moments about by summing moments
the base in the undeformed in the deformed
configuration configuration

D0 P P
DF
Instead of considering P as an
V V applied load, it convenient to replace this
ks ks with the axial force in the column, p.
Vh + PD F - ks D 0 h = 0 Where the column is in compression
the value of p is negative.
(ks - P / h)D F = V
Vh - ks D 0 h = 0
h P The quantity p/h is referred to as the
ks D 0 = V D F = V / (ks - ) geometric stiffness, kg, of the column and
h
is negative when the column is in
D 0 = V / ks compression.
p
D F = V / (ks + )
h

17
 Finding DF in terms of D0 D0 D0
DF = = = lD 0
pD 0 kg
1+ 1+
D0 P
DF
P Vh ks
V V The value l is the displacement
ks ks amplification factor.
Vh + PD F - ks D 0 h = 0
(ks - P / h)D F = V The value pD0/Vh (or kg/ks) is often
Vh - ks D 0 h = 0 referred to as the stability ratio, q.
h P
ks D 0 = V D F = V / (ks - ) This gives
h
D0
D 0 = V / ks DF =
D F = V / (ks +
p
) 1+ q
h

In this formulation q is negative

if the column is in compression.

18
Calculating the Buckling Load

DF P The system buckles when the term ks+p/h is zero.

V This gives P = pcr = ksh = 3EI/h2

Vh + PD F - ks D 0 h = 0
True buckling load for cantilever column is p2EI/(2h)2 = 2.46EI/h2
(ks - P / h)D F = V
The ratio of approximate pcr to true pcr = 3.00/2.46=1.22
P
D F = V / (ks - )
h
The buckling load for the approximate solution is too
p large because the true deflected shape of the column
D F = V / (ks + )
h (assumed as a straight line) is not accounted for in developing
the value of kg =p/h. The true deflected shape is a cubic function.

Not accounting for the true deflected shape artificially

stiffens the system.

19
Calculating the Buckling Load
If the deflected shape of the column is assumed to be
cubic, it can be shown (McGuire et al. 2000) that the geometric
stiffness of the column is 1.2p/h.
DF P
The system buckles when the term ks+1.2p/h is zero.

V This gives P = pcr = khs/1.2 = 2.5EI/h2

True buckling load for cantilever column is p2EI/(2h)2 = 2.46EI/h2

The ratio of approximate pcr to true pcr = 2.50/2.46=1.016

d
The buckling load for the approximate solution is still too
large (marginally) because the true deflected shape of the column
under loads V and P is not exactly cubic (it is transcendental).

The improvement which is due to using a curved deflected shape

is due to the so-called p-d effect (with a lower-case p and delta).

The geometric stiffness of 1.2p/h is referred to as the consistent

geometric stiffness because it is consistent with the cubic shape
of the column under lateral load V (without direct consideration of P).
20
 Expanding Concepts to a Column Modeled with Several Elements
P Each element’s stiffness is composed of two parts:
5
V 1. The elastic stiffness which depends on E, I, A, and L. For
the column shown L=h/4.
4

4 2. The geometric stiffness which depends on p and L. For the

column shown p is the same for all elements and is equal in
3 magnitude to the applied load P. The value of p is negative
because the column is in compression.
h 3

2
There are two forms of the geometric stiffness; linearized
All segment have and consistent. For this structure, the geometric stiffness can be
2 the same E, I, and A. formed prior to analysis because p is known for each element
Segment lengths are and does not depend computed deflections (equilibrium is formed
1 relative to the undeformed configuration).
equal.
Y
1

21
Elastic Element Stiffness
This is the element stiffness matrix used for analyzing planar frame.

é AE / L 0 0 - AE / L 0 0 ù
ê ú
ê 0 12EI / L 3
6EI / L2 0 -12EI / L 6EI / L2
3
ú
ê 0 6EI / L2 4EI / L 0 -6EI / L2 2EI / L ú
ke = ê ú
ê - AE / L 0 0 AE / L 0 0 ú
ê 0 -12EI / L3 -6EI / L2 0 12EI / L3 -6EI / L2 ú
ê ú
êë 0 6EI / L2 2EI / L 0 -6EI / L2 4EI / L úû

y
2 5
3 6
x
1 4
Y i j

X
Z The version shown does not include shear
deformation but this could be easily incorporated.
22
Element Geometric Stiffness

Linearized, based on straight line deformed shape Consistent, based on cubic deformed shape

é 1 0 0 -1 0 0 ù é 1 0 0 -1 0 0 ù
ê ú ê ú
0 1 0 0 -1 0 ú ê 0 6/5 L / 10 0 -6 / 5 L / 10 ú
ê
ê 0 0 0 0 0 0 ú ê 0 L / 10 2L2 / 15 0 - L / 10 - L2 / 30 ú
kgL = p / L ê kgC = p / Lê ú
ê -1 0 0 1 0 0 úú ê -1 0 0 1 0 0 ú
ê 0 -1 0 0 1 0 ú ê 0 -6 / 5 - L / 10 0 6/5 - L / 10 ú
ê 0 0 0 0 0 0 úû ê 0 L / 10 - L2 / 30 0 - L / 10 2L2 / 15 úû
ë ë

y 5
2
3 6
x
Y 1 4
i j
X
Z Axial force in column is p Note: Diagonal terms are negative when the
(compression negative) element is in compression

23
Consistent Geometric Stiffness for a 3-D Frame Element

Note that this matrix depends

on several element forces, (not just
axial force).

