See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235986483

Higher-Order Non-Linear Analysis of Steel Structures. Part II: Refined Plastic

Hinge Formulation

Article  in  Advanced Steel Construction · November 2011

CITATIONS READS

29 654

2 authors, including:

C.K. Iu
Queensland University of Technology
23 PUBLICATIONS   298 CITATIONS   

SEE PROFILE

All content following this page was uploaded by C.K. Iu on 05 June 2014.

The user has requested enhancement of the downloaded file.

 Advanced Steel Construction Vol. 8, No. 2, pp. 183-198 (2012) 183

HIGHER-ORDER NON-LINEAR ANALYSIS

OF STEEL STRUCTURES
PART II : REFINED PLASTIC HINGE FORMULATION
C.K. Iu * and M.A. Bradford

Centre for Infrastructure Engineering and Safety

School of Civil and Environmental Engineering
The University of New South Wales, UNSW Sydney, NSW 2052, Australia
*(Corresponding author: E-mail: [email protected])

Received: 31 May 2011; Revised: 10 October 2011; Accepted: 14 November 2011

ABSTRACT: In the companion paper, a fourth-order element formulation in an updated Lagrangian formulation was
presented to handle geometric non-linearities. The formulation of the present paper extends this to include material
non-linearity by proposing a refined plastic hinge approach to analyse large steel framed structures with many
members, for which contemporary algorithms based on the plastic zone approach can be problematic
computationally. This concept is an advancement of conventional plastic hinge approaches, as the refined plastic
hinge technique allows for gradual yielding, being recognized as distributed plasticity across the element section, a
condition of full plasticity, as well as including strain hardening. It is founded on interaction yield surfaces
specified analytically in terms of force resultants, and achieves accurate and rapid convergence for large frames for
which geometric and material non-linearity are significant. The solutions are shown to be efficacious in terms of a
balance of accuracy and computational expediency. In addition to the numerical efficiency, the present versatile
approach is able to capture different kinds of material and geometric non-linearities on general applications of steel
structures, and thereby it offers an efficacious and accurate means of assessing non-linear behaviour of the structures
for engineering practice.
Keywords: Beam-column, Frames, Interaction surface, Material non-linearity, Refined plastic hinge method, Strain
hardening

1. INTRODUCTION

This paper is concerned with an innovative extension of the geometrically non-linear elastic finite
element formulation with a higher-order element, described in the companion paper [1], to include
steel yielding so as to produce a robust and highly efficient technique for analysing frames with a
multiplicity of component structural members. The material yielding of an infinitesimal steel
element or particle across the element section has been investigated numerically through the plastic
hinge approach in the efficacious manner.

King et al. [2] presented a plastic hinge method which accounts for the degradation of the member
stiffness. The plastic hinge stiffness is formulated in this method by inserting a hinge in the
incremental element stiffness equation if a linear initial yield and fixed full yield condition [3] with
residual stresses are satisfied. In later work, Liew et al. [4, 5] proposed a second-order refined
plastic hinge analysis for steel frame design which included inelastic stiffness degradation, with
reference to a bilinear interaction equation AISC [6]. A tangent modulus technique was applied
which reduced the modulus of elasticity from its elastic value continuously as it entered the plastic
range.

Yau and Chan [7] established a plastic hinge analysis of steel frames in which the hinge stiffness is
formulated into the element stiffness matrix based on the incremental moment equation of
equilibrium at a node, with the full yield criterion of Duan and Chen [3] being adopted in the
formulation. Further, Chan and Chui [8] proposed a method for considering the gradual yielding
184 Higher-Order Non-Linear Analysis of Steel Structures, Part II: Refined Plastic Hinge Formulation

across a cross-section subjected to bending action, for which the axial force was included by
reducing the bending moment capacity. More recently, Iu and Chan [9] developed a refined
plastic hinge method which included strain-hardening for steel structures subjected to elevated
temperatures, while Iu et al. [10] modified a plastic hinge approach in which the interaction of
bending and axial actions on the yield surface at elevated temperature was formulated. Using a
higher-order element representation, Chan and Zhou [11, 12] presented a large displacement
analysis using a plastic hinge approach in which large deflections in the presence of a hinge along
the member can be included.

In the companion paper [1], the geometric non-linearities associated with a second-order analysis of
an elastic framed structure were discussed. Commonly, material yielding is an important
consideration for steel structures at their strength or ultimate limit state, and so generalised
numerical non-linear analysis of steel framed structures at their strength limit state necessitates the
accurate modelling of both geometric and material non-linearities. Many contemporary steel
structures are very large, comprising of a great number members of which many are slender, and
despite advances in computational algorithms and computer hardware capabilities, efficient and
accurate analyses of these structures is still problematic when material non-linearities have to be
considered. In order to provide a solution technique which overcomes these difficulties, the
plastic hinge approach is engaged in this paper to provide a balance between efficiency and
accuracy. Elasto-plastic hinge methods have been developed and reported by many researchers
(e.g. in [2, 7, 13]), but contrary to these, the present paper develops a refined plastic hinge approach
to allow for the gradual development of yielding as distributed plasticity across the section to a
fully plastic hinge, which admits strain hardening and accounts for the interaction of axial and
bending actions. This is an advance on elasto-plastic hinge methods because it allows for a more
general description of the materially non-linear behaviour of the steel in terms of force resultants
(rather than of stresses in a much less-efficient plastic zone approach). This paper therefore offers
a non-linear analysis with accuracy in solution, versatile for vast forms of structures and reliability
in convergence for engineering practice.

