This document is downloaded from DR‑NTU (https://dr.ntu.edu.

sg)
Nanyang Technological University, Singapore.

Residual stress of cold‑formed thick‑walled steel

rectangular hollow sections

Zhang, Xingzhao; Liu, Su; Zhao, Mingshan; Chiew, Sing‑Ping

2016

Zhang, X., Liu, S., Zhao, M., & Chiew, S.‑P. (2016). Residual stress of cold‑formed
thick‑walled steel rectangular hollow sections. Steel and Composite Structures, 22(4),
837‑853. doi:10.12989/scs.2016.22.4.837

https://hdl.handle.net/10356/107125

https://doi.org/10.12989/scs.2016.22.4.837

© 2016 Techno‑Press, Ltd. All rights reserved. This paper was published in Steel and
Composite Structures and is made available with permission of Techno‑Press, Ltd.

Downloaded on 06 Feb 2024 13:19:41 SGT

Steel and Composite Structures, Vol. 22, No. 4 (2016) 837-853
DOI: http://dx.doi.org/10.12989/scs.2016.22.4.837 837

Residual stress of cold-formed thick-walled

steel rectangular hollow sections
Xingzhao Zhang 1, Su Liu 1, Mingshan Zhao ∗2, and Sing-Ping Chiew 3
1
School of Architecture, Hunan University, P.R. China
2
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore
3
Singapore Institute of Technology, Singapore

(Received September 19, 2015, Revised October 18, 2016, Accepted October 31, 2016)

Abstract. This paper presents the experimental and numerical study on the distribution of transverse and
longitudinal residual stresses in cold-formed thick-walled structural steel rectangular hollow sections manufactured
by indirect technique. Hole-drilling method is employed to measure the magnitude of the transverse and longitudinal
surface residual stress distribution, and the effects of the residual stresses are evaluated qualitatively by sectioning
method. It is shown that compared to normal cold-formed thin-walled structural hollow sections (SHS), the cold-
formed thick-walled SHS has similar level of residual stress in the flat area but higher residual stresses in the corner
and welding areas. Both the transverse and longitudinal residual stresses tend to open the section. In order to predict
the surface residual stresses in the corners of the cold-formed thick-walled SHS, an analytical model is developed. 2D
finite element simulation of the cold bending process is conducted to validate the analytical approach. It is shown that
in analyzing bending for thick-walled sections, shifting of neutral axis must be considered, since it would lead to non-
linear and non-symmetrical distribution of stresses through the thickness. This phenomenon leads to the fact that
cold-formed thick-walled SHSs has different distribution and magnitude of the residual stresses from the cold-formed
thin-walled SHSs.
Keywords: cold-formed; thick-walled; structural hollow section; residual stress; analytical model

1. Introduction

Cold-formed steel structural hollow sections (SHS) are widely used as load-bearing members in
welded frame structures. In early stage, common rectangular hollow sections could only be
fabricated as thin-walled sections with thickness ranging from 0.4 to 6 mm (Tong et al. 2012). For
sections with increased thickness, it was difficult to control the through thickness properties at the
cold-forming zones. Theoretically speaking, thin-walled or thick-walled, cold-formed SHSs
(referring to rectangular hollow sections in this paper) are formed by rolling an annealed flat strip
by direct or indirect methods. The direct method bends the strip into the final SHS shape directly,
while the indirect method rolls the flat strip into circular hollow section first and further rolls it
into the rectangular shape (Gardener et al. 2010). The manufacturing processes usually do not
include post-forming stress relief. As a result, there remains complex distributions of yield strength
and residual stress around the section (Key and Hancock 1993). The existence of residual stresses

∗Corresponding author, Ph.D., E-mail: [email protected]

Copyright © 2016 Techno-Press, Ltd.

http://www.techno-press.org/?journal=scs&subpage=6 ISSN: 1229-9367 (Print), 1598-6233 (Online)
838 Xingzhao Zhang, Su Liu, Mingshan Zhao and Sing-Ping Chiew

is mainly due to strain hardening, welding, flame cutting and heat treatment (Abambres and Quach
2016). Through decades of study and practice, the importance of residual stress for designing and
analyzing steel structures has been well recognized, as it is frequently associated with issues such
as brittle fracture, fatigue, stress corrosion, buckling and post-buckling strength reduction (Wang et
al. 2012, Ma et al. 2015). Common design standards have taken the effect of residual stresses into
consideration implicitly (AISC 2005, BSI 2005). However, there are few specific guidelines on
designing and evaluating the distribution of the residual stress directly.
Since the residual stress distribution in cold-formed sections are complex and highly dependent
on the forming techniques, material properties and section profile, experimental approach is more
frequently adopted than theoretical approach in the investigation (Withers et al. 2008). For residual
stress contained in cold-formed thin-walled SHSs, numerous work has been done (Moen et al.
2008, Spoorenberg et al. 2010, Jandera and Machacek 2014). It is well recognized that the
longitudinal residual stresses are in tension at outer surface and in compression at inner surface,
and the distribution is assumed to be linear through the thickness (Mashiri et al. 2014). However,
studies on the cold-formed thick-walled plate subjected to bending (Weng and White 1990, Key
and Hancock 1993) show that the through thickness residual stress distribution pattern is non-
linear. Tong et al. (2012) measured the longitudinal residual stress distribution of a few thick-
walled cold-formed SHSs and found out the magnitude of the residual stress at the corner was
lower than that at flat area, which is also different from the common cold-formed thin-walled
SHSs. Up to date, the literature on the residual stress distribution in cold-formed thick-walled
SHSs is still limited, especially for residual stresses in the transverse direction (Li et al. 2009,
Gardener et al. 2010, Ma et al. 2015).
The main objective of this study is to investigate the distribution of transverse and longitudinal
residual stresses contained in cold-formed thick-walled SHSs by both experimental and numerical
methods. In the experimental investigation, the geometrical and mechanical properties of the
selected cold-formed thick-walled SHS are firstly obtained. Subsequently, the hole-drilling
technique is used to quantitatively measure the magnitude of residual stresses on the outer surface
in both transverse and longitudinal directions. Further, the effect of the locked-in residual stress is
qualitatively evaluated by the sectioning method. Comparison between the experimental findings
and existing literature is also presented. In the second phase, numerical investigation is conducted.
An analytical elastic-plastic large displacement model is developed to predict the surface residual
stress distribution in the corner. 2D finite element (FE) is also employed to simulate the forming of
residual stress in the same area. By comparing the strain and stress distributions in bending and the
following springback stage of these two models, the proposed analytical model is validated.