McGuire et al, 2000 24

Solving Structural Systems
◼ Complete structural systems composed of frame elements can be solved by use of
the following equilibrium equations

(Ke + K g )D = F
where Ke is the elastic stiffness matrix and Kg is the geometric stiffness matrix of the structure.

◼ If the element forces needed to form Kg do not depend on the computed displacements,
the system can be solved without iteration. There are special situations (explained in the next
scene) where this is the case.

◼ An iterative solution is required if the element forces that are needed to form Kg depend on the
computed displacements the system. This is the general case for 3-D systems.

◼ For systems with shell elements the same approach is used wherein the geometric stiffness
is specifically formulated for the shape and properties of the shell element is used.

25
Next Scene: Analyzing 2-D Frames

26
27
2D and 3D P-Delta Effects
PL PG PG PL

PL PG PG PL

29
 Modeling to Include P-D and p-d Effects

◼ The most appropriate way to include P-D and p-d effects is to form the geometric stiffness for
each element and assemble these into the geometric stiffness matrix Kg in the same manner
that the elastic stiffness Ke is formed.

◼ It is also possible to use a “leaning column”, where the P-D effects are separated out into a
fictitious “leaning column” that utilizes linearized geometric stiffness. p-d effects are not
represented with the leaning column.

◼ The leaning column approach should not be used in 3D analysis as it is impossible to accurately
represent spatial distribution of gravity load using a single leaning column.

◼ The use of a leaning column is demonstrated in this scene because it was a common solution in
the past for analysis of planar structures.

30
Analyzing a 2D FRAME (without a leaning column)
0.5P ◼ 100% of the gravity load is resisted by the same
2V elements that resist the lateral load

◼ Each element has an elastic stiffness and a

geometric stiffness
h2
P
◼ Either a linearized or a consistent geometric
V stiffness can be used, but consistent is more
accurate

◼ Where linearized geometric stiffness is used

the analysis can be performed without iteration
h1 Beams: EIb
because the system geometric stiffness Kg
Columns: EIc
depends only on the axial forces in the columns
under gravity load alone

◼ Iteration is required when the geometric stiffness,

which depends on element axial forces, depends
L
on the displacements that are being computed
31
Using a “Leaning Column” ◼ Geometric stiffness is not included
0.5P 0.5P with the frame elements. Instead,
Rigid the geometric stiffness for the
2V Link entire story is assigned to a
“leaning column” that has
“Leaning no elastic lateral resistance.
h2 column” with
P P axial force p=0.5P ◼ Note that the story gravity load
Rigid on the actual frame is replicated in
V
Link
Geometric stiffness the leaning column
= -0.5P/h2
◼ The axial force in the leaning
“Leaning column must be known prior to
Beams: EIb
h1 column” with
Columns: EIc lateral load analysis.
axial force p=1.5P
◼ This force is equal to the total
Geometric stiffness
= -1.5P/h1 gravity force in the story and can be
obtained without iteration

L
32
Using a “Leaning Column” with Inertia
0.5P Node can translate
Rigid horizontally but
2V Link cannot rotate ◼ This approach could
be used where the
software does not
Lateral stiffness =12EI2/h23 have geometric
h2 0.5P
P Geometric stiffness = -0.5P/h2 stiffness
Rigid
V
Link 12EI2/h23 = -0.5P/h2 ◼ It requires the ability to
Use I2 = - 0.5Ph22/12E enter negative
moment of inertia
Beams: EIb (some programs
h1 1.5P Lateral stiffness =12EI1/h13 will not allow this).
Columns: EIc
Geometric stiffness = -1.5P/h1
12EI1/h13 = -1.5P/h1
Use I1 = -1.5Ph12/12E

L
33
Analyzing a 2D System with Gravity-Only Framing (without a leaning
column)
0.5P 0.25P
◼ The gravity-only system is
2V
included in the model, but
provides no lateral load
resistance
h2
P 0.5P ◼ P-D and p-d effects are
V included directly for each
lateral and gravity-only element
by use of geometric stiffness

Beams: EIb Gravity only ◼ Iteration may be required to

h1 Columns: EIc framing obtain the appropriate p-load in
each column

◼ A leaning column is not

needed

L
34
Using a “Leaning Column” where there is Gravity-Only Framing (2D Analysis)
0.5P 0.25P 0.25P

2V Rigid ◼ The lateral system is

Link
modeled with geometric
stiffness
“Leaning
h2 column” with ◼ The gravity-only system
P 0.5P 0.5P axial force 0.25P is not included in the
V Rigid
Link model

◼ The destabilizing effect

Gravity only “Leaning of the gravity-only
Beams: EIb
h1 framing column” with framing must be
Columns: EIc
axial force 0.75P included in the p-load of
the leaning columns

L
35
Using a “Leaning Column” where there is Gravity-Only Framing (2D Analysis)
0.5P 0.25P 0.75P

2V Rigid ◼ The lateral system is

Link
modeled without
geometric stiffness
“Leaning
h2 column” with ◼ The gravity-only system
P 0.5P 1.5P axial force 0.75P is not included in the
V Rigid
model
Link

◼ The destabilizing effect

“Leaning of the lateral framing
Beams: EIb Gravity only
column” with and the gravity-only
h1 Columns: EIc framing
axial force 2.25P framing must be
included in the p-load of
the leaning columns

L
36
Analysis of System with Lateral Load Resisting Frames and
Gravity-Only Frames
2V 2V
V V

Lateral Column
Story P-Load
PL = wBL/6

Gravity
Gravity

Lateral
Rigid
Lateral

L
Diaphragm
Gravity Column
Story P-Load
PG = wBL/3

B
Plan View of typical floor. Gravity uniform load = w (ksf) 37
Using a “Leaning Column” where there is gravity only-framing (2D Analysis)
PL PL 2PL + 2PG
Rigid
Link
6V
“Leaning
column” with
axial force 2PL+2PG
h2
PL PL
2PL + 2PG
“Leaning
6V column” with
Rigid
Link axial force 4PL+4PG

Beams: EIb Beams: EIb

h1 Columns: EIc Columns: EIc In this example the gravity
framing is not included in the
model, but the influence of
the potentially destabilizing
gravity forces are included
by use of the leaning column.