2. INTERACTION EQUATIONS FOR YIELDING CRITERION

In accordance with the so-called “plastic zone method”, the yield condition of the steel is examined
with respect to each point in the member, and so it is convenient to express the behaviour of the
steel in terms of its stress-strain relationship. This technique is valid for all cross-sections along
the member once the domain of the numerical integration has been defined. On the other hand,
the “plastic hinge method” bases its material modelling on the load-displacement relationship for a
beam-column cross-section [3, 6]. This modelling requires interaction equations between bending
and axial actions which defined the capacity its cross-sections.

In general, initial yielding within a member occurs well before the fully plastic cross-sectional
strength is reached. It depends on both the shape factor for the cross-section  and the residual
stresses on the cross-section. The initial yield surface y, or initial yield interaction equation, may
be defined from the bending actions about the major principal axis Mx, minor principal axis My and
axial force P collected in the vector f = {P, Mx, My}T and related by

 M yM y
 y f  
P
 x x  1 (1)
0.8 Py 0.9 M px 0.9 M py
 C.K. Iu and M.A. Bradford 185

in which the numbers 0.8 and 0.9 in the denominator account for residual stresses, Py is the axial
force capacity of the cross-section, and Mpx and Mpy are its full plastic moments about the major
and minor principal axes respectively which have respective shape factors of x and y. When
y(f) < 1, the cross-section is taken as elastic. For a doubly-symmetric I-section, a convenient
formulation used for the first yield surface is

1.25M x 1.25M y
 y f  
P
   1. (2)
0.8 Py M px M py

The AISC-LRFD standard [6 provides bilinear interactions for the criterion of full yield in terms of
a function p(f) = 1, but it is known that this is generally conservative for both short beam-columns
for which material non-linearities are dominant and for biaxial loading cases. The present
formulation therefore adopts the full yield interaction function p(f) proposed by Duan and Chen
[3], which defines a continuous convex full yield surface as shown in Fig. 1, with vertices as its
intersection with the P axis.
P
2 
 M   M 
f     
y

1  P   1  P 
x

 M   M 
1 .3 3
Py Py
P 1 . 25 M 1 . 25 M px py
i   
x y

0 .8 P y M px M py

M x

M y

Figure 1. Initial and Full Yield Surface under Interaction of Force Resultants

For the case of uniaxial bending about the major axis, full yield is taken as

Mx Mx
 p P, M x ,0    1,

M px M px 1  P Py 1.3  (3)

while for uniaxial bending about the minor axis, it is taken as

 p P,0, M y  
My My

M py 
M py 1  P Py 
3
 1 (4)

in which Py = fyA, A is the area of the cross-section, fy is the uniaxial yield stress and the moments
M px and M py are the major and minor axis bending capacities reduced in the presence of axial
force. A modified interaction surface from Tebege and Chen [14] is

2 
M   My 
 p f    x    1 (5)
M  M 
 px   py 

so that using Eqs. 3 and 4,

186 Higher-Order Non-Linear Analysis of Steel Structures, Part II: Refined Plastic Hinge Formulation

2 
   
 Mx   My 
 p f       1
   
(6)
 M px 1  P Py    M py 1  P Py  
1.3 3

in which

P 
  1.2  2 . (7)
P 
 y 

In the refined plastic hinge approach, gradual yielding is defined when the vector of actions on the
cross-section f is such that y(f) > 1 and p(f) < 1; this situation is common for beam-columns at
their strength limit state. The initial and full yield surfaces shown in Fig. 1 have been shown [3]
to be valid for a large range of steel cross-sections under biaxial bending and compression, and
provide an empirical basis for the refined plastic hinge approach of this paper.

3. REFINED PLASTIC HINGE STIFFNESS FORMULATION

3.1. Plastic Hinge Spring Stiffness

The present non-linear analysis for beam-columns relies on plastic hinge springs, as shown in Fig.
2, to be activated when the vector of actions at a node f are such that y(f) exceeds unity. When
this vector is such that p(f) exceeds unity, the cross-section experiences strain hardening. The
stiffnesses of the axial and bending springs are then taken as

EA 1   p f  
Sa    a  (8)
L   y f   1 

and

EI 1   p f  
Sb    b  (9)
L   y f   1 

in which the stiffnesses are such that  > Sa > 0 and  > Sb > 0, a and b are strain hardening
parameters, EI/L is the elastic flexural stiffness of the beam-column and EA/L is its elastic axial
stiffness. The spring formulation of Eqs. 8 and 9 was first introduced by Iu et al. [15].