2. Experimental investigation

2.1 Geometrical dimensions

As shown in Fig. 1, the following characteristic dimensions are measured for the tested SHS:
width b and wall thickness t of every face, inner corner radii (ri) and outer corner radii (ro) of four
corners. Based on the mean values of the above dimensions, the main geometrical parameters
including wall slenderness ratio bm/tm, mean center corner radii rc,m, mean outer corner radii related
to mean wall thickness ro,m/tm and mean center corner radii related to mean wall thickness rc,m/tm
are calculated, as shown in Table 1. Theoretically, the differences between thin-walled and thick
 Residual stress of cold-formed thick-walled steel rectangular hollow sections 839

Face 3
Ro
C3 C2
Ri

Face 4 Face 2
h

Weld Seam

C4 C1
Face 1
b

(a) Definition of the measured geometrical (b) The tested cold-formed SHS (left) in comparison
dimensions with a hot-formed SHS (right)
Fig. 1 Definition of the measured geometrical dimensions and the tested SHS

Table 1 Characteristic geometrical dimensions of the tested SHS

Hollow sections bm (mm) tm (mm) ro,m (mm) ri,m (mm) bm/tm ro,m/tm rc,m/tm
Cold-formed 200.53 12.76 31.00 18.75 15.72 2.43 2.49
EN 10219 200±1.6 12.5±0.5 2.4 to 3.6

walled SHSs come from the rolling process only. There is no strict criterion for determining
whether a cold-formed SHS is thin-walled or thick-walled. For common cold-formed SHSs, the
center corner radius over thickness ratio (rc/t) is restricted to be within 2 to 6 (AISI 1996). Any
section with rc/t close to 2 would behave like the thick-walled, while that with rc/t close to 6 would
behave similar to the thin-walled.

2.2 Mechanical properties

Standard coupon specimens are cut from the center areas of Faces 2, 3 and 4 (Fig. 2) and tested
according to EN 10002-1 (BSI 2001). The dimensions of the specimens are shown in Fig. 2. Non-
proportional gauge length of 80mm is used as the original gauge length. For conversion of
elongation values from non-proportional gauge length to a proportional gauge length 5.65 So ,
the conversion tables from BS EN 2566-1 (BSI 1999) is applied. During tests, both strain gauge
and extensometer are used for stress-strain relationship measurement. The loading rate is set as 1
mm/min and stress-strain data points are captured at frequency of 1 Hz.
The stress-strain curves of the tested coupon specimens are shown in Fig. 3 in comparison with
a typical curve of hot-formed S355J2H. The characteristic mechanical properties of all the tested
specimens are shown in Table 2, in comparison with the requirements by the corresponding
product standard EN 10219 (BSI 2006). It can be seen from Fig. 3 and Table 2 that the mechanical
properties of the cold-formed and hot-formed SHSs differ remarkably, although they consist of
similar contents of elements. Firstly, the cold-formed SHS literally fulfills the mechanical property
specifications of S460NH. Due to the strain hardening by cold forming, the average yield strength
of the cold-formed SHS is 27.1% higher than that of the hot-formed SHS. However, this
improvement in strength is obtained at the expense of ductility. The cold-formed SHS does not
840 Xingzhao Zhang, Su Liu, Mingshan Zhao and Sing-Ping Chiew

T-2
C3 C2

T-3 T-1

Weld Seam
C4 C1
Fig. 2 Sources and dimensions of tensile test specimens

600

500
Stress (MPa)

400

300
Cold-Formed-F2
200 Cold-Formed-F3
Cold-Formed-F4
100
Hot-Formed S355J2H
0
0 5 10 15 20 25 30 35
Strain (%)
Fig. 3 Stress-strain curves of the tested SHS

Table 2 Summary of the tensile test results

fy Ave. fu Ave. Ave. Tensile Elongation
SHS E (GPa) Ave.
(MPa) (MPa) (MPa) (MPa) (GPa) ratio (%)
529.3 562.2 208.3 1.06 21.5
Cold-formed 511.0 521.1 551.6 559.2 209.2 208.4 1.08 1.07 23.6
523.0 563.7 207.7 1.08 21.8
EN 10219 ≥ 540
≥ 460 ≥17
S460NH ≤ 720
Hot-formed 410.6 531.0 208.7 1.29 31

show obvious yield plateau and both tensile ratio and elongation at fracture of the cold-formed
SHS is remarkably lower than the hot-formed SHS.

2.3 Residual stress

2.3.1 Test method

In this study, the hole-drilling method is employed to quantitatively measure the residual
stresses along the perimeter of the SHS. RS200 milling guide and electric strain gauge FRAS-2-11
are employed to measure the residual stresses released by hole drilling. Test positions are arranged
 Residual stress of cold-formed thick-walled steel rectangular hollow sections 841

with equal distance on all the surfaces, and extra test points are added at corners and welding areas.
The arrangement of the drilling points on the SHS is shown in Fig. 4(a). During the test, a hole
with diameter of 2 mm and depth of 2 mm is drilled at the designated position on the special strain
gauge rosette by 8 steps. The released strains at the end of each step are recorded for further
analyzing. By comparing the strains before and after hole-drilling, stress relaxation due to hole
drilling can be determined. On the assumptions that the material is homogeneous and isotropic and
the stress-strain curve is linear, the relieved strain at the tested point can be obtained by
substituting the stress relaxation into the Hooke’s Law (ASTM 2008)