L L
38
The P-Q Effect The P-Q effect captures increases in rotations about the vertical axis due
to gravity loading in the structure.

b Rqf
Torque, T

a R
weight, w
Circular tube
columns

h
qf
EI

E,I,G,J

Column axial force p=abw/4 (compression negative)

Column lateral stiffness = 12EI/h3
Geometric stiffness = p/h
Column torsional stiffness = GJ/h
ép ù
é 12EI 2 GJ ù Structural geometric stiffness = K g = 4 ê R2 ú
Structural torsional stiffness = K e = 4 ê 3 R + ëh û
ë h h úû
q0 q0
q0=T/Ke f = =
K pR 2q 0
1+ G 1+
KE Th
39
Capturing the P-Q Effect in Buildings with Complex Geometry

◼ For more complex structures, the lateral geometric

stiffness and the torsional geometric stiffness are
coupled (as is the elastic stiffness) and cannot be
separated out into single (or multiple) leaning
column(s).

◼ Thus, each column that is part of the lateral system

or of the gravity system must be explicitly
modeled with with sufficient detail to capture the
p-load.

◼ The p-loads cannot generally be determined

without iteration, even if linearized geometric
stiffness is utilized.

40
Use of Leaning Columns in 3D Analysis
PL PG PG PL
◼ Leaning columns
should not be used in
3D analysis because
they cannot accurately
capture torsional
response and the
PL PG PG PL associated P-Q
behavior. Instead, the
geometric stiffness
should be included
with each column

◼ An iterative analysis is
likely be needed
because the column
axial forces cannot be
formed prior to lateral
load analysis

41
Use of Consistent Geometric Stiffness in Analysis

◼ Given that leaning columns should not be used for 3D analysis, and given that iteration will
be required even when linearized geometric stiffness is used, there is no significant
computational penalty for using consistent geometric stiffness.

◼ Hence, it is recommended that consistent geometric stiffness be used for all analysis that is
intended to capture P-D and P-Q effects.

◼ Programs like SAP2000 and ETABS use consistent geometric stiffness by default.

◼ If it is necessary to capture also p-d effects all columns that are intended to resist lateral
load should be modeled with at least two segments. This is also the case if critical buckling
loads are to be determined.

◼ In SAP2000 and ETABS meshing the element allows the element to be subdivided by
meshing the element without physically modeling the element in two (or more) pieces. The
advantage of this is that the design parameters do not need to be adjusted (i.e., 2L instead of L).

42
 Next Scene: Solving Stability Problems

lP1 l
V
lcr ... .
lP2 h
..
V
.
h
. Result for increment of l

Dcontrol
V=0.001P1/h Deflected shape as l
approaches lcr

43
PL PG PG PL

PL PG PG PL

44
Modeling Structures to Include P-Delta Effects
 lP1 l
V
lcr ... .
lP2 h
..
V
.
h
. Result for increment of l

Dcontrol
V=0.001P1/h Deflected shape as l
approaches lcr

46
 Solving Stability Problems

◼ The formulations including geometric stiffness are primarily used to include second order effects
in the structural analysis. Inclusion of such effects is required by building codes in some
circumstances.

◼ The same formulations can be used to determine the elastic buckling behavior of structures.
While buckling analysis may not be required as part of the design process, it can be used to
provide insight into the system behavior.

◼ Two methods for performing a buckling analysis are presented:

◼ A method where the structure is incrementally loaded until buckling occurs
◼ A method where the buckling loads and buckled shape are determined automatically

◼ Both methods require that geometric stiffness be included. While linearized geometric stiffness
can theoretically be used, it can produce unconservative (too high) estimates of the
buckling load.

47
Finding the Buckling Load by Incremental Application of Gravity Load
◼ The system-level buckling load can be determined by solving the equation below with increasing
values of the scalar multiplier l until the displacement at some designated point in the structure
approaches infinity for a very small constant notional lateral loading F0.

(Ke + l K g )D = F0
◼ As l increases towards the “critical” value of l=lcr the displacement will increase rapidly, which is
indicative on an impending instability

◼ As l approaches lcr the displaced shape of the system (based on displacements D) will
approach the buckled shape of the system

◼ The geometric stiffness matrix Kg is constant and should be based on a realistic distribution of
gravity loading in the structure. Accuracy is improved by using consistent geometric stiffness.

◼ Alternately, the value of lcr and the buckled shape D can be determined by solving the eigenvalue
problem (Ke+lcrKg)D={0}. lcr will be equal to the lowest positive eigenvalue and D is the
eigenvector associated with lcr.
48
Illustration on Incremental Approach

Control lP1 l
point
V
lcr .... .
lP2 h
..
V
.
h
. Result for increment of l

Dcontrol
V=0.001P1/h Deflected shape as l
approaches lcr

49
Finding the System Level Buckling Load by Eigenvalue Analysis

1. Form elastic stiffness Ke for the unloaded system

2. Form geometric stiffness Kg under gravity loading only with l=1.0

3. Solve the eigenvalue problem (Ke+lcrKg)D={0}. lcr will be equal to the lowest
eigenvalue and D is the eigenvector associated with lcr. Higher eigenvalues
represent higher ”modes” of buckling.

