A Review on the Phenomena of Linear Analysis,

Large
Deformation Analysis, P-- Effects, Buckling
March 2023 – Ir. Law Chi Wai
 Contents
The discussion comprises :
 Introduction
 First Order Linear Elastic Analysis
 Buckling
 Second Order P- only Non-linear Analysis
 Second Order P-- Large Displacement Non-linear Analysis
 Introduction

 Structural analysis is mostly done by the first order linear elastic analysis which is
the simplest, fastest and most convenient approach;
 The first order linear elastic analysis assumes loads to stay on the un-deformed
shape even though there are node movements. This assumption is generally
considered adequately accurate as we are often dealing with small deformation
structures;
 However, if the structure is undergoing large deformations by which the point of
application of the loads will shift significantly, the first order linear analysis may not
be accurate enough;
 So the second order analysis is required, which takes the deformation of the
structure into account which are better known as the P- and/or P- effects.
Generally large deformations and structural member internal forces are resulted;
 The concept of buckling is employed in the analysis though the phenomenon may
not have much physical meaning.
 First Order Linear Elastic Analysis

Structural Analysis
 Structural analysis is mostly carried out by the “First Order Linear Elastic Analysis”
which is characterized by the followings:
 The force – deformation relation is linear, implying “elastic” behaviour of the

material, i.e. force proportional to deformation;

 The loads are applied at the un-deformed shape of the structure in which

equilibrium is also checked;

 The stiffness matrix of the structure is constant;

 Linear load combinations can be carried out, i.e. the effects due to individual

load cases are calculated and added together as necessary such as 1.4D+1.6L,
i.e. it is not necessary to analyze load combination by load combination which is
time consuming.
First Order Linear Elastic Analysis

Stiffness matrix
of individual
member
First Order Linear Elastic Analysis

Stiffness matrix
(transformed to
inclination )
First Order Linear Elastic Analysis

Assembling matrix
for analysis

Load matrix with

loads applied on
the un-deformed
nodes
 Second Order Non-Linear Analysis – P-- Effect

P- effect
 If the joint movement is taken into account in structural analysis, the most
pronounced effect is the P- effect , by which the axial load in the member will
create additional movements and forces due to the joint movement.
 Second Order Non-Linear Analysis – P-- Effect

P- effect
 P-  arrives from the initial bowing “imperfection” (denoted by ) of the strut
member under compression which creates “eccentric moment”.
 Second Order Non-Linear Analysis – P-- Effect

P-- effect
 P-  is “intrinsic”, i.e. so long the strut is imperfect (all struts are in fact imperfect),
there is P-  effect arriving from the initial “imperfection”. If there are movements
at the struts’ ends , P-  will couple with P- to form P-- effect.
 Second Order Non-Linear Analysis – P-- Effect

P-- effect
 So when the structure is undergoing P-- effect, the geometry of the structure will
change with the change of the deformation under the load application. Thus it
stiffness is also changing (stiffness depends on geometry).
 With the stiffness changing, the stiffness matrix is no longer constant at the loading
history. So in the linear equation , (force = stiffness matrix × deformation) or is no
longer valid and the analysis becomes non-linear.
 Definition of “second order analysis” according to the Steel Code 2011 is “analysis
of structure involving the tracing of equilibrium or the load versus deflection path
up to the formation of the first plastic hinge with consideration of initial member
imperfection.”
 Second Order Non-Linear Analysis – P-- Effect

 Second order non-linear analysis – large displacement analysis

 Buckling – Phenomena and Analysis

 First order linear analysis aims at keeping the stiffness matrix unchanged so that
the analysis is linear but requires provisions to cater for initial imperfections of the
member (P-) and also the node movement (frame sway, P-)
 The effect due to initial imperfections is through the theory of “buckling”.
 “Buckling” is commonly known as the sudden “bulging” out of a strut or plate
member under compression and the bulging is so significant that the member fails
by “breaking”.
 Buckling – Phenomena and Analysis

