ARTICLE IN PRESS

0263-8231/$ - se

doi:10.1016/j.tw

�CorrespondE-mail addr

Thin-Walled Structures 45 (2007) 552–562

www.elsevier.com/locate/tws

Built-up battened columns under lateral cyclic loading

Dipti R. Sahoo, Durgesh C. Rai�

Department of Civil Engineering, Indian Institute of Technology, Kanpur, 208016 UP, India

Received 20 July 2006; accepted 21 February 2007

Available online 31 May 2007

Abstract

Built-up members with battens designed for typical 2–2.5% of axial load may not behave satisfactorily in the presence of lateral seismic

loads. Analytical evaluation of double-channel battened cantilever members designed as per the current practice, and subjected to

constant axial compressive load and gradually increasing lateral load showed that the members failed to reach their expected flexural

capacity due to lateral instability. The design of members was modified by changing the configuration of battens in the expected plastic-

hinge region, i.e., reducing the spacing of battens in end panel by half, and designing battens for a shear demand due to moment capacity

of section. The members with battens designed for moment capacity could able to reach the expected flexural strength. Five half-scale test

specimens of battened members designed as per the current practice and improved design method were subjected to axial load and

gradually increasing cyclic load. The specimens designed as per improved design method showed excellent performance in terms of lateral

strength, lateral stiffness, moment rotation characteristics and energy dissipation capacity.

r 2007 Elsevier Ltd. All rights reserved.

Keywords: Battened; Column; Seismic; Buckling

1. Introduction

A built-up battened column is a kind of compressionmember consisting of two identical longitudinal elementsslightly separated and connected to each other at only afew places along their length by means of battens. Thesemembers are frequently used as light compression mem-bers, such as struts in truss moment frames and as columnsin light steel structures. Double-channel sections are oftenused as built-up battened columns.

Built-up columns are significantly weaker against sheardeflections as compared to solid columns. The reduction inaxial buckling strength of these columns due to the sheardeflection depends on the configurations and the structuralconnections of the connectors. Many researchers [1–3] havedeveloped analytical criterion considering the effects ofaxial load, shear deformation and connection rigidity. Inaddition to the shearing effect, the axial buckling strengthmay also be reduced due to compound buckling, where the

e front matter r 2007 Elsevier Ltd. All rights reserved.

s.2007.02.016

ing author. Tel.: +91 512 259 7717; fax: +91 512 259 7794.

ess: dcrai@iitk.ac.in (D.C. Rai).

localized buckling of components between the battensinteracts with the global buckling mode [4].End plates in battened members distribute the applied

forces or moments to the component elements and maycontribute significantly to the axial buckling strength. Anearlier study of a special battened column, in which thebattens are attached to the longitudinal elements by hingedconnections, revealed that the axial buckling strength ofmember without end plates is no greater than the sum ofthe critical loads of individual elements [5]. In this case, thebattens do not transmit shear between the componentelements. However, the buckling load with end plates is thelower bound buckling strength of battened member.During seismic overloads, the built-up columns are

subjected to excessive compressive strains due to lateralloads in addition to axial compressive load. Under largelateral deformations, plastic hinges are formed at criticalsections of the member, where the maximum moment isexpected. Ductility and energy dissipation capacity ofmembers greatly depend on the performance of theseplastic hinges. The smaller width-to-thickness ratio ofmembers could reduce the severity of local buckling,

ARTICLE IN PRESSD.R. Sahoo, D.C. Rai / Thin-Walled Structures 45 (2007) 552–562 553

leading to an increase in ductility and energy dissipationcapacity [6,7].

The battens of built-up columns are designed for a sheardemand of 2.0–2.5% of total axial load on the column asper the current practice [8,9]. These columns may notbehave satisfactorily when subjected to lateral loads due toearthquake events in addition to axial compressive load. Inthe present study, a nonlinear finite element (FE) analysis,using ABAQUS [10] was carried out to study the effect ofconfigurations of battens on behavior of battened memberssubjected to constant axial compressive load and graduallyincreasing lateral load. In addition, the behavior of half-scale double-channel built-up battened specimens subjectedto constant axial load and gradually increasing cycliclateral load was investigated experimentally. The objectivewas to develop a design of ductile built-up battened beamcolumn, which can reach its expected flexural strengthwithout any instability.