• When calculating buckling loads of complex structures it is recommended that

at least six modes be computed. The analyst should plot (and animate) the deflected
shapes to be sure that the buckled shapes are reasonable. In some cases, only a small
portion of the structure may appear to be buckled, which is indicative of an undesired local
instability.
• Note that is is very common for buckling loads be so high that the structure would be
inelastic under these loads. If this is the case the buckling load is not applicable.

50
Example Buckling Analysis of a Simple Column
P=10k

5
V=0.1
The structure is analyzed as follows:
4
• Range of P Values up to Pcr
4
HSS 16x16x5/8 • 1, 2, 4, or 8 segments
3 Braced out of plane • Geometric Stiffness:
h=480” 3
E=30000 ksi - None
I= 1270 in4 - Linearized
2 A= 35 in2
Pcr=p2EI/4h2=408 k - Consistent
2

1 Exact buckling load

Y
This system can be solved without iteration
1 because the element axial forces do not depend
X on the displacements.

51
 Results Using Eigenvalue Analysis Buckled Shape for
all models

Number of Linearized kg Consistent kg

Segments
lcr lcr Note: Analysis was run
1 496 411 with P=10 k.
2 431 409 Tabulated values
4 413 408 are lcr P
8 409 408

▪ The 1-Segment model using linearized kg

overestimates Pcr by factor of 496/408 =1.216
▪ The 1-Segment model using consistent kg
Y
overestimates Pcr by factor of 411/408 =1.007. 1

X
Analysis run using Mathcad frame analysis program developed by F. Charney

52
 Results Using Incremental Approach
1 Segment 2 Segments
500 500
450 450 ◼ 1 and 2-segment models
400 400
350 350
overestimate Pcr when linearized
Kg is used
P-Load, kips

P-Load, kips
300 300
250 250
200 200
150 150 ◼ For 4 or 8 segments linearized
100 100 and consistent Kg produce
50 50
0 0
very similar results
0 200 400 600 800 1000 0 200 400 600 800 1000
Displacement, inches Displacement, inches
Linearized KG Consistent KG
◼ Both approaches accurately
Linearized KG Consistent KG
8 Segments
predict an elastic buckling load of
4 Segments
500 500 408 kips which is virtually
450
450 equal to the theoretical value
400 400
350 P- Load, kips 350
P- Load, kips

300 300
250 250
200 200
150 150
100 100
Exact Pcr=408 kips
50 50
0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
Displacement, inches Displacement, inches
Linearized KG Consistent KG Linearized KG Consistent KG

53
 Planar Frame Example

0.5P ◼ Analysis run using linearized

P and consistent geometric stiffness
P
◼ Buckling load determined using
P
25’ 0.5P incremental and eigenvalue
approach
V=0.01P 90

80

20’ 70
Columns: W21x122 60

Load Factor Lambda

Rafters: W27x146 50

40

30
60’ 30’ 20

10
Analysis performed for P=10 k
0
lcr=80 for Linearized Kg 0 10 20 30 40 50

lcr=78 for Consistent Kg

Displacement, inches
Linearized KG Consistent KG
54
Notes Regarding Previous Example

◼ Analysis was performed using Mathcad routines without shear deformation. Including shear
deformation will reduce the buckling loads slightly.

◼ The frame is braced out of plane

◼ Members are oriented with the web in the plane of structure

◼ Beams and columns are subdivided as shown

◼ Similar results were obtained using SAP 2000

55
Next Scene: Codes and Standards

56
 lP1 l
V
lcr ... .
lP2 h
..
V
.
h
. Result for increment of l

Dcontrol
V=0.001P1/h Deflected shape as l
approaches lcr

57
Conforming with Requirements within ACI
318
Building Code Requirements for Second Order Analysis

59
 Building Code Requirements for Second Order
Analysis

◼ Prior to the development of modern structural analysis software Effective Length

second order effects were ignored, or were included by use of Alignment Chart
effective length factors determined by hand using alignment
charts.

◼ As presented in the previous scenes it is currently easy to

directly incorporate second order effects into analysis.
Building codes such as ACI 318 and AISC 360 allow second
order analysis, but have numerous requirements for carrying
out the analysis.

◼ Further, ASCE 7 has requirements for incorporation of P-Delta

effects into seismic analysis. (There are no similar
requirements analysis for wind analysis.)

◼ Some of the more important aspects of code requirements for

performing second order analysis are presented in this scene.
60
Conditions Where Second-Order Analysis is Required (Seismic Design)
◼ ASCE 7-22 has requirements for including P-Delta effects is seismic analysis in Section 12.8.7.
Such effects are required to be included where the stability index q of any story exceeds 0.1,
and does not exceed qmax:

Notes:
1. Analysis requirements for computing story drift
D are provided in Section 12.7.3.
2. The analysis used to determine story drifts
D does not include second-order effects.
3. Story gravity loads Px are computed with load
factors of 1.0.
4. Story drifts D should be computed at the edge
of the building if the system has a torsional
irregularity.
5. Final drifts including second-order analysis are
taken as D (1/1-q)
61
Recovering Seismic Stability Index from Second-Order Analysis
◼ Where second-order effects are directly included in the analysis it must be shown that q
does not exceed qmax for any story. In no case shall q exceed 0.25.

