Sie sind auf Seite 1von 7

Structures 9 (2017) 105–111

Contents lists available at ScienceDirect

Structures

journal homepage: www.elsevier.com/locate/structures

A New Codified Design Theory of Second-order Direct Analysis for Steel


and Composite Structures – From Research to Practice
Siu-Lai Chan ⁎, Yao-Peng Liu, Si-Wei Liu
Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China

a r t i c l e i n f o a b s t r a c t

Article history: The terms “Second-order Analysis” and “Direct Analysis” appear in many modern design codes but many engi-
Received 14 June 2016 neers consider this as a method for checking only the second-order sway effect, nonlinear sway moment or a
Accepted 10 October 2016 tool for academic research. This perception is incorrect in modern structural design and in fact this old approach
Available online 13 October 2016
is defined only as the second-order “indirect” analysis or P-Δ-only analysis. When used properly, the full second-
order P-Δ-δ analysis method of design provides a reliable and proficient tool for practical design. Their underlying
Keywords:
Second-order direct analysis
principle also carries a very different philosophy to the conventional design approach. This paper outlines the dif-
Buckling ferences between the two design concepts, namely the first-order linear effective length method (ELM) and the
Nonlinear second-order direct elastic or plastic analysis (SODEA and SODPA). The era for re-thinking our design and anal-
ysis philosophy seems to have arrived. In this paper, the experience and theory on using the new design concept
and method are described with worked example given on their uses. Most previous works on second-order anal-
ysis do not provide reliable solutions to practical engineering problems, because they were not calibrated against
benchmark solutions where the current design code is adequately accurate. As shown in the examples in this
paper, the proposed design is able to reproduce some standard solutions of these benchmark examples so it
can be used for practical design.
© 2016 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.

1. Introduction An alternative, better and more rational design is to simulate the re-
sponse of a structure directly using a nonlinear analysis. In the past cen-
In the last century, design of steel structures relies on the effective tury, theories of nonlinear analysis of structures were studied by
length method (ELM) in conjunction with the first-order linear analysis. numerous researchers. Recent work related to practical applications in-
This conventional design constitutes the major design exercise since clude works by Liew et al. [1–2], Izzuddin and Smith [3], Chen and Chan
one needs to first determine the safety and stability of a structure [4] and White and Hajjar [5]. The works have also been summarised and
under permanent, imposed, wind and seismic loads in ultimate limit reported in detail by Yang and Kuo [6] and Chan and Chui [7]. Applica-
state design. This traditional design approach using the first-order linear tions to special structures like transmission towers are also demonstrat-
analysis in conjunction with the effective length method (ELM) is not ed by Albermani and Kitipornchai [8] and Chan and Cho [9] and
based on a rational theory and it assumes every member buckles simul- performance of various nonlinear numerical methods is also studied
taneously with its own effective length at its undeformed position. This by Clarke and Hancock [10]. The theoretical works have been extensive-
method is a member-based design and it is, strictly speaking, incompat- ly studied with encouraging results reported, but the proposed methods
ible with the computer analysis for a frame system. Because of this as- by these researchers are not easy to apply directly to design of real en-
sumption of drastic buckling in ELM, deflections and stresses cannot gineering structures. For example, most codes require consideration of
be calculated from elastic buckling analysis or the ELM but they are ap- member imperfection in buckling strength determination but the
proximated separately from the first-order linear analysis. In practice, straight cubic beam-column element used in most research and com-
this mode of buckling at undeformed position is unrealistic and valid puter programs for structural analysis does not consider member initial
only for non-existent perfect structures since imperfections due to ma- curvature and imperfection in its formulations, which exists in practical
terial and geometry activate the load vs. deflection behaviour of all prac- members and mandatory in codes. Modelling of a member by several el-
tical structures. ements is not only inconvenient, but it also hinders engineers to visual-
ize the member behaviour from a batch of finite element which is not in
line with their training of structural analysis and design in their courses
⁎ Corresponding author.
of studies. Practicing engineers most commonly use the elastic second-
E-mail addresses: ceslchan@polyu.edu.hk (S.-L. Chan), yp.liu@polyu.edu.hk (Y.-P. Liu), order “indirect” analysis for finding of P-Δ sway moment which in fact
siwei.liu@connect.polyu.hk (S.-W. Liu). has limited contribution to an improved design in terms of safety,