2. Analytical evaluation

Battens of a built-up member are generally subjected toshear force and bending moment due to both axial andlateral load. For design purposes, it is assumed that thepoints of contra-flexure of the main components and

Shear force

Vb =

Max. she

permissible

2 dt

3 Vb =

Calculate

for as

thickn

V

2N

V

2N

V

2N

V

2N

Vb

S

S

C

C

Axial load

Shear force (V)

Channel

sections

Batten

Point of

contra-flexure

She

Desi

Wh

C i

cen

Fig. 1. Flow-chart for design of

battens occur at their mid-points. The configuration ofbattens depends on their center-to-center spacing, which isgoverned by the criteria that the slenderness ratio of themain component over the distance between the battensshall not be greater than 50 or 0.7 times slenderness ratio ofthe battened member as a whole about the axis parallel tothe battens [8]. A flow chart for the design of battens of abuilt-up beam column is given in Fig. 1.

2.1. FE study

Built-up battened cantilever members subjected toconstant axial compressive load and gradually increasedlateral load were studied analytically using a nonlinear FEpackage, ABAQUS [10]. The members consisted of twohot-rolled Indian standard channel sections (ISMC 200) asthe main components and mild steel plates of 10mmthickness as battens. The battened member with loadingsand section properties of main components are shown inFig. 2.Four different designs, namely, current practice, design

option-I, -II and -III were considered in the analysis ofbuilt-up beam columns of equal slenderness ratio. In thecurrent design practice, battens were designed for a sheardemand of 2.0% of axial compressive load and spacing of

in batten,

NC

VS

Bending moment in batten,

VSMb

2N=

ar stress =

shear stress

0.45 fy

depth ‘d ’

sumed

ess‘t ’

Design depth of batten is the

maximum of above two values

Max. bending stress =

permissible bending stress

td20.66 fy

6Mb =

Moment

Designar

gn

Calculate depth ‘d ’

for assumed

thickness‘t ’

ere, N is the number of batten planes;

s spacing of main components; S is the

ter-to-center spacing of battens

battens of built-up members.

ARTICLE IN PRESSD.R. Sahoo, D.C. Rai / Thin-Walled Structures 45 (2007) 552–562554

the battens was uniform for all panels of the member. Thespacing of battens in the panel near fixed end was reducedby half and an additional batten of the same configurationwas placed at the mid-way of the panel in design option-I.In case of design option-II, all the battens were designedfor a shear demand for the reduced plastic moment

Fixed End

Channel Sections

Batten Plates

Lateral Load

Constant

Axial Load

All dimensions are in mm

200

50

200

7575

75

a = 28.21 cm2

Ixx = 1819.3 cm4

Iyy = 140.4cm4

200

11.4

6.1

21.7

x

y

Cross-section

Section properties of ISMC 200

Fig. 2. Built-up battened beam column considered in this study.

Table 1

Properties of built-up battened finite element models

Length, l (mm) 2500

Effective length, leff (mm) 5500

Minimum radius of gyration, rmin (mm) 80

Slenderness ratio, l 68.5

Yield stress, fy (MPa) 250

Modulus of elasticity, E (GPa) 200

Area, A (mm2) 5642

Plastic modulus, Zp (cm3) 442

Axial yield load, Py (kN) 1410

Euler load, Pe (kN) 2374

Axial buckling load, Pcr (kN) 1065

Plastic moment capacity, Mp (kN-m) 110.4

Table 2

Details of configurations of battens of finite element models

P/Py Mpc (kN-m) Current practice

Width of batten (mm) c/c spacing of batte

0.15 111 53 490

0.30 91 105 490

0.45 71 157 490

0.60 57 209 490

capacity (Mpc) of main components. However, only endbattens of the battened member were designed for a sheardemand due to the reduced plastic moment capacity (Mpc)of main components and other battens were designed asper the current practice in design option-III. The propertiesof double-channel built-up battened members are given inTable 1 and the details of configurations of battens ofmembers designed as per the current practice and designoption-II are given in Table 2.Four-noded doubly curved shell (S4R) elements with

6-degrees of freedom at each node were used to model maincomponents and battens. The verification of mesh wascarried out by the element failure criteria, i.e., the facecorner angle less than 30

0

. The material property wasconsidered as elasto-plastic (Young’s Modulus ¼ 200GPa,Poisson ratio ¼ 0.25, and yield stress ¼ 250MPa) andgeometric non-linearity (i.e., large deformation) wasincluded in the analysis. The FE models with threedifferent configurations of battens analyzed in this studyare shown in Fig. 3.