◼ The stability indices may be determined by performing the analysis without second-order
analysis, and again with such effects included. The stability index for any story is computed as
follows:
D0
q = 1-
DF
where D0 is the story drift computed without second-order effects, and DF is the story drift at the same location
with second-order effects included.

In applying this approach, it is essential that both P-D and P-Q effects are included.

Note: This method is not explicitly allowed by ASCE 7 but is theoretically correct.

62
Conditions Where Second-Order Analysis is Required (Wind Design)
◼ ASCE 7 does not have specific requirements for including second-order effects in analysis
for wind loads.

◼ For flexible structures (fundamental frequency < 1.0 Hz or period > 1.0 sec) the gust factor,
Gf, depends on the computed fundamental frequency, h1, of the structure.

◼ While the influence of the frequency on analysis due to second-order analysis is small,
it is recommended that such effects be included in the analysis, if nothing else to provide an idea
of how important second-order effects might be in the response.

63
 Overview of Second Order Analysis Requirements
for Structural Steel and Reinforced Concrete

◼ Therequirements for second order analysis in ACI 318 and

AISC 360 are common with regards to the following:
◼ Analysis is performed for the system at or near ultimate loading
◼ Member properties must reflect cracking and other effects that
reduce stiffness
◼ Both P-D and p-d effects must be included. Although not explicitly stated, this
can be best accomplished by use of consistent geometric stiffness and
subdividing columns and other compression elements into at least two
segments along the length.
◼ For steel structures, the influence of possible out-of-plumb construction must
be included either directly or by use of using notional loads
64
Second Order Effects in Reinforced Concrete Structures
◼ The analysis shall consider the influence of axial loads, presence of cracked regions along the
length of the member, and the effects of load duration (for certain loading). While not
specifically noted, shear deformations, the effect of shear cracking, and beam-column joint
deformations should also be included.

◼ Section properties may be based on ACI 318 Section 6.6.3.1:

65
Second Order Effects in Reinforced Concrete Structures (2)
◼ Slenderness effects along the length of columns (or other compression elements) shall be
considered. This can be accomplished using the requirements of ACI 318 Section 6.6.4.5.

◼ The commentary states that the slenderness requirements can be satisfied by subdividing
members.

◼ Where members are subdivided it is recommended that not less than 4 segments be used
where linearized geometric stiffness is used, and not less than two segments where consistent
geometric stiffness is used.

◼ Increases in design moments due to second order effects shall not exceed a factor of 1.4
(Section 6.2.5.3)

66
Second Order Effects in Steel Structures
◼ In structural steel systems, second order analysis is used within the Direct
Analysis Method (Section C1.1).

◼ Limited 1st order analysis and the Amplified First Order Method are allowed
with several limitations. However, the Direct Analysis Method is appropriate
and can be used for any structure.

◼ The analysis shall consider flexural, shear and axial member deformations,
and other component and connection deformations that contribute to
the displacements of the structure.

◼ The analysis shall incorporate reductions in all stiffnesses that are considered to contribute to the
stability of the structure, as specified in Section C2.3.

◼ The analysis shall be a second-order analysis that considers both P-Δ and p-δ effects. It is
permissible to neglect the effect of p-δ on the response of the structure when certain conditions
are met. p-δ effects must be included in all elements that are in compression and flexure.

Reference: Section C2 of AISC 360-22 67

Second Order Effects in Steel Structures (2)
◼ The analysis shall consider all gravity and other applied loads that may influence the stability of
the structure. Where applicable, leaning columns can be used to include gravity forces that are
associated with gravity-only framing. Use of a full 3D model that includes all gravity resisting
elements is preferred in all cases.

◼ For design by LRFD, the second-order analysis shall be carried out under LRFD load
combinations. For design by ASD, the second-order analysis shall be carried out under 1.6 times
the ASD load combinations, and the results shall be divided by 1.6 to obtain the required
strengths of components.

◼ The effect of initial imperfections in the position of points of intersection of members on the
stability of the structure shall be considered either by direct modeling of these imperfections
in the analysis or by the application of notional loads.

Reference: Section C2 of AISC 360-22 68

Direct Modeling of Imperfections in Steel Structures
◼ Imperfections that need to be considered are related to uncertainty in the exact position of nodal
connection points (due, for example, to construction tolerances) and not in member out-of-
straightness.

◼ The magnitude of the initial displacements shall be the maximum amount considered in the
design; the pattern of initial displacements shall be such that it provides the greatest destabilizing
effect.

◼ Initial displacements similar in configuration to both displacements due to loading and anticipated
buckling modes should be considered in the modeling of imperfections.

◼ The magnitude of the initial displacements should be based on permissible construction

tolerances, as specified in the AISC Code of Standard Practice or other governing requirements,
or on actual imperfections if known.

Reference: Section C2 of AISC 360-22 69

Conceptual Basis of Notional Loads
P
P
Pe/h
Pe/h
P P P
-Pe/h
Pe/h
2Pe/h
P 2P P
-2Pe/h
Pe/h
3eP/h
h P 3P P
-3Pe/h
Pe/h
4Pe/h P
e P 4P
-4Pe/h
Pe/h
5Pe/h P
P 5P
-5Pe/h
Pe/h
6Pe/h
6P
-6Pe/h
6Pe/h

Out of Plumb Structure and Story Story Shears Due to Analysis Structure With
Gravity Forces Out of Plumb Columns and Gravity Load Gravity and
Notional Loads 70
AISC 360 Requirements for Notional Loads
From Section C.2:

71
Adjustments to Stiffness for Steel Structures
◼ All elements must have a basic stiffness reduction factor of 0.8 associated with all sources of
deformation (axial, shear, bending, torsion)

◼ An additional reduction tb shall be applied to flexural properties that contribute to the stability of
the structure

Note: The requirements

presented in this scene for
the Direct Analysis method
are intended as a brief
summary. Refer to Section
C, and Appendix 1 of
AISC 360-22 for details.

