Vous êtes sur la page 1sur 22

Journal of Constructional Steel Research 128 (2017) 579–600

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Behaviour and design of steel members with web openings under


combined bending, shear and compression
M. Najafi, Y.C. Wang ⁎
University of Manchester, UK

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

Article history: This paper examines the behaviour of steel beams with web openings under combined axial compression, bend-
Received 17 May 2016 ing moment and shear force through numerical simulation modelling. The numerical simulation results show
Received in revised form 6 September 2016 that under either pure compression or pure bending, the plastic axial or bending capacities of beams are limited
Accepted 22 September 2016
by buckling of the compressive tee-section with the reduction being much more significant in the case of axial
Available online xxxx
compression. The numerical study results also show that when dealing with the general situation of a beam
Keywords:
under combined axial compression, bending moment and shear force, the effect of compressive force and conse-
Web opening quent tee-section buckling should be included to reduce both the bending moment and shear resistances of the
Bending-compression-shear interaction perforated section.
T-section buckling Based on the numerical simulation results, an analytical method has been derived. The method was developed by
Design modifying existing shear-moment interaction equations for the Vierendeel mechanism to incorporate the influ-
ences of tee-section buckling and additional compressive force in reducing the bending moment and shear
capacities.
To account for the effects of additional compression force on bending resistance, the plastic moment-axial com-
pression interaction equation may be used, however, the plastic bending moment capacity (without axial com-
pression) and the plastic compression resistance (without bending) should be replaced by those under the
influence of T-section buckling. To allow for T-section buckling, an effective T-section buckling length of 0.5L
or L (where L is the T-section length) should be used when calculating the bending moment or compression re-
sistance of the perforated section. The shear resistance of the perforated section is obtained by calculating a crit-
ical shear stress in the T-section. This critical shear stress -direct stress interaction is according to the von Mises
equation, but the square power in the von Mises equation is replaced by a function that reflects the influence of T-
section buckling.
A comparison between the numerical simulation results and the analytical results using the proposed method in-
dicates very good agreement, with the inaccuracy mainly attributed to inaccurate calculation of the bending –
shear interaction of the existing methods which do not consider the effects of additional compression and T-sec-
tion buckling.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction Research on steel beams with web openings dates back many de-
cades starting with early experimental investigations on castellated
Steel sections with web openings are frequently used in construction beams [2,3,4] and tests on beams with isolated circular or rectangular
owing to their many advantages such as attractive architectural appear- openings such as [5,6,7,8,9] attempting to understand the deflection be-
ance when used as façade support (Fig. 1a) and flexibility in accommo- haviour of such beams and to obtain the stresses around the openings.
dating building services when used as floor beams (Fig. 1b). Throughout the years, with more advanced finite element studies on
Fig. 2 shows a section with opening (rectangle, circle or elongated cir- beams with different quantity, shape and size of openings [10,11,12,
cle) with symbols for dimensions. Circular and elongated openings may 13], research on this subject has matured leading to practical design
be treated as equivalent rectangular openings of dimensions (lo’ and do’) methods.
as shown in the figure. In this paper, it is assumed that the beam cross- One important shortcoming of these research studies and design
section and the openings are symmetrical about the centre line. methods is that the steel member is subject to lateral load only resulting
in bending and shear. However, steel members with web openings may
⁎ Corresponding author. be subjected to combined axial compression, bending and shear, as in
E-mail address: yong.wang@manchester.ac.uk (Y.C. Wang). façade support structures.

http://dx.doi.org/10.1016/j.jcsr.2016.09.011
0143-974X/© 2016 Elsevier Ltd. All rights reserved.
580 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

Notations
tf thickness of flange
tw thickness of web
A cross-section area of I-section
UDL uniformly distributed load
Ao cross-section area of perforated I-section
Vo,pl plastic shear capacity of a perforated I-section
Ao,w cross-section area of perforated web between flanges
Vo,pl,N plastic shear capacity of a perforated I-section reduced
At cross-section area of tee-section
for axial compressive force
Bf width of flange
Vo,pl,vi plastic shear capacity of a perforated I-section reduced
d depth between flanges of an I-section
for Vierendeel mechanism
dc distance between centroids of top and bottom tee-
Vo,sd applied shear force at the centreline of an opening
sections
Vt,sd applied shear force on tee-section
dNf height of flange thickness subjected to axial compres-
Wo,pl plastic section modulus of perforated I-section
sive force
Wpl plastic section modulus of I-section
do depth of opening
w uniformly distributed load
do’ depth of equivalent rectangular opening for a circular
xo distance of opening centreline from support
opening
α imperfection factor
do,w,N height of perforated web subjected to axial compressive
λt . non-dimensional slenderness of tee-section
force
ρ axial load ratio
dt depth of tee-section
τ shear stress
dwt depth of web of tee-section
σ direct stress
E modulus of elasticity
ῡ coupled shear capacity ratio
fcrit critical buckling stress of section with opening
χt reduction factor for buckling of tee-section
fn average direct stress over cross-section
fu ultimate strength
fu,f ultimate strength of flange Fig. 3 shows the global and local forces in a steel member with web
fu,w ultimate strength of web openings under lateral load generating bending moments and shear
fv shear strength forces. Under such loading conditions, the opening may fail by either
fvn reduced shear strength due to axial compressive force bending, shear, Vierendeel mechanism or tee-section buckling.
fy yield strength The behaviour and design for bending and shear resistances of
fy,f yield strength of flange beams with web openings is the same as for solid beams, the only differ-
fy,w yield strength of web ence being that the section properties of the perforated section (solid
h overall height of I-section section minus the removed part of the web) are used. They will not be
It second moment of area of tee-section considered any further in this paper.
l span of beam This paper will deal with Vierendeel mechanism, shown in Fig. 4,
lo length of opening which is the most critical failure mode of openings. In this mechanism,
lo’ length of equivalent rectangular opening for circular transfer of a shear force across the opening leads to the generation of sec-
and elongated circle openings ondary bending moments and formation of plastic hinges in the tee-sec-
lt,cr critical length of tee-section tions near the opening corners [14,11,15,16]. This paper will investigate
MAbaqus failure bending moment in ABAQUS how an additional axial compression affects the Vierendeel mechanism.
Mo,b bending moment capacity of section with opening re- Despite numerous available design proposals for Vierendeel mecha-
duced for tee-section buckling nism, they can be all categorised into two main groups. They either con-
Mo,b,N buckling moment capacity of section with opening re- sider the tee-section resistance at the four corners of an opening or the
duced for axial compressive force perforated I-section capacity at the centreline of the opening. In the tee-
Mo,pl plastic bending moment capacity of a perforated I- section method, such as those given by [1,18,19,20,21], the moment resis-
section tances of the four tee-sections at the corners of the opening are reduced
Mo,pl,vi plastic moment capacity of section with opening re- for the co-existing axial loads generated by the applied global moment.
duced for Vierendeel mechanism In the perforated section method [22,23,16], the resistance of the beam
Mo,sd applied bending moment at the centreline of an depends on the section resistance of the perforated section under com-
opening bined global shear and bending moment applied at the centreline of the
Mpl plastic bending moment capacity of solid I-section opening. The influence of Vierendeel moments is taken into account by
Msd applied bending moment reducing the basic shear capacity due to Vierendeel moments.
Mt.,pl Plastic moment capacity of tee-section The SCI publication P355 [1], which is commonly used to design steel
Mt.,sd Vierendeel moments at opening corners on tee-sections beams with openings, uses the tee-section method. The design check for
Mvi,sd total Vierendeel moment over opening due to transfer Vierendeel moment capacity of the section with opening is:
of shear
"   #
NAbaqus axial failure load in ABAQUS Nt;sd 2
No,b axial compressive resistance of section with opening Mo;pl;vi ¼ 4Mt;pl 1− ð1Þ
N t;pl
No,pl plastic axial capacity of perforated I-section
No,w,pl plastic axial capacity of perforated web
Npl plastic axial capacity of solid I-section This resistance should not be exceeded by the total applied
Nsd applied axial force Vierendeel moment caused by the transfer of shear force across the
Nt,b axial bucking resistance of tee-section opening calculated below:
Nt,cr critical elastic Euler buckling load of tee-section
Mvi;sd ¼ V o;sd lo ð2Þ
Nt,pl plastic axial capacity of tee-section
Nt,sd applied axial force on tee-sections
However, Chung and co-workers [24,25,12,10] found the above
P.N.A plastic neutral axis
method conservative because it ignores load redistribution within the
M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600 581

