Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Behavior and Design of Steel Delta Girders

for Flexure and Shear


Omar Y. El Masri 1 and Eric M. Lui, M.ASCE 2

Abstract: A steel delta girder (SDG) is formed by welding two inclined plates in the form of a V from the web to the compression
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

flange of an I-section. Analytical, numerical, and experimental research on SDGs has shown that they possess enhanced flexural and
shear capacities over traditional I-sections with comparable cross-sectional areas. In this paper, three-dimensional nonlinear inelastic
finite-element models that take into consideration the effects of initial geometrical imperfections and residual stresses are developed to
investigate the inelastic lateral-torsional buckling and shear behavior of simply-supported SDGs under uniform bending and pure
shear. Using European H- and I-sections as base cross sections, the analyses cover a wide range of SDG dimensions and inclined
stiffener configurations. The results of these analyses demonstrate that the flexural capacities of SDGs can be represented by two of
the existing EuroCode 3 (EC3) buckling curves. Design equations for the shear capacity of SDGs are then developed and proposed in
accordance with EC3 provisions. Finally, design recommendations for selecting the proper inclined plate dimensions and configu-
rations within the delta region of SDGs are provided. DOI: 10.1061/(ASCE)ST.1943-541X.0003107. © 2021 American Society of
Civil Engineers.
Author keywords: Steel delta girders (SDG); Flexural and shear capacities; Lateral-torsional buckling; Finite-element analysis; Design
guidelines.

Introduction both the compression and tension sides of the cross section)
and proposed design equations for such beams. Khatoni (2016)
A steel delta girder (SDG), shown in Fig. 1, is an enhanced performed numerical and experimental studies on SDGs to show
I-section formed by groove welding two rectangular plates or stiff- their enhanced LTB resistance and determined that the optimal
eners from the compression portion of the web to the compression position to attach the inclined plates was at the ¼ point of the
flange of an I-section. When compared to an I-section, SDG presents web. Rose et al. (2017) demonstrated the improved shear resis-
several advantages such as increased lateral-torsional buckling (LTB) tance of SDGs through numerical and experimental studies. El
and shear capacities, reduced slenderness in the compression flange Masri and Lui (2019a, b) investigated analytically and numeri-
and web, and enhanced resistance to eccentric loading (Arabzadeh cally the behavior of SDGs under flexure and developed equations
and Varmazyari 2009; El Masri and Lui 2019a; Hadley 1961; Hatami for computing bending and torsional cross-section properties,
and Esmaeili 2013). Some applications for SDGs include (1) design verified the elastic LTB capacities of SDGs, and proposed a
of long span laterally unbraced crane bridge and crane runway residual stress pattern for use in inelastic FE modeling. Loaiza
beams, (2) retrofit of existing I-sections to enhance their flexural, et al. (2019) performed nonlinear finite-element analysis of SDGs
torsional, and shear capacities, and (3) possibility for reducing a with inclined plates provided on one or both sides of the web
beam’s depth to allow for more vertical clearance. under patch loading and concluded that the ultimate resistance
In the US, two bridges with superstructures that employed SDG of these girders was significantly increased. They also demon-
were constructed in the 1960s based on experimental results re- strated that for web thickness and height of 4 and 1,000 mm,
ported by Hadley (1961, 1964). After a period of hiatus, research and compressive flange thickness and width of 8 and 250 mm,
on SDGs resumed in 2000s with several studies that numerically the maximum increase in flexural capacities for SDGs occur when
investigated and validated the improved flexural capacities of hd =ðd − tf Þ ¼ 0.3.
SDGs using simple elastic and inelastic finite-element (FE) models While the advantages of SDGs have been experimentally and
(Arabzadeh and Varmazyari 2009; Hatami and Esmaeili 2013; numerically demonstrated, design equations for flexure and shear
Mohebkhah and Azandariani 2015; Sahnehsaraei and Erfani 2014). still need to be developed before this type of girders can become a
Avery et al. (2000) conducted finite-element analyses of thin- routine design option for engineers and designers. The primary ob-
walled cold-formed hollow flange beams (with delta regions in jective of this paper is therefore to propose flexure and shear design
equations for SDGs in accordance with EuroCode 3 (EC3) speci-
1
Adjunct Professor, Dept. of Civil and Environmental Engineering, fications (CEN 2005).
Syracuse Univ., Syracuse, NY 13244-1240 (corresponding author). Email:
oyelmasr@syr.edu
2
Meredith Professor, Dept. of Civil and Environmental Engineering,
Syracuse Univ., Syracuse, NY 13244-1240. ORCID: https://orcid.org Overview of the EC3 Flexural and Shear Capacity
/0000-0002-9459-8204. Email: emlui@syr.edu Equations
Note. This manuscript was submitted on October 20, 2020; approved on
April 19, 2021; published online on July 14, 2021. Discussion period open Flexural Capacities
until December 14, 2021; separate discussions must be submitted for in-
dividual papers. This paper is part of the Journal of Structural Engineer- The lateral-torsional buckling (LTB) capacity curves in EC3
ing, © ASCE, ISSN 0733-9445. (CEN 2005) are based on numerical calibrations of finite-element

© ASCE 04021141-1 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


Table 2. Recommended LTB curves for rolled sections or equivalent
welded sections from CEN (2005)
Cross section Limits Buckling curve
Rolled I-sections d=bc ≤ 2 b
d=bc > 2 c
Welded I-sections d=bc ≤ 2 c
d=bc > 2 d
Source: Data from CEN (2005).

value), respectively. A more thorough discussion of the EC3


approach to flexural design of beams is given by Nethercot and
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

Gardner (2005).
The recommended LTB curves for rolled sections and equiva-
lent welded sections are provided in Table 2. For slenderness
λ̄LT ≤ λ̄LT;0 , the section should be designed for its full plastic mo-
Fig. 1. Typical SDG with dimension notations. ment capacity M p for Class 1 (Plastic) and Class 2 (Compact) non-
composite beams.

Shear Capacity
simulations (Rebelo et al. 2009). The flexural capacity of a laterally
unrestrained beam is computed using the following equation: In the EC3 (CEN 2005) provisions, a member should be designed
for its plastic shear resistance V pl;Rd when
Fy sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
M b;Rd ¼ χLT W y ð1Þ
γ M1 h 72 235
≤ ð5Þ
tW η Fy ½MPa
where M b;Rd is the design flexural strength, χLT is a reduction fac-
tor for LTB, Fy is the nominal yield stress of steel, γ M1 is a partial where η is defined by the National Annex or set equal to 1.2 as
safety factor to account for the member’s resistance to instability, recommended by EC3 (CEN 2006) for steel grades up to S460
and W y is the section modulus about the strong axis and is depen- and equal to 1.0 for higher grades. The National Annex allows each
dent on the cross-section class. nation to independently select the parameters that are most suitable
For the case of rolled sections or equivalent welded sections for its conditions. η is a factor that takes into account the reduced
(Section 6.3.2.3 in EC3), the reduction factor is computed using area and the strain hardening effect of the material. EC3 states that η
the expression may conservatively be taken as 1.0 (CEN 2005). For cases where
 shear buckling is expected to control the design (i.e., when h=tw >
1 1.0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
χLT ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ≤ smaller of ð2Þ ð72=ηÞ 235=Fy ), the code user is referred to Part 5 of EC3 (CEN
2 2 1=λ̄2LT 2006) “Plated Structural Elements” to compute the shear resistance.
ΦLT þ ΦLT − β λ̄LT
For convenience, the plastic shear resistance V pl;Rd will be de-
in which noted as V n and is given by
Fy Av
ΦLT ¼ 0.5½1 þ αLT ðλ̄LT − λ̄LT;0 Þ þ β λ̄2LT  ð3Þ V n ¼ pffiffiffi ð6Þ
3γ M0
and
where γ M0 is a partial resistance factor defined in the National An-
sffiffiffiffiffiffiffiffiffiffiffiffi
nex with a recommended value for buildings equal to 1.0, and Av is
W y Fy
λ̄LT ¼ ð4Þ the shear resistance area determined as follows:
M cr 1. For hot-rolled H- and I-sections loaded parallel to the web

