Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

887

Behaviour of steel single angles subjected to


eccentric axial loads
Yi Liu and Linbo Hui
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by Bangladesh Univ of Engineering & Technol - BUET (PERI) on 08/17/21

Abstract: The response of steel single angles subjected to axial eccentric loading is investigated by means of numerical
modeling based on finite element techniques. Results show that in the case of eccentric compression causing major axis
bending, a critical eccentricity for each slenderness ratio investigated exists and below this eccentricity, any reduction in
the ultimate capacity due to eccentricity is marginal. The critical eccentricity increases with the slenderness ratio of the an-
gle. In contrast, for the case of eccentric compression causing minor axis bending, the reduction in ultimate capacity, as af-
fected by increasing eccentricity, is more pronounced and no similar critical eccentricity is identified. When compared
with values determined from the design equations suggested in the AISC specification 2005, results indicate that the AISC
equations, in general, give a satisfactory estimate of the ultimate capacity for angles subjected to axial compression and
minor axis bending. For the case of eccentric compression causing major axis bending, AISC equations are conservative
over a large range of parameters while they overestimate capacities in some other angles.
Key words: numerical modeling, single angle, eccentric compression, flexure, structural behavior.
Résumé : La réponse des cornières simples en acier soumises à une charge axiale excentrique est examinée au moyen
d’une modélisation numérique basée sur les éléments finis. Les résultats montrent que dans le cas d’une compression ex-
centrique engendrant une grande flexion axiale, une excentricité critique existe pour chaque rapport d’élancement examiné
et, sous cette excentricité, toute réduction de la capacité à la rupture engendrée par l’excentricité est minime. L’excentricité
critique augmente avec le rapport d’élancement de la cornière. Par contre, dans le cas d’une compression excentrique en-
For personal use only.

gendrant une petite flexion axiale, la réduction de la capacité à la rupture engendrée par l’augmentation de l’excentricité
est plus prononcée et aucune excentricité critique similaire n’est signalée. Lorsqu’on les compare aux valeurs déterminées
à partir des équations de conception suggérées par AISC Specification 2005, les résultats indiquent que les équations
AISC en général donnent une estimation satisfaisante de la capacité à la rupture pour les cornières soumises à une com-
pression axiale et à une petite flexion axiale. Dans le cas d’une compression excentrique engendrant une forte flexion
axiale, les équations AISC sont conservatrices pour une grande plage de paramètres alors qu’elles surestiment les capacités
dans certaines autres cornières.
Mots-clés : modélisation numérique, cornière simple, compression excentrique, flexion, comportement structural.

Introduction In building construction involving compression angles,


the capacity of eccentrically loaded single angles may be
Single steel angles are used extensively in a variety of evaluated based on the interaction between axial load and
structures such as steel joists and trusses and braced struc- bending by treating the angles as beam-columns. Experi-
tures. In practical applications, single angles are usually ei- mental research for single angles subjected to eccentric com-
ther welded or bolted to other structural members by one pression with simple support conditions has been conducted
leg. Eccentric compressive loading through this connection by various researchers (Wakabayashi and Nonaka 1965; Ish-
detail leads to bi-axial bending. Due to the complexity asso- ida 1968; Mueller and Erzurumlu 1983; Adluri and Madu-
ciated with combined axial and bending stresses coupled gula 1992; Sakla 2005). Numerical studies based on finite
with typical asymmetric characteristics of angles, research element modeling were also used as an effective method to
into the response of these members has mainly been of an incorporate various boundary conditions, residual stress ef-
experimental nature supplemented by numerical modeling. fects, and geometric imperfections in studying both elastic
and inelastic behavior (Kitipornchai and Lee 1986; Adluri
et al. 1991; Earls and Galambos 1998; Earls and Keelor
Received 4 August 2009. Revision accepted 19 February 2010. 2007). In practice, the design of single angles as beam-col-
Published on the NRC Research Press Web site at cjce.nrc.ca on umns is governed by the American Institute of Steel Con-
3 June 2010.
struction (AISC) Specification for structural steel buildings
Y. Liu1 and L. Hui. Department of Civil and Resource (AISC 2005) in the United States, hereafter referred to as
Engineering, Dalhousie University, Halifax, NS B3J 1Z1, AISC specification and the Canadian standard CAN/CSA
Canada. S16–01 Limit states design of steel structures (CSA 2001)
Written discussion of this article is welcomed and will be in Canada. The AISC specification (2005) suggests the inter-
received by the Editor until 31 October 2010. action of flexure and axial compression applicable to spe-
cific locations on a cross section of a single angle shall be
1Corresponding author (e-mail: yi.liu@dal.ca). limited according to

Can. J. Civ. Eng. 37: 887–896 (2010) doi:10.1139/L10-028 Published by NRC Research Press
888 Can. J. Civ. Eng. Vol. 37, 2010

 
Pr Pr 8 Mrx Mry Fig. 1. Typical finite element mesh.
½1a For  0:2; þ þ  1:0
fc P n fc Pn 9 fb Mnx fb Mny
 