http://dx.doi.org/10.1016/j.istruc.2016.10.002
2352-0124/© 2016 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
106 S.-L. Chan et al. / Structures 9 (2017) 105–111

economy and design efficiency because it considers only partly the non- Selecting a smaller λcr from Eqs. (1) and (2), the amplified moment,
linear and buckling effects. Mamp, is calculated as,
In modern design codes, the requirements for seismic design and
progressive collapse normally go beyond the elastic limit. In many
λcr
steel codes, the second-order direct analysis (SODA) or the direct anal- Mamp ¼ ML ð3Þ
λcr −1
ysis (DA) has been more widely coded as a preferred method. The
Eurocode-3 [11], AISC-LRFD [12] and Code of practice for the structural
uses of steel in Hong Kong [13] further indicate the presence of the three Once we know the effective length, we can determine the slender-
methods as second-order “direct” analysis, second-order “indirect” ness ratio (Le/r) and then we can use the design charts, tables or formu-
analysis and the first-order linear analysis. Note that different codes lae in a code to find the buckling strength pc and buckling resistance Pc
use varied names and the above terminology is adopted from the as,
Hong Kong Steel Code [13].
This paper introduces the theory and development of the new sec-
Pc ¼ pc A ð4Þ
ond-order direct analysis method of design and employs three exam-
ples to demonstrate its applicability and outlines the advantages of
this new method. The detailed beam-column element formulation for in which A is the cross sectional area.
second-order direct analysis can be found in a two-part paper by Chan The ELM above is, strictly speaking, suitable for design of isolated
and Zhou [14]. members since it needs to assume an effective length for each member
and they are assumed to buckle simultaneously and independently
2. The effective length method (ELM) which is far from the actual behaviour. Further, it is not applicable for
design of slender frames because “effective length” is based on unde-
Second-order buckling effects exist whenever a structure is in com- formed geometry and the end nodes of columns in general move with
pression or when it deforms. The effects are more important for slender the loads and the structure is deforming as illustrated in Fig. 1. For argu-
structures and vice versa. The effective length method to reduce the ment sake, if the structure does not deform, it has no displacement, no
buckling resistance is to account for this effect. Fairly speaking, this ap- strain, no stress and no resistance against external loads and therefore
proach is indispensable in the pre-computer age, but becomes more dif- the assumption of using the original geometry for checking is not logi-
ficult to handle design of contemporary structures of irregular geometry cal. The stiffness or sway sensitivity of a frame is then used for assess-
under various scenarios. ment of the effective length factor (Le/r) and amplified moment in Eq.
In the first-order linear analysis used with the effective length meth- (3), but the actual amplification for bending moment is linked to the de-
od (ELM), the critical problem will be on the assessment of the buckling flection of the frame in a nonlinear manner and therefore the method is
strength and the assumption of effective length. Also the linear analysis limited to rather stocky and regular structures which may not be the
is inaccurate near the ultimate limit state that its bending moment and case in practice.
force distribution are insufficiently accurate. The underlying principle for the effective length method is to reduce
The member-based ELM design is only for member design under iso- the resistance of a structure and this is conceptually unacceptable since
lation so its use with frame analysis software is incompatible. To ac- buckling is due to external load and deflection. The mixed use of reduc-
count for system or frame stiffness, some design codes like the ing buckling resistance Pc of a member in Eq. (4) and amplifying the mo-
Eurocode-3 [11] use the classification method of limiting the application ment in Eq. (3) is another inconsistency.
of the method to structures with elastic critical factor (αcr or λcr) less An important message from the simple portal frame in Fig. 1 is the
than 10 or 15 for validity of the first-order linear analysis. fact that the second-order moment is related to the lateral deflection
A building frame is classified as non-sway, sway sensitive and sway- and the axial force as the P-delta moment, the simplified use of effective
ultra-sensitive frames, according to the value of elastic critical load fac- length takes no account of the moment induced by lateral force which
tor, λcr, as, varies continuously with the lateral drift. Therefore the basic assump-
tion with ELM of no pre-buckling deformation until it suddenly buckles
FN h cannot reflect correctly the second-order effects.
λcr ¼ ð1Þ As can be seen in Fig. 2, the stiffness of one bay increases significantly
F V δN
the buckling strength of the complete frame and the buckling resistance
of all vertical columns is affected by the “system” stiffness that the de-
in which FV is the factored dead plus live loads on and above the floor sign of individual columns should not be assessed in isolation.
considered, FN is the notional horizontal force taken typically as 0.5%
of FV for building frames, h is the storey height and δN is the notional hor-
izontal deflection of the upper storey relative to the lower storey due to
the notional horizontal force FN.
For different ranges of elastic buckling load factor, λcr, the structure
is classified as sway and non-sway frames and codes recommend varied P
charts to determine the effective length factor (see, for example, Appen- Δ
dix E in BS5950 [15]).
In the Hong Kong Steel Code [13], an amplification factor as defined
effective length F
in Eq. (2) is further required to ensure the moment amplified along a increasing buckling length
member is properly accounted for. with lateral deflection Δ