 First order linear analysis aims at keeping the stiffness matrix unchanged so that
the analysis is linear but requires provisions to cater for initial imperfections of the
member (P-) and also the node movement (frame sway, P-)
 The effect due to initial imperfections is through the theory of “buckling”.
 “Buckling” is commonly known as the sudden “bulging” out of a strut or plate
member under compression and the bulging is so significant that the member fails
by “breaking”.
 Buckling – Phenomena and Analysis

Consider a strut of flexural rigidity , Length

under an axial compression Load of ,
𝑃 𝑃
O O
At ordinate x, an equilibrium equation can be set up
by considering the free body OX,
𝑥
𝑀
As ; 𝑣X X
Solving 𝑣
where and are constants. 𝐿 𝑃
For the strut pinned at both ends, , ; , ,
; so ;
For , ; ; etc.
So ; ; etc. know as the Euler’s load.

𝑃
 Buckling – Phenomena and Analysis

The equation with ; ; represent different shapes as:

 Buckling – Phenomena and Analysis

In case the ends are both ends fixed, by symmetry, the

𝑃 𝑃
moments at both ends are equal and the lateral shear along O 𝑀 O 𝑀
the strut is zero.
At ordinate x, an equilibrium equation can be set up
𝑥 𝐸𝐼
2
𝑑 𝑣
by considering the free body OX, 𝑑𝑥
2

Put 𝑣X 𝑣
Solving
where and are constants.
𝐿𝑃
The boundary conditions, , ; , ,
,;,,
;
For , ; ; etc.
So ; ; etc. known as the Euler’s load.

𝑃 𝑀
 Buckling – Phenomena and Analysis

 So it follows that buckling load is different with the fixity of the end conditions. For
a strut fixed at one end and pinned at the other, the buckling load can be proven to
be and the deflection equation is with
 In stead of using buckling load directly for a strut under pre-determined end fixities,
it is a practice to express them in terms of an “effective length” as compared with
the simplest strut with pinned conditions at both ends
 So the effective length for both end fixed with is when ;
 Similarly, for one end fixed and one end pinned where ,
 Buckling – Phenomena and Analysis

 The Code asks for more conservative as :

 Buckling – Phenomena and Analysis

 With different and finite end fixities

Buckling – Phenomena and Analysis
Buckling – Phenomena and Analysis
Buckling – Phenomena and Analysis
Buckling – Phenomena and Analysis
 First Order Linear Analysis – Buckling

 With different and finite end fixities

 First Order Linear Analysis – Buckling

 With different and finite end fixities

 Buckling – Phenomena and Analysis

s on the phenomenon of “buckling” by the foregoing analyses:

ilure is related only to the flexural rigidity of the strut. It is not related to
o it is purely a stability problem;
kling” is only a theoretical derivation. It only relates that when the axial
es the discrete values of , etc. the strut can take up any arbitrary shape.
xial load at other values, the strut can only remain straight in order to
equilibrium equation;
heory that when the buckling loads are exceeded, the strut will buckle
ble is not proven.
 Buckling – Phenomena and Analysis

Consider a strut of flexural rigidity , Length with an axial Load of

𝑃 𝑃
and initial “bowing” O O
The differential equation is

𝑥
Assume where is the central initial deflection 𝑀
𝑣0 X
X
Solving 𝑣 𝑣
𝐿 𝑃
where and are constants.
For the strut pinned at both ends, , ; , ,
; so ;
Putting where

𝑃
 Buckling – Phenomena and Analysis

The total deflection is

𝑃 𝑃
O O
So it turns out that and will become infinity if .
By , which is the buckling load.
𝑥
𝑀
So, with the initial imperfection, the deflection of the strut will become 𝑣0 X
X
infinity if This is failure! 𝑣 𝑣
𝐿 𝑃
is the absolute maximum axial load the strut can sustain.

