See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/285518380

Design of steel channel tension members-proposal to IS 800:2007

Article  in  Journal of Structural Engineering (Madras) · June 2011

CITATIONS READS

2 12,182

4 authors:

Bharathi Priya M. Keerthana

CSIR Structural Engineering Research Centre CSIR Structural Engineering Research Centre
19 PUBLICATIONS   58 CITATIONS    26 PUBLICATIONS   23 CITATIONS   

SEE PROFILE SEE PROFILE

Mohit Verma Nawal Kishor Banjara

CSIR Structural Engineering Research Centre CSIR Structural Engineering Research Centre
30 PUBLICATIONS   113 CITATIONS    17 PUBLICATIONS   35 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

tall building View project

Magneto-Rheological Elastomer based Smart Seismic and Vibration Isolators View project

All content following this page was uploaded by M. Keerthana on 08 February 2016.

The user has requested enhancement of the downloaded file.

Journal of Structural Engineering
Vol. 38, No. 2, June - July 2011 pp. 122–130 No. 38-16

Design of steel channel tension members-proposal to IS 800:2007

C. Bharathi Priya∗ , M. Keerthana∗ , Mohit Verma∗ and Nawal Kishor Banjara∗
* E-mail: bharupriya@gmail.com

∗ CSIR-Structural Engineering Research Centre, CSIR Campus, Taramani, Chennai-600 113, India.

Received: 17 March 2010; Accepted: 13 May 2010

Steel tension members are commonly used in bracing members, chords and web members of trusses, in which not all
the elements of the cross section are connected to the gusset plates. The most commonly used cross sections for tension
are angles and channels. These sections are connected only through one leg due to which the unconnected elements
suffer a ’shear lag’ compared to the connected legs. Hence, the whole section is not effective in resisting the tension.
The formula given in IS 800:2007 for angle sections is generally used for the design of channel sections taking analogous
dimensions. The objective of the present investigation is to study the adequacy of the present IS: 800 formula for channels
and accordingly propose new design equation. Experimental investigations and provisions of other international codes
for channel tension members are taken as basis to critically examine the design provisions of IS: 800.

KEYWORDS: Tension members; shearlag; block shear; rupture; end distance; bearing; end spliting.

Steel tension members are widely used for structural ap- are connected by single row of fasteners. The extension
plications. Mostly angles and channels are used to car- of the same equation to channel section is not appro-
ry tensile loads. According to IS 800: 20071 , strength priate. In this paper, details of investigation have been
of the steel tension member is governed by the least of reported to check whether the equation is appropriate
(i) gross section failure (ii) net section rupture and (iii) for channel sections.
block shear mode of failure. These modes of failure are
represented in Fig 1. The present study aims to inve-
stigate the behavior of the channel members connected REVIEW OF LITERATURE
through web, by single row of fasteners. In this parti-
cular case, the gross section yielding may not govern A number of experimental results are reported in the li-
and the failure also does not take place by block shear. terature for tension members. Kulak and Wu2 presen-
Hence, the present studies will concentrate on net secti- ted a good review of such earlier experimental work
on rupture, where the phenomenon of “shear lag” redu- and studied the shear lag effect in bolted angle under
ces the efficiency of the cross section. The net section tension. The design provisions for tension members of
rupture capacity of a steel tension member mainly de- many codes are based on this work. Usha3 also made
pends on the factors such as net area, connection eccen- an analytical study on the nonlinear behavior of steel
tricity and connection length. As per IS 800: 20071 , the angle tension members. A nonlinear finite element mo-
equation for calculating angle section capacity is gene- del was developed to study the behavior of steel angle
rally used for finding the capacity of channel sections. tension members up to failure considering the material
But in reality the behavior of angle sections and channel and geometrical nonlinearities. Gupta and Gupta4 car-
sections are not the same especially when the channels ried out experimental and numerical investigations on
stress distribution in bolted steel angles in tension. Arul