½1b For
Pr
< 0:2;
Pr
þ
Mrx
þ
Mry
 1:0 y x
fc P n 2fc Pn fb Mnx fb Mny
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by Bangladesh Univ of Engineering & Technol - BUET (PERI) on 08/17/21

where Pr and Pn are the required compressive strength and


z
the nominal compressive strength, Mr and Mn are the re-
quired flexural strength and the nominal flexural strength,
fc and fb are resistance factors for compression and flexure,
and x and y refer to the major and minor principal axes
(AISC 2005).
Based on a comparison study of 71 test results, Adluri
and Madugula (1992) pointed out that these interaction
equations are highly conservative. Sakla (2005) evaluated
the efficacy of these equations for both equal- and unequal-
leg angles. Sakla concluded that AISC specification (AISC
2005) provides a better prediction of the ultimate capacity
of equal-leg single angles than it does for unequal-leg single
angles. It underestimates all unequal-leg test results and
overestimates capacities of some equal-leg angles. Earls
(2003) suggested that AISC specification (2005) for single
angle beam design is, in most cases, unnecessarily conserva-
tive while in some other situations capacities of single angle
beams are overestimated.
In Canada, while the interaction approach is recognized in Fig. 2. Residual stress distribution used in the model.
For personal use only.

the current standard CSA S16–01 (2001), the axial force and 0.22Fy
moment interaction factors are different from the AISC spec-
ification (2005). Additionally, this standard does not include -
explicit provisions for determining the nominal flexural 0.24Fy + Tensile stress
+
strength for mono-symmetric or asymmetric angles, but 0.24Fy - Compressive stress
rather, it directs readers to the Structural Stability Research
Council (SSRC) Guide for design methods (Galambos 1998). -
0.25Fy - -
In the research described herein, a numerical study using
finite element modeling was carried out to study the effects 0.22Fy
of several influential parameters on the load-carrying ca- 0.25Fy
pacity of single angles eccentrically loaded with respect to
either the major or minor principal axis. Being the most
commonly used in practice, equal-leg single angles were
considered in this study. Numerical simulation results were applied outside the legs of the angle. This plate was made as
then used to further examine the effectiveness of the interac- a rigid region to avoid possible bending and consequent inter-
tion equations as are presently applicable in design. ference with the behavior of the angle elements. A similar
technique used for load application by Adluri et al. (1991) de-
Finite element model livered satisfactory results in that study. To simulate the sim-
Finite element modeling (FEM) was carried out using the ple end conditions, the centroid of the cross section at the
commercial program ANSYS 10.0 (2006). Nonlinearities re- support was restrained from lateral movement but allowed to
lating to both geometric and material charactistics were con- rotate freely in both x and y directions.
sidered. An iterative approach using the arc-length method Shell 181 elements with a mesh size of 25 mm  6.0 mm
was used in the nonlinear load–displacement analysis and the were meshed in the middle plane of the angle. The end plate
stress–strain relationship for steel was assumed to be elastic– was meshed using 6.0 mm  6.0 mm elements in the middle
perfectly plastic. Angle specimens were discretized into a plane with the nodes on two adjacent edges contacting the
mesh of elements using general-purpose 4-node Shell 181 el- angle legs coinciding with those of the angle legs. A conver-
ements as specified in ANSYS. Considering the symmetry of gence study performed for typical angles with various slen-
the member and loading, only half of the specimen was mod- derness ratios confirmed the described mesh sizes to be
eled and simple support conditions were implemented. Figure satisfactory and were therefore used in the modeling.
1 depicts a characteristic mesh for half an angle specimen Initial imperfections were considered by assuming that the
where the x and y axes intersect at the centroid of the cross initial geometry is in the form of a half sine wave applied
section and represent the major and minor principal axes, re- over the entire length of the specimen about its minor prin-
spectively. A fictitious plate attached to the end of the angle cipal axis. The sinusoidal shape function used in this study
was used to enable the modeling of the eccentric point load is expressed as follows:
Published by NRC Research Press
Liu and Hui 889

pz
½2 w ¼ wo sin corresponding deflections were recorded using a computer-
L controlled data acquisition system. Ultimate load was
where w is the initial imperfection along the angle length, deemed to have been reached when the specimen lost ability
w0 is the amplitude of the imperfection at the mid-height of to sustain increasing load while displaying a significant lat-
the angle, and L is the length of the angle. Experimental eral deflection.
measurements (Adluri and Madugula 1996; Hui 2007) The experimental ultimate load, Pexp, numerical results,
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by Bangladesh Univ of Engineering & Technol - BUET (PERI) on 08/17/21