Thus the buckling load (or effective length) is in use for design for
predetermined initial imperfection.

𝑃
 Buckling – Phenomena and Analysis

Designing a strut with initial imperfection 𝑃𝑐

Consider a strut under an axial load and a moment such that the strut O
reaches its yield stress of at some point in the strut by:

𝑥
where c is the distance of the extreme fibre from centroidal axis
𝑣0 X

If the strut is assumed to be initially curved, then and at the mid-point 𝑣

𝐿
of the strut where , (by )
As , where is the radius of gyration, and ,

Putting where

Expanding

𝑃𝑐
 Buckling – Phenomena and Analysis

To solve the quadratic equation 𝑃𝑐

Put (a property of a strut called “initial curvature parameter” with O
maximum initial imperfection )

𝑥
𝑣0 X

𝑣
𝐿
Factorize
(identical to Eqn (A.30) of Steel Code 2011 Appendix 8.4)
Putting

(identical to Eqn (A.31) of Steel Code 2011 Appendix 8.4)

𝑃𝑐
 Buckling – Phenomena and Analysis

Following BS5950-1:2000 Annex C, is also termed the Perry factor 𝑃𝑐

Let with O
The Steel Code 2011(following BS5950) summarizes different types of sections
bending about major and minor axes into different values of in Table 8.7, 𝑥
bearing in mind that is related to which is related to the maximum
imperfection and and are related in a certain type of section. 𝑣0 X

With known value of and prescribed value of , and then can be calculated 𝑣
𝐿
and finally

𝑃𝑐
 Buckling – Phenomena and Analysis

For example, for a rolled I-section with maximum thickness < 40mm bending about 𝑃𝑐
major axis, curve a) should be followed as per Table 8.7. O
At , N/mm2 by Table 8.8(a) for use of curve a), N/mm2

𝑥
Check by formula, (Appendix 8.4)
𝑣0 X

𝑣
𝐿

and finally
N/mm2
So verified.

𝑃𝑐
Buckling – Phenomena and Analysis

𝜆=𝐿𝐸 / 𝑟
Buckling – Phenomena and Analysis
Buckling – Phenomena and Analysis
 Buckling – Phenomena and Analysis

The values:

 Apparently that different types of sections do have similar values by which we can
calculate the by the foregoing;
 However, instead of measuring the actual values for design, the Code measured
the failure load instead and back-calculate the and values.
 Buckling – Phenomena and Analysis

 The Steel Code 2011 also specifies the () ratio in the “second order” P-- elastic
design with the referenced buckling curve as in its Table 6.1

 But results will not be exactly identical to the code values.

 Buckling – Phenomena and Analysis

Derivation
and Use of
the
Geometric
Matrix
Buckling – Phenomena and Analysis
 Buckling – Phenomena and Analysis

Buckling load analysis can be performed in many software including SAP2000:

 Buckling load analysis is carried out by finding the eigenvalue of which can be proved by
, (Linear problem is )
where is the linear stiffness matrix, is the axial load matrix,
is the deflection matrix, is the external load matrix
To be unstable or is singular
where is the “buckling load factor” by which the axial loads in the members are
increased until the structure becomes unstable by buckling.

Compare with as the eigenvalue problem, the eigenvalue of is

Or
 Buckling load factors are given for the various modes of buckling, implying when all the applied loads are
multiplied by this factor, the structure will fail by buckling;
 A structure generally has many buckling modes, each generally applies to buckling of a certain member.
The member under a certain buckling mode can be identified from the mode shape (deflected shape).
Buckling – Phenomena and Analysis
Buckling – Phenomena and Analysis
Buckling – Phenomena and Analysis
Buckling – Phenomena and Analysis

kN
vs kN
Or kN
m

So the use of length of strut of

3m to determine buckling load
is very wrong in this case!
 kN

Buckling – Phenomena and Analysis

vs kN
Or kN
m

So the Buckling load is kN

A method to find the buckling
load of a member is to apply a
unit axial load to it and carry out which is close to the previous result
buckling load analysis of kN, less than 1%
1000kN