JOURNAL OF STRUCTURAL ENGINEERING 123

Vol.38, No.2, JUNE-JULY 2011
Jayachandran et al5 carried out analytical and pilot ex- If a ‘block’ of the connected element is partially dri-
perimental studies to analyze the design procedure with ven from the remainder of the element as in Fig. 1 (c),
respect to various codes. The non-linear finite element the capacity is calculated by block shear strength which
analysis using ABAQUS software was used to find the is the minimum of the following
stress distribution of the critical cross section during fai-
lure. Many of the factors influencing behavior of chan- Avg fy Atn fu
Td = √ + 0.9 or (3)
nel tension members were taken into account in the 3γmo γm1
work. The design of tension members in IS: 800:2007 Avn fu Atg fy
is based on the work of Kulak and Wu2 and Usha3 . Both Td = 0.9 √ + 0.9 (4)
works are based on the shear lag phenomenon observed 3γm1 γmo
in angle members. However, the same design expres-
sions are extended to the design of channel and other Angles in Tension
tension members using the appropriate shear lag distan-
ce, connection and material characteristics. This paper
presents some investigations into this aspect. When the loading is through the centroid, the design
principle used in case of plates can also be used for the
Tension Area design of angles and channels. In most of the practical
cases channels and angles are connected to other ele-
ments (like gusset plate) with only one element. When
axial load is transferred to only some elements but not
to the whole cross section, there is a variation in the
distribution of stress across the section. As we move
away from the connected surface, the stresses decrease.
Shear Area
This phenomenon is called as shear lag due to which the
(a) (b) (c)
strength of the full section is not availed. Hence, shear
Fig. 1 Modes of failure in tension members (a) Gross Section Yielding
lag should be taken into account in computation of de-
(b) Net section Rupture (c) Block Shear
sign capacity of the section. The stress flow path and
stress distribution in connected as well as unconnected
DESIGN STRENGTH OF TENSION MEMBERS AC-
leg is shown in Fig. 2.
CORDING TO IS 800: 2007
Stress
distribution
The design tensile capacity of the section is the lowest
of (i) Gross Section Capacity (ii) Net Section Capaci-
ty (iii) Block shear capacity. The gross section capacity
can be obtained using the equation

Ag fy
Td = (1)
γmo

When tension members are connected to the other ele- Stress

ments with the fasteners as in Fig. 1(b), the stress distri-
bution will not be uniform because of the stress con- Fig. 2 Shear lag effect in angle tension members
centration around the hole. In such cases, the design
strength governed by net cross section at the hole is gi- An empirical equation which incorporates the shear
ven by lag effect as given in IS 800:2007 is given as follows

An fu Anc fu Ago fy
Td = 0.9 (2) Tdn = 0.9 +β (5)
γm1 γm1 γm0

124 JOURNAL OF STRUCTURAL ENGINEERING

Vol.38, No.2, JUNE-JULY 2011
 w f   b 
β = 1.4 − 0.076
y s
(6) section area by multiplying with a coefficient U. The net
t fu Lc section capacity is given by:
fu γmo
≥ β ≥ 0.7 tn = φ Ae Fu (7)
fy γm1
Ae = U An (8)
x̄
U =1− (9)
L

Fig. 4 Connection eccentricity as per AISC code

Lc AS 4100 - 1998 7
t
The nominal section capacity of a tension member is ta-
ken as the minimum of gross and net section capacities.
When an element is connected to another element, net
w section capacity governs the design. The design equati-
on does not take into account the effect of shear lag and
is given by

bs = w + w 1 - t T = 0.85 kt An fu (10)
w1
Fig. 3 Angle Connected by one leg ANALYSIS OF DESIGN STRENGTH

Channels in Tension Comparison with experimental data

Channel sections are considered as a combination of Results of experimental studies carried out by Arul
two angle sections connected edge to edge. Channel Jayachandran et al6 to study the behavior of single and
section connected through web experiences shear lag double channel tension members connected by two and
phenomenon when they are subjected to tension. For three bolts were taken for our analysis. From the tests
designing channel sections, IS 800:2007 uses the same it has been observed that all the specimens failed by
design expressions Eqs. (6) and (7) as that of angles. end tearing or end splitting. The failed specimens are
When adopting that equation, modified shear lag distan- shown in Fig. 5 from which the end splitting failure can
ce will be used. The use of single design formula for be seen.
channels as well as angles is critically examined. The
results and discussions are presented in the later part of
the paper.