is the nondimensional slenderness for LTB, where αLT is the im- Av ¼ A − 2bf tf þ ðtw þ 2rÞtf but not less than ηhtw ð7Þ
perfection factor provided in Table 1, and M cr is the theoretical
elastic LTB moment. The equation for M cr is not explicitly pro- 2. For welded H- and I-sections loaded parallel to web
vided in EC3. Users are expected to use the appropriate equation Av ¼ ηhtw ð8Þ
from available references. The recommended values for the param-
eters λ̄LT;0 and β are 0.4 (maximum value) and 0.75 (minimum where A is the full cross-sectional area, bf and tf are the width
and thickness of the flange, respectively, h is the web depth, and
r is the root radius.
Table 1. Recommended values of LTB imperfection factors
Buckling curve Imperfection factor, αLT
Finite-Element Model Description and Validation
a 0.21
b 0.34 To simulate the LTB and shear behavior of SDGs, three-
c 0.49 dimensional (3D) nonlinear inelastic FE models were developed
d 0.76 using the commercial finite-element software Abaqus (Simulia
Source: Data from CEN (2005). 2014).

© ASCE 04021141-2 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


FE Modeling of SDGs under Uniform Bending conservative results if hot-rolled I-sections are to be used as base
sections for the SDGs.
Geometry and Material Properties Geometrical imperfection in the form of member out of straight-
S4R shell elements were used to model all the component elements ness was modeled as follows: An eigenvalue analysis was per-
of the girders. The European standard structural steel grade EN formed to obtain the first or lowest global out-of-plane buckling
10025 S355 steel was used. S355 steel has a nominal minimum mode for the girder. This buckled mode shape was then scaled so
yield strength Fy of 355 MPa for plate thickness up to 16 mm, that its maximum out-of-plane deflection would equal Lb =1,000,
but the minimum yield strength reduces to 345 MPa for plate thick- where Lb is the beam’s unbraced length. This scaled deformed
ness between 16 and 40 mm. Since the thickness of some of the shape of the girder was then used in the FE model as initial nodal
component elements for the heavier HEA and IPE sections used displacements.
in the present study exceeds 16 mm, a more conservative value
of 345 MPa was used in all the analyses. The ultimate strength Meshing, Boundary Conditions, and Analysis
Fu for S355 steel is 470–630 MPa for plate thickness between All SDGs were modeled with a maximum element size of 5 cm. In
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

3 and 100 mm. The engineering and true ultimate strengths used cases where the width of the flange is equal to or less than 12 cm,
in the finite-element models were 450 and 504 MPa, respectively. the approximate maximum element size used was 3 cm. In addition,
The modulus of elasticity and Poisson’s ratio were set to 200 GPa smaller elements were used around the welded regions to accurately
and 0.33, respectively. The stress-strain curve used in the FE analy- implement the residual stress values into the models. A conver-
ses was developed using the following assumptions: (1) in the yield gence study was then performed to ensure that the selected mesh
plateau region, the stress is constant; (2) at the onset of strain hard- sizes would provide good solutions.
ening, the strain is ten times the yield strain (Subramanian and All girders were modeled as simply-supported members.
White 2017a, b); and (3) three segments are used to model the Simply-supported boundary conditions were modeled by restraining
strain hardening region. The first initial strain hardening region the nodes located at the centroids at the ends of girder as follows: At
ends at stress and strain values of 415 MPa and 0.03125 m=m, re- one end, all three translational degrees of freedom and the rotational
spectively; the second region ends at stress and strain values of degree of freedom about the longitudinal axis were restrained. At the
450 MPa and 0.12 m=m, respectively, whereas the stress in the other end, the same restraints, except for the longitudinal transla-
third stain hardening region remains constant at 450 MPa. This en- tional degree of freedom, were applied. Because the ends of the
gineering stress-strain curve was then converted into a true stress- girder are not allowed to rotate about its longitudinal axis but allowed
strain curve as depicted in Fig. 2 and used in all FE models. to warp, the tips of both the top and bottom flanges at each end of the
Both material and geometric imperfections were implemented in girder were also restrained against out-of-plane translation in the
the analyses. Material imperfection in the form of residual stresses FE model.
as shown in Fig. 3 was introduced in all the FE models as prede- When concentrated forces or moments are applied directly to the
fined initial stresses. The pattern was developed based on superpo- nodes of a shell element, stress concentration or local buckling
sition of residual stresses for welded I-sections and rectangular steel often occurs. To circumvent these problems, equal and opposite
plates (El Masri and Lui 2019b). When compared to welded I- concentrated moments were applied to reference points located
sections, tensile stress regions are present in the compression flange away from the girder ends. A surface-based kinematic coupling
and web of SDGs due to welding of the inclined stiffeners. The was used to couple the rotational motion of these reference points
pattern satisfies the required condition of cross-section equilibrium to their respective end cross sections of the girder (Subramanian
in the longitudinal direction and the observed condition of material and White 2017c). Additionally, vertical stiffeners were provided
yielding at all weld locations. Because residual stresses are known at both ends to preclude any issues related to stress concentration,
to be higher in welded I-sections when compared to hot-rolled local buckling, and web yielding. The analysis began with an initial
I-sections due to the higher heat input during the welding process, general static stabilization step to allow the girder to reach a state of
the residual stress pattern given in Fig. 3 is believed to yield equilibrium under the applied residual stresses and geometrical

600

500

400
Stress (MPa)

300

200

Engineering Stress and Strain


100
True Stress and Strain
0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Strain (m/m)

Fig. 2. Engineering and true stress-strain curves.