for single angles with various cross sections showed that PFEM, and ratios of Pexp/PFEM for all specimens are included
L/1500 is a reasonable assumption for maximum amplitude in Table 1. The finite element model provided satisfactory
of the imperfection at the mid-height of angles. Unless predictions for the ultimate load with an average ratio of
otherwise noted, w0 was therefore taken as L/1500. Pexp/PFEM of 1.05 and a COV of 4.7%. Typical comparisons
Previous research (Usami and Fukumoto 1972; Temple of analytical and experimental results are illustrated in
and Sakla 1998) revealed that the maximum tensile residual Fig. 5a for load versus mid-height deflection curves and in
stress level in steel angles did not generally exceed 25% of Fig. 5b for failure modes. It shows that the analytical results
the yield stress. While residual stresses had a marked influ- compared well with experimental responses. Similar obser-
ence on the capacity of concentrically loaded steel angle vations were also made for other members (Hui 2007).
columns, a marginal effect of about 5% or less on the ca- The efficacy of the numerical model was further evaluated
pacity was observed for eccentrically loaded columns. In using 103 test results published by others including results of
this research, the residual stress distribution pattern as sug- 36 concentrically loaded angles (Wakabayashi and Nonaka
gested in ECCS (1985) was adopted and is shown in Fig. 2. 1965; Adluri and Madugula 1996), 61 eccentrically loaded
Applied in the longitudinal direction of the member, the angles (Wakabayashi and Nonaka 1965; Ishida 1968; Muel-
self-equilibrating residual stresses were first transformed ler and Erzurumlu 1983), and six angles under flexure only
into a series of discrete uniform stresses. These residual (Madugula et al. 1995). A summary comparison of analytical
stresses were then specified by creating an initial stress field and experimental ultimate loads is presented in Table 2 and
applied to each element of the model during the first sub- detailed graphical comparisons are available in Hui (2007).
step of the first load step of the analysis. In general, reasonably good agreement is obtained between
the numerical and experimental results for angles of various
For personal use only.

configurations subjected to various loading conditions.


Verification of the finite element model
Concurrent with the numerical study presented in this pa- Parametric study
per, an experimental program investigating the behavior and
capacity of single angles under eccentric compression was Included in the parameters investigated by computer anal-
also carried out. In the experimental program (Hui 2007), 28 ysis is the slenderness parameter l, applied eccentricity ex or
L51  51  6.4 single angles were subjected to eccentric ey, initial imperfection w0, and angle leg width-to-thickness
compression causing either major or minor principal axis ratio b/t. The effects of these parameters were studied for
bending with pinned end supports. Table 1 lists the details angles eccentrically loaded with respect to either principal
of test specimens including slenderness parameter, l; yield axis. The test population included nine slenderness parame-
stress, Fy; and eccentricities ex and ey with respect to major ters (l = 0.27, 0.53, 0.80, 1.06, 1.33, 1.60, 1.86, 2.13, and
and minor axis, respectively. Eccentricities, ex and ey, are il- 2.39), six eccentricities (0, x0/4, x0/2, x0, 3x0/2, 2x0) where
lustrated in Fig. 3 together with other key notations used x0 is the distance from the shear center to the centroid of
throughout q this the cross section measured in the major-axis direction, seven
ffiffiffiffiffiffipaper. The slenderness parameter l is calcu- values of initial imperfection (w0 = L/3000, L/2500, L/2000,
F
lated as KL
ry
y
p2 E, where ry is the radius of gyration with re- L/1500, L/1000, L/500, and L/250), and five plate slender-
spect to the minor principal axis and K was taken as 1.0. The ness values of the angle legs (b/t = 5.3, 6.4, 8.0, 10.7, and
initial out-of-straightness along the length of each specimen 16.0). The dimensions of angle sections were taken as the
was measured and the maximum value was less than L/1500 nominal values according to CSA S16–01 (2001). An elas-
with a mean of L/2100 and a coefficient of variation (COV) tic-perfectly plastic material model was assumed for steel
of 8.7% for all specimens. A 500 kN Instron hydraulic uni- with a yield stress of 350 MPa, an elastic modulus of 200
versal testing machine (Fig. 4) was used to load specimens 000 MPa, and a Poisson ratio of 0.3.
in compression. As suggested in the SSRC guide (Galambos
1998), ball bearing supports were used to realize pinned ends Angles eccentrically loaded with respect to
for each specimen. Each support consisted of two 13 mm major principal axis
parallel plates separated by a stainless steel ball seated in
upper and lower sockets suitably milled into the plates. Effect of slenderness parameter, l, and eccentricity, ex
Both vertical deformation and lateral displacements were Normalized ultimate load versus eccentricity curves are
measured during each test. Vertical deformation of a speci- shown in Fig. 6 for a 51  51  6.4 angle with five repre-
men was measured with a built-in linear variable differential sentative slenderness parameters. For angles with a low slen-
transducer (LVDT), which measured the travel of the load- derness (l = 0.27), an increase in eccentricity resulted in an
ing head. Lateral displacements were measured using immediate decrease in ultimate load. As the slenderness in-
LVDTs at the mid-height of each specimen and the mid- creased, the curves showed two distinctive regions. For a
width of each leg. During each test, loading was applied at given slenderness, there existed a critical eccentricity below
a constant rate of 6 kN/min up to failure while loads and which the change of the ultimate load with an increase in

Published by NRC Research Press


890 Can. J. Civ. Eng. Vol. 37, 2010

Table 1. Comparison of experimental and analytical results (Hui 2007).