72
 Summary of Code Provisions for Second-Order
Analysis

◼ Basic guidelines for performing second order analysis within the context of the requirements
of ASCE 7-22, ACI 318-19, and AISC 360-22 are presented.

◼ Where second-order analysis is performed in accordance with the above provisions it is

recommended that the concepts of top-down full-system analysis are strictly adhered to.

◼ For analysis in accordance with ACI 318 and AISC 360, consistent geometric stiffness should be
used where available, with at least two segments for all elements in combined axial force and
flexure.

◼ For members with very high axial load, or for very slender elements, the use of 4 segments per
element is recommended

73
Next Scene: Overview of Second Order and Stability Analysis
Procedures in SAP 2000

Mode 1
Mode 2
lcr=34.16
lcr=89.04

Mode 3 Mode 4
lcr=99.65 lcr=103.3

74
Building Code Requirements for Second Order Analysis

75
Conforming with Requirements in AISC 360
Mode 1
Mode 2
lcr=34.16
lcr=89.04

Mode 3 Mode 4
lcr=99.65 lcr=103.3

77
 Second Order and Buckling Analysis in
SAP 2000

◼ This scene covers second-order and buckling analysis capabilities in SAP 2000, Version 22.
Similar capabilities are available in other programs (e.g. ETABS, LARSA)

◼ SAP has the capability to include second-order effects in structures with linear and nonlinear
materials. This scene will cover only systems with linear elastic materials.

◼ The following analysis procedures are available:

1. P-Delta analysis wherein equilibrium is satisfied in the undeformed configuration, and second-order
effects are included by use of geometric stiffness. This approach would be used for typical building
structures.
2. Large displacement analysis wherein equilibrium is satisfied in the deformed configuration. Geometric
stiffness is used to accelerate convergence. This approach would be used for cable net systems.
3. Buckling analysis. This is basically a sub-set of P-Delta analysis because geometric stiffness associated
with gravity loads is used together with the system’s elastic stiffness to determine the buckling loads and
the buckled shapes

◼ This scene concentrates on items (1) and (3) above

78
Formulation of Element Geometric Stiffness

◼ For frame elements the geometric stiffness is based on a cubic displaced shape. This is known
as consistent geometric stiffness. Linearized geometric stiffness is not directly available but can
be simulated by releasing the moments at each end of the member.

◼ A similar consistent formulation is utilized for shell elements. Plate-bending elements would have
no geometric stiffness because the element has no in-plane forces.

◼ Where it is required to include both P-D and p-d effects, it is recommended that all compression
elements be divided into several segments along their length. It is generally not necessary to
subdivide beam elements unless they are under compression.

◼ Shell elements under considerable compression should also be subdivided. The typical 4 by 4
mesh used for modeling shear walls would be sufficient.

79
 Example (Introduction)
.75P
3V
▪ The 2D frame shown to the left will be
150”

1.5P analyzed using SAP 2000.

5V ▪ This analysis is used to demonstrate basic
150”

concepts. A more realistic 3D system is

.5P analyzed in the next scene.
1.5P
4V ▪ Analyses performed include:
▪ Buckling analysis
150”

P P ▪ Iterative Solution
1.5P
3V
▪ Eigenvalue Solution
▪ Lateral analysis without P-D effects
150”

P 2P ▪ Lateral load analysis with P-D effects

1.5P
2V
180”

Note: the gravity loads and wind loads

on this system are tributary to the system
(using a 30 ft bay width). The system will
be reanalyzed with a leaning column to
represent non-tributary gravity and wind loads.
240” 360” 300” 80
 Example (Features)
.75P P=100k
3V Analysis Features:
V=20k
150”

1. Beams and columns subdivided into

1.5P two segments.
Column:
5V W14x145 2. Consistent geometric stiffness is
150”

.5P used
Beam:
1.5P 3. No modifications to element
W27x114
4V stiffness
150”

P 4. Analyze system for gravity

P
1.5P and lateral loads including
3V P-Delta effects
150”

P 2P Column:
1.5P W14x193
2V
Beam:
180”

W27x146

240” 360” 300” 81

Analysis Setup and Results for Buckling Load Analysis by eigenvalues

Mode 1
Mode 2
lcr=34.16
lcr=89.04
Buckling Analysis

Mode 3 Mode 4
lcr=99.65 lcr=103.3

Load Case for

setting Geometric
Stiffness

82
 Analysis Setup and Results for Buckling Load Analysis by iteration
Static Nonlinear Analysis
with P-Delta lcr=34.16 from eigenvalue analysis
40

35

Roof Displacement, inches

30

25

20

15

10

0
-10 0 10 30 50 70 90 110 130 150
Dead Load Factor

Using Dead and Wind Using Dead and Notional

Scaled factor for dead load increased
from 1.0 to lcr until buckling is apparent Lateral load Lateral load
is 1.0 times wind is 1/1000 wind

