Structures 68 (2024) 107154

Contents lists available at ScienceDirect

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

Steel beam upstands as a strengthening approach for doubly symmetric

I-shaped sections
Kenny Mudenda *, Alphose Zingoni
University of Cape Town, P/Bag X3, Rondebosch, Cape Town, South Africa

A R T I C L E I N F O A B S T R A C T

Keywords: Sustainable design with minimization of carbon footprint requires the consideration of opportunities for optimal
Flange upstand material use as well as for re-use and repurposing existing steel members. For steel beams susceptible to lateral-
Monosymmetry torsional buckling one such opportunity is the use of flange upstands to strengthen existing beams if a change in
Lateral-torsional
use requires that the member have more bending strength or stiffness. Flange upstands employed in the study
Re-use
Strengthen
convert a doubly symmetric section into a monosymmetric section. The monosymmetry effect is studied and
Steel reveals the existence of a range of upstand heights that can be exploited for increased critical elastic buckling
I-section moment. Beyond a given upstand height, the effect of monosymmetry then reduces the critical elastic buckling
moment. It is shown that the critical upstand height relates closely to the point at which the monosymmetric
section attains a coincident shear center and centroid, a property typically associated with doubly symmetric
sections. Elastic and elastic-plastic analysis approaches are used to demonstrate how the stiffeners influence
flexural capacity and stiffness properties of the beams. Application of a desirable range of upstand heights for
strengthening I-shaped doubly symmetric beams is demonstrated with an example.

1. Introduction upstands to an I-shaped doubly symmetric cross section, as proposed in

this study, converts it into a monosymmetric section. This is shown in
Steel beams used in applications where the compression flange is not Fig. 1 with part (a) showing the doubly symmetric section and part (b)
restrained against lateral movement or rotation are susceptible to failure showing the monosymmetric section. Typical applications would be
by lateral-torsional buckling (LTB) when undergoing major-axis flexure. strengthening of existing sections where re-use involves a change in
Based on elastic buckling theory, in this mode of failure at a critical load applied loading, so that the member must now carry an increased load,
the compression flange suddenly moves laterally out of the plane of typically with an accompanying need for greater stiffness. If the doubly
loading accompanied by a twist. Steel beams without lateral restraint symmetric section to be strengthened is in a position where installation
can be found in industrial structures, in crane runway beams and in of flange downstands is particularly challenging then upstands become a
temporary conditions such as building or bridge beams prior to casting viable alternative. Upstands can also be used for cases where the cross
and curing of the composite concrete deck. section needs to be optimized from the start. In a case of the latter a
To avoid lateral-torsional buckling in these beams several in­ smaller sized beam with upstands may be considered in lieu of a larger
terventions can be considered. The list of factors associated with lateral- sized doubly symmetric beam.
torsional buckling failure gives an indication of areas where possible
interventions can be applied to prevent LTB or increase the critical 2. Literature review
elastic buckling load; such a list has been presented by [1]. One such
factor is section geometry. In this study the geometry is altered by An early study on the behaviour of monosymmetric sections by [2]
introducing flange upstands on an I-section beam and the upstand height presented equations for use with unequal width flange I-shaped beams.
adjusted to study improvement in performance against LTB failure. Analytical equations for the case of uniform moment for single span
Recommendations are then made on how this can be implemented to simply supported and fixed end supports were presented. Equations for
obtain higher capacity against LTB failure. The introduction of flange the critical load of an arbitrary section subjected to centric axial load

* Corresponding author.
E-mail address: [email protected] (K. Mudenda).

https://doi.org/10.1016/j.istruc.2024.107154
Received 15 March 2023; Received in revised form 8 August 2024; Accepted 23 August 2024
Available online 5 September 2024
2352-0124/© 2024 The Author(s). Published by Elsevier Ltd on behalf of Institution of Structural Engineers. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
K. Mudenda and A. Zingoni Structures 68 (2024) 107154