© ASCE 04021141-3 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Residual stress pattern for SDG (units in MPa and mm; see Fig. 1 for definitions).

imperfections. A second analysis step was then employed to per- Based on these results for inelastic LTB and those reported by El
form the nonlinear inelastic buckling analysis using the modified Masri and Lui (2019b) for elastic LTB, it is believed the proposed
Riks buckling method. FE modeling technique can be used to provide reasonably good
estimates of the LTB capacities of beams.
FE Model Validation
Test results on nine simply-supported I-beams designed to fail by
inelastic LTB under various loading patterns were reported by Dux Nonlinear FE Modeling of SDG under Pure Shear
and Kitipornchai (1983). The cross-section dimensions, material A nonlinear 3D FE model was created to determine the shear
properties, geometrical imperfections, and residual stresses were resistance of SDG under pure shear. The model was based on
experimentally measured and provided. Using these data, FE models previous studies that examined the shear strength of I-sections
for three test beams (Experiment number 1, 2, and 3) subjected to a (Glassman et al. 2016; Lee and Yoo 1998). In the present study, the
three-point bending load were developed using the methodology de- length of the SDG model was taken as the height of the web below
scribed in the preceding sections. All experimentally measured sec- the delta region (i.e., L ¼ h − hd ) (Fig. 1). The type of element,
tion properties and imperfections were implemented into the FE material properties, and analysis procedure used are similar to those
models. The material stress-strain curve was developed using the described in the section “FE Modeling of SDGs under Uniform
same methodology given in the section “Geometry and Material Bending.” The maximum mesh size used was 3 cm. The FE model
Properties.” The applied load and boundary conditions that simulate
the experimental setup were incorporated in the FE models so the
results can be directly compared. The inelastic LTB loads obtained
using the FE models for the three test beams were 45.6, 60.0, and
66.8 kN, respectively; upon comparison with the test results of 47.1, Translation Rotation
Edge
62.6, and 71.0 kN, an average error of 4.4% was obtained. The small
difference between the experimental and FE results can be contrib- X X X X X
uted to the following modeling approximations: X X X X
1. the fillets at the junction of the web and flanges (which consti- X X X X
tutes about 1% of the total area of a 250UB37 section) were X X X
neglected in the FE models, and
2. the mean values of six sets of measured residual stresses calcu-
Fig. 4. FE model and edge boundary conditions used for SDG shear
lated using an average modulus of elasticity obtained from 12
analysis.
tests were used in the FE simulations.

© ASCE 04021141-4 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


and boundary conditions employed in the analysis are shown in • The SDGs were selected for a range of d=bc ratios that cover the
Fig. 4. The shear force V was applied monotonically along edge highest and lowest values of standard European sections to as-
2 in a vertical direction. Similar load application and boundary con- sess how this ratio could affect the flexural capacities.
ditions have been used and verified by Glassman et al. (2016). The
residual stress pattern shown in Fig. 3 was incorporated into the
model. The initial geometrical imperfections were also included FE Simulation Results
in the FE model as the eigenmode shape associated with the lowest The FE analysis results for SDGs with d=bc ≤ 2 and d=bc > 2,
positive eigenvalue scaled to h=150, where h=150 is the maximum normalized by the plastic moment M p of the cross section, are plot-
allowable web out-of-flatness for girders without intermediate stiff- ted against the nondimensional slenderness given by Eq. (4) in
eners (AWS 2010). An initial eigenvalue analysis was performed to Figs. 5 and 6. The markers represent data points obtained at various
obtain the eigenmode as described in the section “Geometry and unbraced lengths. Also shown in these figures are the four EC3
Material Properties.” Although the magnitude of initial imperfec- flexural capacity curves for rolled sections or equivalent welded
tions mainly affects the ultimate postbuckling shear capacity of sections discussed in the section “Overview of the EC3 Flexural
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

girders and has negligible effect on the shear yielding capacity, ini- and Shear Capacity Equations.” Based on the results, the following
tial imperfections need to be incorporated in the FE model to iden- observations can be made:
tify the occurrence of inelastic shear buckling and to verify that web • the shape of each FE flexural capacity curve closely matches
buckling does not occur below the recommended shear yielding those of the EC3 curves;
capacity. • EC3 curve a or b that corresponds to rolled or equivalent welded
sections gives good approximation to the numerically obtained
flexural capacity curves for all the SDGs used in the present
Flexural Capacities of Simply-Supported SDGs study;
Subject to Uniform Bending • for short SDGs whose slenderness falls in the plateau region, the
FE simulation results are within 2% of the theoretical plastic
moment capacity Mp for all girder sections analyzed; and
Girder Configurations and Variables • the results of the FE simulations are in agreement with EC3 rec-
To determine the LTB flexural capacities of simply-supported ommendations for the slenderness limit of the plateau region.
SDGs under uniform bending, FE simulations were performed us- A comparison of the flexural capacities for all SDGs analyzed in
ing the methodology presented in the section “FE Modeling of this study and the EC3 curves is provided in Table 4. The compari-
SDGs under Uniform Bending.” The girders were selected to cover son is made for d=bc ≤ 2 and d=bc > 2 in the form of M FEM =
various d=bc ratios and four common configurations of inclined M ECðkÞ , where M FEM is the flexural capacity obtained using FE sim-
stiffeners. The base I-sections to which the inclined stiffeners are ulation, and M ECðkÞ is the flexural capacity obtained from EC3
attached are welded European standard H- or I-sections classified curve k, where k ¼ a, b, c, or d. This ratio was computed at each
as compact sections. To avoid local buckling, the selected thickness unbraced length Lb , and the mean, maximum (max), minimum
of the inclined stiffeners was either greater than or at least equal to (min), and standard deviation (SD) values of M FEM =M ECðkÞ for
the web thickness of the base I-sections. The dimensions and the all computed values are given. The results show that EC3 curves
c and d provide buckling moments with average values of 10%
d=bc ratios used for the 14 representative sections are provided in
(conservative) and 23% (conservative), respectively. On the other
Table 3.
hand, EC3 curves a and b provide results with average values of
The following variables were considered in this study:
6% (unconservative) and 2% (conservative), respectively.
• Each SDG was modeled using 8 to 12 different unbraced lengths
In another study (El Masri and Lui 2019b), a detailed compar-
so the full range of the flexural capacity curve could be ob-
ative analysis using different levels of residual stresses and geomet-
tained. The lengths were selected to cover the plastic moment,
rical imperfections has shown that a rather large discrepancy occurs
inelastic LTB, and elastic LTB regions.
between the FE simulation results and the AISC (2016) flexural
• Four different inclined stiffener configurations were utilized
design equations, with the AISC equations overestimating the
to evaluate how these configurations could affect the flexural
LTB capacity of SDGs by up to 20% when the levels of imperfec-
capacity.
tions comparable to those used in the development of these design
equations were used. This discrepancy can be attributed to the
method used to develop the AISC equations. The AISC flexural
Table 3. Dimensions of representative SDG used in FE analysis design equations were developed primarily from curve-fitting of
experimental data, and since no data for SDGs were used in the
FE model no. Base section d=bc bd hd td (mm)
calibrations, these design equations do not represent the LTB
1 IPE 360 2.12 bc =2 h=3 8 behavior of SDGs well.
2 IPE 550 2.62 bc =2 h=5 12
3 IPE 550 2.62 2bc =3 h=5 12
4 IPE 550 2.62 bc =2 h=3 12 Proposed Flexural Design Equations for SDGs
5 IPE 550 2.62 2bc =3 h=3 12
6 HEA 400 1.3 bc =2 h=5 12 The data presented in Table 4 show that EC3 curves a and b provide
7 HEA 400 1.3 2bc =3 h=5 12 relatively good estimates for the flexural capacities of SDGs with
8 HEA 400 1.3 bc =2 h=3 12 an average error of 2% (unconservative) when compared to the FE
9 HEA 400 1.3 2bc =3 h=3 12 simulation results. The maximum differences are 6% (unconserva-
10 HEA 600 1.97 2bc =3 h=3 14 tive) for d=bc ≤ 2 and 7% (unconservative) for d=bc > 2. These
11 HEA 280 0.96 bc =2 h=5 8 maximum errors occur at the largest Lb value for each SDG
12 IPE 240 2.00 2bc =3 h=3 8 (i.e., in the elastic range). As a result, EC3 curve a is recommended
13 IPE 300 2.00 2bc =3 h=3 8 for use in the design of Class 1 (compact) SDG with d=bc ≤ 2, and
14 HEA 220 0.95 bc =2 h=3 8
EC3 curve b is recommended for use in the design of Class 1 SDG