π2 EI
λcr ¼ ð2Þ
Fc L2e

where, Le is the effective length, EI is the flexural constant and Fc is the


member compression force due to design loads. Fig. 1. Effective length is taken as the buckling length before deformation.
S.-L. Chan et al. / Structures 9 (2017) 105–111 107

4. Imperfections

A valid second-order analysis must consider imperfections at both


element and frame levels. Ignoring either one of them will provide an
under-designed solution to a design exercise. To this, the following
two aspects of imperfections are required for consideration in an analy-
sis and design software to code requirement.

4.1. Global P-Δ imperfection

Fig. 2. Cross braces affect the buckling resistance of all columns which should not be Linear analysis uses the moment amplification to enlarge the linear
designed in isolation. moment for sway effect, which can be due to wind load or notional
force normally taken as 0.5% for permanent structures and 1% for tem-
porary structures.
3. Type of analysis In second-order analysis, wind load or notional force can still be
used, but an alternative and more reliable and convenient method is
An over view of design and analysis method is shown graphically in to use the elastic buckling mode as the imperfection mode with ampli-
Fig. 3. Various types of buckling loads and analysis represent different tude set equal to the out-of-plumbness normally taken as 1/200 of the
levels of accuracy in modelling. building height for permanent structures or other justified values.
P-Δ-only analysis is the simplest method in the family of non-linear Other imperfection mode assumption such as use of post-buckling
analyses. In this method, the only considered non-linear effect is the shape can be used. When designing a structure of unconventional
change of structural geometry due to nodal displacements. The process shape, the second method of using elastic buckling mode as imperfec-
is to simply add displacements to coordinates of all nodes in a structure tion mode is more rational since the locations and direction of applica-
during an analysis. The member curvature or the P-δ effect is ignored tion of notional force become difficult or controversial to determine.
and the member is assumed to remain perfectly straight in an analysis There have been arguments on which buckling mode should be used
and as a result design formulae in a design code must be applied in and, in general, the actual imperfection mode is random and unknown.
this method for checking the buckling resistance of a member with its Using the elastic buckling mode in the plane of buckling recommended
length either of an effective member length or multiplied by a rationally in Eurocode-3 [11] as imperfection mode is adequate in general and
determined effective length factor (Le/L). studies by the authors show that the difference is small and much less
P-Δ-δ elastic analysis allows for the non-linear effects due to nodal than assuming an incorrect effective length. To this, the program NIDA
displacement and member bowing. The consideration of these two ef- [16] contains options of using a selected elastic buckling mode and
fects is equivalent to reducing the member resistance by the correct using the displacement mode as imperfection mode.
value of effective length which is unfortunately unknown for most
structures. The term “elastic” here indicates that the method disallows 4.2. Member P-δ imperfection
stress or moment re-distribution.
P-Δ-δ plastic analysis allows for buckling and material yielding and it Linear analysis uses various buckling curves to represent different
extends the method of elastic analysis to plastic range utilizing the values of member imperfections and this leads to the production of dif-
mechanism of moment or stress re-distribution after the first-plastic ferent design tables in various notional codes, with the exception of the
hinge formation. This is in line with the ultimate limit state design AISC-LRFD [12] which uses only one buckling curve.
which requires the structural resistance allowing for the effects of buck- In a second-order or P-Δ-δ elastic analysis, the initial curvature can
ling and plastic yielding larger than the factored ultimate load. be pre-set to the code values which can be calibrated from the buckling
Naturally, a more simple method gives a less accurate solution and curves. The member curvature can be varied with an increasing axial
leaving the part of safety check to the manual design part which con- force such that the P-δ effect can be considered correspondingly (see
tains assumption and assessment with lesser theoretical verification. Example 1 of this paper).