for doubly symmetric section beams, the difference being accounted for
only in the value of the critical elastic moment. [17] also observed that
the shape of the inelastic buckling curves of monosymmetric sections
were not well described by current code regulations. This study was
specific to Eurocode 3 with comparisons to results from Geometrical and
Material Nonlinear with Imperfections Analysis (GMNIA). These results
were compared to design curves generated based on the conservative
‘general method’ in the code.
An overview of the developments on structural member verification
in prEN 1993–1-1:2020 as part of the second generation of Eurocodes is
presented by [18]. Monosymmetric sections are also addressed. The
state of whether the member is in single or double curvature is recog­
nized as having a significant effect on the critical elastic moment. This
Fig. 1. Cross section under consideration. complexity presents a challenge unlike the simplicity with which this is
dealt with in the case of doubly symmetric sections. It also means that
and end moments about three orthogonal axes were presented by [3]. moment gradient factors for doubly symmetric sections should not be
The case of a monosymmetric section subjected to end moment could be indiscriminately applied to monosymmetric sections as safety cannot be
obtained by selecting only applicable parameters while neglecting the guaranteed for all cases when this is done. For monosymmetric sections
ones that do not apply. Studies presented by [4] and [5] were referenced cases must be distinguished and care must be taken when the moment
in a seminal paper by [6] who, among other things, undertook an diagram changes signs within an unrestrained length. Due to this
evaluation of the correct expression for the monosymmetry constant, βx , complexity, it becomes difficult to develop generalized equations. One
which had been presented in different forms by [3,4,5]. It was shown in approach suggested by the authors is to assume no change in the
the study that the form presented by [3] was the correct one and that the moment sign when this produces conservative outcomes. The other,
other two were in error. The monosymmetry constant encapsulates the more accurate yet laborious option, is to carry out second-order analysis
‘Wagner effect’ which manifests in flexural members where the shear with imperfections or GMNIA calculations.
center and centroid do not coincide. In addition, the study also consid­ It should be noted that, while I-beams of present interest will be
ered the case of transverse loading on a beam in the form of a point load assumed to be of constant cross-section along the member, steel beams
or a uniformly distributed load (UDL). Previous studies had focused may have tapered webs for enhanced structural efficiency [19]; such
largely on the case of beams with end moments. Equations for critical beams could also be assigned flange upstands or downstands, but this
loads were obtained using the finite integral method to solve the type of problem is not within the scope of the present study.
nonlinear governing differential equations.
Because section properties required to obtain the critical elastic 3. Study and method
moment of monosymmetric sections were generally not easy to evaluate,
[7] developed simplified equations for I-shaped monosymmetric sec­ The current study is on the lateral-torsional buckling behaviour of a
tions as well as for I-shaped sections with flange end downstands. The monosymmetric section produced from introducing stiffeners to the top
simplified equation for the monosymmetry constant presented has been flange of a doubly symmetric beam. The prefixes ‘NF’ and ‘WF’ are used
widely adopted and remains in use to this day in many design guides and on the standard rolled section sizes to denote that in the current study
literature on monosymmetric sections. The only drawback to this the radii are neglected to simplify calculations. The prefixes are for
equation is that is has a validity range. It also seems that it needs to be narrow flange and wide flange sections respectively. The member
modified for geometries dissimilar to the ones its derivation was based elevation and cross-section considered is shown in Fig. 1 with distance
on. For instance, [8] had to modify this equation for the case of mono­ from load point of application to shear center (SC) shown as ‘a’. In Fig. 1
symmetric crane beams consisting of an I-section with a channel cap. (a) is the doubly symmetric or bi-symmetric beam prior to introduction
Various aspects of monosymmetric beams have been the subject of of stiffeners and (b) is the monosymmetric beam after stiffeners are
studies over the years. Effects of moment gradient for instance have been introduced.
studied by [9] and [10]. Complexities and peculiarities with mono­ To study the effect of monosymmetry the requisite equations for
symmetric sections have long been recognized and reported on. The geometrical properties are first derived. This is followed by a study on
behaviour of these sections differs in many respects to that of doubly the effects of upstands on the relative position of the shear center and
symmetric sections. This has been reported in studies by [11] and [12]. centroid. The study is then extended to the effect of upstands on the
It was shown, for instance, in the latter study that the case of uniform critical elastic moment. The critical elastic moment study is carried out
moment along the beam length was not the critical case as would be using equations developed based on the Rayleigh-Ritz energy method.
expected for doubly symmetric beams. These studies validate the need to The upstand height denoted by ‘α’ in the study is of particular interest.
better understand behaviour of monosymmetric sections for their opti­ Application of the study findings as a strengthening method is presented.
mum utilization. Due to the residual stress pattern for the section considered not yet being
Design guidance on monosymmetric sections is limited compared to available the member strength is approximated by carrying out
that on doubly symmetric members and in some cases when provided it nonlinear analysis that considers only the geometrical out of straight­
appears support from research is limited. Early work by [13] and [14] ness and material constitutive behaviour. This is in effect a geometrical
had already shown how the inelastic buckling behaviour of mono­ and material nonlinear analysis (GMNA) with the ‘I’ part omitted due to
symmetric sections differed to that of doubly symmetric sections. [15] the non-inclusion of residual stress as an imperfection. Inelastic behav­
highlighted the seeming lack of significant research underpinning the iour due to parts of the member yielding before others is accounted for
design approaches adopted by North American, European and Austra­ with the inclusion of the material constitutive relationship for a selected
lian codes of practice. The study on inelastic buckling showed some of steel grade. It is shown in Section 7 that this analysis approach is valid
the code short comings. The study then presented accurate approxima­ for slender beams whose behaviour is largely unaffected by the presence
tions for uniform bending and less accurate but conservative approxi­ of residual stress. The GMNA analysis is carried out using finite element
mations for moment gradient case. In the specific case of the Eurocode 3, analysis software.
it was stated by [16] that no specific design guidance is provided for
monosymmetric sections. The code applies the same general method as