(a) Steel façade support with openings (b) Floor beams with openings of various shapes [1]

Fig. 1. Different uses of steel sections with web openings.

opening and have proposed an alternative method. In their method, the It is possible to introduce the influence of additional axial load in the
resistance of the beam with web opening depends on the resistance of above equations by reducing the moment and shear capacities used in
the net perforated section under combined global shear and bending the interaction curves. However, the design method of AISC is based
moment applied at the centreline of the opening. Their interaction on studies of rectangular openings and conversions are needed to calcu-
equations are given in Eq. (3). Note that in their interaction equations, late the resistance of circular and elongated circular opening shapes.
the influence of Vierendeel moments is taken into account by reducing This may lead to some conservative results. On the other hand, the
the basic plastic shear capacity of the perforated section (in the absence method of Chung et al. [16] has the limitation of being specific to the
of global bending moment) using reduction factor v referred to as openings of certain shapes and sizes. Furthermore, both methods ignore
coupled shear capacity ratio). Values of v or different opening types buckling of the compression tee-section which may occur in the case of
are given in Table 1. large openings with slender tees [27,22] as shown in Fig. 5.
This research is to resolve the above shortcomings of existing design
    methods by incorporating the effects of additional global axial compres-
2 V o;sd 2 Mo;sd 2
For vb þ ¼1 sion on bending-shear interaction of steel members with web openings,
3 vV o;pl Mo;pl
0V  12 ð3Þ and by considering the effect of buckling in the compression Tee-sec-
o;sd
− v− 23  2
For v≥
2 @V o;pl A þ Mo;sd ¼1
tion. The investigation starts by developing a method to calculate the
3 2= M o;pl axial and moment resistance of openings under the influence of tee-sec-
3
tion buckling. This is then incorporated into the shear and Vierendeel
moment interaction curve. The investigation will be carried out through
A similar approach is given by Darwin in the AISC design guide [22] numerical simulation using the general finite element package ABAQUS
which gives a cubic interaction curve (Eq. (4)) based on plastic analysis [28].
of beams with rectangular openings carried out by [26]. The reduced
shear capacity in the interaction equation is defined based on curve
fitting a simplified version of the von Mises yield criterion using avail- 2. Validation of numerical modelling
able experimental data.
2 3 Since the findings of this research are purely based on numerical
    pffiffiffi simulations, it is imperative that the numerical models are verified
M o;sd 3 V o;sd 3 6 6 7
þ ¼1 where κ ¼ 6
4 lo pffiffiffi5
7 ð4Þ against experimental results. Here, the experimental tests used for the
Mo;pl κV o;pl validation study include the tests by [9] on steel beams with one or
þ 3
dt two openings and two tests on castellated beams [4]. Although these

Fig. 2. Section with rectangular, circular or elongated circle opening.


582 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

Fig. 3. Forces and moments at an opening and its tee-sections [12].

Fig. 4. Vierendeel mechanism [17] [9].

tests do not have axial compression in the steel section, they exhibit the connected to the I-section through coupling constraints provided in
main effects of axial compression/shear and bending interactions. ABAQUS using a reference point at the middle of the plate and mid-
The main features of the numerical model are: height of the I-section. The boundary conditions are then applied to
the reference point. For the simply supported condition, the vertical
• Riks method: To allow post peak load behaviour to be traced. and longitudinal restraints are applied at one end (to the reference
• Fully integrated, general purpose, quadrilateral shell elements with point) while the other end is only restrained from vertical movement.
four integration points (S4): preferred over solid elements owing to This type of modelling can allow for any external axial load or bending
their superior computational efficiency. moment to be applied at the support reference point without causing
• The simulations were carried out using mesh sizes of 10 or 20 mm. any eccentric loading condition.
Some mesh sensitivity studies were performed [30] and the results
showed that as long as there were four or more elements above and
below the openings, the numerical simulation results converged. For 2.1. Tests of Redwood and McCutcheon [9]
circular opening, there was no advantage in reducing the mesh size
below 20 mm (4 elements above and below the opening) but for Redwood and McCutcheon carried out 19 tests on a series of beams
the rectangular opening, a mesh size of 10 mm was needed to have containing one or two openings. The experiments were carried out on
four elements above and below the opening and convergence of the simply supported beams consisting of W8x17 sections (flange width
results. 133.4 mm, flange thickness 7.8 mm, overall height 203.2 mm, web
• Measured material steel yield and ultimate stress, Young's modulus of thickness 5.8 mm) of A36 steel grade (nominal yield strength 248 N/
210,000 N/mm2, Poisson ratio of 0.3 were used. The standard mm2). The test paper [9] has reported the actual measured values of
Eurocode 3 strain-strain relationship with strain hardening was used cross-sectional dimensions and steel properties for each test and they
to simulated the material. were used in the simulation models here.
• Geometric non-linearity. Three opening shapes of circle, rectangle and elongated circle were
• Concentrated loads acting on a small strip on top of the top flange included in the experimental program. The circular and rectangular
were applied using the coupling constraint in ABAQUS. shapes were tested individually and in pairs with wide (114.3 mm)
and narrow (57.2 mm) spacing. All openings were located at the mid-
depth and had maximum depths of 57% of the overall beam's height.
In order to create the support boundary conditions, a rigid plate was Openings were placed at different positions along the beams for dif-
defined at the ends of the beam as shown in Fig. 6. The plate was ferent values of shear/moment ratio, from zero for pure bending to
0.425 for high shear/moment interaction. The shear to moment ratio
was defined as the ratio of the average shear stress in the web τ to the
maximum normal stress in the outside fibre σ, assuming elastic condi-
Table 1
Coupled shear capacity ratio v [16]. tions and using the un-perforated section properties. Variable shear/
moment ratio was achieved by varying the span, loading position and
Opening shape Opening size

do/h = 0.50 do/h = 0.67 do/h = 0.75

Circle 0.95 0.87 0.80


Square 0.66 0.42 0.32
Elongated circle (lo = 2do) 0.56 0.35 0.26
Rectangle (2:1) 0.38 0.21 0.16
Fig. 5. Buckling of compression tee-section of an opening under bending moment.
M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600 583