Load factor λ
π EI/L =λ
2 2
Elastic critical load= cr cr
Pcr
λ cr=2.2
First-order Second-order P −Δ− only elastic analysis
Linear Analysis

Second-order P −Δ−δ elastic analysis

λ p=1.2
Rigid plastic load=p S=
y λ p

Design resistance or collapse load by


λ U=0.85
Second-order Plastic Analysis
4m where P=1000kN
λ y=0.76 Δ λP
Braced & hori. members 152x152x37

Design resistance by
4m Others : 203x203x60

"First-plastic-hinge" 4m Yield stress : 355MPa

3m
Δ Deflection

Fig. 3. Types of structural analysis.


108 S.-L. Chan et al. / Structures 9 (2017) 105–111

4.3. Structural resistance check

In the codified linear analysis and design, a member is required for


checking against member buckling and sectional strength. In the sec-
ond-order analysis for design, only the section capacity check is re-
quired and checking can be conducted in the following expression.
 
P My þ PΔy þ Pδy ðMz þ PΔz þ Pδz Þ
þ þ ¼ φ≤1 ð5Þ
py A MCy MCz

in which py is the design strength, A is the cross sectional area, My, Mz,
MCy and MCz are respectively the design moment and moment capacity
about the principal axes, P is the axial force, P-Δ and P-δ are the second-
Fig. 4. Modelling of circular and rectangular concrete-filled sections.
order delta moments and ϕ is the section capacity factor.
In dealing with members with possibility of lateral torsional buck- along a member is allowed to have a sectional capacity factor greater
ling, the minor axis moment can include the component of the major than 1 in Eq. (5) as strain hardening is not considered.
axis moment (Trahair and Chan [17]) and the twist angle. The use of one-element per member with a smooth extension of
elastic to plastic analysis represents a consistent design procedure
4.4. Design of concrete in-filled beam-columns from the first-order linear elastic analysis to second-order plastic analy-
sis since the analysis model using one element per member is un-
Concrete in-filled beam-columns have an advantage of higher capac- changed. The consistent design philosophy allows engineers to
ity, economy and elimination of using concrete formworks which al- compare the results by the new and the old method, which allows a
lows faster construction (as the encasing steel tube can function as a simpler evaluation of the advantages of the new approach.
formwork). Its design and analysis are numerically more complicated,
but the concept is the same as bare steel members. To this, the yield
or plastic function shown in Fig. 4 for a typical infilled section is devel- 6. Examples
oped and used to assess the plastic capacity of a section for use in Eq.
(5). The plastic function can be produced by the following equations. Many structural analysis programs claim their capability to do a sec-
ond-order direct analysis. Examples are provided below to confirm their
nL nL nv ðiÞZ uijþ1 Z vðnÞ validity for this sort of analysis. Its application could also be used in con-
Nxc ¼ Σ NLxci ¼ Σ Σ σ ic dudv ð6Þ junction with the section capacity of composite columns (An and Han
i¼1 i¼1 j¼1 uij 0
[18]).

nL nL nv ðiÞZ uijþ1 Z vðnÞ h i


Muc ¼ ρ Σ M Luci ¼ ρ Σ Σ −σ ic ðv þ vn Þ dudv ð7Þ 6.1. Example 1: verification example for testing of element and software
i¼1 i¼1 j¼1 uij 0