2
K. Mudenda and A. Zingoni Structures 68 (2024) 107154

4. Geometrical properties
4αtα H + 2α2 tα + 2Bt1 H − Bt21 + 2tw hw t2 + h2w tw + bt 22
y= (1.6)
4αtα + 2Bt1 + 2hw tw + 2bt2
Key geometrical properties that need to be determined are the
centroid and shear center position, the second moments of area, the These developed algebraic expressions are straightforward to apply
monosymmetry constant, the torsional constant and the warping con­ with a spreadsheet.
stant. Overall and mid-line dimensions that were used in this study are
given in Fig. 2. 4.2. Torsional constant
Fig. 2(a) gives overall dimensions as H = height, B, b = breadth of
top and bottom flanges respectively. Symbols t1 , t2 , tw and tα are thick­ The determination of the torsional constant follows the approaches
ness of top flange, bottom flange, web and upstands respectively. Sym­ given by [22] and [20] largely based on the membrane analogy pre­
bols ‘SC’ and ‘C’ are for the shear center and centroid positions sented by [23]. For a section that can be considered as an assembly of
respectively. The web height is denoted by hw . Fig. 2(b) gives the mid- plates, the torsional constant can be suitably approximated with Eq.
line dimensions with b1 , b2 being the breadth of top and bottom (1.8) for plate width ‘b’ and plate thickness ‘t’.
flanges respectively, lα the upstand height to flange midline, e the dis­
n
tance to shear centre from bottom flange and h the midline height. 1∑
J= bi t3 (1.8)
3 i=1 i
4.1. Shear center and centroid
∑bi t3 2lα t 3 b1 t 3 ht3 b2 t 3
J= i
= α
+ 1+ w+ 2 (1.9)
To find the shear center an approach by [20] is combined with that 3 3 3 3 3
presented by [21] to locate the shear center based on longitudinal stress
Where plates intersect, the mid-line dimension for ‘b’ should be
equilibrium as well as cross-section moment equilibrium when consid­
considered. For the section considered in the current study midline di­
ering transverse load acting at the shear center and the resulting shear
mensions with reference to Fig. 2 are utilized in the torsional constant
flow in a thin-walled arbitrary section. This consideration results in the
Eq. (1.9).
location ‘ysc’ of the shear center along the axis of symmetry and
measured from the base of the member. After evaluating the integral
terms from the derived expressions, algebraic expressions are obtained 4.3. Warping constant
for the location and are presented in Eqs. (1.1) to (1.4).
The warping constant is obtained by considering approaches given
b31 by [22]. The resulting equation for the cross-section considered in the
Iωx1 = (ht1 ) (1.1)
12 study is given in Eq. (1.10).
( )
lα b1 tα b1 lα t1 b31 (h − e)2 tα b21 lα 3 tα b21 (h − e)l2α tα b21 (h − e)2 lα t2 e2 b32
Iωx2 = b1 h − (1.2) Cw = + − + +
2 2 12 6 2 2 12
(1.10)
Iωx = Iωx1 + Iωx2 (1.3)
Midline dimensions are utilized in Eq. (1.10).
Iωx t2
ysc = + (1.4)
Iy 2 4.4. Monosymmetry constant

The second moment of area about the minor axis is given in Eq. (1.5)
The monosymmetry constant is evaluated using the approach pro­
and the centroid location in Eq. (1.6) measured from the base of the
posed by [24] where the cross section as an assembly of plate elements is
member.
divided into its separate elements. The monosymmetry constant equa­
[ )2 ]
1 [ 3 ] αt 3α
(
B tα tion verified by [6] as given in Eq. (1.11) is then applied to each element.
Iy = t1 B + t2 b3 + hw t 3w + + 2αtα − (1.5) These individual contributions are then summed up. The integration in
12 6 2 2
(1.11) can be simplified into the algebraic expression in Eq. (1.12) which

Fig. 2. Dimensions for I-Section with flange upstands.

3
K. Mudenda and A. Zingoni Structures 68 (2024) 107154

is applied for the section considered in the study.

∫
1 ( )
βx = y y2 + x2 dA − 2y0 ) (1.11)
Ix A
{∑[ ]}
1 1( 3 )( ) 1 ( )
βx = x − x31 y22 − y21 + (x2 − x1 ) y42 − y41 − 2y0
Ix 6 2 4
(1.12)
Eq. (1.12) is applied with reference to Fig. 3. The integration in
(1.11) is applied over each plate through application of Eq. (1.12) with
(x1 , y1 ) and (x2 , y2 ) being coordinates over which the plate is integrated
and y0 the distance between the centroid and the shear center.
Summation of values from Eq. (1.12) can be conveniently imple­
mented in a spreadsheet.

Fig. 4. Movement of shear center and centroid with upstand height for a
5. Shear centre movement study
NF203 × 133 × 25 beam.

Having obtained the geometrical properties, the movement of the

shear center and centroid with change in upstand height was investi­ The values for γ 1 , γ2 , x1 , x2 , x3 , x4 and x5 can be determined from the
gated. A trend was observed where, as the upstand height increases the equations (A.1) to (A.7) in the Appendix. Equating the two equations
shear center position moves away from the bottom datum, reaches a and solving for ‘α’ yields the desired value of the upstand height at the
peak value and then starts to drop towards the datum. The trend is not coincidence point as a root of the third-degree polynomial obtained. The
linear. The centroid on the other hand is observed to move away from values for ‘αc ’ were determined for all hot rolled ‘I’ and ‘H’ sections in
the bottom datum linearly the whole time. The trend reveals two (2) the South African Steel Construction Handbook [25] simply referred to
peculiarities. The first is that at a given upstand height the mono­ in this study as ‘the steel handbook’. This was also done for selected
symmetric section attains a property typically associated with bisym­ sections from the North American historical database as well as selected
metric or doubly symmetric sections in that the shear center and the sections from the United Kingdom ‘UKB’ and ‘UKC’ sections. Based on a
centroid are coincident. This is shown for a NF203 × 133 × 25 narrow statistical analysis of the results it was found that:
flange beam in Fig. 4. The point of coincidence is indicated by the (i) For narrow flange (NF) sections the coincidence point occurs at an
symbol ‘αc ’. upstand height of approximately one third of the member height, for the
The second peculiarity is that at a certain upstand height the mon­ case of flange thickness equal to stiffener thickness.
osymmetry constant attains a value of zero for the monosymmetric (ii) For wide flange (WF) sections the upstand height at which the
section. This is another property typically associated with doubly sym­ point of coincidence occurs on average is at a value of 0.26 times the
metric cross sections. It is recognized that the monosymmetry constant is member height. The variation for WF sections is greater than for NF
also dependent on the shear center position relative to the centroid. sections.
Therefore, the study focuses on the effect of the shear center movement. The coincidence points given in (i) and (ii) are values that can be
To determine the upstand height ‘αc ’ at which coincidence of the used for initial estimate. The actual coincidence point upstand height
shear center and centroid takes place (coincidence point) the equations should be determined from Eqs. (1.13) and (1.14). This determination is
for the centroid (1.13) and shear center (1.14) location need to be critical for optimizing the use of upstands to increase flexural capacity.
equated and the value of ‘α’ determined. The shear center and centroid movement can be plotted as a function
of the ratio of the upstand height to I-section member height (α/H) to
2α2 tα + 4Hαtα + γ 1 show how the coincident point falls around the one-third point for
y= (1.13)
4αtα + γ2 Narrow flange sections as shown in Fig. 5.