1 + 𝑣𝑣 𝜎𝜎∅ + 𝜎𝜎𝜃𝜃 1 𝜎𝜎∅ − 𝜎𝜎𝜃𝜃 1

𝜀𝜀 = 𝑎𝑎� + 𝑏𝑏� cos2𝛼𝛼 + 𝑏𝑏�𝜏𝜏∅𝜃𝜃 sin2𝛼𝛼 (1)
𝐸𝐸 2 𝐸𝐸 2 𝐸𝐸

In Eq. (1), E is the Young’s modulus, 𝑣𝑣 is the Poisson’s ratio and 𝜀𝜀 is the relieved strain. 𝜎𝜎∅
and 𝜎𝜎𝜃𝜃 are the stresses in the ∅ and 𝜃𝜃 directions (Fig. 4(b)), respectively. 𝛼𝛼 is the angle from
the x-axis to the maximum principle stress, and 𝑎𝑎� and 𝑏𝑏� are almost material-independent
calibration constants indicating the relieved strains due to unit stresses within the hole depth. Since
the holes are drilled directly on the surface of the tubes (Fig. 4(b)) and the stress/strain status of the
SHS can be treated as plane strain condition, 𝜎𝜎∅ and 𝜎𝜎𝜃𝜃 correspond to the transverse and
longitudinal stresses. It should also be noted that the relieved strains are mostly influenced by the
near-surface residual stresses. Interior stresses have influences that diminish with their depth from
the surface (ASTM 2008). For the strain rosettes employed in this study (Fig. 4(b)), the sensitivity
diminishes to near zero for stresses beyond 1mm depth. Therefore, the measurement actually
indicates a weighted average of the residual stresses within the near-surface layer, i.e., 1 mm deep
from the measured surface. Besides, the surface strains that will be relieved by drilling a hole
depend only on the stresses that originally existed at the boundaries of the hole. The stresses
beyond the hole boundary do not affect the relieved strains, even though the strains are measured
beyond the hole boundary. In this study, the outer radius, 31.0 mm, is relatively large compared to
the hole size. Therefore, it is assumed that Eq. (1) is still valid for the corners.
Besides the hole-drilling method, the sectioning test is also employed to evaluate the effect of
the residual stresses. The sectioning test is carried out on SHS specimens with length of 200 mm.
Three cuts with equal spacing and depth (170 mm) are applied on the flat area of each face, as
shown in Fig. 5. By measuring the changes in the geometry before/after sectioning, the effects of
the locked-in residual stress are evaluated.

Face-2
7 6 5 4
8 45 ▲
3
45
▲

46.66 46.66 46.67

9 2
46.67

10 1
46.66

Face-3 Face-1
11 20
46.67

12 19

13 18
14 15 16 17
Face-4

(a) Test positions (b) Strain gauge rosette and coordinate system of stresses
Fig. 4 Hole drilling test
842 Xingzhao Zhang, Su Liu, Mingshan Zhao and Sing-Ping Chiew

180

30 60 60
180

(a) Cutting positions (b) Sectioning in progress

Fig. 5 Sectioning test

249.9 （48.1%）
291.6 （56.1%）
129.4 （24.9%）

155.7（29.9%）
-89.4 (-17.2%)
168.1 (32.2%)

148.2 (28.5%)
-45.6 (-8.8%)

-30.6 (-5.9%) -41.9 (-8.1%) 163.9 （31.5%） 103.5（19.9%）

-58.8 (-11.3%) -33.9 (-6.5%)

175.5（33.8%） 169.9 （32.7%）
400
300
200
200
149.9 (28.8%) 100 113.4 (21.8%) 292.5 （56.3%） 100 283.3 （54.5%）
Outer Surface Outer Surface
-100 -100

90.5 (17.4%) 127.3 (24.5%) 314.4 （60.5%） 299.2 （57.5%）

-39.4 (-7.6%) -104.5 (-20.1%) 136.9（26.3%） 135.2（26.0%）

-60.0 (-11.5%) 148.4 （28.5%） 151.0 （29.0%）

-74.9 (-14.4%)
162.5 (31.3%)
5.7
57.5 (11.1%)

112.5 （21.6%）
-28.8

-89.3 (-17.2%)

395.6 （76.1%）
351.9 （67.7%）
163.0（31.3%）

414.9 （79.8%）
(1.1%)
(-5.5%)

(a) Residual stress in the transverse direction (b) Residual stress in the longitudinal direction
Fig. 6 Residual stress distribution (MPa and %) of the cold-formed SHS

2.3.2 Test results

The residual stress distributions in both transverse and longitudinal directions of the cold-
formed SHS are shown in Fig. 6. The results for the flat area, corner and welding zone are also
summarized in Table 3. In the above figures and tables, residual stresses are evaluated using both
absolute value (MPa) and percentage over the yield strength (residual stress/actual yield strength ×
100%). It can be seen from Fig. 6 that the distributions of the residual stress in both directions are
generally symmetrical about the neutral axes.
In the transverse direction, the residual stresses are compressive stresses at the corners and
tensile stresses elsewhere. The residual stress levels of the corner range from -20.1% to -5.9%;
those of the flat area range from 17.4 to 32.2%; while those of the welding zone range from 1.1%
to 31.3%. On the general level, the stress level is relatively low. Only 2 out of 21 positions show
residual stress level higher than 30%. Based on the weighted average stresses in different zones,
 Residual stress of cold-formed thick-walled steel rectangular hollow sections 843

100 Longitudinal Stress Tong et al. (2012)

Transverse Stress Poposed Longitudinal stress 50

Thickness (%)
18%
Proposed Transverse Stress
75

52%
50
Stress (%)

25
0
-100 -50 0 50 100

0 𝜎𝜎∅⁄𝑓𝑓𝑦
0 3 6 9 12 15 18 21
−52%
-25

C2 C3 C4 C1
-50 -18% -50
Position (No.)

(a) Surface residual stress distribution (b) Through thickness stress

distribution (Tong et al. 2012)
Fig. 7 Residual stress distribution (%) of the cold-formed SHS