The interaction between the three force resultants in the vector f has a twofold effect in the refined
hinge formulation; it firstly defines the initial yield and fully plastic surfaces by y(f) = 1 and p(f)
= 1, and it secondly defines the spring stiffness by Eqs. 8 and 9 which are degraded when y(f) > 1.
This kind of gradual yielding by proposed refined plastic hinge approach heralds distributed
plasticity across the member section. The spring formulation is therefore able to capture the
non-linear material behaviour, including its elastic domain, gradual or partial yielding, full
plasticity, strain hardening as well as residual stresses, in the load-deformation relationship for the
quartic beam-column finite element. These hinges may be incorporated into the second-order
elastic stiffness formulation of the companion paper [1] using the procedures described in [7, 16].
In reference to [15], it is also worth mentioning that the plastic hinges in Eqs. 8 and 9 are versatile
 C.K. Iu and M.A. Bradford 187

for different kind materials, once the interaction equations y(f) and p(f) are replaced by the
corresponding material failure surfaces.

3.2. Secant Stiffness in Plastic Hinge Formulation

In order to combine the plastic hinge spring stiffness formulation with the second-order elastic
stiffness formulation for a beam-column element, the incremental equilibrium equation of the
element is written as

 Pa   S a   u a 
 P   K11   u 
 e     e
 M s1   S b1  S b1    s1 
    (10)
 M e1    S b1 K 22  S b1 K 23    e1 
M s 2   K 32 K 33  S b 2  S b 2   s 2 
     
M e 2    Sb2 S b 2   e 2 

Ms2
ue
Ms1 Pe Me2  s2
ua s1 Me1 e2
e1
EI 1 f
Pa 1 Sb2  L  b
Sb1  EI f  b i 1
L i 1

1 f L
Sa  EA  a
L i 1

Figure 2. Beam-column element with axial and rotational springs

and for which it can be assumed that moment and axial force equilibrium of the beam-column
element can be considered separately in regard to material yielding behaviour. Since a plastic
hinge is inserted at the ends of the beam-column element, additional rotational degrees of freedom
at the nodes s1 and s2 are introduced, as shown in Fig. 2. Using matrix condensation of the
internal degree of freedom of the beam-column element (Me1 = Me2 = 0), Eq. 10 can be
decomposed into

 M s1   Sb1    s1   Sb1    e1 
     
Sb 2   s 2    Sb 2   e 2 
(11)
M s 2  

and

0  Sb1    s1   K11  Sb1 K12    e1 

     .
 Sb 2   s 2   K 21 K 22  Sb 2   e 2 
(12)
0 

After matrix condensation as in Eqs. 12 and 13, the incremental rotational deformations of the
beam-column element e1 and e2 can be evaluated from the incremental joint rotations s1 and
s2 from the equation
188 Higher-Order Non-Linear Analysis of Steel Structures, Part II: Refined Plastic Hinge Formulation

1
  e1   K11  Sb1 K12   Sb1    s1 
   ,
K 22  Sb 2   Sb 2   s 2 
(13)

 e 2   K 21

or

  e1  1  S b1 K 22  S b 2   S b 2 K12    s1 
    
S b 2 K11  S b1    s 2 
(14)
 e 2  K   S b1 K 21

where |K| = (K11 + Ss1)(K22 + Ss2) – K12K21. Once the element incremental rotations e1 and
e2 are known from Eq. 14 which includes material non-linearity, the incremental bending
resistance can be evaluated from

 M e1   K11 K12    e1 
   ,
K 22   e 2 
(15)
M e 2   K 21

which is based on the elastic stiffness formulation based on elastic stiffnesses Kij because it is
assumed that the element is elastic. Hence the equation

 M e1  1  K 11 K 12   S b1 K 22  S b 2   S b 2 K 12    s1 
    
  S b 2 K 11  S b1    s 2 
(16)
M e 2  K  K 21 K 22    S b1 K 21

is equivalent to the incremental secant stiffness in the second-order elastic stiffness formulation in
the companion paper [1]. For axial actions, the incremental force equilibrium equation is written
separately as

Pa   S a 0  u a   S a 0  u a 

     ,
K11   u e   0 EA L   u e 
(17)
 Pe   0

where the incremental axial force in the linear spring Pa is equal to the axial force in the
beam-column element Pe. The total incremental axial deformation u is the sum of the
incremental axial deformation in the axial spring ua and in the element ue, and consequently the
axial resistance of an element can be expressed as

 1 
P   u , (18)
 a
S  EA L 

which describes the axial resistance of the element in terms of the axial displacement and so is
equivalent to the secant stiffness for a plastic axial spring.

3.3. Tangent Stiffness for Plastic Hinge Formulation

For bending actions, the incremental moment-rotation relationship needs to be reformulated with
respect to the nodal or local coordinate system, and the axial spring stiffness superimposed into the
stiffness formulation in a similar way; it being assumed that the moment and axial force
equilibrium conditions can be formulated separately. This produces
 C.K. Iu and M.A. Bradford 189

 P   K11   u 
    
 M s1    
K 22     s1 
K 23 (19)

M       s 2 
 s2   K 32 K 33

where

S a K11
K11  , (20)
K11  S a

 
K 22
1
K

S b1 K 22 K 33  S b 2   S b 2 K 232 , (21)

  K 32
K 23  
1
S b 2 K 23 K 22  S b1   S b 2 K 22 K 32  , (22)
K

 
K 33
1
K

S b 2 K 33 K 22  S b1   S b1 K 322 .  (23)

Eq. 19 is equivalent to the tangent stiffness in the second-order elastic stiffness formulation in the
companion paper [1]. This non-linear equilibrium path due to material non-linearities can be
traced using the non-linear solution procedures described in the companion paper [1].