2.2. FE results

Nonlinear static collapse analysis was carried out for thebuilt-up battened FE models preloaded with axial com-pressive load. Four levels of axial compressive loads,namely, 15%, 30%, 45% and 60% of axial yield load, wereconsidered. Table 3 compares the peak lateral loads carriedby the battened FE models at different levels of axial load.The models designed as per the current practice carriedlesser lateral loads at each axial load level as compared tomodels designed as per the three improved design methods.The lateral load carrying capacity of the model designed asper design option-I was found to be slightly higher than themodels designed as per the current practice. The increase inlateral strength of models designed as per design option-IIand -III was about 30–50% of that designed as per thecurrent practice. However, the models designed as perdesign option-II and option-III carried nearly same peaklateral loads for all levels of axial load.

2.3. Comparison

The plastic moment capacity (Mp) of a member isreduced in the presence of axial compressive load (P). The

Design option-II

ns (mm) Width of batten (mm) Clear spacing of battens (mm)

545 420

446 355

350 290

253 155

ARTICLE IN PRESS

Fig. 3. Finite element models of cantilever battened member.

Table 3

Lateral strength of finite element models at different axial load levels

P/Py Peak lateral load (kN)

Current practice Improved design

Option-I Option-II Option-III

0.15 38.0 38.8 50.4 49.3

0.30 28.0 28.9 39.2 38.8

0.45 18.3 21.1 26.9 26.9

0.60 10.2 15.9 18.6 17.7

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1 1.2

Mpc /Mp

P/P

y

Equation (1)

Analytical

xx

y

y

Fig. 4. Comparison of axial load–moment capacity interaction.

0

0.2

0.4

0.6

0.8

1

1.2

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Mpc /Mp

P/P

y

Design curveDesign option-IDesign option-IIDesign with reduced spacingCurrent practice

Fig. 5. Comparison of moment capacity of finite element models.

D.R. Sahoo, D.C. Rai / Thin-Walled Structures 45 (2007) 552–562 555

reduced moment capacity (Mpc) for the value of axial load(P) greater than or equal to 15% of axial yield load (Py)can be given by the following expression [8,9]:

Mpc ¼ 1:18Mp 1:0�P

Py

� �. (1)

Eq. (1) is primarily intended for determining the reducedmoment capacity of beam column of I-section bendingabout its major axis. However, this relation can beapplicable to the double-channel built-up beam column.The axial load-bending moment interaction for the built-upbattened beam column is derived analytically assumingthat the components can reach their plastic momentcapacity without any instability. The analytical resultmatched very well with the interaction curve given byEq. (1) as shown in Fig. 4.

The flexural strength of the FE models at different levelsof axial load was compared with the design momentcapacity given by Eq. (1) as shown in Fig. 5. FE modelswith the battens designed as per current practice failed to

ARTICLE IN PRESSD.R. Sahoo, D.C. Rai / Thin-Walled Structures 45 (2007) 552–562556

reach their expected flexural capacity of components.However, the models designed as per design option-IIand option-III showed higher plastic moment-carryingcapacity and reached the design moment capacity of beamcolumn for axial load of 15% and 30% of yield load. Asthe axial load in the models increased beyond 40% of yieldload, the moment capacity was found to be less than thedesign moment capacity. The reason may be due to the factthat as the axial load level in the member increased, themoment capacity of the member decreased. Hence, theshear demand (Vb) on the battens was reduced requiringsmaller depth of battens. Since, the designers generallyrestrict the axial load in the member to the 40% of yieldload, designing the battens for the axial load beyond thislimit is of lesser importance. The model designed as perdesign option-I could not reach its expected flexuralstrength; however, it carried the higher lateral load ascompared with the model designed as per the currentpractice.