a prediction model is proposed, as shown in Fig. 7(a) and Table 3. To maintain the stress
equilibrium through the thickness, it is obvious that the residual stresses at the inner surface of the
SHS would be tensile stresses at the corners and compressive stresses elsewhere. It is predictable
that this distribution would tend to open the section.
In the longitudinal direction, the residual stresses are all tensile stresses and the stress level is
generally higher than the level of transverse residual stresses. It can be seen from Table 3 that the
residual stress levels of the corner vary from 19.9% to 32.8% while most of the rest areas have
residual stress level higher than 50%. Further, the longitudinal residual stress distribution is
compared with the surface longitudinal residual stress model for indirectly cold-formed SHS
proposed by Tong et al. (2012), as shown in Fig. 7. Tong et al. (2012) suggested values of 55%
and 18% for the surface longitudinal stresses for the corner and flat area, respectively, and a
symmetrical through thickness distribution (Fig. 7(b)). From the comparison (Fig. 7(a)), it is
shown that this prediction model agrees well with the test results for the flat area, but is too low for
the corner and welding zone. The possible deduction is that the model proposed by Tong et al.
(2012) underestimates the stress level for thick-walled cold-forming processes and the residual
stress induced by highly localized welding heat input. In fact, the residual stress caused by welding
should be much higher than the flat area, and the assumption of symmetrical stress distribution in
the thickness direction (Fig. 7(b)) is no longer accurate for thick-walled cold formed members.
Herein, a new model is proposed for the distribution of surface longitudinal residual stresses in
thick-walled cold-formed SHS. Special caution is given to the welding zone and corner area, as
shown in Fig. 7(a) and Table 3.
Fig. 8(a) shows the specimen before and after sectioning. It can be seen from Fig. 8(a) that the
section opens after sectioning and the section width increases by 4.5%. Both the through thickness
transverse and longitudinal residual stresses contribute to this phenomenon through the membrane
and bending actions (Gardener et al. 2010). While the transverse residual stresses are more prone
to open the section (Fig. 6(a)), the longitudinal residual stresses are more likely to bend the section
wall. As shown in Fig. 8b, these 300mm long and 70mm wide coupons cut from the flat area show
convexity of about 1% after cutting.
844 Xingzhao Zhang, Su Liu, Mingshan Zhao and Sing-Ping Chiew

Table 3 Residual stress in the transverse direction

Flat area Proposed Corner Proposed Welding zone Proposed
(%) (%) (%) (%) (%) (%)
Transverse 17.4% to 32.2% 26% -20.1 to -5.9% -11% 1.1% to 31.3% 15%
Longitudinal 48.1% to 60.5% 55% 19.9% to 33.8% 28% 67.7% to 79.8% 75%

(a) Sectioning test (a) Coupons bent due to residual stresses

Fig. 8 Sectioning test results of the cold-formed thick-walled SHS

3. Analytical model for residual stress induced by cold bending

3.1 Theory of residual stress formed during bending

Despite that there are many technologies to form a structural SHS, the residual stresses are
always the result of mechanical and thermal actions. For cold-formed SHS, residual stresses can be
induced by (1) cold bending; (2) welding; and (3) flame cutting (Ma et al. 2015). Since the
welding and flame cutting only affect the material locally, this study focus on the residual stress
formed by cold bending only. Fig. 9 shows a typical roll-forming process for the corners of
rectangular SHSs. During bending, the material between the roller die reactions is expected to
undergo certain level of yielding as the stress distribution transits from elastic to plastic. After the
plate becomes fully plastic, the engineering strain continues to increase as the rolling radius keeps
increasing. When the final bending radius is reached and the imposed radial displacement is
removed, an elastic springback occurs and unload the corner. At the meantime, certain level of
stress is locked-in as residual stress (Moen et al. 2008). It should be pointed out that in the above
processes, the distribution of residual stress is dependent only on the geometry and mechanical
properties including the yield strength/tensile strength and elastic/plastic modulus (Liu et al. 2015).
Compared to the cold-formed SHS, the hot-formed SHSs are less concerned regarding the residual
stress because the hot-formed SHSs are rolled at high temperature conditions where both the
strengths and elastic/plastic modulus are significantly lower than at room temperature.
In this study, a general analytical model for predicting the surface residual stresses in the
corners of the cold-formed thick-walled SHSs is proposed. In the case of thick-walled (small
radius) bending, many commonly used beam analysis assumptions are no longer valid such as the
symmetrical through thickness stress distribution assumption, middle surface neutral axis
assumption and engineering strain method (Amouzegar et al. 2016). Therefore, a series of
analytical assumptions must be defined first.
 Residual stress of cold-formed thick-walled steel rectangular hollow sections 845

𝜎𝜎𝑟
𝑀𝑀𝑖𝑖𝑏𝑏𝑏𝑏𝑙𝑙𝑏𝑏 𝑠𝑠𝑟𝑟𝑟𝑟𝑓𝑓𝑎𝑎𝑐𝑐𝑏𝑏 𝜎𝜎𝜃 𝑡𝑡

𝑟𝑟𝑖 = 𝑎𝑎
𝑁𝑁𝑏𝑏𝑟𝑟𝑡𝑡𝑟𝑟𝑎𝑎𝑙𝑙 𝑎𝑎𝑥𝑖𝑖𝑠𝑠

M 𝑟𝑟𝑐 𝑟𝑟𝑜 = 𝑏𝑏
M

Fig. 9 Roll-forming for corners of SHS Fig. 10 Stress-strain coordinate system

3.2 Assumptions for the analytical model

The stress-strain coordinate system is shown in Fig. 10. Based on this system, the following
assumptions are established.
(1) This problem is simplified as plane strain problem and the thickness of the section remain
the unchanged throughout the process. On the inner and outer surfaces, the following
boundary conditions apply: for 𝑟𝑟 = 𝑟𝑟𝑖𝑖 and 𝑟𝑟 = 𝑟𝑟𝑜𝑜 , 𝜎𝜎𝑟𝑟 = 0. For further simplification, the
through thickness stresses in the radial direction are all neglected, i.e., 𝜎𝜎𝑟𝑟 = 0 for
𝑟𝑟𝑖𝑖 < 𝑟𝑟 < 𝑟𝑟𝑜𝑜 . This assumption avoids the tri-axial stress status analysis, since the radial
stress is only significant near the neutral axis. This approach is widely adopted in the
measurement of residual stresses by techniques such as hole drilling and sectioning
methods (Ma et al. 2015).
(2) Logarithmic stress-strain (true stress-strain) relationship is adopted in the analysis, since
large deformation is involved. As indicated by Yu and Zhang (Yu and Zhang 1996), for
pure bending of wide plates with center line bending radius 𝑟𝑟𝑐𝑐 ≤ 10 𝑡𝑡, the adopting of true
stress-strain is necessary.
(3) The neutral axis remains unchanged during cross-section bending. For large curvature
bending analysis, the neutral axis no longer coincides with the middle surface. It is slightly
shifted to the inner surface side and the distance would further increase when plasticity
takes place (Zhu 2007). In this study, only the initial position of the neutral axis is
calculated and the second shift caused by plasticity is ignored to simplify the derivation.
(4) This approach only predicts the residual stresses formed by cold-bending, which are
between the roller die reactions (Fig. 9). In reality, some yielding is expected to occur
outside the roller reactions as the stress distribution transits from fully plastic to fully
elastic, and the stress distribution may not be uniform throughout the corners.