No Specimen ID l ex (mm) ey (mm) Fy (MPa) Pexp (kN) PFEM (kN) Pexp /PFEM
1 E900–1 1.23 0.0 0.0 330 117.0 105.5 1.11
2 E900–2 1.23 4.2 0.0 330 110.9 105.1 1.06
3 E900–3 1.24 10.0 0.0 330 105.3 104.9 1.00
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by Bangladesh Univ of Engineering & Technol - BUET (PERI) on 08/17/21

4 E900–4 1.22 16.8 0.0 330 84.5 77.6 1.09


5 E900–5 1.22 29.4 0.0 330 59.0 56.1 1.05
6 E900–6 1.22 50.4 0.0 330 46.1 39.7 1.16
7 E900–7 1.25 0.0 8.4 348 65.6 59.5 1.11
8 E900–8 1.25 0.0 16.8 348 46.4 44.4 1.10
9 E900–9 1.25 0.0 29.4 348 34.4 31.9 1.05
10 E900–10 1.25 0.0 50.4 348 23.3 22.7 1.08
11 E1200–1a 1.62 4.2 0.0 330 68.1 66.5 1.03
12 E1200–1b 1.62 4.2 0.0 330 69.7 66.1 1.02
13 E1200–2 1.62 10.0 0.0 330 66.7 66.0 1.05
14 E1200–3a 1.65 16.8 0.0 348 66.4 65.3 1.01
15 E1200–3b 1.62 16.8 0.0 330 71.3 65.2 1.02
16 E1200–4 1.62 29.4 0.0 330 57.0 52.9 1.09
17 E1200–5 1.65 50.4 0.0 348 42.5 38.3 1.08
18 E1500–1a 2.06 0.0 0.0 348 45.3 43.8 1.11
19 E1500–1b 2.05 0.0 0.0 348 41.1 43.9 1.03
20 E1500–1c 2.06 0.0 0.0 348 42.8 43.8 0.94
21 E1500–2 2.05 8.4 0.0 348 42.9 42.9 0.98
22 E1500–3 2.05 16.8 0.0 348 41.3 41.1 1.00
23 E1500–4 2.05 29.4 0.0 348 40.1 40.9 1.01
24 E1500–5 2.05 50.4 0.0 348 37.0 36.0 0.98
For personal use only.

25 E1500–6 2.05 0.0 8.4 348 35.2 32.4 1.03


26 E1500–7 2.06 0.0 16.8 348 27.8 26.5 1.09
27 E1500–7 2.06 0.0 29.4 348 23.4 21.5 1.05
28 E1500–7 2.05 0.0 50.4 348 18.0 16.3 1.09
Avg. 1.05
C.O.V 4.7%

Fig. 3. Description of the cross section details of the angle. ex0, the maximum stress occurred simultaneously at both the
y heel and the toe. A further increase in the eccentricity re-
sulted in maximum stress occurring at the toe.
ey Numerical results showed that angles subjected to eccen-
Loading point
Toe tric compression resulting in major axis bending failed in
torsional–flexural buckling mode. This failure mode is be-
lieved to attribute to the phenomenon of critical eccentricity.
ex The flexural deflection and twist of the cross section
reached maximum at the mid-height of the angle and results
Heel O (C.G) x in re-orientation of the principal axes. To satisfy equilibrium
on the deformed shape, additional stresses due to the twist
of the cross section result at both the heel and toe of the
S.C cross section. These additional stresses either add to or sub-
tract from the stresses caused by axial and flexural deforma-
xo
tions at the heel and toe locations. The critical eccentricity is
that at which the combined stresses at both the heel and the
eccentricity was marginal to almost unnoticeable. This indi- toe simultaneously reach a maximum value. The value of
cates that an eccentrically loaded angle can reach its concen- this critical eccentricity is dependent on the magnitude of
tric compressive capacity as long as the applied eccentricity flexural deflection and twist of the angle.
was within the critical eccentricity. Defined as ex0, the crit-
ical eccentricity assumed an increasingly greater value with Effect of leg width-to-thickness ratio, b/t
increased slenderness. Figure 7 illustrates the numerical von Five cross sections, L51  51  3.2, L51  51  4.8,
Mises stress distribution on the cross section at mid-height L51  51  6.4, L51  51  7.9, and L51  51  9.5
of a specimen where the location of the maximum stress is with the leg width-to-thickness ratios, b/t, ranging from 5.3
indicated in a circle. Figure 7 shows that when the applied to 16.0 were used in this study. The relationship of ultimate
eccentricity is below ex0, the maximum stress occurred at load versus b/t ratios is shown in Fig. 8 for four representa-
the heel of the angle. As the applied eccentricity approached tive l values with ex taken as x0/2. As the b/t ratio increased,

Published by NRC Research Press


Liu and Hui 891

Fig. 4. Test set-up. Fig. 5. Comparison of analytical and experimental results: (a) load
versus deformation response; (b) failure modes.
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by Bangladesh Univ of Engineering & Technol - BUET (PERI) on 08/17/21
For personal use only.