4. NUMERICAL VERIFICATIONS

In order to verify the accuracy and efficiency of the present non-linear second-order inelastic
analysis, material non-linearities for an encastré beam with a concentrated load are compared with
other inelastic analyses. Two benchmark solutions for frame structures are also investigated,
including both material and geometric non-linearities, while a calibration frame is studied for
investigating the effect of spreading of plasticity. Finally, a large-scale space framed structure is
analysed using the proposed method.

4.1. Encastré Beam with An Asymmetric Point Load

A beam with fixed or encastré ends has been investigated to isolate a case which is governed by
material yielding. The inelastic behaviour of this fixed beam with a residual stress of 60% of the
yield stress was studied by Liew et al. [17], in which a point load was applied at one-third of the
beam, as shown in Fig. 3. Plastic hinges then form in sequence, until a plastic mechanism is
achieved (the load factor p = PL/Mp quoted in Trahair et al. [18] is 9.0 for a full plastic
mechanism). Liew et al. [17] analysed this problem by both a refined plastic hinge method and a
hinge-by-hinge method. The load factor p = PL/Mp is plotted against the dimensionless
deflection EI/MPL2 at the point of application of the load in Fig. 3. For the proposed non-linear
analysis, two elements were used for the beam with the load being applied at span. It can be seen
that the results of the present method are in accord with those of Liew et al. [17]. In the present
non-linear analysis, initial yielding initiates at location 1 with a load factor p1 = 5.41; this hinge
becomes fully plastic when p1 = 8.81. Fully-yielded plastic hinges form at locations 2 and 3 at
values of p2 = 8.91 and p3 = 9.01.
190 Higher-Order Non-Linear Analysis of Steel Structures, Part II: Refined Plastic Hinge Formulation

10 [p1=6.75, p2=8.63, p3=8.91 (elasto-plastic [p1=6.81 p2=8.71, p3=9.11 (elasto-plastic

8 [p1=8.81, p2=8.91, p3=9.01

7 P 
[p1=8.31, p2=8.76, p3=8.91 (elasto-plastic
Load factor PL/Mp

6 1 2 3

5

4 L/3 2L/3

3 Liew et al. [42] Refined plastic hinge method

Liew et al. [42] Hinge by hinge method

2
Present analysis (4rd-order element - Refined plastic hinge method)

1 Present analysis (4th-order element - Elastic-perfectly-plastic hinge method)

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Dimensionless d eflection EI/M pL2

Figure 3. Normalized Load-displacement Curve of a Fixed-end Beam

Figure 3 also summarises the plastic load factors from various approaches. This numerical
verification illustrates that the proposed plastic hinge method is able to model residual stresses,
gradual yielding and fully-yielded material behaviour.

4.2. Single-bay Frame of Vogel

P P=2800kN

H=35kN
HEA340
HEB300

HEB300

5m

E  2.05  108 kN m 2

 y  2.35 kN m 2

4m

Figure 4. Geometric Configurations and Loading Pattern of Vogel’s Frame

The portal frame shown in Fig. 4 was analysed numerically in 1985 by Vogel [19], and this frame
has been used by several researchers (e.g. Chan and Zhou [12] and Toma and Chen [20]) as a
benchmark solution for including material non-linearities including residual stresses, gradual
yielding and full-plasticity, as well as geometric non-linearities. Figure 4 shows the applied
loading, material properties, member cross-section and geometry of the simple portal frame. The
initial out-of-straightness  = 1/400 of the column height was engaged in the plastic zone method
[19]. The horizontal load on the top of the frame produces sway in the frame, and the vertical
loads generate second-order effects in the columns.
 C.K. Iu and M.A. Bradford 191

1.2

1.05 1.02
1
1.017


P P
0.8
H
Load factor

0.6

0.4

0.2 Vogel [44] plastic zone method

Vogel [44] plastic hinge method
Present analysis (4th-order element-plastic hinge method)

0
0 2 4 6 8 10 12 14 16 18
Lateral displacement (mm)

Figure 5. Load-displacement Curve at the Top of Vogel’s Portal Frame

In the model herein, only one element was used to model each of the members, whereas Clarke [21]
used a refined mesh with 50 elements for each column and 20 elements for the beam in his
numerical modelling. More recently, Teh and Clarke [22] produced a more computationally
efficient approach in their plastic zone method for which 4 elements were used in each column to
model geometric non-linearity and stability effects, and one element for the beam. The present
technique is still more computationally efficient than the latter modelling of Teh and Clarke [22].

The load versus displacement curves for the top of the column from both the approach of this paper
and that of Vogel [19] are shown in Fig. 5. Generally, the load-displacement behaviour from the
present method is reasonably consistent with that of Vogel, the deviation between the present
analysis and Vogel’s plastic zone method being attributed mainly to the different onsets of initial
yielding at load factors of 0.8 and 0.6 respectively. The plastic hinge formulation of Vogel does
not include gradual yielding. The ultimate load factor from the analysis of this paper is 1.05,
whereas that from Vogel’s plastic zone analysis is 1.02 and Vogel’s plastic hinge analysis is 1.07.
All of these three methods produce a consistent ultimate load factor.