Coupling Constraint
Reference point at mid-height

Rigid plate

Fig. 6. Use of rigid plate and coupling constraint for support boundary conditions.

locations of the openings. For instance, to test an opening under pure 2.2. Tests of Hosain and Speirs [4]
moment, two- point loading was applied and the opening was placed
at mid-span between the loads. Table 2 summarises the main test pa- Fig. 9 shows the overall dimensions and test arrangements of the
rameters and the failure mechanisms of the tests. two test beams (Beams B3 and G2) of Hosain and Speirs [4]. For ABAQUS
All 19 tests were simulated in ABAQUS and the simulation results modelling, the actual section dimensions, yield strength and ultimate
were compared to the test results. Fig. 7 compares the moment deflec- strength were used as reported in the test paper (Beam B3: yield
tion curves for all nineteen tests between the ABAQUS simulation and strength of flange and web = 355 N/mm2, ultimate strength of web
experimental tests. In all cases, the applied moment at the openings is and flange = 447 N/mm2–Beam G2: yield strength of flange and
plotted against the mid-span deflection of the beams. Overall, the agree- web = 314 N/mm2, ultimate strength of flange = 474 N/mm2 and ulti-
ment is excellent. It is clear from these graphs that the numerical curves mate strength of web = 384 N/mm2). Beam B3 failed due to formation
follow the corresponding test results closely with both the numerical of Vierendeel mechanism and failure in beam G2 was due to bending
and experimental results. followed by local flange buckling. Fig. 10 and Fig. 11 show the deformed
The small differences in some of the cases may be attributed to dif- shapes of the two test specimens simulated by ABAQUS. As shown, the
ferences in the simulation and experimental material properties. The failure modes and deformation patterns from the ABAQUS models are
test paper only provided measured yield and ultimate stresses of steel the same as reported for the tests.
with no information on Young's modulus (a value of 210,000 N/mm2 Fig. 12 shows the test load-deflection curves of the beams B3 and G2
was used in the simulations) and stress-strain relationships. and they are compared with the simulation graphs. In both cases the
Fig. 8 compares the deformed shapes and yielding patterns around ABAQUS results are very close to the test results.
the openings of the test specimens against the simulation models. The
black patterns in the photographs of the specimens represent the 3. Bending moment and axial compression capacity of openings un-
yielded areas. The yielding regions predicted by ABAQUS are very simi- der the influence of tee-section buckling
lar to the test observations in all cases.
3.1. Simulation results

Fig. 13 shows the loading and boundary conditions of the numerical


parametric study models for openings under axial load and bending
moment respectively. Note that the axial load is applied at the end
Table 2
with no longitudinal restraint. A total of 99 models were simulated to
Test details and failure modes for the experimental program of [9].
investigate the effects of changing the following variables:
Test Opening shape τ/σ Failure mechanism

1A Single circle 0 Typical bending failure


• Three different section sizes of UB254 × 146 × 43, UB457 × 152 × 82
1B Elongated circle 0 Typical bending failure and UB914 × 305 × 82 with spans of 3 m, 3 m and 4 m respectively
1D 2 circles – close spacing 0 Typical bending failure (Note: Since the focus of this study was to investigate the effects of
11H 2 rectangles – wide 0 Typical bending failure tee-section failure, the member was deliberately made short to
spacing
avoid global buckling failure).
2A Single circle 0.425 4 hinges with web yielding
2B Elongated circle 0.425 4 well-defined hinges • Opening shapes of circle, rectangle (square in the case of rectangle
2C 2 circles- wide spacing 0.425 Some evidence of four hinges with 1:1 aspect ratio) and elongated circles.
2D 2 circles – close spacing 0.425 4 hinges – two openings failed as one • Three different opening depths of 0.50 h, 0.67 h and 0.75 h.
unit
• The length of the opening for rectangles and elongated circles was
2F Single rectangle 0.425 4 well-defined hinges
21G 2 rectangles – close 0.319 4 hinges – two openings failed as one
varied from short (0.5do) to long (3do). The selected lengths for rect-
spacing unit angular openings were 0.5do, 1do, 1.5do, 2do, 2.5do and 3do; and for
21H 2 rectangles – wide 0.319 8 hinges – 4 at each opening elongated circles they were 1.5do, 2do, 2.5do and 3do.
spacing • The steel grade was S275 with a Young's modulus of 210,000 N/mm2
3A Single circle 0.255 Evidence of 4 hinges with web yielding
and Poisson's ratio of 0.3.
4A Single circle 0.160 Evidence of 4 hinges with associated
yielding • Initial imperfections were specified based on the shape of the tee-sec-
4B Elongated circle 0.160 4 hinges tion buckling mode from eigenvalue analysis prior to loading the
4C 2 circles – wide spacing 0.160 Failure only at opening nearest to the structure. The maximum initial imperfection was lo/1000 where lo is
load the overall length of the opening. This value was chosen because the
4D 2 circles – close spacing 0.160 Failure only at opening nearest to the
load
tee-section was treated as an axially loaded column for which an im-
4F Single rectangle 0.160 4 well-defined hinges perfection value of Length/1000 is frequently adopted.
4G 2 rectangles – close 0.160 4 hinges – 2 openings failed as one unit
spacing
4H 2 rectangles – wide 0.160 Failure only at opening nearest to the The simulation results for circular openings all indicate that their
spacing load
plastic axial and bending moment capacities were reached in all cases.
584 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

X-axis: Mid-span deflection (mm)


Y-axis: Moment at opening (kNm)

Fig. 7. Comparison between simulation results against test results of [9].

Therefore, the tee-section in a circular opening is stocky and is not sus- The plastic moment capacity is given as:
ceptible to buckling. The following discussions will focus on rectangular ! !
2 2
d tw d tw
opening. Mo;pl ¼ W o;pl f y ¼ W pl − o f y ¼ Mpl − o fy ð6Þ
4 4
Fig. 14 and Fig. 15 show the reduction in failure loads and bending
moments, expressed as a ratio of the numerical failure axial compres- Comparing the results in Fig. 14 for axial compression and in Fig. 15
sion load and bending moment to the respective plastic axial compres- for bending, it is clear that the reductions under axial compression can
sion and plastic bending moment capacity of the section with opening. be quite high, with the maximum being about 50% for sections with
The plastic axial capacity No,pl is calculated as: large openings. However, the reductions in the same sections under
bending moment are much lower. This can be explained by considering
the different buckling modes of steel sections with web openings under
No;pl ¼ Ao f y ð5Þ axial compression and bending moment.
M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600 585

Test 1A Test 2A

Test 4A Test 2B

Test 4B Test 4F

Test 2C Test 4C

Test 2D Test 4D

Test 4G Test 4H

Fig. 8. Comparison of deformed shapes and yielding patterns for test specimens of [9].

Flange Width = 101.6mm

Flange thickness = 6.8mm

Fig. 9. Tests beams of Hosain and Speirs [4].

Under compressive axial load, the Tee-section buckling mode is sim- 3.2. Development of an analytical approach
ilar to the sway failure mode of a column with rotationally fixed ends as
shown in Fig. 16. For this boundary condition, the effective length of the The axial compressive resistance of a section with web opening, No,b,
tee-section can be taken as the opening length. is the sum of the buckling resistances of its two tee-sections Nt,b. For a
Under bending, the buckling mode of the top tee-section is similar symmetric perforated section:
to that of a non-sway column with rotationally fixed ends as shown in
Fig. 17. The critical length of the tee-section is half their overall opening
length. No;b ¼ 2Nt;b ð7Þ

Fig. 10. Vierendeel failure of test beam B3 of Hosain and Speirs [4] in ABAQUS.
586 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

Fig. 11. Bending and local flange buckling of test beam G2 of Hosain and Speirs [4] in ABAQUS.