the ultimate load capacity decreased for all investigated


slenderness parameters. An increase in b/t ratios from 5.3 to
16.0 resulted in the reductions in the ultimate load capacity
of 66.2%, 63.1%, 62.2%, and 62.2% for angles with l of
0.80, 1.33, 1.86, and 2.39, respectively. Furthermore, the
critical eccentricities as affected by slenderness were also
determined for these five b/t ratios. Expressed in terms of
the critical eccentricity ratio m0, defined as the ratio of crit-
ical eccentricity, ex0 and x0, results are listed in Table 3 and
m0 versus l curves are plotted for five b/t ratios in Fig. 9. Effect of initial imperfection, w0
Figure 9 shows that for all b/t ratios, m0 and thus ex0 in- The ultimate load versus initial imperfection curves for a
creased with an increase in the slenderness parameter. How- 51  51  6.4 angle with l = 1.07 and eccentricities vary-
ever, for a given slenderness parameter, the variation of ex0 ing from 0 to 2x0 are plotted in Fig. 10. In the case of con-
with b/t ratios was more complex. For l < 1.0, ex0 remained centric loading (ex = 0), the ultimate load showed a
almost constant for all b/t ratios. For l > 1.0, the variation consistent decrease as the initial imperfection increased and
in ex0 for different b/t ratios became more pronounced and the total decrease was 35.9% when the initial imperfection
an increase in the b/t ratio resulted in an increase in ex0 val- varied from L/3000 to L/250. In the case of ex = x0/2, the
ues. It suggests that when the failure of the angle shifted ultimate load of the angle was essentially the same as its
from inelastic to elastic, the effect of the b/t ratio on the concentric axial capacity for the range of the initial imper-
critical eccentricity values increased. It has been established fection values that were studied. The proximity of the two
that for angles failing in elastic flexural–torsional buckling, curves for ex = 0 and x0/2 is attributed to the fact that the
the b/t ratio affected the flexural deflection and the twist of critical eccentricity for angle 51  51  6.4 with l =
the angle (Dabrowski 1961), which in turn would affect the 1.07 was determined to be 0.5x0 (Table 3), which was coin-
critical eccentricity values. As the angle became stocky in cident with the applied eccentricity. Therefore, the ultimate
the length direction, it seemed that the b/t ratio had a dimin- load of the angle loaded at an eccentricity of 0.5x0 reached
ishing effect on the flexural deflection and twist and thus the its approximate concentric loading capacity. For ex = x0 and
critical eccentricity values. 2x0, the magnitude of the initial imperfection had no notice-

Published by NRC Research Press


892 Can. J. Civ. Eng. Vol. 37, 2010

Table 2. Comparison of experimental and analytical results for single equal-leg angles.

No. of Slenderness Avg. of


Source specs ratio Angle size Fy (MPa) Loading condition Pexp /PFEM COV (%)
Wakabayashi and 10 20–150 90907 303 Concentrically loaded 1.17 9.3
Nonaka (1965)
Adluri and Madugula 26 68–188 64649.5 335–411 Concentrically loaded
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by Bangladesh Univ of Engineering & Technol - BUET (PERI) on 08/17/21

(1996) 76764.8
767612.7
1021027.9
1271279.5
Wakabayashi and 40 20–150 90907 294–323 Eccentrically loaded 1.02 8.4
Nonaka (1965) causing major,
minor and bi-axial
bending
Ishida (1968) 7 20–100 75756 58.8–63.9 Eccentrically loaded 0.96 3.0
causing minor axis
bending
Mueller and Erzurumlu 14 60, 110, 75756 349–422 Eccentrically loaded 1.00 10.4
(1983) 192 causing bi-axial
bending
Madugula et al. (1995) 6 53.3, 100 75759 295–433 Pure bending about 0.98 1.7
757512 minor principal axis
35035035

Fig. 6. Normalized load versus eccentricity for various slenderness For pure moment capacities corresponding to P = 0,
For personal use only.

ratios (major axis bending). although both the analytical and AISC moment capacities
decreased with an increase in the slenderness ratio, the ana-
lytical values were consistently higher than the AISC results.
For most slenderness ratios, the analytical moment capaci-
ties were above 1.5My, the upper bound moment capacity
specified by the AISC specification (2005). This is in line
with findings reported by Earls (2003) for major axis bend-
ing. The AISC results, on the other hand, showed a continu-
ous decreasing moment capacity from 1.5My as the
slenderness ratio increases.
In comparing combined axial load and moment cases, a
significant deviation of analytical results from results using
AISC specification (2005) equations was evident where the
analytical results, in general, provided higher estimates of
capacity. Consistent with the critical eccentricity discussion,
able effect on the ultimate load, which was also noted by the analytical load–moment interaction curves demonstrated
Temple and Sakla (1998). two distinctive regions where, in one region the axial load
remained almost constant as the moment increased, and in
Evaluation of AISC specification 2005 the other region the axial load decreased as the moment in-
The load–moment interaction (eq. [1]) as suggested in the creased. The first region became increasingly larger as the
AISC specification (2005) was compared to the numerical slenderness ratio increased. For l = 0.80, the AISC specifi-
results for angles eccentrically loaded with respect toa major cation (2005) underestimated the load capacity by an
principal axis. Figure 11 shows one such comparison for an- amount as large as 15% when M < 0.8 My. For l = 1.33,
gle 51  51  6.4 (b/t = 8.0) with various l values where 1.86, and 2.39, the underestimation by AISC specification
the load was normalized by the yield strength of the cross (2005) was evident over an increasingly wide moment
section, Py, and the moment was normalized by the yield range. In the region of critical eccentricities, AISC values
moment, My. In comparing pure axial capacities (M = 0), can be as low as 70%, 64%, and 57% of the numerical re-
the analytical and AISC results agreed well with each other sults, respectively. On the other hand, there were instances
except in the case of l = 0.80. Residual stresses and initial where the AISC specification (2005) overestimated the an-
imperfections tend to have a larger effect on angles with in- gle compressive capacity. For example, in the region of
termediate slendernesses. It is therefore believed that the 0.6My < M < 1.3 My for l = 0.27, the AISC overestimated
discrepancy noted above was due to the influence of the as- the capacity by a maximum amount of 8% whereas the
sumed values for residual stress and initial imperfection overestimation was as high as 20% in the region of M > 0.8
magnitudes that were used in the numerical model. My for l = 0.80.