3.3 Residual stress from cross-section bending

It is common to use simplified stress-strain curves such as linear elastic-perfectly plastic and
elastic-plastic with linear strain hardening in theoretical study, as shown in Fig. 11(a). The
following effective stress (𝜎𝜎�) and effective strain (𝜀𝜀̅) relationship is firstly established:

𝜎𝜎� = 𝑓𝑓𝑦𝑦 + 𝐸𝐸𝑝𝑝 �𝜀𝜀̅ − 𝜀𝜀𝑦𝑦 � = �𝐸𝐸 − 𝐸𝐸𝑝𝑝 �𝜀𝜀𝑦𝑦 + 𝐸𝐸𝑝𝑝 𝜀𝜀̅ (2)
846 Xingzhao Zhang, Su Liu, Mingshan Zhao and Sing-Ping Chiew

𝑦𝑦 𝑦𝑦
𝑡𝑡⁄2 𝑡𝑡⁄2
𝑓𝑓𝑢
Loading

𝑓𝑓𝑦 𝐸𝐸 𝑝 Plastic

𝐸𝐸 𝐸𝐸 Unloading
Geometrical Center
S
Neutral Axis Elastic
Residual
stress
Plastic
𝜀𝜀𝑦 𝜀𝜀𝑚 −𝑡𝑡⁄2 −𝑡𝑡⁄2
(a) Stress-strain model (b) Transverse stress
Fig. 11 Stress-strain behavior during bending

where 𝑓𝑓𝑦𝑦 and 𝜀𝜀𝑦𝑦 are the true stress and strain at yielding, respectively. 𝐸𝐸 and 𝐸𝐸𝑝𝑝 are the elastic
and plastic modulus, respectively. Based on the engineering stress-strain curves (Fig. 3), the true
stress-strain relationship is obtained: 𝑓𝑓𝑦𝑦 = 521.1 MPa, 𝜀𝜀𝑦𝑦 = 0.00251 and 𝐸𝐸𝑝𝑝 = 585.1 MPa.
Under the situation of pure bending for curved beam system (Fig. 10), there are two possible
patterns of stress/strain distribution over the cross-section, as shown in Fig. 11(b).
(a) When the cross section is not yet yielded, the non-linear stress distribution can be
expressed by (Timoshenko and Goodier 1970)

4𝑀𝑀 𝑎𝑎2 𝑏𝑏 2 𝑏𝑏 𝑟𝑟 𝑎𝑎
σ𝜃𝜃 = − �− 2 𝑙𝑙𝑙𝑙𝑙𝑙 + 𝑏𝑏 2 𝑙𝑙𝑙𝑙𝑙𝑙 + 𝑎𝑎2 𝑙𝑙𝑙𝑙𝑙𝑙 + 𝑏𝑏 2 − 𝑎𝑎2 � (3a)
𝑁𝑁 𝑟𝑟 𝑎𝑎 𝑏𝑏 𝑟𝑟

4𝑀𝑀 𝑎𝑎2 𝑏𝑏 2 𝑏𝑏 𝑟𝑟 𝑎𝑎
σ𝑟𝑟 = − � 2 𝑙𝑙𝑙𝑙𝑙𝑙 + 𝑏𝑏 2 𝑙𝑙𝑙𝑙𝑙𝑙 + 𝑎𝑎2 𝑙𝑙𝑙𝑙𝑙𝑙 � (3b)
𝑁𝑁 𝑟𝑟 𝑎𝑎 𝑏𝑏 𝑟𝑟

𝜏𝜏𝑟𝑟𝑟𝑟 = 0 (3c)
𝑏𝑏
where 𝑁𝑁 = (𝑏𝑏 2 − 𝑎𝑎2 )2 − 4𝑎𝑎2 𝑏𝑏 2 (𝑙𝑙𝑙𝑙𝑙𝑙 )2 is a constant. As stated previously, the σ𝑟𝑟 is
𝑎𝑎
only significant near the neutral axis and equals to zero on the surface. Therefore, it is
neglected in this study. By substituting geometrical data into (3a), the neutral axis position
(shift from middle surface by s) can be easily determined by solving σ𝜃𝜃 (𝑟𝑟) = 0. Besides,
since 𝜏𝜏𝑟𝑟𝑟𝑟 = 0, σ𝜃𝜃 and σ𝑟𝑟 become principal stresses in the bending plane (Fig. 10)
naturally.
(b) When the outer and inner surfaces are yielded while the core remains elastic, an arbitrary
point in the sheet undergoes transverse straining with the amount depending on its location
y away from the current neutral axis (Fig. 11(b)). By the assumption of σ𝜃𝜃 = 0 at the
neutral axis, the true transverse strain due to cold bending can be obtained (Quach et al.
2004)
𝑠𝑠 − 𝑦𝑦
ε𝜃𝜃 = ln⁡�1 + � �� (4)
𝑟𝑟𝑐𝑐 − 𝑠𝑠
 Residual stress of cold-formed thick-walled steel rectangular hollow sections 847

For the elastic core, the transverse and longitudinal stresses are governed by Hooke’s Law. Due
to assumption (1) made in section 3.2, the stress and strain in the through thickness direction are
ignored, i.e., 𝜎𝜎𝑟𝑟 = 0 and ε𝑟𝑟 = 0. The stress-strain state can be expressed as
𝐸𝐸 𝐸𝐸
𝜎𝜎𝜃𝜃 = (ε𝜃𝜃 + 𝑣𝑣ε∅ ) = ε (5a)
1 − 𝑣𝑣 2 1 − 𝑣𝑣 2 𝜃𝜃
𝑣𝑣𝑣𝑣
𝜎𝜎∅ = 𝑣𝑣𝜎𝜎𝜃𝜃 = ε (5b)
1 − 𝑣𝑣 2 𝜃𝜃
For the point just reaches yielding, the stress state is governed by Von Mises principle

𝜎𝜎� = �𝜎𝜎𝜃𝜃2 + 𝜎𝜎∅2 − 𝜎𝜎𝜃𝜃 𝜎𝜎∅ (6)

By substituting Eqs. (5a) and (5b) into (6)