The first example is to demonstrate the performance of an element


nL nL nv ðiÞZ uijþ1 Z vðnÞ
Mvc ¼ ρ Σ MLvci ¼ρΣ Σ σ ic ududv ð8Þ to determine the buckling strength and behaviour of a column under
i¼1 i¼1 j¼1 uij 0 compression. The success of the element in capturing the buckling be-
haviour of a column implies a consistent extension on the use of a struc-
in which Nxc, Muc and Mvc are the axial force, bending moments about u- tural model from linear analysis to second-order direct analysis.
axis and v-axis respectively,ρ is equal to +1 when Nxc ≥ 0 and equal to Modelling of a member by several elements to simulate initial imperfec-
− 1 when Nxc b 0; nL is the number of sectional layers; nv(i) denotes tion results not only in unnecessary requirement of longer computer
the number of intersection points in the corresponding layers; v is the
linear equation of the boundary line equal to v(u) − vn, with vn being
the coordinate of the neutral v-axis.
The cross sections are discretized in Fig. 4 and Eqs. (6) to (8) are then
applied to obtain the yield surface.
The method for direct analysis of composite structures is applied to
Example 4 in this paper.

5. First-plastic hinge and plastic analysis (SODEA and SODPA)

In the first-plastic hinge design approach or the second-order direct


elastic analysis (SODEA) where the design load is limited to the load
causing the formation of the first-plastic hinge, the checking of Eq. (5)
is carried out for all members and the design resistance for the complete
structure is then checked and monitored. The design load should be
taken as the load causing the formation of the first-plastic hinge in a
member of the structure. When using the second-order direct plastic
analysis (SODPA), the analysis is continued even after the first-plastic
hinge and the design load is limited only by the load causing the com-
plete structure to collapse as indicated as a limit point or a peak in the
geometry and material nonlinear load vs. deflection plot, which may
occur after formation of a series of plastic hinges. Eq. (5) is used again
for checking and insertion of plastic hinge in a member. Nowhere Fig. 5. Design and analysis of a pin-fix column.
S.-L. Chan et al. / Structures 9 (2017) 105–111 109

Table 1 Member buckling check to non-sway mode effective length under


Comparison of Euler's buckling load and design load capacity. amplified moment.
Methods Euler's buckling load (kN) Design load capacity (kN)
Fc mx Mx my My
Analytical 131 – þ þ ≤1
π2EI/(0.7L)2 Pc Mcx Mcy
NIDA 131 108
BS5950 – 109
For Pc , effective length factor (Le/L) = 1, Le = 10 m

Le Le 10; 000
time, but also in a complication in modelling and an inconsistency in lin- ¼ 1:0; ¼ ¼ 36:1
L r 277
ear and nonlinear computer models for structural analysis.
This example is a column of cross section CHS88.9 × 3.2, length of From design table, the buckling strength pc = 327.9 N/mm2, Pc =
5 m and under an axial force shown in Fig. 5. The results from analytical 327.9 × 17,800/103 = 5837 kN.
method, BS5950 [15] and NIDA [16] are listed in Table 1. The analytical The amplified moment at joint B and C due to P-Δ effect is obtained
Euler's buckling load and design resistance from the code BS5950 [15] as follows.
are respectively 131 kN and 109 kN, compared well with the solutions
for buckling and design resistance by the proposed method and pro-
FN h π2 EI
gram NIDA [16] of 131 kN and 108 kN using only one element in the λcr ¼ smallerof and ¼ smaller of 6:49 and 26:9; use 6:49
FV δN Fc L2E
model. In the design by the present method, no effective length has
been assumed and it gives a similar solution as ELM assuming an accu-
rate effective length factor which is only possible for single columns but λcr 6:49
M¼M ¼ 500  ¼ 591 kN‐m
not for complex and practical frames. For complex structures, the as- λcr −1 6:49−1
sumption of effective length is unreliable whereas the second-order di-
rect analysis (SODA) is still applicable as it is free from effective length Fc mx Mx my My 1000 0:6  591
þ þ ¼ þ ¼ 0:397b1; O:K:
assumption. Pc Mcx Mcy 5837 1573