α(2x1 + t1 x4 ) − α2 2x2 + (2x3 + t1 x5 ) 6. Critical elastic moment with upstand height

ysc = (1.14)
α2x4 + 2x5
The critical elastic buckling load equation for the monosymmetric
section for two load cases, a central point load and a uniformly

Fig. 3. Coordinate values for I-Section with flange upstands for use in Fig. 5. Movement of shear center and centroid with upstand height - UKB
Eq. (1.12). section (NF203 × 102 × 23).

4
K. Mudenda and A. Zingoni Structures 68 (2024) 107154

distributed load (UDL), was formulated using the Rayleigh-Ritz energy The trend is similar across all sections studied. The trend is shown in
method. The approximating function used in the energy equation is a Table 1 which presents part of the results for the NF305 × 102 × 25
half sine wave for both the displacement as well as the rotation. section whose plot is presented in Fig. 6 for the central point load case,
Boundary conditions were selected for a typical ‘fork type’ support. This with load applied on the top flange.
entails simple supports with rotation about axis restrained at ends and In addition, it was found that the percentage increase is greater for
these ends having freedom to warp. Solution of the energy equation for the narrow flange sections than it is for the wide flange sections. This
the critical elastic load leads to quadratic equations for the point load could be attributed to wide flange sections having a high minor axis
and UDL cases given in (1.15) and (1.16) respectively. second moment of area. Even though the percentages are less the nu­
( )2 [ ( ( merical values are relatively high compared to the narrow flange
1 1 EIy π4 1 sections.
P2 + 2 − P β x π 2
16 4π 2L3 16 Table 2 shows this trend for a NF305 × 102 × 25 beam under central
) )] ( )
1 EIy π4 GJπ 2 ECw π4 point load for a 2 m span. It was already shown in Table 1 how the in­
− +a − +
4π 2 2L 3 2L 2L 3
crease was initially diminishing up to an upstand height of 90 mm. The
= 0 (1.15) values highlighted show a similar critical elastic moment for an upstand
height of 80 mm and at an upstand height between120mm and 130 mm.
(
w 2 L2 π 2 + 3
)2 [
EIy π 4
( (
βx π 2 − 3
)
a
)]
EIy π4
(
GJπ2 Therefore, the loss in critical elastic moment gain with increasing
− w + − upstand height becomes apparent.
64 3 2L2 8 3 2 2L3 2L
) To establish the region where monosymmetry has positive effect the
ECw π 4
+ critical elastic moment of the cross section was then plotted for upstand
2L3
height in comparison to an ‘equivalent doubly symmetric’ section (Eq.
=0 DS). This is a section which at each upstand height would have the same
(1.16) geometrical properties as the monosymmetric section but with the
monosymmetry constant, βx , having a value of zero. The concept is
In the equations ‘P’ is the applied point load and ‘w’ the applied UDL.
illustrated in Fig. 7.
The output from these equations was verified against the results pre­
This was done for selected NF and WF beams and the outcome is
sented in [26] and [27] for the point load and UDL cases respectively.
similar for all sections considered. The result reveals the presence of a
The results presented in the two studies were based on different nu­
region where the monosymmetry contributes to an increase in critical
merical methods to the one used in the current study and thus suitable to
moment. At a critical upstand height, the moment trend then switches
be used for verification. One was based on the Galerkin method of so­
into a region where monosymmetry contributes to a reduction in the
lution for a nonlinear differential equation, whereas the other used a
critical load. This shows that for these sections there exists a range of
Simpson rule approach to solve the integrals obtained from applying an
upstand heights were the monosymmetry works beneficially and one
energy method. The results were with a maximum of 4 % difference in
where it gives an undesirable outcome. This is illustrated in Fig. 8.
all cases considered. These equations derived in the current study were
The area of interest in Fig. 8 represents an area where the critical
used in a spreadsheet to obtain the critical elastic moment. The critical
elastic moment for the monosymmetric section is not only changing due
elastic moment vs upstand height relationships where established and
to geometrical section properties but also the effect of monosymmetry
suitable plots generated. The upstand height was varied from no upstand
contributing to a higher critical load. Beyond the intersection point of
to a member with upstands in 10 mm increments up to a value of
the two curves the effect of monosymmetry is to decrease the critical
130 mm. This upper limit was arbitrarily selected to suitably capture the
load.
observed trend.
This trend is like that observed for the movement of the shear center
The trends for both the point load case and the UDL are similar. At
vs upstand height. To compare with the conclusions from Section 5 the
low upstand heights the critical elastic moment increase is significant.
plots had the values on the horizontal axis converted to upstand height
As the upstand height increases the increase in critical elastic moment
divided by member height. This is illustrated for a NF254 × 146 × 37
reduces and, in some cases, reaches a peak value before it starts to
beam in Fig. 9. The acronym ‘DS’ is for doubly symmetric and ‘MS’ for
reduce with further increase in upstand height. This shows a ‘dimin­
monosymmetric while ‘Eq.’ is ‘equivalent’.
ishing return’ with increase in upstand height. An example for a narrow
Results for selected narrow flange sections showed similar trends as
flange sections from the steel handbook [25] is shown in Fig. 6. The
did those for wide flange sections. It can be seen that results in Figs. 8
critical elastic moment is normalized by expressing the moment at an
and 9 compare closely to the ‘α/H’ coincidence point from Section 5
upstand height (Mcr ) divided by the critical moment for the cross section
(Figs. 4 and 5). This shows that the point of coincidence of the shear
without upstands (Mcr,0 ).
center and centroid is also a proxy indicator of where the beneficial ‘area
of interest’ ends. This can thus be used to determine the optimal upstand
height for these cross sections so that the benefit of monosymmetry is
exploited and the region of upstands where the monosymmetry works to
reduce critical elastic moment is avoided. This critical upstand height
can be determined from equating Eqs. (1.13) and (1.14) given in Section
5. The procedure laid out in Fig. 10 leads to the option of optimum use of
flange upstands as stiffeners.
The concept described here can be used for a practical example. This
is given in Section 8. Before looking at an example the study is extended
to member resistance based on a nonlinear analysis that incorporates
geometrical out-of-straightness and material behaviour so that the
application example is based not only on consideration of elastic critical
load but also on some estimation of actual member flexural strength. To
obtain this approximated member strength GMNA was utilized. The
finite element modelling and analysis software Abaqus was used for this.
Fig. 6. Change in critical elastic moment with upstand height - (NF305 × Model details and analysis approach are given in the section that
102 × 25). follows.