ε𝜃𝜃 = ±𝜎𝜎�(1 − 𝑣𝑣 2 )/(𝐸𝐸 �1 − 𝑣𝑣 + 𝑣𝑣 2 ) (7)

where ε𝜃𝜃 > 0 if the strain is tensile (𝑦𝑦 > 𝑠𝑠) and ε𝜃𝜃 < 0 if the strain is compressive (𝑦𝑦 < 𝑠𝑠).
Together with Eq. (4), the range of the elastic zone can be defined
��1−𝑣𝑣 2 �
𝜎𝜎 ��1−𝑣𝑣 2 �
−𝜎𝜎
(8)
𝑠𝑠 − (𝑟𝑟𝑐𝑐 − s)(e 𝐸𝐸 �1−𝑣𝑣+𝑣𝑣 2 − 1) ≤ 𝑦𝑦 ≤ 𝑠𝑠 − (𝑟𝑟𝑐𝑐 − s)(e 𝐸𝐸 �1−𝑣𝑣+𝑣𝑣 2 − 1)

For the plastic zone, the stresses are subjected to plastic straining. The following stress ratio can
be defined
𝜎𝜎∅ = 𝜔𝜔𝜎𝜎𝜃𝜃 (9)

where 𝜔𝜔 = 0.5 is adopted in this analysis (Massonnet et al. 1979). Combining Eqs. (6) and (9),
the stress of any point undergoing plastic straining due to cold bending can be expressed as
𝜎𝜎�
𝜎𝜎𝜃𝜃 = ± (10a)
√1 − 𝜔𝜔 + 𝜔𝜔 2
𝜔𝜔𝜎𝜎�
𝜎𝜎∅ = ± (10b)
√1 − 𝜔𝜔 + 𝜔𝜔 2
where 𝜎𝜎𝜃𝜃 and 𝜎𝜎∅ > 0 when 𝑦𝑦 > 𝑠𝑠; and 𝜎𝜎𝜃𝜃 and 𝜎𝜎∅ < 0 when 𝑦𝑦 < 𝑠𝑠.
For the equivalent plastic strain, the following relationships exist

√3
���
𝜀𝜀 𝑝𝑝 = �𝜀𝜀𝜃𝜃2 + 𝜀𝜀∅2 − 𝜀𝜀𝜃𝜃 𝜀𝜀∅ = 𝜀𝜀 (11a)
2 𝜃𝜃
𝜎𝜎�
𝜀𝜀̅ = 𝜀𝜀 𝑝𝑝 + (11b)
𝐸𝐸
With the true stress-strain relationship governed by Eq. (2), the through thickness distribution
848 Xingzhao Zhang, Su Liu, Mingshan Zhao and Sing-Ping Chiew

of 𝜎𝜎𝜃𝜃 and 𝜎𝜎∅ can be solved by Eqs. (10) and (11), while the bending moment is calculated as

𝑡𝑡/2
𝑀𝑀𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 = � 𝜎𝜎𝜃𝜃 𝑦𝑦𝑦𝑦𝑦𝑦 (12)
−𝑡𝑡/2

When the final bending radius is reached and the imposed radial displacement is removed, an
elastic springback would occur and elastically unload the corner (Fig. 9). To simulate the elastic
rebound of the bending process, this plastic moment is applied elastically through the thickness by
using Eqs. (3a) and (5b). It should be noted that unlike cold-bending for open sections, the elastic
springback for cold-formed SHS does not fully unload the bending moment 𝑀𝑀𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 (Liu et al.
2015). This springback moment can be defined as

𝑀𝑀𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = 𝜑𝜑𝑀𝑀𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 (13)

where 𝜑𝜑 is a reduction factor dependent on the constraint that the corner is subjected to during
the elastic springback. In this study, 𝜑𝜑 = 0.9 is used according to the 2D plane strain FE analysis
result.
The final transverse stress state is the summation of the plastic stress distribution caused by
bending and the unloading stress from the elastic springback

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆
𝜎𝜎𝜃𝜃𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = 𝜎𝜎𝜃𝜃𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 − 𝜎𝜎𝜃𝜃 (14a)

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆
𝜎𝜎∅𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = 𝜎𝜎∅𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 − 𝜎𝜎∅ (14b)

4. Validation of the analytical model

4.1 FE simulation

In order to verify the analytical model derived in Section 3, a 2D plane strain FE model is built
to simulate the forming of residual stress in the corner of the tested cold-formed thick-walled SHS.
As shown in Fig. 12(a), this model consists of 2D deformable quarter section (flat strip) and
analytically rigid roller and roller die. Dense mesh is attributed to the corner area and transition
mesh is set at the boundaries between the coarse and fine mesh.
Fig. 12(b) shows the Mises stress distribution of the section during bending. It can be seen from
Fig. 12(b) that the neutral axis of the flat area is at the middle surface while that of the corner area
is slightly shifted towards the inner surface. The stresses are generally symmetrically distributed
about the neutral axis. Although large area of the corner is undergoing plastic straining, there is
still a small-depth elastic core. However, this elastic core diminishes when the bending radius 𝑟𝑟𝑐𝑐
further decreases. When the final bending profile is reached, 0.5mm displacement load (to fully
unload the bending moment, 0.56 mm is needed) is applied on the boundaries to simulate the
constrained springback. The final transverse and longitudinal residual stress distributions are
shown in Figs. 13(a) and (b), respectively. It should be noted that the transverse residual stress (Fig.
13(a)) is not necessarily compressive on the outer surface. This amount is dependent on the
amount of springback action the corner is subjected to.
 Residual stress of cold-formed thick-walled steel rectangular hollow sections 849