Initial yielding occurs firstly in the right column at its bottom at a load factor of 0.825, and all four
joints of the frame have reached initial yield at a load factor of 0.875. When the frame is loaded
to a load factor of 1.05, a fully yielded plastic hinge forms in the right column at its bottom, and
with an infinitesimal increase of this load factor, a failure mechanism develops in the frame. The
portal frame exhibits no strength reserve from its threefold redundancy because of the symmetric
loading distribution in this frame; when the fully yielded plastic hinge is formed, the other three
hinges for which full yielding is imminent form fully plastic hinges in the next iteration with
numerical divergence being detected at a load factor of 1.05. This represents the ultimate capacity
of the frame. The lateral behaviour of the frame from the present approach is stiffer than the
results of Vogel [19], probably because of the different initial yielding conditions which are
engaged in both analyses. This benchmark numerical example is used commonly as a
countercheck of gradual yielding behaviour.
192 Higher-Order Non-Linear Analysis of Steel Structures, Part II: Refined Plastic Hinge Formulation

4.3. Six Storey Rigid-jointed Frame of Vogel

The two-bay six-storey European calibration frame shown in Fig. 6 was subjected to proportionally
applied distributed gravity loads and to concentrated loads, and its behaviour was reported by Vogel
[19]. The member sections and frame geometry are depicted in Fig. 6, and the elastic modulus
and yield stress of the steel were taken as 2.05108 kN/m2 and 2.35105 kN/m2 throughout. The
initial out-of-straightness of the frame was assumed to be  = 1/450, with all members being
rigidly connected at their joints. In the modelling of this frame, each column was discretised as
one element whilst the beams were subdivided into four elements because only nodal loads are
admitted, although the present formulation does allow for a quadratic bending moment distribution
to be simulated in by the fourth-order finite element. It is worth noting that using the plastic zone
approach [21], each column was subdivided into 20 elements and each beam into either 20 or 40
elements, depending on the degree of non-linearity.
F 1=10.23kN 31.7kN/m

HEB200

HEB160
IPE240 IPE240

F 2=20.44kN HEB200 49.1kN/m

HEB160
IPE300 IPE300

F2 49.1kN/m

6
HEB240

IPE300 IPE300 HEB220

3.75 = 22.5m
F2 49.1kN/m
HEB240

HEB220

IPE330 IPE330

F2 49.1kN/m
HEB260

HEB220

IPE360 IPE360

F2 49.1kN/m
HEB260

HEB220

IPE400 IPE400


2 6 = 12m

Figure 6. Geometric Configuration and Loading Pattern of Vogel’s six-storey Frame

1.2
3 3
2 2
1 1 
1

0.8 
Load factor

0.6

0.4

Vogel [44] Numerical results (Plastic zone method) [at roof]

Vogel [44] Numerical results (Plastic zone method) [at 4th floor]
0.2
Present analysis (4th-order element - plastic hinge method) [at roof]

Present analysis (4th-order element - plastic hinge method) [at 4th floor]
Lateral displacement (mm)
0
0 50 100 150 200 250 300 350

Figure 7. Lateral Drifts of the Vogel’s six-storey Frame

 C.K. Iu and M.A. Bradford 193

Figure 7 shows the lateral drifts of the 4th and 6th floors of the frame determined from the analysis
of this paper, and those reported by Vogel [19], and it can be seen that they are in excellent
agreement. The ultimate load factor predicted by the current method is 1.14, which is slightly
higher than the value of 1.12 given by Vogel [19].

In the approach of this paper, initial yielding occurs at a load factor of 0.62 at the beams adjacent to
the interior column joints at the 4th and 6th floor levels. Both of these hinges have fully yielded
when a load factor of 0.94 is reached, which heralds the onset of non-linearity which can be seen in
Fig. 7. At a load factor of 1.02 in Fig. 7, the number of fully-yielded plastic hinges increases to
four, and these are denoted by the numeric index 1 in Fig. 8a and in Fig. 7. When the frame is
loaded further to a load factor of 1.10, a total of 10 fully-yielded plastic hinges have formed and the
additional ones are denoted by the index 2 in Fig. 8a and in Fig. 7. When the ultimate load of the
frame has been reached, there are a total of 23 fully-yielded plastic hinges, those formed at the
beams’ mid-spans are due mainly to gravity loading, whereas those formed at the beams’ ends are a
result of the effect of sway in accordance with the distribution of bending moment at ultimate
loading which is shown in Fig. 8b. This verification example indicates that the present inelastic
approach can accurately replicate the effects of gradual yielding and full plasticity in a large-scale
multi-storey steel frame. Ziemian [23] also carried out a calibration study of this frame in his
numerical analysis in 1992. It is worth noting that the following comparison is for information
only because of different computational technologies being used. While Ziemian’s plastic zone
approach took 8 minutes to perform the analysis, the present analysis took 4 seconds on a desktop
personal computer. The present plastic hinge analysis is more efficient by comparison and can
capture the real inelastic and stability behaviour of large-scale frames accurately, including initial
yield, gradual or partial yield and full plasticity, as well as frame sway and member bowing.
2 3 1 3 3
3
86kN m 86.2kN m