Channel tension capacity by International codes

AISC-20056
In AISC-2005, the shear lag effect is accounted for by
the reduction in the net section area. They reduce the net Fig. 5 End splitting failure in channel sections5

JOURNAL OF STRUCTURAL ENGINEERING 125

Vol.38, No.2, JUNE-JULY 2011
 By taking tf = tw

The rolled steel channel sections have thickness of flan-

ge usually greater than the thickness of web. The chan-
nels are connected to gusset plate or connecting element
through the web and flanges are unconnected. The de-
sign equation in IS 800 predicts greater strength than
the experiments even after incorporating shear lag coef-
ficient β to reduce the effectiveness of the unconnected
flanges. If the thickness of flange is taken to be equal
to the thickness of web, the β is reduced which results
in lower tension capacity of the sections leading to con-
servative design of the section. The modified β equation
is
Fig. 6 Comparison of channel capacity as per different
codes with experimental capacity6    
w fy bs
β = 1.4 − 0.076 (12)
tw fu Lc
Proposed methods for conservative design of Channel
tension members

The design strength of the tested channel members were ξ factor method
calculated according to IS 800:2007 and other interna-
tional codes. Observations from Fig. 6 clearly indicate The entire connected leg is not stressed to ultimate level
that AISC-2005, AS 4100-2005 and IS 800:2007 over in tension members. Only the regions around the bolts
predict the capacity of channel section. The Indian code will be subjected to maximum stress. Taking into ac-
provision of extending the formula for angles to chan- count this variation of stresses across the leg, a factor
nels is unconservative. One of the possible reasons may ξ is introduced to reduce the capacity of the connected
be that IS 800:2007 equation and other codal equations leg. The proposed equation with ξ factor is given below,
are based on net section behavior whereas the actual
experimental mode of failure is end splitting. Since the  
Anc fu Ago fy
mode of failure of all the tested experimental specimens Tdn = ξ 0.9 +β (13)
was end splitting, three possible methods are proposed γm1 γm0
to alleviate the discrepancy.
The ξ factor given in Eq. (14) reduces the effecti-
Capacity based on end splitting failure
ve area of connected leg. Though the understanding of
For failure of member to take place, the web of the connected leg behavior is more rational, when experi-
channel has to open out at the end, crossing the end di- mental results were used to calculate the value of ξ, it
stance provided and the distance between two fasteners varied over a wide range (from 0.25 to 0.8). The va-
(Lc ). Hence the stressed area is the sum of the above lues are given in the Table 1. A more general value of
mentioned distances, multiplied to the thickness of the ξ can be arrived based on further experiments. Hence
member in that section. As the failure is net section fai- this equation is not taken into consideration for further
lure, the material reaches to the stress level of ultimate. studies.
Hence the capacity of the member will be the product of The proposed Eqs. (11) and (12) are taken for com-
stressed area and ultimate stress of the material. Based parison and results are presented in Fig. 7 where the
on these, the following equation has been formulated. capacity as per current codal provision is also plotted
to check the benefit of the proposed equation. It is seen
tw f u that both the proposed equations are predicting conser-
Tne = (LC + end distance). (11)
γm1 vative values for the tension capacity.