© ASCE 04021141-5 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


1.1

1.0

0.9

0.8
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

Series1 Curve b

0.7
Curve c Curve d

0.6

0.5

0.4

0.3
0.20 0.40 0.60 0.80 1.00 1.20 1.40

Non-dimensional slenderness

Fig. 5. Comparison of normalized FE and EC3 flexural capacity curves (d=bc ≤ 2).

with d=bc > 2. These recommendations are summarized in Table 5. EC3 (CEN 2005) equations for I-section with the recommendations
The parameters λ̄LT;0 and β discussed in the section “Overview of given in the section “Proposed Flexural Design Equations for
the EC3 Flexural and Shear Capacity Equations” are also recom- SDGs.” The result of Test 2 was not utilized in the comparison
mended for use since the FE results show good agreement with since the inclined stiffeners of this girder were welded to the tension
EC3 in the plateau region of the flexural capacity curves. Note that portion of the web which is not the focus of the current study.
the flexural capacity curve selection is independent of the inclined For the I-section of Test 1, a conservative 9.5% difference in
stiffener configurations. This is because when one compares the flexural capacities was observed [8.6 kN·m (EC3) versus 9.5 kN·m
results shown in Fig. 5 for SDGs 6–9, which use HEA 400 (experimental)]. For the SDG of Test 3, a conservative 8.0% differ-
(d=bc ≤ 2) as the base section with four different inclined stiffener ence in flexural capacities was observed [20.6 kN · m (section “Pro-
configurations, it can be seen that they all cluster around EC3 curve posed Flexural Design Equations for SDGs” recommendations)
a. Similarly, when one compares the results shown in Fig. 6 for versus 22.4 kN · m (experimental)]. For purpose of comparison,
SDGs 2–5, which use IPE 500 (d=bc > 2) as the base section with the experimental result of this test after being normalized by the plas-
four different inclined stiffener configurations, it can be seen that tic moment Mp of the section is shown in Fig. 5. Given the uncer-
they all cluster around EC3 curve b. tainties associated with any tests, the results obtained herein
demonstrate that the recommendations given in the section “Pro-
posed Flexural Design Equations for SDGs” can be used to estimate
Evaluation of Proposed Equations the flexural capacities of SDGs with reasonable accuracy.
Khatoni (2016) performed experimental tests on three girders that
failed by inelastic LTB. All three girders were IPE14 sections with
Guidelines for Inclined Stiffener Configurations
st37 steel properties. Test 1 used an I-section without any inclined
stiffeners, whereas Tests 2 and 3 used I-sections with inclined stiff- Several previous studies have provided recommendations for place-
eners placed at hd ¼ h/2 and h/4, respectively. The buckling mo- ment of the inclined stiffeners on an I-section to elicit the most ben-
ments of the girders used in Test 1 and Test 3 were computed and eficial effect (Hatami and Esmaeili 2013; Khatoni 2016; El Masri
compared to the flexural moment capacities calculated using the and Lui 2019a; Loaiza et al. 2019). In particular, the location where

© ASCE 04021141-6 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


1.1

1.0

0.9

0.8
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

0.7

0.6

0.5

0.4

0.3
0.20 0.40 0.60 0.80 1.00 1.20 1.40

Non-dimensional slenderness

Fig. 6. Comparison of normalized FE and EC3 flexural capacity curves (d=bc > 2).

Table 4. Results comparison for SDGs with different d=bc


d=bc Statistics M FEM =MECðaÞ MFEM =M ECðbÞ MFEM =M ECðcÞ M FM =M ECðdÞ
≤2 Mean 0.98 1.06 1.15 1.29
Maximum 1.03 1.14 1.26 1.46
Minimum 0.94 1.00 1.00 1.01
Standard deviation 0.01 0.04 0.07 0.13
>2 Mean 0.91 0.98 1.06 1.18
Maximum 1.01 1.02 1.11 1.28
Minimum 0.84 0.93 0.98 0.99
Standard deviation 0.04 0.01 0.03 0.08

Table 5. Recommended EC3 curves for the design of Class 1 SDG guidelines to determine the optimum inclined stiffener configura-
Cross section Limits Buckling curve tions, a parametric study that takes into account inelasticity effect
and design economy is performed. In the present study, a parameter
SDG d=bc ≤ 2 a
κ is introduced to examine various inclined stiffener configurations
d=bc > 2 b
and their effect on the flexural capacities of SDGs. κ is defined as
the ratio of the percent increase in flexural capacity to the percent
the stiffeners are attached to the web of the base I-section (i.e., hd increase of cross-section area when inclined stiffeners are added to
shown in Fig. 1) is of importance. Based on elastic FE models and the I-section. In other words, a κ value of 10 means there is a 10%
limited experimental tests, each of these studies has resulted in a increase in flexural capacity for a 1% increase in cross-section area
different recommendation for hd. To provide designers with (weight) when inclined stiffeners are added to the section.

© ASCE 04021141-7 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


Table 6. H- and I-sections used in the analysis the thickness of the inclined stiffeners beyond that of the web is
Base section d=bc td (mm) not an effective method to increase the LTB capacity of SDGs.
As a result, the thickness of the inclined stiffeners was selected
HEA 240 0.96 8.0
HEA 280 0.96 8.0
to be equal to or slightly larger than the web thickness based on
HEA 400 1.30 12.0 plate thickness that is available at 2-mm increment. The κ values
HEA 600 1.97 14.0 for all these girders were then calculated and compared. The con-
IPE 300 2.00 8.0 figuration that gives the largest κ value is considered the most
IPE 360 2.12 8.0 desirable.
IPE 450 2.37 10.0 In addition, to investigate the effect of the girder’s length on the
IPE 550 2.62 12.0 optimum inclined stiffener configurations, κ values for each girder
with a specific stiffener configuration were evaluated at seven lat-
eral unbraced lengths Lb . The computations were made at Lp (plas-
In the parametric study, flexural capacities of eight standard tic moment capacity), LrI (i.e., Lr for the base H- and I-sections),
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

welded H- and I-sections shown in Table 6 along with their LrD (i.e., Lr for the SDG), selected lengths between Lp and LrI ,
d=bc ratio and the thickness of the inclined stiffeners were com- and lengths higher than LrI and LrD . The length Lr is the limiting
puted. Each used 16 different inclined stiffener configurations that unbraced length for the limit state of inelastic lateral-torsional
cover the cases bd ¼ 2bc =5;bc =2;2bc =3;3bc =4, and hd ¼ h=5; h=4; buckling (limit between elastic and inelastic LTB) and is given
h=3; 2h=5. El Masri and Lui (2019a) have shown that increasing for monosymmetric beams by White and Jung (2003) as