Fig. 12. Comparison of simulation and test results of Hosain and Speirs [4].

Fig. 13. Details of openings subjected to axial load (a) or bending moment (b).

Fig. 14. Ratios of ABAQUS failure axial compression loads to their respective plastic axial capacities.
M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600 587

Fig. 15. Ratios of ABAQUS failure bending moments to their respective plastic bending moment capacities.

(a) Buckling failure mode of (b) Sway failure mode of a column


opening under axial load with rotationally fixed ends

Fig. 16. Buckling mode of T-section in steel section with opening under compression.

The buckling resistance of one tee-section is calculated using the section. Since this rotational stiffness depends on the length of the
Eurocode [29] equation: beam segment on the sides of the opening, it is expected that the overall
beam span/depth ratio and the location of the opening along the beam
Nt;b ¼ χ t Nt;pl ð8Þ span would have some effect.
To examine these effects, additional numerical simulations were car-
where the plastic axial capacity of the tee-section Nt,pl is given as: ried out, including 25 cases covering different span/depth ratios and 8
   cases covering different opening positions. In all cases, a large rectangu-
Nt;pl ¼ At f y ¼ B f t f þ ðdwt t w Þ f y ð9Þ lar opening (do = 0.75 h and lo = 3do) was used to ensure Tee-section
buckling.
The reduction factor χt is calculated as: Fig. 18 compares typical failure load – span/depth ratio relationships
between simulation and calculation results. The simulation results clearly
1 show the beam span/depth ratio has some effect on the axial buckling ca-
χt ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi where Φt
2 pacity of the tee-section. With very small span/depth ratios, the tee-sec-
Φt þ Φt 2 −λt
h   2
i tion behaves as a non-sway column with rotationally fixed ends. In this
¼ 0:5 1 þ α λt −0:2 þ λt ð10Þ
case, the effective length of the tee-section approaches 0.5 times the
opening length. However, such span/depth ratios are not practical.
The value of the imperfection factor α depends on the adopted buck- For more practical dimensions, as a simple approximation, the effect
ling curve which is explained later. of the span/depth ratio may be dealt with in the following way: for 10 ≤
The value of the non-dimensional slenderness of the tee-section λt span/depth b 15: the effective length of the tee-section is 1.0lo (or 1.0lo’
depends on the critical (or effective) length of the tee-section for buck- for elongated circular openings); for 15 ≤ span/depth ≤ 25, the effective
ling. According to Fig. 16, the buckling length of the tee-section can be length of the tee-section is 1.05lo.
taken as the opening length. However, the rotational restraint to the To determine the column buckling curve that gives the best agree-
tee-section comes from the beam section on either side of the tee- ment with the numerical simulation results, Eurocode column buckling

(a) Compressive buckling mode of tee- (b) Buckling mode of a non-sway


section in beam under bending fixed-ended column

Fig. 17. Buckling mode of T-section in steel section with opening under bending.
588 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

Fig. 18. Comparison between simulation and analytical results for using different effective length values for members with varying span/depth ratios.

curves “a” (α = 0.21), “b” (α = 0.34), and “c” (α = 0.49) were used to The full plastic moment capacity Mpl and axial load capacity Npl of
calculate the tee-section compression resistance. The calculation results the solid section without considering the opening are:
showed that curve “b” gives the best agreement, with the calculation re-
sults still being conservative compared with the simulation results. Npl ¼ A f y ð12Þ
Using buckling curve “b” is in according with the Eurocode buckling
curve for I-section buckling about the major axis. Fig. 19 shows this Mpl ¼ W pl f y ð13Þ
agreement by comparing the compressive capacities obtained from
simulations with the calculation results for different opening shapes
The full plastic axial force and bending moment capacities of the per-
and sizes and beam section sizes using Eurocode curve “b”.
forated section (No,pl and Mo,pl respectively) are calculated as:
It should be noted that the imperfection values used in the Eurocode
curves take account of residual stresses and geometrical imperfections
N0;pl ¼ A0 f y ð14Þ
whereas only geometrical imperfections were introduced into the
ABAQUS models. However, the geometrical imperfections in the au- ! !
2 2
thors' ABAQUS models were quite severe and may be considered to do t w do t w
Mo;pl ¼ W o;pl f y ¼ W pl − f y ¼ Mpl − fy ð15Þ
compensate for the lack of residual stress. 4 4
For an opening under pure bending moment, the bending capacity
Depending on the location of plastic neutral axis (PNA), whether in
“Mo,b” is limited to the buckling resistance of the top tee-section
web or in flange, the bending moment – axial compression interaction
which is under compression. Therefore:
is different. Detailed equations have been derived [30] and the final cal-
Mo;b ¼ Nt;b  dc ð11Þ culation equations are:
For P.N.A. is in the web (when Nsd ≤No,w,pl) reduced plastic bending
moment capacity is given as:
The compressive resistance of the tee-section can be calculated in
the same way as explained in the previous section. However, the effec- "    #
tive length of the tee-section can be taken as half the opening length. Al- A2o Nsd 2 Ao do Nsd
Msd ¼ Mo;pl 1− − ð16Þ
though the span/depth ratio may affect the Tee-section effective length, 4t w W o;pl No;pl 2W o;pl No;pl
since the change in Tee-section buckling capacity is relatively small
when the member is in bending, no refinement in Tee-section effective For P.N.A. in the lower flange (when Nsd N No , w , pl) reduced plastic
length is required. bending moment capacity is calculated as:
Fig. 20 compares the bending moment capacities between simula-
 

tion and calculation results for different opening shapes and sizes and Ao 0:5h−0:5t f N
beam section sizes. The agreement is very good. Msd ¼ Mo;pl 1− sd ð17Þ
W o;pl No;pl

4. Combined axial compression and bending moment


The applied axial compressive force Nsd calculated as:
4.1. Plastic interaction  
Nsd ¼ ðd−do Þt w þ 2dNf B f f y ð18Þ
Fig. 21 shows a perforated I-section with an opening depth of do sub-
jected to combined bending moment Msd and axial force Nsd and the
corresponding plastic stress distributions at yield. The position of the 4.2. Effects of tee-section buckling on bending moment – axial compression
plastic neutral axis (P.N.A.) changes as the magnitude of the axial interaction
force changes. With no axial force, the P.N.A. is in the middle with the
top tee-section under compression and the bottom tee-section under To account for Tee-section buckling, the full plastic axial and mo-
tension. As the applied axial force increases, the P.N.A. moves further ment capacities of the openings in Eqs. (16) and (17) can be replaced
down so an increasingly larger part of the bottom tee goes under com- by the buckling capacities No,b and Mo,b calculated using the proposed
pression in addition to the top tee-section. methods in Section 3. The resulting interaction equations are:
M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600 589

Fig. 19. Comparion between ABAQUS results (solid curves) with calculation results (dotted curves) for openings under axial compression.