126 JOURNAL OF STRUCTURAL ENGINEERING

Vol.38, No.2, JUNE-JULY 2011
 Comparison with the gross capacity
TABLE 1
ξ VALUES FOR VARIOUS TESTED STEEL Failure of gross section is always preferred over other
CHANNEL SPECIMEN modes of failure because the entire section will be ef-
Specimen Channel Description Failure load ξ fectively utilized. Comparison has been made between
(kN) the capacities calculated using different codes and the
Specimen-1 ISMC075 Single channel 151.6 0.26 gross section capacity in Fig. 8. In addition to these, the
on one side of capacity of channel sections is also determined from the
gusset plate
preliminary design equation given by the IS 800:2007
Specimen-2 ISMC075 Single channel 218.33 0.7
for sizing of members, to check for the ability of the
on one side of
gusset plate equation to predict capacity close to gross section capa-
Specimen-3 ISMC075 Double channel 300.6 0.25 city
on one side of
gusset plate An fu
Tn = α (14)
Specimen-4 ISMC075 Double channel 453.18 0.8 γm1
on one side of
gusset plate The capacities obtained as per various codes are di-
vided by the gross section capacity, to normalize them.
The pattern of the Fig. 8 makes it very clear that AS
4100 - 2005 and AISC 2005 predict capacities greater
than gross section capacity. IS 800:2007 predicts capa-
cities close to gross section capacity for channels with
low depths (up to 150 mm) because the effect of shear
lag is not predominant in such sections. The prelimi-
nary equation taken for initial sizing of members in IS
800: 2007 over predicts the capacity. It cannot be used
for conservative design of channel sections despite its
simplicity. The capacities as per the proposed equations
are also calculated and compared with gross section ca-
Fig. 7 Comparison of channel capacity as per the proposed methods with pacity. The results of the comparison are presented in
experimental capacity6 Fig. 9.

Fig. 8 Comparison of channel capacity as per different codes with gross section capacity

JOURNAL OF STRUCTURAL ENGINEERING 127

Vol.38, No.2, JUNE-JULY 2011
Fig. 9 Comparison of channel capacity as per proposed equations with gross section capacity

When methods proposed in section conservation present work is on hot rolled steel members, for the sa-
design of channel are used, the capacities obtained are ke of a comparative study, the data from the experimen-
less than gross section capacity and hence they are con- tal work carried out by Ling was used to compare the
servative. End splitting which is the actual mode of fai- strength with the different codes and the comparison
lure for channels with one row of bolts is not taken into chart is presented in Figs. 10 and 11.
account in the present code. As this behavior is incor- The methods proposed in the section tension mem-
porated in the end splitting equation proposed in section bers were also taken for comparison with the experi-
4.2, the capacity gets closer to the gross section capaci- mental values. The net section capacity of the cold for-
ty. med sections taken from Ling’s experiments was cal-
Comparison with the experimental data of cold- culated using the formula given in code for hot rolled
formed steel sections. It is inappropriate to extend observation on be-
havior of cold formed sections to hot rolled sections be-
Shear lag effect in the cold formed steel channels was cause in the former the ratio of ultimate and yield stress
extensively studied by Chi Ling Pan8 . Even though the is less compared to the latter. But still, Ling’s (2004)

Fig. 10 Comparison of cold formed steel channel capacity as per different codes with experimental capacity8

128 JOURNAL OF STRUCTURAL ENGINEERING

Vol.38, No.2, JUNE-JULY 2011
Fig. 11 Comparison of cold formed steel channel capacity as per proposed methods with experimental capacity8