1000kN
1000kN

1000kN
Buckling – Phenomena and Analysis
Buckling – Phenomena and Analysis
 Second Order Non-Linear Analysis – P- Effect

P P P
3
𝑆𝐿
3
𝑆𝐿
S S
∆=
S
∆>
3 𝐸𝐼 3 𝐸𝐼 kN;
kN;
L L L
m
m4
𝑀=𝑆𝐿 𝑀=𝑆𝐿+𝑃 ∆ kN/m2

(a) First Order Linear (b) Second Order Analysis (P-)

Analysis
Figure 2 – Illustration of First Order Linear and Second Order P- Analysis

Top Base Axial Base

Deflection Moment Load Shear
Linear 0.01206 300 1000 100
P- 0.01320 313.20 1000 100
 Second Order Non-Linear Analysis – P- Effect

P- effect can be calculated in many software including SAP2000:

 P- effect can be calculated by the following equation

, (First Order Linear problem is )
where is the linear stiffness matrix, is the axial load matrix,
is the deflection matrix, is the external load matrix

 However, is not generally known at first as in the simple example of the

cantilever. So the software has to carry out iteration. Initially the P values
are first obtained by the first order linear analysis and the analysis carried
out by the above equation with the initial Ps. Then a new of Ps will be
obtained. Iteration is carried out until the successive Ps are within certain
errors.
Second Order Non-Linear Analysis – P- Effect

Cantilever Deflection with

and without P-Delta
No P-Delta With P-Delta

3.5

3
Height of Cantilever (m)

2.5

1.5

0.5

0
0 2 4 6 8 10 12 14

Deflection (mm)
 Second Order Non-Linear Analysis – P- Effect

For a high-rise building by ETABS

 Second Order Non-Linear Analysis – P- Effect

For a high-rise building by ETABS

 Second Order Non-Linear Analysis – P- Effect

Deflections at
roof level

No P-
P- under
WX1
D+L+WX1
 Second Order Non-Linear Analysis – P- Effect

Check Reactions
Load FX FY FZ MX MY MZ
Case/Combo kN kN kN kN-m kN-m kN-m
Dead -5.054E-07 0 305890.665 3668994.54 -3675768 -0.0006
Live 0 0 107520 1290240 -1290240 0.0001
WX1 -14068.6913 0 -9.382E-07 -0.0001 -1354166 168824.2954
P-
D(D+L+WX1) -14068.6894 2.405E-05 413410.6639 4959148.9295 -6458475 168820.3112
Max
P-
D(D+L+WX1) -14068.6894 2.405E-05 413410.6639 4959148.9295 -6458475 168820.3112
Min
The WX1 wind loads is in the X-direction creating MY moment. From the above
table, the sum of moments due to dead, live and WX1is
3675768+1296240+1354166=6320174kNm. So the moment created by the P- effect
is 6458475 – 6320174 = 138301kNm.
As the total D+L load is = 305891+107520=413411kN. Assuming the average  on
the floors to be 669.2/2 = 334.6mm, the P- moment is then
413411×0.3346=138327=kNm which is very close to 138301kNm
 Second Order Non-Linear Analysis – P- Effect

Deflections at
roof level

P- under
P- under
D+L+WX1
D+L+WX1
with large
deformation
 Effect of Construction Sequence on Building

 A building is constructed floor by floor;

 So its stiffness is gradually built up instead of one single stiffness for all dead loads
 Theoretically, the dead load of the nth floor is resisted by the structure comprising
n-1th floor and below.
Second Order Non-Linear Analysis – P- Effect
Second Order Non-Linear Analysis – P- Effect
 Second Order Non-Linear Analysis – P- Effect

Building Deformation at
Various Stages