For P.N.A. in the web: section was UB475x152x82 with a span of 6 m to give a realistic span/
" depth ratio of 12.9. The beam was subjected to combined axial load
   #
A2o Nsd 2 Ao do Nsd and end moments as shown in Fig. 22.
Msd ¼ Mo;b 1− − ð19Þ
4t w W o;pl N o;b 2W o;pl N o;b Fig. 23 compares the numerical interaction curves with the analyti-
cal buckling interaction Eqs. (19) and (20) and the unmodified plastic
interaction Eqs. (16) and (17). The results show that the simulation
For P.N.A. in the lower flange:
graphs are in between the plastic interaction curves and the new buck-
  ling interaction curves, with the new buckling interaction curves always

Ao 0:5h−0:5t f N giving safe results and the plastic interaction curve over-predicting the
Msd ¼ Mo;b 1− sd ð20Þ
W o;pl No;b capacity of the openings by not taking tee-section buckling into account.
To check the accuracy of these two interaction equations, a set of As expected, the proposed buckling interaction equations give very
simulations were carried out on large rectangular openings (do = accurate results when the opening length is short because the reduction
0.75 h) with variable opening lengths of 1do, 2do and 3do. The beam in tee-section resistance due to buckling is small. For a more practical
590 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

Fig. 20. Comparion between numerical simulation results (solid curves) with calculation results (dotted curves) for opening under bending moment.

longer opening length (lo = 2do), the new buckling interaction equa- distributed load combined with a compressive axial load. Table 3 lists
tions provide results that are reasonably accurate. Because the proposed the details of the opening and its location in all the simulation cases.
buckling interaction equations are simple to use, they are considered The size and location of the opening was selected such that it was not
adequate for design calculations. susceptible to tee-section buckling, and that failure occurred at the loca-
tion of the opening rather than the mid-span of the beam where the ap-
5. Influence of compression on shear and Vierendeel mechanism plied bending moment was at its maximum.
Fig. 25 shows the deformed shapes and yielding patterns for circular
5.1. Plastic interaction between shear, axial compression and bending and rectangular openings under different combinations of the vertical
moment and axial loads. When there is no axial load, clearly defined plastic
hinges form at the corners and the failure mode is a clear Vierendeel
A set of simulations were carried out on a 6 m UB457 × 152 × 82 mechanism. As the compressive axial load increases, yielding moves
beam. A single opening was placed at different locations along the away from the corners towards the centreline of the openings.
span of the beam as shown in Fig. 24 so that they would experience dif- In order to develop an analytical solution for the general case of axial
ferent moment/shear ratios. The beam was simulated under a uniformly compression, bending moment and shear force (including the effect of
M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600 591

Fig. 21. Plastic interaction of bending moment and axial force in a perforated section.

Vierendeel mechanism), the existing shear – moment interaction equa- However, the above only incorporates the effect of axial compression on
tions for Vierendeel action (AISC [22] and Chung et al. [16]) can be mod- the bending moment capacity of the section and it does not take account
ified using the proposed M-N interaction equations as follows. of its influence on the plastic shear capacity of the section (N-V interac-
The bending moment capacity of the openings Mo,b is first reduced to tion). Therefore, a further modification is made following the von Mises
Mo,b,N for a given value of the applied axial load Nsd by using the axial yield equation by reducing the shear strength fv of the material to fvn to
load - bending moment interaction equations (Eqs. (19) and (20)). include the effect of the direct stress from the axial force as follows:
This reduced value of bending moment capacity Mo,b,N is then used in
pffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
the formulations for Vierendeel mechanism instead of the full bending f vn ¼ 1= 3
2
f y −f n
2
ð23Þ
moment capacity of the perforated section, such that the Vierendeel
equations become:
Using the AISC method [22]: Where fn is the average direct stress over the perforated section
    caused by the axial force and is calculated as:
Mo;sd 3 V o;sd 3
þ ¼1 ð21Þ
M o;b;N κV o;pl Nsd
fn ¼ ð24Þ
Ao
Using the method of Chung et al. [16]:
      The reduced value of the shear strength is then used instead of the
V o;sd 2 M o;sd 2 2 full shear strength in the Vierendeel action Eqs. (21) and (22) when cal-
þ ¼ 1 for vb
v V o;pl Mo;b;N 3 culating the plastic shear capacities.
2V  32  ð22Þ
o;sd
− v− 2 2   Fig. 26 to Fig. 28 compare the results of the above proposed method
4 V o;pl 3
5 þ M o;sd ¼ 1 for v≥
2
with the simulation results. It can be seen that the analytical and simu-
2 Mo;b:N 3
3
lation results are in reasonably good agreement. The accuracy of the

Fig. 22. Model for ABAQUS simulations of openings susceptible to buckling under combined axial load and bending moment.
592 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

Fig. 23. M-N interaction curves for rectangular opening (do = 0.75 h) and variable opening lengths, beam section UB457 × 152 × 82.

proposed methods is largely dependent on the accuracy of the existing Fig. 31 and Fig. 32 compare the ABAQUS interaction curves with the
design methods for Vierendeel mechanism without considering the ef- calculated values using the method in the previous section (labelled as
fect of the compressive axial load. A comparison of the calculated mod- proposed method). The results are reasonable, but two further modifi-
ified Vierendeel capacities based on using the AISC and Chung et al. cations to account for the effects of buckling are necessary. Without
equations suggests that the Chung et al. method is slightly more accu- these further modifications, the numerical interaction curves are
rate. However, there is a limitation with the method of Chung et al. be- below the analytical calculation results, indicating that the analytical re-
cause the coupled shear capacity ratio v is defined for only a limited sults are not safe. This is simply due to the buckling influence. As the
number of sizes and shapes of the opening (as given in Table 1). There- slenderness of the tee-section of the opening increases, the opening be-
fore, the method cannot be used for every opening. Consequently, the come more instable and even a small increase in the axial load will cause
authors suggest using the Chung et al. method if v is available for the ex- them to buckle without experiencing much yielding.
amined opening but otherwise the AISC method should be used. Consequently, the proposed Method of the previous section cannot
produce accurate results. This is because the shear capacity is signifi-
cantly reduced due to flexural buckling of the tee-section and the von
5.2. Buckling interaction between shear, axial compression and bending Mises yield criterion is no longer valid to account for buckling interac-
moment over the opening tion between the shear and axial force. Therefore, a more severe interac-
tion equation is needed.
In order to extend the applicability of the proposed method in the To include this effect of buckling, the first step is to replace the yield
previous section to openings with slender tee-sections that may be sus- strength fy in the von Mises equation by the critical buckling stress of the
ceptible to buckling, a set of new numerical simulations were carried opening fcrit as follows:
out in ABAQUS on large rectangular openings with perforation depth
do = 0.75 h and variable lengths placed at 1 m distance from the support
in a 6 m beam made of UB457 × 152 × 82 section (see Fig. 29). The pffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
f vn ¼ 1= 3 f crit − f n ð25Þ
length of the opening varied from 1.0do to 3.3d.