work was taken to understand the phenomenon of shear can resolve the problem of over prediction of tensile ca-
lag in case of channels connected to gusset plate by two pacity of channel section.
rows of bolts. With the Figs. 10 and 11 obtained based Though shear lag phenomenon is taken into account
on experiments carried out by Ling8 as basis, it can be in the IS code provisions, the effect of shear lag is not
clearly inferred that predominant in case of channels connected to another
element through web with single row of bolts. In such
1. AISC-2005, AS 4100-2005 and IS 800:2007 pre- case, the sections fail by bearing and end splitting.
dict capacity of channel section unconservatively. i. The end splitting equation proposed in this paper
AISC 2005 and IS 800:2007 codes have taken in- can be used for the design of channel section. This
to account the shear lag phenomenon in the design mode of failure can be avoided, if the minimum
equation, due to which their capacities have been end distance is increased by a factor 1.2. The revi-
found to be closer to the experimental capacity. sed minimum end distance that is to be adopted in
2. Preliminary design equation in IS 800:2007 code is that case will be 2 do.
found to be inconsistent in predicting the strength. ii. The present equation can be modified by taking
thickness of flange equal to the thickness of web
3. The shear lag phenomenon is actually taking place
due to which the value of β and area of unconnec-
in the channels tested by making connection with
ted leg will decrease.
two rows of bolts.
iii. Since the entire part of the connected leg does not
CONCLUSION reach the ultimate stress, an efficiency factor (ξ)
can be adopted to account for the contribution of
connected leg to the tension capacity. The value of
Based on the extensive investigations carried out, it has ξ has to be normalized based on intensive experi-
been found that American (AISC 2005), Australian (AS mental and analytical studies.
4100-2005) and Indian code (IS 800: 2007) over pre-
dicts the tensile capacities of the channel section. The It is recommended that appropriate modifications
use of equation given in IS 800:2007 for angles cannot based on the above may be incorporated in the IS
be extended to channels based on the assumption that 800:2007 for design of channel tension members. The
behavior of channel section is close to the behavior of proposed methods can be taken into consideration for
angle section. This is well-proven from the unconserva- revision of existing code as all of them are proven to
tive design capacities. Equations proposed in this paper give basis for conservative design.

JOURNAL OF STRUCTURAL ENGINEERING 129

Vol.38, No.2, JUNE-JULY 2011
 NOTATION ACKNOWLEDGEMENT

This paper is published with the kind permission of Di-

γmo = partial safety factor for failure in rector, SERC, Chennai 600 113, India.
tension by yielding
γm1 = partial safety factor for failure at REFERENCES
ultimate stress
Φ = resistance factor for tension 1. IS: 800:2007, “Code of practice for general con-
Ago = gross area of outstanding leg struction in steel”, Bureau of Indian Standards,
New Delhi.
2. Kulak G. L. and Wu E. Y, “Shear Lag in Bolted An-
An = net area of the cross section obtained gle Tension Members”, Jl. of Struct. Engg., ASCE,
by deducting from gross area, the Vol. 123, No. 9, 1997, pp 1144–1152.
sectional area of all penetration and
holes 3. Usha, “Analytical Study of Nonlinear Behaviour of
Anc = net area of connected leg Steel Angle Tension Members”, M.S Thesis Re-
Atg , Atn = minimum gross and net area in tension port, Indian Institute of Technology, 2003, Madras.
from the bolt hole to the toe of the 4. Gupta Mohan and Gupta L. M., “Evaluation of
angle end bolt line, perpendicular to Stress Distribution in Bolted Steel Angles under
the line of force, respectively Tension”, Elec. Jl. of Struct. Engg., Vol. 4, 2004,
Avg , Avn = minimum gross and net area in shear pp 17–25.
along bolt line parallel to external
force respectively 5. Arul Jayachandran. S, Marimuthu. V, Prabha .P
bs = shear lag width and Seetharaman. S, “Evaluation of Design Pro-
fu = ultimate stress of the material and visions for Eccentrically Loaded Channel Tensi-
fy = yield stress of the material on Members”, 5th Intl. Conf. on Advances in Steel
kt = correction factor for distribution of Structs., 2007, Singapore.
forces
L = length of the connection in the 6. AISC, “Specification for Structural Steel Buil-
direction of loading dings”, American Inst. of Steel and Cons., Illinois,
Lc = length of the end connection, i.e., 2005, USA.
center to center distance between the
end fasteners measured along the 7. AS 4100, “Steel Structures”, 1998, Standards of
length direction Australia.
Tdb = design tensile strength 8. Chi-Ling Pan, “Prediction of the strength of bol-
t = thickness of the leg ted cold-formed channel sections in tension”, Thin-
w = outstand leg width Walled Structures, Vol. 42, Issue 8, August 2004,
x̄ = connection eccentricity pp 1177–1198.

130 JOURNAL OF STRUCTURAL ENGINEERING

Vol.38, No.2, JUNE-JULY 2011

View publication stats