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffi u  2  ffi
1.38E I y J u t 2.6β x FL Sxc 2.6β x FL Sxc 27.0Cw FL Sxc 2
Lr ¼ þ1þ þ1 þ ð9Þ
Sxc FL EJ EJ Iy EJ

where E is the modulus of elasticity, J and Cw are the torsional and values appear to be more effective at shorter unbraced lengths
warping restraint constants, and β x is given by Kitipornchai and and vice versa. For SDGs with d=bc ≤ 1 and d=bc ≥ 2, the results
Trahair (1979) as are more consistent in that regardless of the unbraced lengths, the
   2  configuration with the smaller hd and bd is the most effective for
2I yc Iy girders with d=bc ≤ 1, and the configuration with the smaller hd
β x ¼ 0.9ho −1 1− ð10Þ
Iy Ix and larger bd is the most effective for girders with d=bc ≥ 2. This
is because girders with d=bc ≥ 2 have a relatively smaller flange
in which ho is the distance between the flange centroids, I yc is mo- width when compared to girders with d=bc < 2. The only exception
ment of inertia of the compression flange about the minor axis (axis is when Lb ¼ Lp wherein the cross section develops the full plastic
of symmetry), I y is the moment of inertia of the full cross section moment M p before LTB occurs. Since the main purpose of adding
about the minor axis, I x is the moment of inertia of the full cross inclined stiffeners is to increase the LTB capacity of an I-section and
section about the major axis, and FL is the magnitude of flexural because LTB does not occur at Lb ≤ Lp (or λ̄LT ≤ λ̄LT;0 ), low values
stress in the compression flange at which LTB is influenced by of κ are obtained at or near M p. For this situation, other alternative
yielding and is given by solutions (e.g., the use of cover plates) should be considered. The
8 Sxt results also show that the use of inclined stiffeners becomes more
>
< 0.7Fy when ≥ 0.7 effective (i.e., gives higher κ values) when the unbraced length of
Sxc
FL ¼ ð11Þ the beam increases and that a hd (Fig. 1) value of h=5 or h=4 often
>
: Fy Sxt ≥ 0.5Fy S
when xt < 0.7 provides better results than when hd is equal to h=3 and 2h=5.
Sxc Sxc
To gain more insight into the effectiveness of the use of inclined
where Sxt and Sxc are the elastic section moduli with respect to the stiffeners to increase the LTB capacity of H- and I-sections, the
tension flange and the compressive flange, respectively, and Fy is highest κ factors for all the SDGs analyzed at Lb ¼ LrI were ex-
the yield stress of steel. amined. The average κ value for these H- and I-sections is 5.43.
The total number of SDGs used for this comparison is 896 (eight This means a 10% increase in cross-section weight due to the ad-
girders, each with 16 different inclined stiffener configurations, dition of inclined stiffeners will result in a 54% increase in the LTB
evaluated at seven unbraced lengths). Table 7 shows sample results capacity of the base I-girder.
of κ values evaluated for a welded HEA 400 section. The bold val- Based on the above observations, the recommended values for
ues represent the most effective stiffener configuration for each un- hd and bd are given in Table 8. The values are given as a range so
braced length Lb (i.e., the highest κ value). designers can select the most appropriate configuration based on
Based on the analysis results, it was observed that the most ef- other design constraints and requirements. For example, lateral de-
fective inclined stiffener configurations differ for sections with flection is an important design consideration for crane runway
d=bc ≤ 1, 1 < d=bc < 2, and d=bc ≥ 2. For SDGs with 1 < d=bc < 2, beams; thus, a larger bd value would provide a higher moment
the highest values of κ occur for different inclined stiffener configu- of inertia about the axis of symmetry, allowing for a better resis-
rations at different unbraced lengths. However, smaller bd and hd tance to lateral deflection. These recommendations can also be used

© ASCE 04021141-8 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


2.60
Table 7. Effect of inclined stiffener configurations on a welded HEA 400
2.40
bd 2.20
Lb hd 2bc =5 bc =2 2bc =3 3bc =4 2.00

Lp h=5 0.39 0.54 0.49 0.46 1.80


h=4 0.48 0.46 0.42 0.40 1.60
h=3 0.37 0.35 0.33 0.32 1.40
2h=5 0.30 0.29 0.28 0.27 1.20
5m h=5 1.96 1.87 1.70 1.62 1.00
h=4 1.74 1.69 1.59 1.53
0.80
h=3 1.44 1.44 1.39 1.36 SDG1 SDG2 SDG3
0.60
2h=5 1.25 1.27 1.25 1.23 SDG4 SDG5 SDG6
LrI h=5 5.47 5.35 4.95 4.73 0.40 SDG7 SDG8 SDG9
SDG10 SDG11 SDG12
h=4 5.03 5.00 4.73 4.55 0.20
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