Fig. 24. ABAQUS simulation model for openings with stocky tee-sections placed at different locations along the span under combined UDL and compressive axial load.
M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600 593

Table 3 the opening experiencing buckling, the stress interaction equation


Opening details for simulation of the influence of compressive axial load on Vierendeel should also depend on the slenderness of the tee-section.
mechanism.
Therefore, a final modification is made to Eq. (28). The idea is to
Opening shape Opening depth Opening Opening distance from change the power “2” to a value dependent on the buckling capacity
(do) length (lo) support (xo) of the opening such that the power decreases according to the reduction
Circle 0.75 h – 1.0 m in the axial capacity of the opening. The new format of Eq. (28) is:
Circle 0.75 h – 1.5 m
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Circle 0.75 h – 2.0 m pffiffiffi q f q −f q
crit n
Circle 0.75 h – 2.5 m f vn ¼ f y = 3 q ð29Þ
Square 0.5 h 1do 2.0 m f crit
Rectangle (2:1) 0.5 h 2do 1.0 m
Rectangle (2:1) 0.5 h 2do 1.5 m
The power q is a function of the reduction in the axial capacity of the
Rectangle (2:1) 0.5 h 2do 2.0 m
Rectangle (2:1) 0.5 h 2do 2.5 m section with opening. Therefore, q is defined in terms of the axial load
Elongated Circle 0.5 h 2do 1.5 m ratio ρ, where ρ is calculated as:
(2:1)
Elongated Circle 0.5 h 2do 2.0 m No;b
(2:1)
ρ¼ ð30Þ
No;pl
Elongated Circle 0.5 h 2do 2.5 m
(2:1)
Based on the simulation results, a curve fitting exercise was carried
out and the best fitting equation is (Fig. 30):
The critical buckling stress fcrit is calculated by dividing the axial
compression resistance of the section with opening by its cross-section- q ¼ 0:75 þ 1:3  10−8 e18:3ρ ðfor 0:5≤ρ≤1:0Þ ð31Þ
al area as given below:
The results incorporating the above further modification are labelled
No;b as “Final Proposed Method” in Fig. 31 and Fig. 32. This further modifica-
f crit ¼ ð26Þ
Ao tion gives analytical calculation equations that are in very good agree-
ment with the ABAQUS results. The main source of inaccuracy of the
Note that the format of Eq. (25) is only suitable for openings that are analytical method is from the original inaccuracy of the Vierendeel ac-
not susceptible to buckling; hence for tee-sections that buckle, the effect tion equation which does not consider the effect of axial force.
of tee-section flexural buckling is included by replacing the absolute
stress values in Eq. (25) by the relative values on both sides as follows: 6. Overall comparison of analytical and simulation results
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u Fig. 33 presents an overall comparison between the analytical resis-
f vn u f 2 −f 2
pffiffiffi ¼ t crit 2 n ð27Þ tance (including axial compression, bending, combined axial compres-
f y= 3 f crit sion/bending and combined axial compression/bending/shear) with
all the respective ABAQUS results. The results show that the theoretical
Rearranging gives: values are close to the simulation results in the majority of cases, with
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi an average error of 11.8% and standard deviation of 16.7%.
u
pffiffiffi u f crit 2 − f n 2 Observation of Fig. 26, Fig. 27, Fig. 28, Fig. 31 and Fig. 32 (for open-
f vn ¼ f y= 3 t 2
ð28Þ ings under combined axial compression, shear and bending) show
f crit that accuracy of the analytical results is mostly dependent on accurate
calculations of the two end points: the theoretical axial compressive ca-
Using Eq. (28), the shear strength will be equal to the full plastic pacity when there is no bending and/or shear and the reduced moment
shear strength when there is no axial force. However, the reduction is capacity due to shear when there is no axial load. In cases where the two
more severe than the original von Mises equation since the critical di- ends are close to the simulation results, the analytical and simulation
rect stress is used instead of the yield strength. However, even using curves are almost identical.
Eq. (28), the reduction would not be adequately severe. This is due to As shown earlier, when there is no bending moment in the member,
the quadratic form of the von Mises equation, which is derived based the axial compression capacity of the member with openings is calculat-
on the theory of plasticity and complete yield, whereas in the case of ed accurately. Therefore, to a large extent, inaccuracy of the analytical

Fig. 25. Failure modes of openings under different combinations of vertical and axial loads.
594 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

ABAQUS ABAQUS
Proposed Method (AISC) Proposed Method (AISC)
Proposed Method (Chung) Proposed Method (Chung)

ABAQUS ABAQUS
Proposed Method (AISC) Proposed Method (AISC)
Proposed Method (Chung) Proposed Method (Chung)

Fig. 26. Comparison between ABAQUS results with analytical results – circular openings.

method, when compared to the simulation results, comes from inaccu- The other source of the big difference in Fig. 33 is related to the way
racy of the adopted shear/bending interaction equations (AISC or Chung the difference is calculated. It is calculated by comparing the applied lat-
et al.) for calculating the opening resistance when there is no axial load. eral loads on the beam at different axial load levels. Therefore, the ana-
For confirmation, Fig. 34 compares the analytical and numerical lytical to numerical ratio in Fig. 33 refers to the ratio of the calculated/
values for the cases under axial load only. The average difference is simulated results of the reduced moment capacity for a given value of
only 3.5% with a standard deviation of 4.2%. Fig. 35 compares the calcu- the applied axial load. For example, Fig. 36 shows the analytical and
lated values of bending moment resistance in the presence of shear but ABAQUS numerical interaction curves for a beam with a rectangular
without axial load using the adopted shear/bending interaction equa- opening of do = 0.75 h, lo = 1do placed at xo = 1 m under combined
tions (AISC or Chung et al.) with the ABAQUS results. The average differ- axial compression, bending moment and shear force. One can say that
ence is 10.6% and the standard deviation is 10%. these two curves are very close. The calculated axial compressive

ABAQUS
Proposed Method (AISC) ABAQUS
Proposed Method (Chung) Proposed Method (AISC)
Proposed Method (Chung)

ABAQUS
Proposed Method (AISC)
Proposed Method (Chung)

Fig. 27. Comparison between ABAQUS results with analytical results - elongated circular openings (2:1).
M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600 595

ABAQUS ABAQUS
Proposed Method (AISC) Proposed Method (AISC)
Proposed Method (Chung) Proposed Method (Chung)

ABAQUS ABAQUS
Proposed Method (AISC) Proposed Method (AISC)
Proposed Method (Chung) Proposed Method (Chung)

ABAQUS
Proposed Method (AISC)
Proposed Method (Chung)

Fig. 28. Comparison between ABAQUS results with analytical results - rectangular openings.

resistance is also very close to the numerical value, with a difference of circumstance, the analytical result is very close to the ABAQUS simula-
only 5%. However, if the axial load is high, for example the first point in tion results.
Fig. 36, the ratio of analytical value/ABAQUS simulation value is 0.46
even though the two curves are close. This difference between the 7. Summary and overview of the proposed analytical method
two sets of results drops down rapidly when the axial load level de-
creases. In realistic beams, the applied axial load level is relatively Fig. 37 provides an overview of the final analytical method in which
small so the bending moment resistance is high. Under this the existing shear-moment interaction equations are modified such that

Fig. 29. ABAQUS simulation model for large rectangular opening with variable lengths under combined axial and vertical load.
596 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