Published by NRC Research Press


Liu and Hui 893

Fig. 7. Numerical von Mises stress distribution on the cross section.


Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by Bangladesh Univ of Engineering & Technol - BUET (PERI) on 08/17/21

(a) ex < exo (b) ex = exo (c) ex > exo

Fig. 8. Ultimate load versus b/t ratio for various slenderness ratios Fig. 9. Critical eccentricity ratio versus slenderness parameter for
(major axis bending). various b/t ratios.
200 4.0
3.5
160
Ultimate load (kN)

λ=0.80 3.0
120 mo =exo /xo b/t=16.0
2.5
1.33
b/t=10.7
80 1.86 2.0
2.39 b/t=8.0
1.5
For personal use only.

40 b/t=6.4
1.0
0 b/t=5.3
0.5
3 6 9 12 15 18
b/t 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Table 3. Critical eccentricity ratio for various cross sections. Slenderness parameter λ
Critical eccentricity ratio, m0 = ex0/x0
b/t l=0.27 l=0.80 l=1.07 l=1.33 l=1.86 l=2.40 Fig. 10. Effect of initial imperfection w0 on the angle ultimate load
5.3 0.01 0.18 0.42 0.58 1.25 2.45 (major axis bending).
6.4 0.01 0.19 0.44 0.61 1.34 2.79 160 ex=0
8.0 0.01 0.20 0.49 0.66 1.48 3.07
10.7 0.01 0.21 0.51 0.71 1.59 3.32 xo/2
Ultimate load (kN)

120
16.0 0.01 0.21 0.53 0.77 1.75 3.58
xo
80
Angles eccentrically loaded with respect to
2xo
minor principal axis 40
Effect of slenderness parameter l, and eccentricity ey
Minor principal axis bending occurs when the load eccen- 0
tricity is along the major principal axis. Depending on the 0 1/1000 1/500 3/1000 1/250
sense of the eccentricity either the toe or the heel of the an-
gle will be in compression. Numerical results showed that Initial imperfection at mid-height wo (xL)
the angle behavior and capacity remained approximately the
same for a given magnitude of eccentricity causing minor
axis bending regardless of its sense. Figure 12 illustrates the the critical eccentricity phenomenon as discussed in the
normalized ultimate load versus eccentricity curves for vari- case of major axis bending. Numerical results showed that
ous slenderness ratios for a 51  51  6.4 angle with the angles subjected to eccentric compression resulting in minor
toe in compression. Compared with a 51  51  6.4 angle axis bending failed in flexural buckling but without twisting
subjected to eccentric compression causing major axis bend- of the cross section. Failure was then controlled by compres-
ing (Fig. 6), the ultimate loads associated with the minor sive stress occurring either at the heel or toe of an angle de-
axis bending were lower for all investigated eccentricities pending on whether the applied moment causes the heel or
and the curves for minor axis bending did not demonstrate toe in compression.

Published by NRC Research Press


894 Can. J. Civ. Eng. Vol. 37, 2010

Fig. 11. Comparison of normalized load–moment interaction dia- Fig. 13. Effect of initial imperfection w0 on the angle ultimate load
gram of analytical and AISC specification (2005) results (major (minor axis bending).
axis bending). 160 ey =0

Ultimate load (kN)


120
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by Bangladesh Univ of Engineering & Technol - BUET (PERI) on 08/17/21

xo /4
80 xo /2
xo
40

0
0 1/2000 1/1000 3/2000 1/500

Initial imperfection at mid-height wo(xL)

Fig. 14. Comparison of normalized load–moment interaction dia-


gram of analytical and AISC specification (2005) results (minor
Fig. 12. Normalized load versus eccentricity for various slenderness axis bending): toe in compression.
ratios (minor axis bending): toes in compression. 1.0
1.0 FEM

Normalized load P/Py


0.8 λ=0.27 AISC specification 2005
Normalized load P/Py

0.8 0.6 0.80


λ=0.27
For personal use only.