83
 P-Delta Analysis under Lateral Loads
Static Nonlinear Analysis Deflected Shape
with P-Delta

Full dead load followed by wind

84
Results for Lateral Load Analysis
Story displacements and drift without
Analysis run without P-Delta effects P-Delta
SP SV h Total Drift Story Drift D0 q=PD0/Vh
Level (k) (k) (in) (in) (in)
R 75 60 150 4.788 0.624 0.0052
5 225 160 150 4.164 1.088 0.0102
4 425 240 150 3.076 0.869 0.0103
3 775 300 150 2.207 0.750 0.0129
2 1225 340 180 1.457 1.457 0.0292

Analysis run with P-Delta effects

SP SV H Total Drift Story Drift DF q=1-D0/DF
Level (k) (k) (in) (in) (in)
R 75 60 150 4.871 0.629 0.0079
5 225 160 150 4.242 1.099 0.0100
4 425 240 150 3.143 0.879 0.0114
3 775 300 150 2.264 0.762 0.0157
2 1225 340 180 1.502 1.502 0.0300
85
Results for Lateral Load Analysis
Analysis run without P-Delta effects
SP SV h Total Drift Story Drift D0 q=PD0/Vh
Level (k) (k) (in) (in) (in)
R 75 60 150 4.788 0.624 0.0052
5 225 160 150 4.164 1.088 0.0102
4 425 240 150 3.076 0.869 0.0103
3 775 300 150 2.207 0.750 0.0129
2 1225 340 180 1.457 1.457 0.0292

Analysis run with P-Delta effects

SP SV H Total Drift Story Drift DF q=1-D0/DF
Level (k) (k) (in) (in) (in)
R 75 60 150 4.871 0.629 0.0079
5 225 160 150 4.242 1.099 0.0100
4 425 240 150 3.143 0.879 0.0114
3 775 300 150 2.264 0.762 0.0157
2 1225 340 180 1.502 1.502 0.0300

Increase in Displacement and Drift due to P-Delta 86

Results for Lateral Load Analysis
Analysis run without P-Delta effects
SP SV h Total Drift Story Drift D0 q=PD0/Vh
Level (k) (k) (in) (in) (in)
R 75 60 150 4.788 0.624 0.0052
5 225 160 150 4.164 1.088 0.0102
4 425 240 150 3.076 0.869 0.0103
3 775 300 150 2.207 0.750 0.0129
2 1225 340 180 1.457 1.457 0.0292

Analysis run with P-Delta effects

SP SV h Total Drift Story Drift DF q=1-D0/DF
Level (k) (k) (in) (in) (in)
R 75 60 150 4.871 0.629 0.0079
5 225 160 150 4.242 1.099 0.0100
4 425 240 150 3.143 0.879 0.0114
3 775 300 150 2.264 0.762 0.0157
2 1225 340 180 1.502 1.502 0.0300

Stability factors computed from first-order and second-order analysis are similar 87
Next Scene: 3D Example in SAP 2000

88
Mode 1
Mode 2
lcr=34.16
lcr=89.04

Mode 3 Mode 4
lcr=99.65 lcr=103.3

89
Other Elastic Stability Considerations
91
 3D Frame Example for P-D and Stability

◼ 6-Story Steel Braced Frame under seismic loads (ASCE 7-16)

◼ Lateral load applied with and without eccentricity

◼ Analyses to be performed
◼ Compute critical buckling loads
◼ Determine Stability Coefficients, q, by hand and using second order analysis
◼ Compute lateral displacements with and without P-Delta, and with and without torsional loads

◼ Modeling Details
◼ Fully 3D model with accurate distribution of gravity loads
◼ Semirigid diaphragms modeled with membrane elements, 2-by-2 mesh per bay
◼ All columns except braced frame columns resist gravity load only
◼ Columns and diagonals of the braced frames are subdivided into two elements per story
◼ All columns are pinned based
◼ Analyzed using SAP 2000 V22
92
Overview of Building to be Analyzed

A B C D E F G

I I 1 Frame 1

I
I

I
Frame 1
I I I I I I I 2
3@20’=60’

5@10’=50’
Frame 1

Frame 1
I I I I I I I 3
Frame 1

e
I I 4
I
I

I
Y B=6@20’=120’
Z
X
Note Direction and Eccentricity of Loading X

93
Loads Acting on Model

Lateral Loads:
Seismic Design Category D Ie=1.0
Gravity Loads: SDS=0.75 SD1=0.3
wd=110 psf wl=20 psf Cladding=500 plf Special Steel Braced Frame, R=6, Cd=5
T=1.0s W=5796k V=290k k=1.25
Column Gravity Load on Column (kips)
Dead Live Total Level F (k) T (in-k)
(unfactored)
R 87.4 6290
Roof: Corner 19 2 21
5 74.1 5332
Roof: Edge 30 4 34
4 56.0 4034
Roof: Interior 44 8 52
3 39.1 2815
Floor: Corner 21 4 25
2 23.6 1696
Floor: Edge 32 8 40
1 9.9 713
Floor: Interior 44 16 60
Building weight=11.1 pcf

94
Member Sizes (Steel is 50 ksi)
◼ Gravity Columns ◼ Braced Frame Columns
◼ Stories 1 and 2 ◼ Stories 1 and 2 W14x132
◼ Corner W14x61 ◼ Stories 3 and 4 W14x120
◼ Edge W14x74 ◼ Stories 5 and 6 W14x109
◼ Interior W14x90
◼ Stories 3 and 4 ◼ Braced Frame Diagonals
◼ Corner W14x48 ◼ Stories 1 and 2 HSS5x5x3/8
◼ Edge W14x62 ◼ Stories 3 and 4 HSS5x5x5/16
◼ Interior W14x74 ◼ Stories 5 and 6 HSS5x5x1/4
◼ Stories 5 and 6
◼ Corner W14x48 ◼ Braced Frame Beams
◼ Edge W14x48 ◼ Levels R and 6 W21x93
◼ Interior W14x48 ◼ Levels 5 and 4 W21x93
◼ Levels 3 and 2 W21x93
◼ Gravity Floor Beams
◼ All locations W24x76
95
SAP2000 Model