8. Conclusion

This paper has presented extensive numerical simulation results for


steel beams under combined axial compression, bending moment and
shear force, including the effects of T-section buckling. The numerical
simulation results showed that under either pure compression or pure
bending, the maximum axial or bending capacities of the beams were
limited by buckling of the compressive tee-section. Under pure axial
compression, the compressive tee-section buckled as sway columns;
thus the effective length of the tee-section was very close to the opening
length. Under pure bending moment, the compressive tee-section buck-
Fig. 30. Eq. (31).
led as a non-sway column with rotationally restrained ends so the effec-
tive length of the tee-section was close to half of the opening length. The
numerical study results also showed that when dealing with the general
the plastic moment and shear capacities are reduced to take account of situation of a beam under combined axial compression, bending mo-
axial compression and tee-section buckling. ment and shear force, the effect of compressive tee-section buckling
In summary, for any opening under combined axial compression, should be included to reduce both the bending moment and shear resis-
bending moment and shear force the following steps are taken to tances of the perforated section.
check the resistance: Based on the results of numerical simulations in ABAQUS, this paper
has proposed an analytical method to calculate the capacity of steel
1. Calculate the compressive resistance “No,b” and bending moment re- members at the location of openings, for different shapes and sizes,
sistance “Mo,b” of the section with opening using Eqs. (7) and (11) re- under a general loading condition of combined axial compression,
spectively and the recommended buckling curves (imperfection bending moment and shear force. The proposed method is based on
factors) and critical lengths in Section 3. using an existing bending moment – shear force interaction equation
2. Calculate the reduced bending moment capacity of an opening Mo,b,N for Vierendeel mechanism in the absence of additional compression
due to axial force Nsd using the bending moment –axial compression and T-section buckling, but replacing the bending moment resistance
interaction Eqs. (19) and (20): and shear resistance of the T-section with reduced values to incorporate
3. Determine the reduced shear capacity Vo,pl,N due to axial force using the influences of tee-section buckling and additional compressive force.
Eq. (29) for the calculation of the shear stress. To account for the effects of additional compression force on bending
4. Substitute the reduced values for bending and shear capacities resistance, the plastic moment-axial compression interaction equation
(Mo,b,N and Vo,pl,N respectively) into the shear-moment interaction may be used, however, the plastic bending moment capacity (without
equations of either AISC [22] or Chung et al. [16] and find the resis- axial compression) and the plastic compression resistance (without
tance of the opening (Eqs. (21) and (22)). bending) should be replaced by those under the influence of T-section

ABAQUS ABAQUS
Proposed Method (AISC) Proposed Method (AISC)
Final Proposed Method (AISC) Final Proposed Method (AISC)

ABAQUS ABAQUS
Proposed Method (AISC) Proposed Method (AISC)
Final Proposed Method (AISC)
Final Proposed Method (AISC)

Fig. 31. Comparison between ABAQUS results with different analytical results.
M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600 597

ABAQUS
Proposed Method (AISC)
Final Proposed Method (AISC)
Final Proposed Method (Chung)
Proposed Method (Chung)

Fig. 32. Comparison between ABAQUS results with different analytical results.

Fig. 33. Comparison between analytical and simulation results.


598 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

Fig. 34. Comparison between analytical and numerical axial compression resistance.

buckling. To allow for T-section buckling, an effective T-section buckling Step 1: Calculate the bending moment and axial compression capac-
length of 0.5L or L (where L is the T-section length) should be used ity of the opening under the influence of tee-section buckling (Mo,b and
when calculating the bending moment or compression resistance of No,b respectively).
the perforated section. The shear resistance of the perforated section is Using Eq. (7): No,b = 2 × Nt,b = 2 × 723.5= 1447 kN.
obtained by calculating a critical shear stress in the T-section. This crit- Nt,b is calculated using Eurocode curve b (α = 0.34) and equations as
ical shear stress -direct stress interaction is according to the von Mises given below:
equation, but the square power in the von Mises equation is replaced Nt;cr ¼ π
2 EI
t
¼ 1899:4 kN where lt , cr = lo = 698.7mm (since 10 ≤
lt;cr 2
by a function that reflects the influence of T-section buckling.
span/depth ≤ 15)
A comparison between the numerical simulation results and the an-
alytical results using the proposed method indicates very good agree- sffiffiffiffiffiffiffiffiffi
ment, with the inaccuracy mainly attributed to inaccurate calculation Nt;pl
Nt;pl ¼ At f y ¼ 920:7kN; λt ¼ ¼ 0:696 Φt
of the bending – shear interaction of the existing methods (AISC [22] Nt;cr
h   2
i
or Chung et al. [16]) which do not consider the effects of additional com- ¼ 0:5 1 þ α λt −0:2 þ λt ¼ 0:827
pression and T-section buckling.
1
χt ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:786; Ntb ¼ X t Nt;pl ¼ 723:5 kN
2
Appendix 1. Worked example Φt þ Φt 2 −λt

A 6 m span simply supported beam of section size UB457 × 152 × 82 Using Eq. (11): Mo , b = Nt , b × dc = 871.2 × 0.4397 =
is subjected to a uniformly distributed load of 10 kN/m and axial com- 383.1 kN m.where Nt,b is calculated using Eurocode curve b (α =
pressive load of 500 kN. The steel grade is S275 and the Young's Modu- 0.34) equations as given below:
lus is 210,000 N/mm2. A rectangular opening (do = 0.75 h and lo = 2do)
is placed 1 m from the support. Check the capacity of the beam at the π2 EIt
N t;cr ¼ 2
¼ 7597:6 kN where lt;cr ¼ 0:5  lo ¼ 349:35 mm
opening location. lt;cr
sffiffiffiffiffiffiffiffiffi
Data: fy = 275 N/mm2, E = 210,000 N/mm2, l = 6 m, xo = 1 m, h = N t;pl h   2
i
465.8 mm, Bf = 155.3 mm, tf = 18.9 mm, tw = 10.5 mm, d = 428 mm, λt ¼ ¼ 0:348; Φt ¼ 0:5 1 þ α λt −0:2 þ λt ¼ 0:5858
Nt;cr
do = 349.35 mm, lo = 698.7 mm, l/h = 12.9, It = 4.4738 × 105 mm4,
1
At = 3348.1 mm2, dc = 439.7 mm, Wo,pl = 1.4722 × 106 mm3, Ao = χt ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:946; N t;b ¼ χ t Nt;pl ¼ 871:2
2
6696.2 mm2, No,pl = 1841.4 kN, Mo,sd = 25 kN m, Vo,sd = 10 kN. Φt þ Φt 2 −λt

Fig. 35. Comparison between analytical (AISC or Chung et al.) and simulated bending moment resistance.
M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600 599

Fig. 36. Comparison between ABAQUS results with different analytical results for opening height do = 0.75 h and opening length lo = do.

Step 2: Reduce the bending moment capacity Mo,b for axial load to N o;b
Critical buckling stress of the section with opening f crit ¼ Ao ¼ 216:
Mo,b,N using interaction Eq. (19) or (20). 2
1 N=mm .
Axial capacity of the perforated web No , w , pl = [tw(d − do)]fy =
Average direct stress on the perforated section caused by the com-
227.1kN.
pressive force
sd = 500kN.
Nsd N No,w ,pl Therefore, P.N.A is in the bottom flange (Fig. 38). There- f n ¼ NAsdo ¼ 6696:2
500
¼ 74:7
mm2
N
.
¼ 0:79; q ¼ 0:75 þ 1:3  10−8 e18:3ρ ¼ 0:77
N o;b
fore, the linear Eq. (30) is used. ρ ¼ No;pl ¼ 1841:4
1447
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffi q q q

! Reduced shear strength f vn ¼ ðf y = 3Þ f critf −qf n ¼ 74:5 N=mm2 .


crit
W o;pl M o;b;N Nsd
  þ ¼1 Shear area of the perforated section in the method of Chung et al.
Ao 0:5h−0:5t f M o;b No;b
[16]:

Aov ¼ ðh−do Þt w þ 1:5t f 2 ¼ 1758:5 mm2


Substituting Mo , b = 383.1 kN m, No , b = 1447 kN and Nsd =
500 kN into the above interaction equation gives Mo,b,N = 254.8 kN m.
Step 3: Reduce the shear capacity Vo,pl for axial load to Vo,pl,N. The reduced shear capacity Vo,pl,N = Aovfvn =131 kN.