SDG13 SDG14 HEA400


h=3 4.32 4.37 4.24 4.13 0.00
2h=5 3.83 3.91 3.85 3.78 0.00 0.02 0.04 0.06 0.08 0.10
15 m h=5 7.85 7.99 7.72 7.49 γ
h=4 7.52 7.76 7.62 7.42
h=3 6.75 7.04 7.05 6.93 Fig. 7. Comparison of normalized shear capacities.
2h=5 6.13 6.42 6.49 6.42
20 m h=5 8.99 9.39 9.43 9.30
h=4 8.83 9.40 9.61 9.51
h=3 8.22 8.87 9.22 9.18 (V FEM =V n;I ), are plotted against the shear strains γ. For comparison
2h=5 7.65 8.27 8.65 8.65 purposes, the figure also includes the results of the welded I-section
LrD h=5 8.13 8.90 9.56 9.76 HEA 400 which was used as the base I-section for SDGs 6, 7, 8,
h=4 8.36 9.42 10.56 10.95 and 9. V n;I is the EC3 (CEN 2005) nominal shear resistance calcu-
h=3 8.31 9.73 11.53 12.22 lated using Eqs. (6)–(8) with η ¼ 1.0. The shear strain γ was ob-
2h=5 8.15 9.72 11.88 12.78 tained by dividing the vertical displacement of the loaded side
25 m h=5 9.51 10.10 10.38 10.36
(i.e., Edge 2 in Fig. 4) by the length of the FE model.
h=4 9.50 10.30 10.84 10.87
h=3 9.04 10.00 10.74 10.84 The results show the expected increase in nominal shear
2h=5 8.55 9.49 10.28 10.40 capacities of SDGs when compared to the corresponding base
I-sections. Shear yielding initiates in SDGs at a V FEM =V n;I ratio
Note: Bold values represent the most effective stiffener configuration for higher than one, and a yield plateau is absent for all SDGs. The
each unbraced length Lb .
lack of a yield plateau can be explained by the difference in shear
stress distribution of an SDG when compared to an I-section.
Fig. 8(a) shows a typical shear stress distribution over the cross
as a starting point for the design of SDGs. It should be noted that to section of an I-section. Note that the shear stress τ ave does not vary
prevent local buckling from occurring in the inclined stiffeners, the appreciably over the depth of the web and is much higher than that
following width-to-thickness ratio based on EC3 (CEN 2005) Class of the flanges (Beer et al. 2015). As a result, for practical design
1 cross sections should be checked considerations, the shear force V is often assumed to be carried
sffiffiffiffiffiffiffiffi only by the web. A typical shear stress distribution over the cross
wd 235 section of an SDG is shown in Fig. 8(b). In the figure, τ ave is nor-
≤ 72 ð12Þ
td Fy malized by the maximum shear stress τ m obtained at the centroid of
the cross section under a unit shear force V. Fig. 8(b) shows that the
where Fy is the yield stress of the inclined stiffeners in MPa, and wd maximum shear stress in the inclined stiffeners and in the web re-
and td are the width and thickness of the inclined stiffeners, gion between the stiffeners is approximately one-third the maxi-
respectively. mum shear stress in the web portion below the delta region.
However, depending on the cross section, this shear stress ratio
was found to vary from 0.2Fy to 0.4Fy . This difference in the mag-
Shear Capacities of SDGs nitude of shear stress in the cross section explains why a yield pla-
teau is absent in SDGs as yielding occurs gradually over the cross
FE Analysis Results and Discussions section, starting with the web portion below the delta region and
Finite-element analyses were performed on the same SDGs given progressing toward the delta region as depicted in Fig. 9. Moreover,
in Table 3 using the model described in the section “Nonlinear FE the analysis results show that the configuration of the inclined stiff-
Modeling of SDG under Pure Shear” to determine their shear eners affects the shear capacity only after yielding initiates in the
capacities. The results are given in Fig. 7 where the shear capacities cross section. In this regard, the FE results show that SDGs with
of the SDG sections, normalized by those of the base I-sections higher hd =h ratio provide higher post yield shear capacities while
the bd =bc ratio has a negligible effect on the shear capacities. Addi-
tionally, web buckling is not a concern as it occurs at a very high
Table 8. Recommendations for inclined stiffener configurations shear load V and shear strain γ (shown as a dip in Fig. 10), after the
stresses in the cross section have reached the ultimate stress Fu .
Limits Length hd bd
d=bc ≤ 1 Any h=5 to h=4 2bc =5 to bc =2
1 < d=bc < 2 Lb ≤ LrI h=5 to h=4 2bc =5 to bc =2 Proposed Shear Design Equations for SDG
Lb > LrI h=5 to h=4 bc =2 to 3bc =4 In line with the design equations for flexure, the proposed design
d=bc ≥ 2 Any h=5 to h=4 2bc =3 to 3bc =4
equations for the shear of SDG sections will follow the format of

© ASCE 04021141-9 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


Web Inclined stiffener
600 600

500 500

400 400

h (mm)
h (mm)
300 300

200 200

100 100

0 0
0 0.10.20.30.40.50.60.70.8 0.9 1 1.1 0 0.1 0.2 0.3 0.4
τave/τm τave/τm
(a) (b)
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Typical shear stress distribution of (a) an I-section; and (b) SDG model No. 5.

4
3.5
3
2.5
2
1.5
1
0.5
0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
γ

Fig. 10. Web buckling identified on the shear resistance curve of SDG
model No. 12.

width of the inclined stiffeners, and td and tw are the thicknesses of


Fig. 9. von Mises shear stress distribution in SDG model No. 5 (unit the inclined stiffeners and the web, respectively.
in kPa). The 0.5 factor in front of Ad in Eqs. (14) and (15) accounts for
the nonuniform shear stress distribution in SDGs as explained in
the section “FE Analysis Results and Discussions” where it has
EC3 (CEN 2005) provisions. The following equations are proposed been shown that lower shear stresses are obtained in the Ad region
for determining the shear capacities of Class 1 SDG sections: when compared to the Awl region. Furthermore, while it is assumed
that both the web and the inclined stiffeners have reached the yield
F y Av stress when η is obtained using Eq. (15), only the lower portion of
V n ¼ pffiffiffi ð13Þ
3γ M0 the web Awl is assumed to have yielded when η is conservatively
taken as 1.0. Table 9 shows the shear capacities of SDGs normal-
in which Fy is the nominal yield stress of steel, γ M0 is a partial ized by the shear capacities of the corresponding I-sections
resistance factor as defined in Eq. (6), and (η ¼ 1.0), obtained using η calculated from Eq. (15) and a
conservative value of η ¼ 1. The table also shows the shear strain
Av ¼ ηðAwl þ 0.5Ad Þ ð14Þ γ that corresponds to the shear capacity V n. The results show that
using the conservative value of 1 for η provides an 11% to 26%
0.5Ad increase in the nominal shear capacity with respect to the base
η ¼1þ ð15Þ
Awl þ 0.5Ad I-section, whereas using η computed using Eq. (15) provides a
41%–89% increase in the nominal shear capacity (18%–58% in-
Awl ¼ ðh − hd Þtw ð16Þ crease based on η ¼ 1.2 for I-sections). The shear strains obtained
using a η value of 1.0 are relatively small, whereas shear strains of
Ad ¼ hd tw þ 2wd td ð17Þ up to 5.3% are obtained using the proposed equation for η.

where Av is the shear resistance area, η is a factor that accounts for


strain hardening and nonuniform shear stress distribution and may Conclusions
conservatively be taken equal to 1.0, Awl is the area of the web
portion below the delta region, Ad is the area of the delta region, This paper addresses the design of Class 1 SDGs for flexure and
h is the web height, hd is the height of the delta region, wd is the shear. Various prismatic simply-supported SDGs made from some

© ASCE 04021141-10 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


Table 9. Proposed normalized shear capacity and the corresponding shear SDGs under monotonically increasing loads. The equations pro-
strain γ vide flexibility for designers by either allowing strain hardening
SDG No. V n =V n;I (η ¼ 1.0) γ (%) V n =V n;I γ (%) in the cross section [i.e., use Eq. (15) to calculate η] or conserva-
tively ignoring it (i.e., use η ¼ 1). Within the range of the girders
1 1.15 0.38 1.63 2.8
2 1.11 0.43 1.41 2.9
analyzed, a 41%–89% increase in shear capacity for the SDG
3 1.13 0.50 1.46 4.3 over that of the corresponding base I-section has been observed.
4 1.16 0.44 1.65 4.3 A detail design example of the problem in Fig. 11 is provided in
5 1.18 0.48 1.70 5.3 the Appendix. In the example, an HEA 1,000 section is replaced
6 1.17 0.42 1.54 2.8 with an HEA 500 section with a pair of inclined stiffeners, which
7 1.23 0.60 1.65 4.0 results in a reduction of 95 kg=m (35%) and 50 cm (51%) in the
8 1.21 0.38 1.76 2.9 cross-section weight and depth, respectively, as illustrated in
9 1.26 0.51 1.86 3.9 Fig. 12.
10 1.20 0.46 1.74 4.4
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