5
K. Mudenda and A. Zingoni Structures 68 (2024) 107154

Table 1
Change in critical elastic moment with upstand height.
2 m Span NF305 × 102 × 25

Upstand Height (mm)

0 10 20 30 40 50 60 70 80 90
Critical Moment (kNm)
86 117 144 167 186 201 212 220 225 228
% Increase - 36 % 24 % 16 % 11 % 8% 6% 4% 2% 1%
Cumulative % Increase - 36 % 68 % 95 % 117 % 134 % 147 % 156 % 162 % 166 %

Table 2
Change in critical elastic moment with upstand height.

2m Upstand Height

Span 0 10 20 30 40 50 60 70 80 90 100 110 120 130

Mcr
86 117 144 167 186 201 212 220 225 228 229 228 226 224
(kNm)

in the model. Introduction of upstands on the flange is likely to affect the

existing residual stress pattern of the member and subsequently the
member strength. However, since this new residual stress pattern has
not yet been determined a nonlinear analysis that incorporates only the
effects of the out-of-straightness, as a geometrical imperfection, and the
material constitutive relationship is carried out as a reasonable
approximation. A nonlinear analysis of an I-section beam for cases of the
model incorporating residual stress and one that omits residual stress is
Fig. 7. Concept of ‘equivalent doubly symmetric’ (Eq.DS) section.
presented to demonstrate that omitting the residual stress pattern still
gives a reasonable indication of member strength particularly for
7. Nonlinear analysis slender beams. This is acceptable for proving the strengthening concept
presented in the current study.
Once a suitable section has been identified based on the critical The RIKS analysis method was implemented in the software for the
elastic moment consideration outlined in Section 6 it then becomes nonlinear analysis. The results are utilized in the application of the ‘area
essential to check the member strength values for a final decision. of interest’ outlined in Section 6. The numerical model was verified
Typically, to get the member strength a geometrical and material against results from Eqs. (1.15) and (1.16) with the largest difference
nonlinear analysis needs to be carried out with initial geometrical out- being in the order of 10 % for critical elastic buckling loads. An
of-straightness, material constitutive law and residual stress included

Fig. 8. Change in critical elastic moment with upstand height for monosymmetric section and equivalent doubly symmetric section.

6
K. Mudenda and A. Zingoni Structures 68 (2024) 107154

Fig. 12. Finite element model boundary conditions.