3 3 1 3 3 3 150kN m 149kN m

151kN m 148.6kN m

3 3 1 3 3 1 150kN m 150kN m

147.9kN m 150kN m

2 2 191kN m 191kN m

165.3kN m 158kN m
2 3 239.4kN m 240kNm

131kN m
156.6kN m
144.4kNm 228.5kN m

114.2kN m
168.4kN m
2

106kN m 127.5kN m 107kNm

a) Locations of the fully plastic b) Moment distribution on the

Figure 8. Plastic Hinge Pattern and Moment Distribution on the Frame

4.4. Two-storey Asymmetric Calibration Frame

The two-storey two-bay asymmetrical frame subjected to vertical loading only shown in Fig. 9 was
studied in 1982 by Iffland and Birnstiel [24] as part of the American Institute of Steel Construction
report on frame stability with plasticity. This calibration frame was chosen as a benchmark for
validating the loading redistribution due to plasticity. The material modelling used was
elastic-perfectly-plastic (without strain hardening), and the geometry, material properties, loading
and section configurations are shown in Fig. 18. Ziemian [23] and Clarke [21] also analysed this
frame in the verification of their numerical analyses. Ziemian [23] presented both plastic zone
194 Higher-Order Non-Linear Analysis of Steel Structures, Part II: Refined Plastic Hinge Formulation

and hinge methods, in which he used 60 elements per beam and 50 elements per column in his
modelling, while Clarke [21] undertook a plastic zone method analysis of the frame with each
column being subdivided into 20 elements, beams B1 and B3 into 20 elements and beams B2 and
B4 into 40 elements. However, modelling the frame using the approach of this paper necessitated
subdividing the columns into one element beams into 4 elements to include the distributed loading,
as discussed previously. Both arc-length [25] and residual load [26] strategies are proposed to
trace the non-linear equilibrium solution of this two-storey asymmetrical frame.
 71.5kN/m

B3 - W 21 44 B4 - W 27 102

C6 - W14 109
C5 - W14 120
C4 - W8 13

4.567m
 153.3kN/m

B1 - W 27 84 B2 - W 36 170

C2 - W14 132

C3 - W14 120

6.069m
C1 - W8 15

E  2  10 8 kN m 2
 y  2 . 234  10 5 kN m 2

6.069m 14.6404m

Figure 9. Configurations of the American Calibration Frame

1
0.999

 0.8

Load factor

0.6

0.4

Ziemian [47] (Plastic zone method)

Ziemian [47] (Plastic hinge method) 0.2
Clarke et al. [45] (Plastic zone method)
Present analysis (Refined plastic hinge method)

0
-40 -35 -30 -25 -20 -15 -10 -5 0 5 10
Lateral displacement (mm)

Figure 10. Lateral Displacement at Top of the Two-storey Frame Against Load Factor

The lateral displacement of the frame at the roof level  is plotted in Fig. 10 against the load factor.
The lateral displacement from the method of this paper agrees almost completely with the plastic
hinge method [23] throughout the entire loading range. Columns C1 and C4 are much more
flexible than the other columns of the frame, and so initially the frame sways towards its weaker
bay as shown in Fig. 10. At a load factor of about 0.808, three plastic hinges are formed as shown
in Fig. 11, with the differential yielding between the stronger and weaker bays causing their relative
stiffnesses to change, with a consequent change in the direction of the sway , as shown in Fig. 10.
Using the present non-linear analysis, the plastic hinges form subsequently with an increase in load
until the ultimate load factor (u = 0.999) is reached, with the hinges shown in Fig. 11, but before
doing so the sway reverses its direction again due to the significant yielding of the stronger bay, in
which plastic hinges form at the mid-span of beams B2 and B4. The ultimate load factor of u =
0.999 compares very well with that of Ziemian [23] (u = 1.01) and Clarke [21] (u = 0.985). The
numerical analysis proposed herein was completed in 6 seconds for 300 load cycles.
 C.K. Iu and M.A. Bradford 195

p = 0.736 p = 0.976

p = 0.808 p = 0.998

p = 0.85 p = 0.999

Figure 11. Sequence of Plastic Hinges Formed at Corresponding Plastic Load Factors

4.5. 6-storey Large Space Frame Structure

The rigid-jointed 6-storey asymmetrical space frame shown in Fig. 12 was proposed and analysed
by Orbison et al. [27], and later by Liew et al. [28] and Jiang et al. [29]. The yield strength and
elastic modulus were taken as 250105 kN/m2 and 2.07108 kN/m2 respectively, and the frame was
subjected to both uniform gravity loads of 9.6 kN/m2 on each floor and lateral loads of 53.376 kN
which were applied at every beam-column joint in the transverse z-direction, as shown in Fig. 12.
The member sections and plan configuration of the floors are also shown in Fig. 12. In the present
numerical modelling, one element was used for each member to replicate the overall
load-displacement behaviour of the space frame, which is the same as used in [28, 29]. The main
differences in these analyses is in the treatment of inelasticity; a plastic hinge model using a plastic
interaction function was employed by Orbison et al. [27] and by Liew et al. [28], whereas a plastic
zone approach with gradual member yielding was used by Jiang et al. [29]. The refined plastic
hinge approach in the present method allows for gradual yielding at the hinge under both axial and
bending actions. y
x
z Point B