96
Buckling Loads and Shape

lcr,1 = 13.93 (Primarily Torsional Mode) lcr,2 = 19.04 (Sway-Torsional Mode)

• l Values shown are multipliers on dead load. The structure would be
inelastic before these loads occurred.
• Modes 3-6 are localized buckling in braces
97
Computed Lateral Displacements with and without P-D
1000
◼ Drifts reported at edge of the
D No Acc. 0.05B Torsion 0.12B Torsion
900 Torsion Eccentricity Eccentricity building
+
800
◼ Analysis run without accidental
700
and with 0.05 and 0.05Ax
eccentricity
Height, inches

600

◼ P-Delta effects significantly

500
increase lateral displacements for
400 each torsion case
300
◼ Percent increase of drift due
200 to P-Delta effects are
approximately the same for each
100
Dotted Line: no P-D
Solid line: with P-D torsion case
0
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50
Lateral Displacement, inches

98
Effect of Torsional Eccentricity on Stability Coefficient
Computed Using ASCE 7: Computed from analysis with and without PD:
Torsional Eccentricity Story Torsional Eccentricity
Story None 0.05B 0.05AxB None 0.05B 0.05AxB
6 0.0292 0.0372 0.0459 6 0.0423 0.0136 0.0598
5 0.0402 0.0501 0.0637 5 0.0610 0.0604 0.0622
4 0.0517 0.0641 0.0814 4 0.0625 0.0700 0.0585
3 0.0592 0.0733 0.0930 3 0.0683 0.0670 0.0710
2 0.0602 0.0752 0.0960 2 0.0746 0.0729 0.0718
1 0.0588 0.0735 0.0940 1 0.0698 0.0714 0.0748
q=PD/Vh q=1-(D0/DF)
Stability Coefficients increase with Stability Coefficients somewhat
increasing torsional eccentricity insensitive to torsional eccentricity

The stability coefficient should theoretically be insensitive to torsional eccentricity

99
Effect of Torsional Eccentricity on Stability Coefficient
Computed Using ASCE 7: Computed from analysis with and without PD:
Torsional Eccentricity Story Torsional Eccentricity
Story None 0.05B 0.05AxB None 0.05B 0.05AxB
6 0.0292 0.0372 0.0459 6 0.0423 0.0136 0.0598
5 0.0402 0.0501 0.0637 5 0.0610 0.0604 0.0622
4 0.0517 0.0641 0.0814 4 0.0625 0.0700 0.0585
3 0.0592 0.0733 0.0930 3 0.0683 0.0670 0.0710
2 0.0602 0.0752 0.0960 2 0.0746 0.0729 0.0718
1 0.0588 0.0735 0.0940 1 0.0698 0.0714 0.0748
q=PD/Vh q=1-(D0/DF)

Stability Coefficients are very similar when the torsional eccentricity

is 0.05B (which is specified by ASCE 7).

100
 Second Order Effects Summary

◼ Most modern programs can directly include P-D and p-d effects in the analysis.

◼ Where the system is modeled in 3D with all vertical load elements being included and carrying
their tributary vertical load, P-Q effects are included automatically.

◼ Accuracy is improved where the program uses consistent geometric stiffness.

◼ Stability ratios (as specified in ASCE 7-22 Section 12.8.7) can be computed by use of lateral
displacements computed with and without P-Delta effects. These values will be similar to the
values (within a few percent) determined using Equation 12.8-16.

◼ For performing buckling analysis (which is a recommended step in the the model validation), it is
usually necessary to subdivide the vertical elements into at least two segments.

101
 Next Scene: Week 9 Review and Week 10
Preview

40

35

Roof Displacement, inches

30

25

20

15

10

0
-10 10 30 50 70 90 110 130 150
Dead Load Factor

Using Dead and Wind Using Dead and Notional

102
103
Conclusion
 40

35

Roof Displacement, inches

30

25

20

15

10

0
-10 10 30 50 70 90 110 130 150
Dead Load Factor

Using Dead and Wind Using Dead and Notional

105
 Recap of Week 9: Second Order Effects and
Stability Analysis

◼ Introduced the concept of geometric stiffness

◼ Defined and analyzed systems for P-D, P-Q, and p-d effects

◼ Summarized the requirements for including second order analysis in ASCE 7,

ACI 318, and AISC 360

◼ Presented the concept of a leaning column (which should be avoided in 3D

analysis)

◼ Demonstrated the use of SAP 2000 in the second order and buckling analysis of
framed structures

106
 Week 9 References

◼ McGuire, W., Gallagher, R., and Ziemian, R. (2000), Matrix Methods of Structural Analysis, John
Wiley and Sons

107
 What’s Coming in Week 10?

◼ Overview of Serviceability Issues

◼ Sensitivity analysis and optimization

◼ Use of “Drift Damage Indices” to assess serviceability compliance

108
 40

35

Roof Displacement, inches

30

25

20

15

10

0
-10 10 30 50 70 90 110 130 150
Dead Load Factor

Using Dead and Wind Using Dead and Notional

109