(a) FE model with mesh (b) Mises stress during bending

Fig. 12 Quarter SHS model for FE analysis

(a) Transverse residual stress (b) Longitudinal residual stress

Fig. 13 Residual stress distribution by FE analysis

4.2 Comparison between the FE and analytical model

Figs. 14(a) and (b) show the equivalent (Mises) strain and stress distribution of the corner at
full bending state, respectively. It can be seen from Figs. 14(a) and (b) that the analytical model
agrees reasonably well with the FE model. The differences (calculated as Analytical model/FE
model×100%-100%) for the surface equivalent strains are 4.7% and -5.8% for the outer and inner
surfaces, respectively, while the differences for the surface equivalent stresses are -0.56% and 5.9%
for the outer and inner surfaces, respectively. The differences between the two models increase
from the surfaces towards the neutral axis due to the absence of the radial stress σ𝑟𝑟 in the
analytical model. Under elastic status, σ𝑟𝑟 is zero on the outer and inner surfaces and maximum
850 Xingzhao Zhang, Su Liu, Mingshan Zhao and Sing-Ping Chiew

near the neutral axis (Eq. (3b)). As a result, the precision of the analytical model in the core is not
as good as that near the outer and inner surfaces. Fig. 14c shows the final equivalent residual stress
distribution. Similar to the equivalent strain and stress distribution at full bending, the equivalent
residual stress distribution of the analytical model agrees well with the FE model near the surface
but not good enough near the neutral axis. It should be noted that due to the absence of σ𝑟𝑟 , the
equivalent stress in the elastic core obtained by the analytical model is close to zero (Figs. 14(b)
and (c)). Besides, the position of the neutral axis of the FE model is slightly lower than that of the
analytical model and further away from the middle surface. The neutral axis in the analytical
model is 0.7 mm (5.5% of thickness) away from the middle surface, while that in the FE model is
1.06 mm (8.3% of thickness) from the middle surface. This difference is mainly caused by the
second shifting of neutral axis when plasticity takes place, which is ignored in the analytical model.

7 7 7 Analytical

Thickness (mm)
6 6 6 FEM
5 5 5
4 4 4
3 3 3
Thickness (mm)

Analytical
Thickness (mm)

2 2 Analytical 2
1 FEM 1 1
FEM
0 0 0
-1 0.0 0.1 0.2 0.3 0.4 -1 0.0 0.5 1.0 1.5 2.0 -1 0.0 0.5 1.0 1.5 2.0
𝜀𝜀𝑒𝑞 𝜎𝜎𝑀⁄𝑓𝑓𝑦 𝜎𝜎𝑀⁄𝑓𝑓𝑦
-2 -2 -2
-3 -3 -3
-4 -4 -4
-5 -5 -5
-6 -6 -6
-7 -7 -7

(a) Equivalent strain at (b) Equivalent stress at (c) Equivalent stress

full bending full bending after springback
Fig. 14 Equivalent plastic strain and stress

7 7 7 Analytical
Thickness (mm)

Thickness (mm)
Thickness (mm)

Analytical
6 6 6 FEM
5 5 5
4 4 4
3 3 3
2 2 2
1 1 1
0 0 0
-2.0 -1.5 -1.0 -0.5-1 0.0 0.5 1.0 1.5 2.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
-1 -1
-2
𝜎𝜎𝜃⁄𝑓𝑓𝑦 𝜎𝜎𝜃⁄𝑓𝑓𝑦 𝜎𝜎𝜃⁄𝑓𝑓𝑦
-2 -2
-3 -3 -3
-4 -4 -4
-5
-5 -5
-6 Analytical -6 -6
-7
FEM -7 -7

(a) Plastic bending (b) Elastic springback (c) Residual stress

Fig. 15 Self-equilibrating transverse residual stress
 Residual stress of cold-formed thick-walled steel rectangular hollow sections 851

7 7 7 Analytical
Thickness (mm)

Thickness (mm)
Thickness (mm)
Analytical
6 6 6 FEM
5 5 5
4 4 4
3 3 3
2 2 2
1 1 1
0 0 0
-2.0 -1.5 -1.0 -0.5-1 0.0 0.5 1.0 1.5 2.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
-1 -1
-2
𝜎𝜎∅⁄𝑓𝑓𝑦 𝜎𝜎∅⁄𝑓𝑓𝑦 𝜎𝜎∅⁄𝑓𝑓𝑦
-2 -2
-3
-3 -3
-4
-4 -4
-5
-5 -5
-6 Analytical -6 -6
-7
FEM -7 -7

(a) Plastic bending (b) Elastic springback (c) Residual stress

Fig. 16 Self-equilibrating longitudinal residual stress

Figs. 15(a) and 16(a) show the predicted transverse and longitudinal stress distribution at full
bending, respectively. The unloading elastic springback stress distributions in the transverse and
longitudinal directions are show in Figs. 15(b) and 16(b), respectively, while the final transverse
and longitudinal residual stress distributions are shown in Fig. 15(c) and 16(c), respectively. It can
be seen from the above Figs. that all the through thickness stress distribution patterns are nonlinear
and not symmetrical to the neutral axis, which is essentially different from the cold-formed thin-
walled SHSs. Although the analytical model is not able to predict the stress near the neutral axis as
precisely as the FE model does, the accuracy near the outer and inner surfaces are reasonably good.

4.3 Comparison between test results and the analytical model

Since the surface residual stresses can be treated as generally linearly distributed (Figs. 15(c)
and 16(c)), the surface stress (AM-surface), 1mm depth stress (AM-1mm), and the mean stress of
these two (AM-mean) are calculated and compared with the results by the hole-drilling test, as

Position (No.) Test Test-mean AM-surface

0 3 6 9 12 15 18 21 AM-1mm AM-mean
0
35
-5
-10 30
Stress (%)

Stress (%)

-15
25
-20
-25 20
C2 C3 C4 C1 C2 C3 C4 C1
-30 15
Test Test-mean AM-surface
0 3 6 9 12 15 18 21
AM-1mm AM-mean Position (No.)

(a) transverse stress (b) longitudinal stress

Fig. 17 Comparison between test result and analytical solution
852 Xingzhao Zhang, Su Liu, Mingshan Zhao and Sing-Ping Chiew

shown in Fig. 17. For the transverse residual stress, it can be seen from Fig. 17(a) that only 1 data
point is outside the predicted range and the difference between the mean values of the analytical
model and test is only 0.9% of the yield strength. As for the longitudinal stress distribution, the
diversity is larger (Fig. 17(b)). Only half of the data points are within the predicted range (AM-
surface to AM-1mm). However, the mean stresses from the test agree well with the analytical
model prediction. The AM-mean is only 0.7% (of the yield strength) lower than that of the Test-
mean. Therefore, it is confirmed that the proposed surface residual stress distribution model in Fig.
7(a) is valid.