Point A
W10 60

W10 60
W10 60

W10 60
6 @ 3.685m=21.948m

W12 26 W12 26
W12 120
W12 87

W12 87
W12 53

W12 87

W12 53

7.315m

W12 120
W12 87

W12 87

W12 26 W12 26

7.315m 7.315m

Wind load
direction

Figure 12. Section and geometric configuration of 6-storey space framed structure
196 Higher-Order Non-Linear Analysis of Steel Structures, Part II: Refined Plastic Hinge Formulation

1.2
z-directi x-directi x-directio
on on n

1
0.8

Load factor
0.6
0.4
Liew et al. [52] (Refined plastic hinge method)

Jiang et al. [53] (Plastic zone method)

0.2
Present analysis (Refined plastic hinge method) [Pt. A]

Present analysis (Refined plastic hinge method) [Pt. B]

0
-350 -300 -250 -200 -150 -100 -50 0 50 100 150
Lateral drift (mm)

Figure 13. Lateral drifts of points A and B in x- and z-directions vs load factor

Figure 13 plots the lateral drifts at points A and B, which are shown as black dots in Fig. 12, in the
x and z-directions against the load factor. In [28, 29], the lateral displacements of point A are
completely consistent in all ranges with the results of the present non-linear analysis. The present
analysis predicts initial frame yielding at a load factor of 0.835; the ultimate load factor of the space
frame from [28] is 1.005 whereas a value of 1.0344 is predicted from the present approach. In Fig.
13, it can be seen that the lateral displacement of point B in the x-direction is antisymmetric to that
of point A, whilst the lateral displacements for points A and B are the same in the z-direction.
This implies that the structure twists as the applied loads and structural form are asymmetrical
because of the change of the structural plan at the fourth floor level. In addition, after sufficient
plastic hinges have formed at a load factor of about 0.92, the space frame undergoes large lateral
displacements in both transverse directions and the present non-linear analysis predicts the
formation of 16 plastic hinges at the ultimate load factor of 1.0344. The locations of the plastic
hinges are shown by the dots in Fig. 13, indicating the yielding which occurs at the ends of the
beams in the 6-sotrey space frame due to the torsional effect on the whole structure. In [28], a
total of 20 plastic hinges developed and most of them were located at the fourth floor owing to the
twisting of this space structure. In summary, the general material and geometric non-linear
behaviour of the large-scale space structure is adequately captured using the present non-linear
analysis.

5. CONCLUDING REMARKS

In this paper, a refined plastic hinge method has been proposed to account for material
non-linearities in steel framed structures. The formulation makes recourse to a gradual yielding
from prescribed initial and fully-yielded interaction criteria in terms of force resultants (axial and
bending actions), so that distributed plasticity across the element section is adequately modelled.
This proposed refined plastic hinge formulation is incorporated into the non-linear elastic stiffness
analysis described in the companion paper. The technique has been applied to a number of
independent sample problems, with good agreement being obtained, and moreover the solutions
were obtained very rapidly.
 C.K. Iu and M.A. Bradford 197

Both this and the companion paper demonstrate that the present fourth-order stiffness formulation
with a refined plastic hinge methodology can accurately tackle material yielding and structural
instability, using a minimum of element discretisations by comparing with the conventional finite
element in deference to the plastic zone method. The technique provides an efficacious and
reliable means for the analysis of general steel structures numerically, when they have many
members, accounting adequately for geometric and material non-linear effects. Further, the
present plastic hinge stiffness formulation is able to evolve itself accounting for the various kinds of
materials versatilely, such as concrete or composite, once the failure surfaces in the hinge stiffness
are replaced correspondingly. This paper therefore offers a non-linear analysis with accuracy in
solution, versatile for vast forms of structures and efficiency in convergence for engineering
practice.

ACKNOWLEDGEMENT

The work in this paper was supported by the Australian Research Council through a Federation
Fellowship awarded to the second author.

REFERENCES

[1] Iu, C.K. and Bradford, M.A., “Higher-order Non-linear Analysis of Steel Structures”, Part 1:
Elastic second-order formulation, (???? Year).
[2] King, W.S., White, D.W. and Chen, W.F., “On Second-order Inelastic Analysis Methods for
Steel Frame Design”, Journal of Structural Engineering, ASCE, 1992, Vol. 118, No. 2, pp.
408-428.
[3] Duan, L. and Chen, W.F., “Design Interaction Equation for Steel Beam-columns”, Journal
of Structural Engineering, ASCE, 1989, Vol. 115, No. 5, pp. 1225-1243.
[4] Liew, J.Y.R., White, D.W. and Chen, W.F., “Second-order Refined Plastic-hinge Analysis
for Frame Design”, Part I, Journal of Structural Engineering, ASCE, 1993, Vol. 119, No. 11,
pp. 3196-3216.
[5] Liew, J.Y.R., White, D.W. and Chen, W.F., “Second-order Refined Plastic-hinge Analysis
for Frame Design”, Part II, Journal of Structural Engineering, ASCE, 1993, Vol. 119, No. 11,
pp. 3217-3237.
[6] American Institute of Steel Construction, Load and Resistance Factor Design Specification
for Structural Steel Buildings, AISC, Chicago, 1986.
[7] Yau, C.Y. and Chan, S.L., “Inelastic and Stability Analysis of Flexibly Connected Steel
Frames by Springs-in-series Model”, Journal of Structural Engineering, ASCE, 1994, Vol.
120, No. 10, pp. 2803-2819.
[8] Chan, S.L. and Chui, P.P.T., “A Generalized Design-based Elastoplastic Analysis of Steel
Frames by Section Assemblage Concept”, Engineering Structures, 1997, Vol. 19, No. 8, pp.
628-636.
[9] Iu, C.K. and Chan, S.L., “A Simulation-based Large Deflection and Inelastic Analysis of
Steel Frames under Fire”, Journal of Constructional Steel Research, 2004, Vol. 60, pp.
1495-1524.
[10] Iu, C.K., Chan, S.L. and Zha, X.X., “Material Yielding by Both Axial and Bending Spring
Stiffness at Elevated Temperature”, Journal of Constructional Steel Research, 2007, Vol. 63,
pp. 677-685.
[11] Zhou, Z.H. and Chan, S.L., “Elastoplastic and Large Deflection Analysis of Steel Frames by
One Element per Member I: One Hinge along Member,” Journal of Structural Engineering,
ASCE, 2004, Vol. 130, No. 4, pp. 538-544.
 198 Higher-Order Non-Linear Analysis of Steel Structures, Part II: Refined Plastic Hinge Formulation