Having validated the model, the study was then progressed to the
nonlinear analysis.
Fig. 9. Normalized critical moment vs upstand to member height ratio
A quad-linear stress-strain curve developed by [28] was employed as
(NF254 × 146 × 37).
the constitutive relationship for a grade S355JR steel. The following
properties were used for this grade of steel as given the Southern African
Steel Design Handbook [25]: E = 200GPa, fy = 355MPa, fu = 470MPa.
Strains εy , εsh and εu are strains at yield, strain hardening commence­
ment and fracture respectively. Plastic engineering stress and strain used
for the Abaqus software input are given in Table 4. Plastic strain is ob­
tained by subtracting the elastic strain from the total strain.
An elastic analysis was first carried out to obtain the half-sine wave
buckling mode. This was then scaled in the nonlinear analysis as an
initial geometric imperfection with a maximum out-of-straightness
value obtained from either Eqs. (1.17) or (1.18). The maximum out-of-
straightness value specified in the South African standard SANS2001-
Fig. 10. Model for choice whether to use flange upstands or replace with a CS1 [29] based on allowable manufacturing tolerance was used to
bigger size DS member. obtain the value of the geometrical imperfection. The larger output from
applying Eqs. (1.17) and (1.18) was used.
illustration of the boundary conditions employed in the finite element
L
model is shown in Fig. 11 and Fig. 12. These are the ‘fork type’ support u= or 3mm, whichever is greater (1.17)
1000
conditions with a simply supported beam, rotation restrained at the
supports and warping allowed to occur freely. One support end is L
modelled on rollers (Uz ∕ = 0), while the other is restrained with respect u= or 25mm, whichever is lesser (1.18)
500
to displacement (Ui ).
The beam failure at ultimate load is illustrated in Fig. 14.
A shell element model was utilized for this analysis with S4R ele­
Slender beams are considered in the application example presented
ments in the Abaqus library used for this. This is a four-node element
in the section that follows. For these beams the influence of residual
with reduced integration and having both displacement and rotational
stress is not significant in comparison to intermediate length beams. This
degrees of freedom at the nodes.
approach is taken in order to demonstrate the optimum application of
Based on a mesh convergence study a global mesh size of 8 mm was
flange upstands as presented in the study, despite not having the residual
found to be optimal. An elastic buckling analysis was then carried out to
stress pattern information available. The determination of member
validate the shell model. The loading case considered was that of the
strength for beams undergoing lateral torsional buckling in the Cana­
central point load applied at the shear centre of the cross section.
dian and South African codes of practice for hot-rolled steel member
Boundary conditions were modelled as indicated in Fig. 11 with stiff­
design ([30] and [31]) is based on an interpolation equation for beams
eners added at loading and support reaction points to prevent web
whose critical elastic moment value is greater than 0.67 times the plastic
distortion. The finite element analysis (FEA) software output from the
moment i.e. 0.67Mp (for compact sections) and for a critical moment less
current study was compared to results from Eq. (1.15) as well as from
than 0.67 times the plastic moment the critical elastic moment is used as
[26] and presented in Table 3.
the flexural resistance moment. This means that beams whose critical
The results from the finite element analysis are almost identical to
elastic moment falls in this range are slender enough for the strength of
those from [26] and both sets within 4 % of predictions from the
the member to be based on the critical elastic moment. The
equation developed in the current study. The lateral-torsional buckling
failure mode for the 4 m span beam is shown in Fig. 13. non-dimensional slenderness (λLT ) associated with this cut-off moment
has a value of 1.22. This is illustrated in Fig. 15. Therefore, beams uti­
lized for demonstrating the application of the study findings are selected
to be slender and have a non-dimensional slenderness value greater than
1.22.
( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ )
With λLT = Mp /Mcr , Mp is the plastic moment, Mcr the critical
elastic moment and Mr the typical resistance moment with all imper­
fections considered.
In order to verify that the resistance of slender beams can indeed be
acceptably approximated from a shell model that omits residual stress
distribution such a check was carried out based on results for a doubly
symmetric section (IPE500 universal beam) presented in [18]. Two
models, one with a residual stress pattern as presented by [32] and one
Fig. 11. Boundary conditions for the ‘fork type’ support at pinned end.

7
K. Mudenda and A. Zingoni Structures 68 (2024) 107154

Table 3
Material properties utilized in finite element model nonlinear analysis.

SC
P
C

H = 300 mm Elastic Critical Moment – Mcr (kNm)

B = 150 mm SPAN (m) Eq. (1.15)− 1- FEA(Current study)− 2- Mohri et al. [26]− 3- Ratio:(2)/(1) Ratio:(3)/(1)
b = 75 mm 4 139.9 134.0 134.4 0.96 0.96
tf = tb = 10.7 mm 6 80.2 77.6 77.5 0.97 0.97
tw = 7.1 mm 8 55.9 54.2 54.2 0.97 0.97
E = 210 GPa

Symbols: H = member height, B = width of top flange, b = width of bottom flange, tf = thickness of top flange, tb = thickness of bottom flange, tw = thickness of web,
E = modulus of elasticity.

Fig. 13. Shell element model (a) and first positive buckling mode for a 4 m span beam (b).

Table 4
Material properties utilized in finite element model nonlinear analysis.
No. Stress Plastic Strain
1 355 0.0000 4
2 355 0.0188
3 427 0.0503
1 2 3
4 470 0.1450
Stress

Strain

Fig. 14. Beam failure mode during nonlinear analysis.

without residual stress, were analyzed and the results compared. The
residual stress model used employs a triangular stress distribution with
compression at flange tips as well as central area of web, and tension at
the web to flange junctions. Maximum stress values are 30 % of the yield
stress. The model with residual stress included gave lower results than Fig. 15. Non-dimensional slenderness value at moment cut-off for
the model without residual stress, as expected. The difference between slender beams.
the two model results was found to be about 6 % at (λLT = 1.0) and 3.9 %
at (λLT = 2.0). When compared to [18] the model with residual stress This shows that for an I-section beam the omission of residual stress
included gave values slightly lower but within 6 % whereas the model in the nonlinear analysis model for a slender beam, while deviating
without residual stress gave a slightly higher value for the less slender slightly from the values when residual stresses are included, still gives a
beam and a slightly lower value for the slender beam but all within 4 % good indication of member resistance. The use of nonlinear analysis
difference Table 5. with a model that omits the as yet unavailable residual stress pattern is
thus employed in the section that follows to demonstrate practical

8
K. Mudenda and A. Zingoni Structures 68 (2024) 107154

Table 5 Table 7
Moment ratio values for an IPE 500 beam from FEA compared to [18]. Critical elastic moment and second moment of
area - NF305 × 165 × 40 beam.
λLT(1.0) λLT(2.0)
6 m Span
FEA Model without residual stress Moment Ratio 0.650 0.240
Model with residual stress (Mr /Mp ) 0.613 0.231 Mcr (kNm) 87
*Knobloch et al.[18] 0.630 0.250 Ix (x106 mm4) 84
* (λLT ) 1.6
Scaled from Fig. 3 of publication.

application of the aspects discussed in Section 6.