5. Conclusions

The magnitude and effect of the transverse and longitudinal surface residual stresses contained
in cold-formed thick-walled SHS is studied. Compared to cold-formed thin-walled SHS, the tested
cold-formed thick-walled SHS contains similar magnitude of longitudinal residual stresses in the
flat area but higher stress levels in the corner and welding area. Although the residual stress level
in the transverse direction is much lower than that in the longitudinal direction and is less
frequently studied, these stresses tend to open the section as well. In order to theoretically predict
the residual stress distribution in the corners of cold-formed thick-walled SHSs, an analytical
elastic-plastic large displacement model is proposed. 2D FE model is employed to simulate the
forming of residual stresses in the corners due to cold bending and verify the proposed analytical
model. It is shown that good agreement has been achieved for the surface residual stresses in both
transverse and longitudinal directions. By analyzing the through thickness stresses, it is found out
that the residual stress distribution pattern of the thick-walled SHS is totally different from that of
the thin-walled SHS. It is also shown that during bending the thick-walled SHS, shifting of neutral
axis must be considered, since it would lead to non-linear and non-symmetrical distribution of
stresses through the thickness. Finally, the analytical model is validated by comparing the surface
stress distribution prediction with the experimental results.

References

Abambres, M. and Quach, W.M. (2016), “Residual stresses in steel members: a review of available
analytical expressions”, Int. J. Struct. Integrity, 7(1), 70-94.
AISC (2005), Specification for structural steel buldings, ANSI/AISC 306-05; Americal Institute of Steel
Construction, Chicago, IL, USA.
AISI (1996), Cold-formed steel deseign manual; American Iron and Steel Institute, Washington D.C., USA.
Amouzegar, H., Schafer, B.W. and Tootkaboni, M. (2016), “An incremental numerical method for
calculation of residual stresses and strains in cold-formed steel members”, Thin-Wall. Struct., 106, 61-74.
ASTM (2008), ASTM E837-08: Standard test method for determining residual stresses by the hole-drilling
strain gauge method; ASTM International, West Conshohocken, PA, USA.
BSI (1999), BS EN 2566-1: Steel - Conversion of elongation values - Part 1: Carbon and low alloy steels,
British Standards Institution, London, UK.
BSI (2001), BS EN 10002-1: tensile testing of metallic materials: part1 method of test at ambient
temperature. London, British Standards Institution.
BSI (2005), Eurocode 3: design of steel structures: part 1-1 general rules and rules for buildings, BS EN
1993-1-1, British Standard Institution, London, UK.
BSI (2006), BS EN 10219: cold formed welded structural hollow sections of non-alloy and fine grain steels,
 Residual stress of cold-formed thick-walled steel rectangular hollow sections 853

part 1: Technical delivery conditions, British Standards Institution, London, UK.

Gardener, L., Saari, N. and Wang, F. (2010), “Comparative experimental study of hot-rolled and cold-
formed rectangular hollow sections”, Thin-Wall. Struct., 48(7), 495-507.
Jandera, M. and Machacek, J. (2014), “Residual stress influence on material properties and column
behaviour of stainless steel SHS”, Thin-Wall. Struct., 83, 12-18.
Key, P.W. and Hancock, G.J. (1993), “A theoretical investigation of the column behaviour of cold-formed
square hollow sections”, Thin-Wall. Struct., 16(1-4), 31-64.
Li, S.H., Zeng, G., Ma, Y.F., Guo, Y.J. and Lai, X.M. (2009), “Residual stresses in roll-formed square
hollow sections”, Thin-Wall. Struct., 47(5), 505-513.
Liu, T., Wang, Y., Wu, J., Xia, X., Wang, J., Wang, W. and Wang, S. (2015), “Springback analysis of Z &
T-section 2196-T8511 and 2099-T83 Al–Li alloys extrusions in displacement controlled cold stretch
bending”, J. Mater. Process. Technol., 225, 295-309.
Ma, J.L., Chan, T.M. and Young, B. (2015), “Material properties and residual stresses of cold-formed high
strength steel hollow sections”, J. Construct. Steel Res., 109, 152-165.
Mashiri, F.R., Paradowska, A., Uy, B., Tao, Z., Khan, M. and Dayal, P. (2014), “Residual stresses
distribution measured by neutron diffraction in fabricated square high strength steel tubes”, Proceedings
of the 7th International Conference on Mechanical Stress Evaluation by Neutrons and Synchrotron
Radiation, MECA SENS 2013, Sydney, NSW, Australia, September.
Massonnet, C., Olszak, W. and Phillips, A. (1979), Plasticity in Structural Engineering Fundamentals and
Applications, Springer-Verlag Wien, New York, NY, USA.
Moen, C.D., Igusa, T. and Schafer, B.W. (2008), “Prediction of residual stresses and strains in cold-formed
steel members”, Thin-Wall. Struct., 46(11), 1274-1289.
Quach, W.M., Teng, J.G. and Chung, K.F. (2004), “Residual stresses in steel sheets due to coiling and
uncoiling: a closed-form analytical solution”, Eng. Struct., 26(9), 1249-1259.
Spoorenberg, R.C., Snijder, H.H. and Hoenderkamp, J.C.D. (2010), “Experimental investigation of residual
stresses in roller bent wide flange steel sections”, J. Construct. Steel Res., 66(6), 737-747.
Timoshenko, S. and Goodier, J.N. (1970), Theory of Elasticity, McGraw-Hill, New York, NY, USA.
Tong, L., Hou, G., Chen, Y., Zhou, F., Shen, K. and Yang, A. (2012), “Experimental investigation on
longitudinal residual stresses for cold-formed thick-walled square hollow sections”, J. Construct. Steel
Res., 73, 105-116.
Wang, Y.B., Li, G.Q. and Chen, S.W. (2012), “The assessment of residual stresses in welded high strength
steel box sections”, J. Construct. Steel Res., 76, 93-99.
Weng, C. and White, R. (1990), “Cold‐bending of thick high‐strength steel plates”, J. Struct. Eng., 116(1),
1(40), 40-54.
Withers, P.J., Turski, M., Edwards, L., Bouchard, P.J. and Buttle, D.J. (2008), “Recent advances in residual
stress measurement”, Int. J. Press. Vessel. Pip., 85(3), 118-127.
Yu, T.X. and Zhang, L.C. (1996), Plastic Bending: Theory and Applications, World Scientific Publishing
Co., NY, USA.
Zhu, H.X. (2007), “Large deformation pure bending of an elastic plastic power-law-hardening wide plate:
Analysis and application”, Int. J. Mech. Sci., 49(4), 500-514.

BU