[12] Chan, S.L. and Zhou, Z.H., “Elastoplastic and Large Deflection Analysis of Steel Frames by
One Element per Member II: Three Hinges along Member”, Journal of Structural
Engineering, ASCE, 2004, Vol. 130, No. 4, pp. 545-553.
[13] Kassimali, A., “Large Deformation Analysis of Elastic-plastic Frames,” Journal of
Structural Engineering, ASCE, 1983, Vol. 109, No. 8, pp. 1869-1886.
[14] Tebedge, N. and Chen, W.F., “Design Criteria for H-columns under Biaxial Loading”,
Journal of the Structural Division, ASCE, 1974, Vol. 100(ST3), pp. 579-598.
[15] Iu, C.K., Bradford, M.A. and Chen, W.F., "Second-order Inelastic Analysis of Composite
Framed Structures based on the Refined Plastic Hinge Method”, Engineering Structures,
2009, Vol. 31, No. 3, pp. 799-813.
[16] Iu, C.K., “Inelastic Finite Element Analysis of Composite Beams on the Basis of the Plastic
Hinge Approach”, Engineering Structures, 2008, Vol. 30, pp. 2912-2922.
[17] Liew, J.Y.R., White, D.W. and Chen, W.F., “Second-order Refined Plastic Hinge Analysis of
Frames”, Structural Engineering Report CE-STR-92-12, Purdue University, West Lafayette,
Indiana, 1992.
[18] Trahair, N.S., Bradford, M.A., Nethercot, D.A. and Gardner, L., “The Analysis and
Behaviour of Steel Structures to EC4”, 4th edn., Taylor and Francis, London, 2008.
[19] Vogel, U., “Calibrating Frames”, Der Stahlbau 1985, Vol. 10, pp. 296-301.
[20] Toma, S. and Chen, W.F., “European Calibration Frames for Second-order Inelastic
Analysis”, Engineering Structures 1992, Vol. 14, No. 1, pp. 7-14.
[21] Clarke, M.J., “Plastic Zone Analysis of Frames”, In: Advanced Analysis of Steel Frames,
Chen, W.F. and Toma, S., eds., CRC Press, Boca Raton, Florida, 1994, pp. 259-320.
[22] Teh, L.H. and Clarke, M.J., “Plastic-zone Analysis of 3D Steel Frames using Beam
Elements”, Journal of Structural Engineering, ASCE, 1999, Vol. 125, No. 11, pp.
1328-1337.
[23] Ziemian, R.D., “A Verification Study for Methods of Second-order Inelastic Analysis”,
Proceedings of Annual Technical Session, Structural Stability Research Council, Pittsburgh,
Pennsylvania, 1992, pp. 315-326.
[24] Iffland, J.S.B. and Birnstiel, C., “Stability Design Procedures for Building Frameworks”,
Research Report, AISC Project No. 21.62, American Institute of Steel Construction,
Chicago, Illinois, 1982.
[25] Crisfield, M.A., “An Arc-length Method including Line Searches and Accelerations”,
International Journal for Numerical Methods in Engineering, 1983, Vol. 19, pp. 1269-1289.
[26] Chan, S.L., “Geometric and Material Nonlinear Analysis of Beam-columns and Frames
using the Minimum Residual Displacement Method”, International Journal for Numerical
Methods in Engineering 1988, Vol. 26, pp. 2657-2669.
[27] Orbison, J.G., McGuire, W. and Abel, J.F., “Yield Surface Applications in Nonlinear Steel
Frame Analysis”, Computer Methods in Applied Mechanics and Engineering 1982, Vol. 33,
No. 1, pp. 557-573.
[28] Liew, J.Y.R., Chen, H., Shanmugam, N.E. and Chen, W.F., “Improved Nonlinear Plastic
Hinge Analysis of Space Frame Structures”, Engineering Structures, 2000, Vol. 22, No. 2,
pp. 1324-1338.
[29] Jiang, X.M., Chen, H. and Liew, J.Y.R., “Spread of Plasticity Analysis of
Three-dimensional Steel Frames”, Journal of Constructional Steel Research, 2002, Vol. 58,
pp. 193-212.

View publication stats