6.2. Example 2: calibration for design of a simple portal


6.2.2. By second-order direct “elastic” analysis or P-Δ-δ elastic analysis
This example is to check the structural adequacy of the portal shown The section capacity factor ϕ is 0.544. This is close to the value for
in Fig. 6. The section is 686 × 254 × 140 UB of grade S355 steel. The using Le/L = 2.9 above, showing the present design is more economical
frame is restrained out-of-plane, rigid-jointed and pin-supported with and the design is more efficient. Here, the elastic analysis is referred to
dimensions shown in Fig. 6. The frame is designed by the linear, the sec- the first-plastic-hinge design using plastic sectional modulus in mem-
ond-order direct analysis for comparison on their efficiency and ber, but not moment re-distribution after yielding.
accuracy.
6.2.3. By second-order direct “plastic” analysis or P-Δ-δ plastic analysis
6.2.1. By effective length method (ELM) The complete load vs. deflection plot of the portal is shown in Fig. 7.
The maximum bending moment at top of column is 500 kN-m. It can be seen that the collapse load factor is about 1.54. This method of
The section capacity check in a linear analysis is carried out as fol- design, when applied to a larger structure, normally requires longer
lows. computing time as the load steps are required to be smaller. But this
computer time is justifiable in modern computers. Here, the factor of
safety is 1.54 which indicates the portal is structurally adequate with
Fc Mx My 1000  103 500 similar findings by methods above.
þ þ ¼ þ ¼ 0:476b1; O:K:
Py Mcx Mcy 355  17; 800 1573

The member check using the effective length method

Fc mx Mx my My 1000 0:6  500


þ þ ¼ þ ¼ 0:548b1; O:K:
Pc Mcx Mcy 2800 1573

Fig. 6. The portal designed by different methods. Fig. 7. The load vs. deflection plot in plastic analysis.
110 S.-L. Chan et al. / Structures 9 (2017) 105–111

Fig. 8. The structure under construction.

Fig. 10. Utilization plot for all members.


6.3. Example 3: design of a practical irregular steel frame

The structure shown in Fig. 8 is designed against permanent, im- 6.4. Example 4: design of composite columns and composite truss
posed, wind and seismic loads as well as temperature loads. As can be
seen in the design process, the checking was completed by a fraction The same concept can be applied to analysis and design of com-
of time required for a manual design with accuracy improved and posite frames. However, the yield and failure surfaces for a compos-
weight reduced. If one uses the ELM, the material weight could be in- ite section are more complicated and allowance must be made for
creased by 23% with some critical members under designed due to weak tension stress in concrete. Fig. 11 shows typical yield and fail-
error in assuming an effective length. ure surfaces of a composite column used to control the formation
The computer model and the load vs. deflection plot are shown in of plastic hinges.
Fig. 9. Totally 32 combined load cases have been considered with 35 sec- The results using the proposed method agree well with test, as seen
tion types used in the structure. The plastic hinges and the load vs. de- in Fig. 12. Below is a typical load deflection plot of a composite concrete
flection plots are indicated for the permanent and imposed load infilled square column and the full set of comparison is reported by Liu
combined load case in the plot diagram inside Fig. 9. The collapse load et al. [19]. The application was further carried out in the design of a 4
factor is about 2.05 under this particular load case. level composite truss at Tamar Government Headquarters in Hong
This can be seen that the SODPA (second-order direct plastic design) Kong with significant saving of material.
can be done in a very convenient manner using the same model for the
first-order linear and second-order direct elastic analysis (SODEA) be- 7. Conclusions
cause refinement of finite element model of using several elements
per member is not needed. To date, the second-order direct analysis (SODA) is a tool for direct
For practical design for this frame, the checking for section capacity design of steel and composite structures without assumption of effec-
factor was done at design load factor equal to 1. For SODEA, no member tive length. However, most engineers only use the P-Δ-only type of sec-
should have a section capacity greater than 1 and if this happens, the ond-order analysis for finding of the sway P-Δ moment which is only
member size needs to be increased. The utilization ratio is also shown the primitive P-Δ-only analysis named by ASIC-LFRD [12] and Hong
in Fig. 10 with every spot representing the section capacity factor for a Kong Steel Code [13]. This paper presents an introduction to this new
member in a particular combined load case. When some section types and better design method with its advantages illustrated. It is evidenced
are under-utilized, their size or thickness can be reduced for material that the key issues for the practicality of the design method lie on (1) ac-
weight saving. Although the section capacity check is also available in ceptable accuracy of using one element per member with initial curva-
software using ELM, the checking becomes meaningless because of ture set equal to initial imperfection of a member specified in various
the gross error associated with the error of assumed effective length. codes to model a member and (2) automatic use of buckling mode as

Fig. 9. The load vs. deflection plot allowing for plastic hinge and buckling effects with red spots representing plastic hinges.
S.-L. Chan et al. / Structures 9 (2017) 105–111 111

(PolyU 152012/14E)” and “Second-Order Analysis of Flexible Steel Cable


Nets Supporting Debris (PolyU 152008/15E)”. This last author would
like to appreciate the financial support by the Faculty of Construction
and Environment through the project “FCE Postdoctoral Fellow Scheme”.