11 1.20 0.55 1.60 3.4


12 1.26 0.59 1.85 3.9 Appendix. Design Example: A Simply-Supported
13 1.21 0.53 1.76 4.1 Beam under a Uniformly-Distributed Load
14 1.28 0.50 1.89 3.2
The girder shown in Fig. 11 is to be designed to carry a factored
uniformly-distributed load qu of 57 kN=m (which includes a
225 kg=m allowance for self-weight and 22 kN=m service live
commonly used European standard H- and I-sections were used in
load). The load is applied on the top flange. The span L of the
the study. Different inclined stiffener configurations in the delta
simply-supported girder is 12.0 m. Due to specific site limitations,
region of the girder were also used. 3D nonlinear inelastic finite-
lateral supports that prevent lateral deflections and rotations can
element (FE) models that take into consideration geometrical im-
only be provided at the two ends of the girder. The girder is to
perfections and residual stresses were developed to determine the be designed for moment and shear, then checked for allowable de-
flexural and shear capacities of these girders. The following are the flection. The following material properties are used: Fy ¼ 345 MPa,
main conclusions of this work: E ¼ 200 GPa, and G ¼ 75 GPa.
• Based on the FE simulation results, the flexural capacities of Solution:
simply-supported SDGs subjected to uniform moment can be An initial analysis has shown that the only available hot-rolled
determined using EC3 (CEN 2005) buckling curves a and b section that can carry the applied loads is the heaviest H-section
(for rolled sections or equivalent welded sections) for SDGs HEA 1,000. An alternative solution is to add inclined stiffeners
with d=bc ≤ 2and d=bc > 2, respectively. The results have to an HEA 500 section to enhance its flexural capacity. The thick-
shown that these EC3 buckling curves predicted the flexural ness of the stiffeners is selected to be equal to the web thickness of
capacity of SDGs with an average error of 2% (unconservative) 12 mm. The configuration of the inclined stiffeners will be selected
for both d=bc ≤ 2 and d=bc > 2 and a maximum difference of based on the recommendations of Table 8. For the HEA 500,
7% (unconservative). d=bc ¼ 1.63, and LrI is calculated to be 9.4 m. Thus, for 1 <
• Using a parametric study of 896 SDGs, design guidelines for d=bc < 2 and Lb > LrI , try the following inclined stiffeners con-
selecting effective inclined stiffener configurations are provided figuration of hd ¼ h=5 and bd ¼ 2bc =3. The dimensions and prop-
in Table 8. These recommendations are expressed in terms of the erties of the newly formed cross-sections are obtained as follows:
d=bc ratio of the SDGs and the Lr value of the corresponding bc ¼ bt ¼ 300 mm, bd ¼ 200 mm, d ¼ 490 mm, hd ¼ 88.8 mm,
I-sections. wd ¼ 117.3 mm, tc ¼ tt ¼ 23 mm, tw ¼ 12 mm, td ¼ 12 mm,
• A comparison between the shear stress distribution in I- and ȳ ¼ 269.79 mm, I x ¼ 9.27 × 108 mm4 , I y ¼ 1.15 × 108 mm4 , Sxc ¼
SDG sections have shown that while the shear stress is more 4.21 × 106 mm3 , J ¼ 2.00 × 107 mm4 , ey ¼ −17.07 mm, Cw ¼
or less uniform over the depth of an I-section, it is nonuniform
5.47 × 1012 mm6 , Zx ¼ 4.16 × 106 mm3 , I yc ¼ 6.32 × 107 mm4 ,
for an SDG. The maximum shear stress at the centroid of an
and β x ¼ 41 mm.
SDG section is much larger than the maximum shear stress
The moment gradient coefficient that takes into account the load
in the delta region of the cross section. Hence, shear yielding
height is (Helwig et al. 1997)
is a gradual process in SDGs that starts in the part of the
web below the delta region and progresses into the delta region Cb ¼ Cb ð1.42ylc =ho ÞRm ¼ 1.14ð1.4−2×245=467 Þð1Þ ¼ 0.801 ð18Þ
under an increasing shear force.
• Design equations for shear that follow the EC3 (CEN 2005) The elastic critical buckling moment is calculated as (El Masri
format are proposed for the nominal shear capacities of Class 1 and Lui 2019a)

Fig. 11. Design example.

© ASCE 04021141-11 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


( sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  )
π2 EI y β x βx Cw GJ L2
M cr ¼ 2  þ þ ¼ 1,288 kN · m
L 2 2 I y EI y π2
ð19Þ
The flexural capacity of the SDG is computed using Eqs. (1)–(4)
presented in the section “Overview of the EC3 Flexural and Shear
Capacity Equations” and the recommendations given in the section
“Proposed Flexural Design Equations for SDGs” as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð4.16Þð345Þ
λ̄LT ¼ ¼ 1.06 ð20Þ
1,288
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

For an SDG with d=bc ≤ 2, use buckling curve a and thus


αLT ¼ 0.21 Fig. 12. Design alternatives.

ΦLT ¼ 0.5½1 þ 0.21ð1.06 − 0.4Þ þ 0.75 × 1.062  ¼ 0.991 ð21Þ

1 Check the dead load of the SDG: 180 kg=m ≤ 225 kg=m ∴ ok
χLT ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0.733 ð22Þ The final design shows that an SDG formed by welding two
0.991 þ ð0.991Þ2 − 0.75ð1.06Þ2
inclined stiffeners to a standard hot-rolled HEA 500 section is ad-
equate. The delta stiffener configurations are hd ¼ h=5 and bd ¼
M b;Rd ¼ ð0.733Þð4.16Þð345Þ ¼ 1,052 kN · m ð23Þ 2bc =3 with a plate thickness of 12 mm.
The required flexural strength is
1 Data Availability Statement
M Ed ¼ qu L2 ¼ ð1=8Þð57Þð12Þ2 ¼ 1,026 kN · m ð24Þ
8
Models and data that support the findings of this study are available
M Ed =M b;Rd ¼ 1,026=1,052 ¼ 0.975 ≤ 1.0 ∴ ok ð25Þ from the corresponding author upon reasonable request.