Fig. 37. Interaction curve of the proposed analytical design method for openings under combined axial load, bending moment and shear force including tee-section buckling.
600 M. Najafi, Y.C. Wang / Journal of Constructional Steel Research 128 (2017) 579–600

Fig. 38. Worked example beam.

Step 4: Check for Vierendeel mechanism using the interaction equa- [10] T.C. Liu, K.F. Chung, A.C. Ko, Finite element modelling on vierendeel mechanism in
steel beams with large circular web openings, Structural Engineering, Mechanics
tions of Chung et al. and the reduced bending moment and shear and Computation 2 (2001) 1567–1574.
capacities. [11] K.F. Chung, T.C.H. Liu, A.C.H. Ko, Investigation on Vierendeel mechanism in steel
The interaction equation of Chung et al. for openings with different beams with circular web openings, J. Constr. Steel Res. 57 (2001) 467–490.
[12] T.C.H. Liu, K.F. Chung, Steel beams with large web openings of various shapes and
coupled shear capacity ratio v is given as: sizes: finite element investigation, J. Constr. Steel Res. 59 (2003) 1159–1176.
 2  2   [13] K.D. Tsavdaridis, C. D'Mello, Web buckling study of the behaviour and strength of
V o;sd Mo;sd 2 perforated steel beams with different novel web opening shapes, J. Constr. Steel
þ ¼ 1 for vb Res. 67 (10) (2011) 1605–1620.
v V o;pl;N M o;b;N 3
[14] R.M. Lawson, Design for Openings in the Webs of Composite Beams (P68), Steel
Construction Institute (SCI), Ascot, 1987.
2V  32 
o;sd
− v− 23 2   [15] R. Redwood, S.H. Cho, Design of steel and composite beams with web openings, J.
4V o;pl;N 5 þ Mo;sd ¼ 1 for v≥
2 Constr. Steel Res. 25 (1993) 23–41.
2 M o;b:N 3 [16] K.F. Chung, C.H. Liu, A.C.H. Ko, Steel beams with large web openings of various
3
shapes and sizes: an empirical design method using a generalised moment-shear
interaction curve, J. Constr. Steel Res. 59 (2003) 1177–1200.
From Table 1, the coupled shear capacity ratio v for a rectangle (2:1) [17] D. Kerdal, D.A. Nethercot, Failure modes for castellated beams, J. Constr. Steel Res. 4
(1984) 295–315.
opening is 0.16. Therefore, the first equation is used here. Substituting [18] R.M. Lawson, Design for Openings in the Webs of Composite Beams, Steel
Mo , b , N = 254.8 kN m Vo , pl , N = 131 kN , Mo , sd = 25 kN m and Vo , - Contruction Institute (SCI), Ascot, 1987.
[19] J.K. Ward, Design of Composite and Non-composite Cellular Beams, vol. 100, Steel
sd =20 kN, into the above equation gives:
Construction Institute (SCI), 1990.
2 2
ð0:16131
20
Þ þ ð254:8
25
Þ ¼ 0:92b1. Therefore, the opening is ok. [20] K.H. Liew, Theoretical Modelling of Cellular Steel Beams and Steel/Concrete Com-
posite Beams With Single Openings, University of Manchester, Manchester, 2005.
Note that the ABAQUS failure load w for the applied axial load of
[21] R.M. Lawson, D. Osdatogbe, G.M. Newman, Design of FABSEC Cellular Beams in Non-
Nsd = 500 kN is 13.4 kN/m which gives the Mo,sd of 33.5 kN m and composite and Composite Applications for Both Normal Temperature and Fire Engi-
Vo,sd of 26.8 kN. neering Conditions, Steel Construction Institute (SCI), 2006.
[22] D. Darwin, Steel and Composite Beams With Web Openings, American Institute of
Steel Construction (AISC), Chicago, 1990.
References [23] R. Redwood, H.C. Soon, Design of steel and composite beams with web openings, J.
Constr. Steel Res. 25 (1993) 23–41.
[1] R.M. Lawson, S.J. Hicks, Design of Composite Beams with Large Web Openings [24] T.C.H. Liu, K.F. Chung, Practical design of universal steel beams with single web
(P355), Steel Construction Institute (SCI), Ascot, 2011. openings of different shapes, 2nd, European Conference on Steel Structures; Eurosteel
[2] J.E. Gibson, W.M. Jenkins, An investigation of the stresses and deflections in castel- '99, Prague, 1999.
lated beams, The Structural Engineer 35 (1957) 467–497. [25] C.H. Ko, K.F. Chung, A review of recent developments on design of perforated beams,
[3] M.D. Altfillisch, B.R. Cooke, A.A. Toprac, An investigation of welded open-web ex- Advances in Steel Structures 1 (2002) 121–128.
panded beams, Welding Journal 36 (2) (1957) 77. [26] D. Darwin, R.X. Donahey, LRFD for composite beams with unreinforced web open-
[4] M.U. Hosain, W.G. Speirs, Experiments on castellated steel beams, Weld. J. 52 (1973) ings, J. Struct. Eng. 114 (1988) 535–552.
329–342. [27] R. Redwood, S.C. Shrivastava, Design recommendations for steel beams with web
[5] J.E. Bower, Elastic stresses around holes in wide-flange beams, Journal of the Struc- holes, Can. J. Civ. Eng. 7 (4) (1980) 642–650.
tural Division of ASCE 92 (2) (1966) 85–102. [28] Simulia, ABAQUS User's Manual Version 6.10, Simulia, RI, 2010.
[6] J.E. Bower, Experimental stresses in wide-flange beams with holes, Journal of the [29] CEN, BS EN 1993-1-1, Eurocode 3 Design of Steel Structures, Part1.1: General Rules
Structural Division on ASCE 92 (5) (1966) 167–186. and Rules for Buildings, British Standards Institution, London, 2005.
[7] J.O. McClutcheon, W.C. So, B. Gersovitz, A Study of the Effects of Large Circular Open- [30] M. Najafi, Behaviour of Axially Restrained Steel Beams With Web Openings at Ele-
ings in the Webs of Wide Flange Beams, McGill University, Montreal, 1963. vated TemperaturesManchester 2014.
[8] J.O. McCutcheon, J.F. Dickie, S.Y. Cheng, Experimental Investigation of Large Extend-
ed Openings in the Webs of Wide Flange Beams, McGill University, Montreal, 1965.
[9] R. Redwood, J.O. McCutcheon, Beam tests with unreinforced web openings, Journal
of the Structural Division of ASCE ST1 (1968) 5706–5723.

Vous aimerez peut-être aussi