Table 8
Resistance moment comparison based on FEA results.
8. Practical application
6 m Span Mcr Mr (λLT )
(kNm) (kNm)
Application of the ‘area of interest’ from Section 6 in selecting a
NF− 25414637 (20 mm Upstand) 94 89 1.4
suitable upstand height is demonstrated here. The scenario considered is NF− 30516540 84 79 1.6
where a decision needs to be made whether to replace an existing doubly
symmetric section with another (larger size) such section or to
strengthen the existing section with flange upstands. Beams considered weighed together with the complementary carbon footprint for an
here all had a (λLT ) value greater than 1.22. A NF254 × 146 × 37 with a economical and sustainable option to be selected. All this in addition to
6 m span is used in this example. The critical elastic moment and second time considerations.
moment of area of the beam are shown in Table 6 based on a point load
applied at midspan on the top flange. Elastic properties are considered 9. Conclusions
first.
In the Southern African Steel Construction Handbook [25] the next In this study the behaviour of a monosymmetric section formed by
larger size doubly symmetric section in terms of flexural capacity is a introducing upstands to a doubly symmetric section was investigated.
305 × 165 × 40 beam. The requisite member properties are given in Such a section would arise in the event of a need to strengthen an
Table 7. existing doubly symmetric section instead of removal and replacement
The coincidence point, ‘αc ’ for the NF254 × 146 × 37 is found to be at with a larger size doubly symmetric section. The section may also be
an upstand height of 77 mm or (α/H) of 0.30 based on Eqs. (1.13) and considered as part of an optimization process at member selection stage
(1.14). From Table 6 the NF254 × 146 × 37 beam with an upstand in a new design. Requisite equations for geometric properties of the
height of 20 mm has an Mcr of 99 kNm which is greater than that for the section considered have been developed and presented in algebraic form
section in Table 7 at 87 kNm. The 20 mm upstand is below the coinci­ making it straight forward to apply. From the developed equations a
dence point upstand value of 77 mm and therefore the section with this study of the shear center movement as well as centroid movement
upstand height would be in the desirable area for exploiting benefits of measured against increase in upstand height has revealed a peculiarity
monosymmetry for flexure. In the case of the second moment of area at of a coincident shear center and centroid for the monosymmetric section
62 × 106 mm4 this is less than that for the section in Table 7. This means considered in the study at a specific upstand height. This is a property
that although requirements for moment value (flexure) are met, if ser­ typically associated with doubly symmetric sections. The critical load
vice limit state governs the member choice, then a higher stiffener behaviour against upstand height has revealed the presence of a region
height would be required. A choice of a 70 mm upstand height would where monosymmetry can be used beneficially. The upstand height
meet the stiffness requirement since at 86 × 106 mm4 this would be corresponding to the coincident shear center and centroid point repre­
greater than that of the section in Table 7 at 84 × 106 mm4. The height of sents an indicator for the end of the region where monosymmetry can be
70 mm is still within the range where monosymmetry provides benefi­ beneficially exploited. Therefore, upstand heights up to this point only
cial effects in flexure. A comparison based on member resistance of the should ideally be utilized for optimum returns. This approach promises
two beams is given in Table 8. to help design engineers utilize monosymmetry beneficially and gives
It can be seen from Table 8 that the results from the nonlinear them an option to consider that would potentially offer cost savings and
analysis validate what the comparison based on elastic analysis already a carbon offset as well by avoiding having to dispose of a member and
shows. The monosymmetric member selected on the basis of the elastic then replace it with another member. In the case of first time use it offers
critical moment also shows higher member resistance in flexure the design engineer an option for an optimized member when compared
(89 kNm) than the doubly symmetric alternative (79 kNm) when the to larger size doubly symmetric members suitable for similar
nonlinear analysis is carried out for the member strength. This example application.
helps to illustrate how the ‘desirable area’ approach can be implemented In the final stages of the revision of this paper, two very recent
in practice. For the case of requiring greater flexural capacity for an publications, which have relevance to some of the considerations in this
existing doubly symmetric beam the monosymmetric member option paper, have come to the attention of the authors. One article concerns
involves only introduction of upstands to the existing beam, whereas the the addition of flange upstands and downstands to I-section steel col­
alternative option would involve temporary works to remove the umns by wire-arc additive manufacturing (WAAM) while preserving the
existing beam, disposal of the existing beam, purchase of the new beam double symmetry of the cross-section [33], while the other article pre­
and installation of the new beam. The cost of the two options can thus be sents a numerical formulation for monosymmetric I-sections taking into

Table 6
Critical elastic moment and second moment of area - NF254 × 146 × 37 beam.
6 m Span Upstand Height

0 10 20 30 40 50 60 70 80 90 100

Mcr (kNm) 73 86 99 110 119 128 135 142 147 152 155
Ix (x106 mm4) 55 58 62 67 71 76 81 86 92 97 104
(λLT ) = 1.5

9
K. Mudenda and A. Zingoni Structures 68 (2024) 107154

account distortional modes [34]. Declaration of Competing Interest

CRediT authorship contribution statement The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
Alphose Zingoni: Writing – review & editing, Supervision, the work reported in this paper.
Conceptualization. Kenny Mudenda: Writing – original draft, Software,
Methodology, Investigation, Formal analysis, Conceptualization.