0.6 0.4 1.33

0.80 1.86
0.4 0.2
2.39
1.33 0.0
0.2
1.86 0.0 0.5 1.0 1.5 2.0
2.39
0.0 Normalized moment M/M y

0.0 0.4 0.8 1.2 1.6 2.0


e y (× xo) l = 1.06 and with the toe in compression. Initial imperfec-
tions showed a marked influence on the ultimate load only
for concentrically loaded specimens. For eccentrically
loaded specimens, the ultimate load remained practically
Effect of leg width-to-thickness ratio, b/t
unchanged as the initial imperfection increased.
The variation of ultimate loads as affected by the b/t ratios
between cases of the toe in compression and the heel in com-
pression was insignificant. The trend of the reduction in the Evaluation of AISC specification (2005)
ultimate load capacity with an increase in b/t values was sim- Load–moment interaction curves for combined axial load
ilar to that exhibited in the major axis bending case (Fig. 8). and minor axis bending obtained from the AISC specifica-
For example, as b/t increased from 5.3 to 16.0, the ultimate tion (2005) and numerical model are compared for a 51 
load capacity decreased by 66.8%, 64.1%, 63.2%, and 63.2% 51  6.4 angle in Fig. 14 for the case of the toe in compres-
for angles with l of 0.80, 1.33, 1.86, and 2.39, respectively, sion. The comparison for the case of the heel in compression
for the toe in compression. Similar percentage reductions are is almost identical and thus is not presented. The interaction
also observed for the case of the heel in compression. curves in this case did not show a moment region where the
ultimate load remained almost constant, as in the case of
Effect of initial imperfection, w0 major axis bending. Rather, it showed that, the ultimate
In the case of eccentric compression causing minor axis load decreased when the moment and thus the applied ec-
bending, the direction of the initial imperfection was as- centricity increased. While the load–moment interaction val-
sumed to increase the moment effect in cases of either toe ues obtained from AISC were generally lower than the
or heel compression. Results showed that the effects of ini- analytical values, the correlation between the two was rea-
tial imperfections on ultimate capacities varied little regard- sonably good especially for angles with l values greater
less of whether the toe or heel was in compression. Ultimate than 0.8. For angles with l = 0.27, AISC specification
load versus initial imperfection curves for various eccentric- (2005) was increasingly conservative as the moment became
ities are shown in Fig. 13 for a 51  51  6.4 angle with larger. At the extreme case, when P = 0, the numerical

Published by NRC Research Press


Liu and Hui 895

flexural capacity was shown to be a function of the slenderness References


ratio where it decreased with increased slenderness whereas the Adluri, S.M.R., and Madugula, M.K.S. 1992. Eccentrically loaded
AISC specification (2005) limited the flexural strength for mi- single angle struts. Engineering Journal, AISC, 29(2): 59–66.
nor axis bending to 1.5 My without considering the slenderness Adluri, S.M.R., and Madugula, M.K.S. 1996. Flexural buckling of
effect. This lower bound estimate in the flexural strength is be- steel angles: experimental investigation. Journal of Structural
lieved to result in the conservatism for short angles in this case. Engineering, 122(3): 309–317. doi:10.1061/(ASCE)0733-
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by Bangladesh Univ of Engineering & Technol - BUET (PERI) on 08/17/21

9445(1996)122:3(309).
Conclusions Adluri, S.M.R., Madugula, M.K.S., and Monforton, G.R. 1991. Finite
element failure analysis of schifflerized angles. Computers &
A parametric study was conducted to investigate the ef- Structures, 41(5): 1087–1093. doi:10.1016/0045-7949(91)90303-4.
fects of several parameters on the load-carrying capacity of AISC. 2005. Specification for structural steel buildings. American
equal-leg single angles eccentrically loaded resulting in ei- Institute of Steel Construction, Chicago, IL.
ther the major or minor principal axis bending. Simple sup- ANSYS 10.0. 2006. Release 10.0 Documentation for ANSYS:
port conditions were assumed and angles were not restrained Guide to the ANSYS Documentation. ANSYS, Inc., and
to bend about any particular axis. The essential conclusions ANSYS Europe, Ltd.
listed below should be viewed in this context. CAN/CSA S16–01. 2001. Limit states design of steel structures.
For eccentric compression resulting in major axis bend- Canadian Standards Association, Mississauga, Ont.
ing, the ultimate load of equal-leg single angles did not Dabrowski, R. 1961. Thin-walled members under biaxial eccentric
show significant variation from the pure axial capacity as compression. Der Stahlbau. Wilhelm Ernst and Sohn, Germany.
long as the applied eccentricity was within the critical ec- Vol. 30. pp. 360–365.
centricity. Values of the critical eccentricity increased with Earls, C.J. 2003. Design of single angles bent about the major prin-
increasing angle slenderness. For eccentric compression re- cipal axis. Engineering Journal, AISC, 40(3): 159–166.
Earls, C.J., and Galambos, T.V. 1998. Practical compactness and
sulting in minor axis bending, the effects of all the parame-
bracing provisions for the design of single angle beams. Engi-
ters studied on angle ultimate capacities showed negligible
neering Journal, AISC, 35(1): 19–25.
variations for cases of either the angle toe or heel in com- Earls, C.J., and Keelor, D.C. 2007. Toward a simplified design of sin-
pression. In contrast with the major axis bending, the critical gle-angle beam-columns. Engineering Journal, AISC, 44(1): 55–63.
For personal use only.