The shear resistance is obtained using equations and recommen-


dations of the section “Proposed Shear Design Equations for SDG” Acknowledgments
as follows:
The authors are grateful that the first author was awarded the
η ¼ 1.0 ðconservativeÞ ð26Þ Yabroudi and Al-Bitar Fellowship, which made it possible for him
to pursue advanced study and conduct this research.
Av ¼ ð444 − 88.8Þð12Þ þ 0.5ð88.8 × 12 þ 2 × 117.3 × 12Þ
¼ 6,203 mm2 ð27Þ
References
1
V n ¼ pffiffiffi ð345 × 6,203 × 10−3 Þ ¼ 1,236 kN ð28Þ AISC (American Institute of Steel Construction). 2016. Specification for
3 structural steel buildings. ANSI/AISC 360-16. Chicago: AISC.
Arabzadeh, A., and M. Varmazyari. 2009. “Strength of I-girders with Delta
The shear design value is stiffeners subjected to eccentric patch loading.” J. Constr. Steel Res.
65 (6): 1385–1391. https://doi.org/10.1016/j.jcsr.2009.01.009.
qu L ð57 × 12Þ Avery, P., M. Mahendran, and A. Nasir. 2000. “Flexural capacity of hollow
V Ed ¼ ¼ ¼ 342 kN ð29Þ
2 2 flange beams.” J. Constr. Steel Res. 53 (2): 201–223. https://doi.org/10
.1016/S0143-974X(99)00067-X.
V Ed AWS (American Welding Society). 2010. Structural welding code-steel.
¼ 0.28 ≤ 1.0 ∴ ok ð30Þ AWS D1.1:D1.1M. Miami: AWS.
Vn
Beer, F. P., E. R. Johnston Jr., J. T. Dewolf, and D. F. Mazurek. 2015.
Mechanics of materials. 7th ed. New York: McGraw-Hill.
Check beam deflection under service live load as follows:
CEN (European Committee for Standardizaion). 2005. EuroCode 3:
L 12,000 Design of steel structures—Part 1.1: General rules and rules for
Δall ¼ ¼ ¼ 33.3 mm ð31Þ buildings. EN 1993-1-1. Brussels, Belgium: CEN.
360 360 CEN (European Committee for Standardizaion). 2006. EuroCode 3:
Design of steel structures—Part 1.5: Plated structural elements.
5qL4 5ð22Þð10−3 Þð12,000Þ4
ΔL ¼ ¼ ¼ 32.0 mm ≤ Δall ∴ ok EN 1993-1-5. Brussels, Belgium: CEN.
384EI x ð384Þð200Þð9.27 × 108 Þ Dux, P. F., and S. Kitipornchai. 1983. “Inelastic beam buckling experi-
ments.” J. Constr. Steel Res. 3 (1): 3–9. https://doi.org/10.1016/0143
ð32Þ
-974X(83)90011-1.
Check the width-to-thickness ratio of the inclined stiffeners El Masri, O. Y., and E. M. Lui. 2019a. “Cross-section properties and elastic
using Eq. (12) as follows: lateral-torsional buckling capacity of steel delta girders.” Int. J. Steel
Struct. 19 (3): 914–931. https://doi.org/10.1007/s13296-018-0175-y.
rffiffiffiffiffiffiffiffi El Masri, O. Y., and E. M. Lui. 2019b. “Influence of imperfections on
wd 117.3 235
¼ ¼ 9.8 ≤ 72 ¼ 59.4 ∴ ok ð33Þ the flexural resistance of steel delta girders.” Adv. Steel Constr.
td 12 345 15 (2): 157–164. https://doi.org/10.18057/IJASC.2019.15.2.5.

© ASCE 04021141-12 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141


Glassman, J. D., M. E. M. Garlock, E. M. Aziz, and V. K. Kodur. 2016. Nethercot, D. A., and L. Gardner. 2005. “The EC3 approach to the design
“Modeling parameters for predicting the postbuckling shear strength of of columns, beams and beam-columns.” Steel Compos. Struct. 5 (2–3):
steel plate girders.” J. Constr. Steel Res. 121 (Jun): 136–143. https://doi 127–140. https://doi.org/10.12989/scs.2005.5.2_3.127.
.org/10.1016/j.jcsr.2016.01.004. Rebelo, C., N. Lopes, L. Simoes da Silva, D. Nethercot, and P. M. M. Vila
Hadley, H. M. 1961. Exploratory test on a steel delta girder, 50–52. Real. 2009. “Statistical evaluation of the lateral–torsional buckling re-
Reston, VA: ASCE. sistance of steel I-beams, part 1: Variability of the Eurocode 3 resistance
Hadley, H. M. 1964. “The bridge delta girder–single-webbed and double- model.” J. Constr. Steel Res. 65 (4): 818–831. https://doi.org/10.1016/j
webbed.” AISC Eng. J. 1 (Oct): 132–136. .jcsr.2008.07.016.
Hatami, F., and N. Esmaeili. 2013. “Optimization of height at delta stiff- Rose, V. S., D. A. Jebitha, and D. Ramalingam. 2017. “Shear behaviour of
ened in steel girders by numerical modeling.” Supplement, J. Am. Sci. delta hollow flange beam with and without web stiffener.” In Proc., 4th
9 (S2): 1–5. Int. Conf. on Recent Advances in Engineering Science and Manage-
Helwig, T. A., K. H. Frank, and J. A. Yura. 1997. “Lateral-torsional buck- ment, 5. Chandigarh, India: Institution of Electronics and Telecommu-
ling of singly symmetric I-beams.” J. Struct. Eng. 123 (9): 1172–1179. nication Engineers.
https://doi.org/10.1061/(ASCE)0733-9445(1997)123:9(1172). Sahnehsaraei, M. J., and S. Erfani. 2014. “Analysis of elastic buckling
Khatoni, S. Y. 2016. “Experimental investigation of buckling and lateral
Downloaded from ascelibrary.org by Anna University - Chennai on 11/06/22. Copyright ASCE. For personal use only; all rights reserved.

behavior of steel delta girders.” Int. J. Eng. Technol. 3 (3): 372–377.


torsion barre enforced by delta under pure bending load.” Turk. Online https://doi.org/10.14419/ijet.v3i3.3059.
J. Des. Art Commun. 6 (Jul): 814–834. https://doi.org/10.7456/1060JSE Simulia. 2014. Abaqus 6.14 documentation. Providence, RI: Simulia.
/050. Subramanian, L., and D. W. White. 2017a. “Improved noncompact web-
Kitipornchai, S., and N. S. Trahair. 1979. Buckling properties of monosym- slenderness limit for steel I-girders.” J. Struct. Eng. 143 (4): 04016216.
metric I-beams. Rep. No. CE 4. Brisbane, Australia: Dept. of Civil https://doi.org/10.1061/(ASCE)ST.1943-541X.0001722.
Engineering, Univ. of Queensland. Subramanian, L., and D. W. White. 2017b. “Reassessment of the lateral
Lee, S. C., and C. H. Yoo. 1998. “Strength of plate girder web panels under torsional buckling resistance of rolled I-section members: Moment gra-
pure shear.” J. Struct. Eng. 124 (2): 184–194. https://doi.org/10.1061 dient tests.” J. Struct. Eng. 143 (4): 04016216. https://doi.org/10.1061
/(ASCE)0733-9445(1998)124:2(184). /(ASCE)ST.1943-541X.0001722.
Loaiza, N., C. Graciano, and E. Casanova. 2019. “Nonlinear finite element Subramanian, L., and D. W. White. 2017c. “Resolving the disconnects be-
analysis of delta hollow flange girders subjected to patch loading.” tween lateral torsional buckling experimental tests, test simulations and
In Proc., Int. Colloquia on Stability and Ductility of Steel Structures, design strength equations.” J. Constr. Steel Res. 128 (Jan): 321–334.
708–716. Prague, Czech Republic: Stability and Ductility of Steel https://doi.org/10.1016/j.jcsr.2016.08.009.
Structures. https://doi.org/10.1201/9780429320248. White, D. W., and S. K. Jung. 2003. Simplified lateral-torsional buckling
Mohebkhah, A., and M. G. Azandariani. 2015. “Lateral–torsional buckling equations for singly-symmetric I-section members. Structural Engineer-
of Delta hollow flange beams under moment gradient.” Thin-Walled ing, Mechanics and Materials Rep. No. 24b. Atlanta: School of Civil
Struct. 86 (Jan): 167–173. https://doi.org/10.1016/j.tws.2014.10.011. and Environmental Engineering, Georgia Institute of Technology.

© ASCE 04021141-13 J. Struct. Eng.

J. Struct. Eng., 2021, 147(9): 04021141

You might also like