References

[1] Liew JYR, Chen WF. Implications of using refined plastic hinge analysis for load and
resistance factor design. Thin-Walled Struct 1994;20(1–4):17–47.
[2] Liew JYR, Chen WF, Chen H. Advanced inelastic analysis of frame structures. J Constr
Steel Res 2000;55(1–3):245–65.
[3] Izzuddin BA, Smith DL. Large-displacement analysis of elastoplastic thin-walled
frames. Parts 1 and 2. J Struct Eng ASCE 1996;122(8):905–25.
[4] Chen WF, Chan SL. Second order inelastic analysis of steel frames using element with
mid-span and end springs. J Struct Eng ASCE 1995;121(3):530–41.
[5] White DW, Hajjar JF. Design of steel frames without consideration of effective
length. Eng Struct 1997;19(10):797–810.
[6] Yang YB, Kuo SR. Theory and analysis of nonlinear framed structures. N.Y.: Prentice
Hall; 1994.
[7] Chan SL, Chui PPT. Non-linear static and cyclic analysis of semi-rigid steel frames.
Elsevier Science; 2000 336.
[8] Albermani FGA, Kitipornchai S. Numerical simulation of structural behaviour of
transmission towers. Thin-Walled Struct 2003;41(2–3):167–77.
[9] Chan SL, Cho SH. Second-order P-delta analysis and design of angle trusses allowing
for imperfections and semi-rigid connections. Int J Adv Steel Constr 2005;1(1):
157–72.
[10] Clarke MJ, Hancock GJ. A study of incremental-iterative strategies for nonlinear anal-
ysis. Int J Numer Methods Eng 1990;29:1365–91.
Fig. 11. The concrete crushing, yield and failure.
[11] CEN, Eurocode 3-1-1. Design of steel structures. European Standard; 2005.
[12] AISC. Load and resistance factor design specification for structural steel buildings.
Chicago: American Institute of Steel Construction; 2011.
[13] Code of practice for structural uses of steel. Hong Kong SAR Government: Buildings
Department; 2011.
initial imperfection mode. Both of these considerations have been [14] Chan SL, Zhou ZH. Elastoplastic and large deflection analysis of steel frames by one
allowed for in various steel design codes. The design software NIDA, element per member. Parts 1 and 2. J Struct Eng ASCE 2004;130(4):538–53.
which has been used for a decade in Hong Kong and Macau region of [15] BSI. Structural use of steelwork in building – part 1: code of practice for design –
rolled and welded sections, BS5950. London: BSI; 2000.
China, accumulated extensive practical and theoretical experience that [16] NIDA. Non-linear integrated design and analysis user's manual, NAF-NIDA series,
it could assist to design safer, more economical and sustainable struc- version 9; 2015[http://www.nidacse.com].
tures as illustrated in the examples in this paper. [17] Trahair NS, Chan SL. Out-of-plane advanced analysis of steel structures. Eng Struct
2003;25:1627–37.
[18] An YF, Han LH. Behaviour of concrete-encased CFST columns under combined com-
Acknowledgement pression and bending. J Constr Steel Res 2004;101:314–30.
[19] Liu SW, Liu YP, Chan SL. Advanced analysis of hybrid steel and concrete frames,
part 2: refined plastic hinge and advanced analysis. J Constr Steel Res 2012;70:
The authors are grateful to the financial supports by the Research 337–49.
Grant Council of the Hong Kong SAR Government on the project “Sec- [20] Bridge RQ. Concrete filled steel tubular columns research report no. R283. Sydney
ond-order and Advanced Analysis of Arches and Curved Structures University; 1976.

Fig. 12. The load vs. deflection plot of a composite column with validation with experiment by Bridge [20].

Das könnte Ihnen auch gefallen