Appendix

Expanded terms for seven variables in Eqs. (1.13) and (1.14):

γ1 = 2Bt1 H − Bt 21 + 2tw hw t2 + h2w tw + bt22 (A.1)

γ2 = 2Bt1 + 2hw tw + 2bt2 (A.2)

b21 tα h t1 tα b21
x1 = − (A.3)
2 4

b21 tα
x2 = (A.4)
4

b21 tα t1 h b21 t21 tα b31 t1 h

x3 = − + (A.5)
4 16 12
[ ]2
t 3α B tα
x4 = + 2tα − (A.6)
6 2 2

t1 B3 t2 b3 hw t 3w
x5 = + + (A.7)
12 12 12

References [18] Knobloch M, Bureau A, Kuhlmann U, da Silva LS, Snijder HH, Taras A, et al.
Structural member stability verification in the new Part 1-1 of the second
generation of Eurocode 3. Steel Constr 2020;13:208–22.
[1] Ziemian RD. Guide to Stability Design Criteria for Metal Structures. Hoboken, New
[19] Quan C, Kucukler M, Gardner L. Design of web-tapered steel -section members by
Jersey: John Wiley & Sons Inc; 2010.
second-order inelastic analysis with strain limits. Eng Struct 2024;306:117763.
[2] Hill HN. The lateral instability of unsymmetrical I beams. J Aeronaut Sci 1942;9:
[20] Heins CP. Bending and Torsional Design in Structural Members. Lexington, MA:
175–80.
Lexington Book Co; 1975.
[3] Goodier JN. Cornell University Engineering Experiment Station. Flexural-Torsional
[21] Jenkins W. Longman House. Structural analysis using computers. Burnt Hill,
Buckling Bars Open Sect 1942.
Harlow, Essex CM 20 2JE, England: Longman Scientific & Technical; 1990.
[4] Bleich F. Buckling Strength of Metal Structures. New York: Mc Graw-Hill Book
[22] Timoshenko SP, Gere JM. Theory of Elastic Stability. New York: McGraw-Hill;
Company, Inc; 1952.
1961.
[5] O’Connor C. The buckling of a monosymmetric beam loaded in the plane of
[23] Prandtl L. Zeitschrift des Vereins Deutschner Ingenieure., 1907.
symmetry. Aust J Appl Sci 1964:191–203.
[24] Hsu WT, Lue DM, Chen YF. Design aid for moment strength of built-up crane
[6] Anderson JM, Trahair NS. Stability of monosymmetric beams and cantilevers.
runway girders. Int J Steel Struct 2012;12:403–17.
J Struct Div 1972;98:269–86.
[25] SAISC. Southern African Steel Construction Handbook. 6th ed. 43 Empire Road,
[7] Kitipornchai S, Trahair NS. Buckling properties of monosymmetric I-beams.
Parktown West, Johannesburg: The Southern African Institute of Steel
J Struct Div 1980;106:941–57.
Construction; 2016.
[8] Ellifritt DS, Lue D. Design of crane runway beam with channel cap. Eng J 1998;35:
[26] Mohri F, Brouki A, Roth JC. Theoretical and numerical stability analyses of
41–9.
unrestrained, mono-symmetric thin-walled beams. J Constr Steel Res 2003;59:
[9] Wang CM, Kitipornchai S. Buckling capacities of monosymmetric I-beams. J Struct
63–90.
Eng 1986;112:2373–91.
[27] Aydin R, Gunaydin A, Kirac N, Ozbasaran H. On the evaluation of critical lateral
[10] Helwig TA, Frank KH, Yura JA. Lateral-torsional buckling of singly symmetric I-
buckling loads of prismatic steel beams. Proc Fourteenth Int Conf Civ Struct
beams. J Struct Eng 1997;123(9):1172.
Environ Eng Comput 2013.
[11] Bradford M, Woolcock S, Kitipornchai S. Lateral buckling design of crane runway
[28] Yun X, Gardner L. Stress-strain curves for hot-rolled steels. J Constr Steel Res 2017;
beams. Proceedings of the Second International Conference on. Struct Stab Dyn
133:36–46.
2003.
[29] SANS (South African National Standards). Construction Works - Part CS1:
[12] Camotim D, Andrade A, Basaglia C. Some thoughts on a surprising result
Structural Steelwork. SANS 2001-CS1. 1 Dr Lategan Road. Groenkloof, Pretoria:
concerning the lateral-torsional buckling of monosymmetric I-section beams. Thin-
South African Bureau of Standards; 2012.
walled Struct 2012;60:216–21.
[30] CAN/CSA S16-19. Limit State Design of Steel Structures. Mississagua, Ontario:
[13] Nethercot DA. Inelastic buckling of monosymmetric I-beams. J Struct Div 1973;99:
Canadian Standards Authority; 2019.
1696–701.
[31] SANS (South African National Standard). Structural Use of Steel. Part 1: Limit-state
[14] Lindner J. The ultimate load of monosymmetric sections due to lateral-torsional
Design of Hot-rolled Steelwork. SANS 10162-1:2011. 1 Dr Lategan Road.
buckling. Final Rep, IABSE 1976:353–7.
Groenkloof, Pretoria: South African Bureau of Standards; 2011.
[15] Trahair NS. Inelastic buckling design of monosymmetric I-beams. Eng Struct 2012;
[32] Valeš J, Stan T. FEM modelling of lateral-torsional buckling using shell and solid
34:564–71.
elements. Procedia Eng 2017;190:464–71.
[16] Trahair NS, Bradford MA, Nethercot DA, Gardner L. The Behaviour and Design of
[33] Gardner L, Li J, Meng X, Huang C, Kyvelou P. I-section steel columns strengthened
Steel Structures to EC3. 2 Park Square. Milton Park, Abingdon, Oxon OX14 4RN:
by wire arc additive manufacturing-concept and experiments. Eng Struct 2024;
Taylor & Francis; 2008.
306:117763.
[17] Taras A, Gonzalez Puig M, Unterweger H. Behaviour and design of members with
[34] Lui YZ, Yang YB, Lui XH, Guo DZ, Xu H. A novel straight beam element for lateral-
monosymmetric cross-section. Proc Inst Civ Eng-Struct Build 2013;166:413–23.
distortional deformation analysis of frames and curved beams made of
monosymmetric I-sections. Int J Struct Stab Dyn 2024:2471007.

10