eccentricity was not evident and the ultimate capacity con- ECCS. 1985. Recommendations for angles in lattice towers. Tech-
sistently decreased as the eccentricity increases from zero. nical Committee 8, Publication No. 39, European Convention
For eccentric compression resulting in either major or minor for Constructional Steelwork, Brussels, Belgium.
axis bending, the effect of leg width-to-thickness ratios was Galambos, T.V. 1998. Guide to stability design criteria formetal
similar in that as the b/t ratio increased, the ultimate capacity structures. 5th ed. Structural Stability Research Council. John
decreased for all investigated slenderness ratios. However, in Wiley and Sons, Inc., New York.
the case of major axis bending, it was found that an increase Hui, L.B. 2007. Beam-column behaviour of steel single angles.
in b/t ratios resulted in an increase in the critical eccentricity MASc. thesis, Department of Civil and Resource Engineering,
especially in the region of l > 1.0. For l < 1.0, the critical ec- Dalhousie University, Halifax, NS.
centricity values remained practically unaffected by b/t ratios. Ishida, A. 1968. Experimental study on column carrying capacity
For eccentric compression resulting in major axis bending, of ‘SHY steel’ angles. Yawata Technical Report 265, Yawata
the initial imperfection had no significant effect on the ulti- Iron and Steel Co. Ltd., Tokyo, Japan.
mate capacity of the angle provided that the applied eccentric- Kitipornchai, S., and Lee, H.W. 1986. Inelastic buckling of single
angle tee and double angle struts. Journal of Constructional
ity was greater than the critical eccentricity. Otherwise, an
Steel Research, 6(1): 3–20. doi:10.1016/0143-974X(86)90018-0.
increase in the initial imperfection resulted in a decrease in Madugula, M.K.S., Kojima, T., Kajita, Y., and Ohama, M. 1995.
the ultimate capacity of the angle. However, in the case of mi- Minor axis bending strength of angle beams. Proceedings of the
nor axis bending, the initial imperfection had a negligible ef- International Conference on Structural Stability and Design,
fect on the ultimate capacity of an eccentrically loaded angle. Sydney, Australia, pp. 73–78.
Compared with the numerical results, AISC specification Mueller, W.H., and Erzurumlu, H. 1983. Behaviour and strength of
(2005), in general, markedly underestimated the ultimate angles in compression: an experimental investigation. Research
load capacity of single angles under eccentric loading caus- Report of Civil-Structural Engineering, Division of Engineering
ing major axis bending. Results also identified a few instan- and Applied Science, Portland State University, Oregon, USA.
ces where AISC overestimated ultimate capacities of angles Sakla, S.S.S. 2005. Performance of the AISC LRFD specification
in the case of major axis bending. For eccentric loading in predicting the capacity of eccentrically loaded single-angle
causing minor axis bending, AISC specification (2005) struts. Engineering Journal, AISC, 42(40): 239–246.
showed a marked underestimation of compression capacities Temple, M.C., and Sakla, S.S.S. 1998. Single-angle compression
for angles with a low-l value (l = 0.27) but it provided members welded by one leg to a gusset plate, II, a parametric
lower bound and reasonably accurate predictions for angles study and design equation. Canadian Journal of Civil Engineer-
with relatively large slenderness. ing, 25(3): 585–594. doi:10.1139/cjce-25-3-585.
Usami, T., and Fukumoto, Y. 1972. Compressive strength and de-
Acknowledgement sign of bracing members with angle or tee sections. Proceedings
of Japanese Society of Civil Engineering, Vol. 201. pp. 31–50.
The authors wish to recognize the contribution of finan- Wakabayashi, M., and Nonaka, T. 1965. On the buckling strength
cial assistance by the Steel Structures Educational Founda- of angles in transmission towers. Bulletin of the Disaster Pre-
tion (SSEF) as well as in kind assistance from Marid vention Research Institute, Kyoto University, Japan, 15(91),
Industries Ltd. for the donation of specimens. Part 2, pp. 1–18.

Published by NRC Research Press


896 Can. J. Civ. Eng. Vol. 37, 2010

Symbol list Prrequired compressive strength


Py yield capacity of angle cross section
b/t angle leg width-to-thickness ratio execcentricity with respect to major principal axis
E modulus of elasticity of angle steel ex0critical eccentricity with respect to the major princi-
Fy yield stress of steel pal axis
K effective length factor = 1.0 ey eccentricity with respect to minor principal axis
w initial imperfection along the length of angle
Can. J. Civ. Eng. Downloaded from cdnsciencepub.com by Bangladesh Univ of Engineering & Technol - BUET (PERI) on 08/17/21

KL/ry slenderness ratio of angle


L length of angle w0 amplitude of the imperfection at the mid-height of
angle qffiffiffiffiffiffi
M applied moment Fy
Mn nominal flexural strength l slenderness parameter = KL ry p2 E
Mr required flexural strength fb resistance factors for flexure
My yield moment about the axis of bending fc resistance factors for compression
P applied compressive load x0 distance between shear centre and centroid of angle
Pexp experimental compressive capacity cross section
PFEM finite element results of compressive load capacity
Pn nominal compressive strength
For personal use only.

Published by NRC Research Press

You might also like