3. Experimental evaluation

Half-scale test specimens of double-channel built-upbattened cantilever members were investigated experimen-tally to study the behavior under constant axial compres-sive load of 32% of axial yield load and gradually increasedcyclic lateral load. The main parameters of interests werethe spacing and the configuration of battens in the expectedplastic hinge region of built-up battened members.

3.1. Test specimens

Five test specimens with three different configurationsof battens consisted of ISMC 200 were used as main

Batten Plates

(140x120x10mm)

Channel Section

(ISMC 200)

Rib-plates

End plate

(16mm Thick)

Gusset Plate

(150x150x10mm)

Base Plate

(400x400x16mm)100

120

120

120

120

1320

370

370

400

200

DCC

7575

DC

20

4

75

Fig. 6. Details of t

components and mild steel plates as battens. The minimumspecified yield stress of channel sections and mild steelplates was 250MPa. The overall length of all specimenswas 1.32m and the cyclic lateral loading was applied at adistance of 1.2m from the fixed end. All specimens were ofequal slenderness ratio.The specimens were designated as DC1C, DC1MR,

DC2MR, DC1MB and DC2MB. DC represents double-channel battened built-up specimens and numeric 1 or 2stands for number of specimens tested experimentally. C,MR, and MB represent the design of battens as per thecurrent practice, modified with reduced spacing andmodified with ‘box’ configuration, respectively (Fig. 6).The battens of specimen DC1C were placed at uniformspacing of 490mm center-to-center; where as an additionalbatten of the same configuration was placed at mid-way ofthe panel near fixed end of specimen DCMR. SpecimenDCMB had a very wider batten near the fixed end so that a‘box’ configuration was formed in the expected plastichinge region. The details of configuration and spacing ofbattens of specimens are given in Table 4.

3.2. Test setup

Test setup shown in Fig. 7 consisted of a reaction frame,two double-acting servo-hydraulic actuators, and a reac-tion block. The axial compressive load was applied bymeans an actuator (force rating ¼7500 kN and strokelength ¼7125mm), whereas the cyclic lateral load wasapplied by means of another actuator (force ra-ting ¼7250 kN and stroke length ¼7125mm). Boththe actuators were connected at the free end of thespecimen, whereas the fixed end of specimen was connectedto a reaction block. Four linear variable differential

MR

1320

0

00

100

120

120

120

120

370

120

125

125

75

DCMB

1320

200

100

120

120

120

370

365

125

400

7575

est specimens.

ARTICLE IN PRESS

Table 4

Configuration of battens of test specimens

Specimen Batten configuration (mm) Number of battens Panel length (mm) Comments

DC1C 140� 120� 10 06 370 Current practice

DC1MR 140� 120� 10 08 370 Reduced spacing of battens

DC2MR 125

DC1MB 140� 120� 10 04 370 ‘‘Box’’ configuration

DC2MB 140� 365� 10 02 125

Strong Floor

Reaction Block

Test Specimen500 kN Actuator

Reaction Frame

250 kN Actuator

Reaction Beam

Fig. 7. Test setup.

Number of Cycles

12 16 20 24 28-60

-45

-30

-15

Dis

pla

cem

ent Level (m

m)

32

0

15

30

45

60

0 4 8-5.0

-3.8

-2.5

-1.3

0.0

1.3

2.5

5.0

3.8

Drift L

evel (%

)

Fig. 8. Displacement history.

D.R. Sahoo, D.C. Rai / Thin-Walled Structures 45 (2007) 552–562 557

transducers were used to measure of the lateral displace-ment of specimens. The state of strain at the mid-points ofwebs and flanges of channel sections near fixed end andmid-points of end battens were measured by means ofstrain gauges.

3.3. Displacement history

A simple, multiple-step displacement history (Fig. 8)consisting of symmetric cycles of increasing amplitude wasused for the purpose of assessment of seismic performanceof battened specimens [11]. The displacement (or driftratio) levels in the displacement history were 73mm(0.25%), 76mm (0.5%), 79mm (0.75%), 712mm (1%),715mm (1.25%), 718mm (1.5%), 721mm (1.75%),730mm (2.5%), 740mm (3.3%) and 760mm (5%). Thedisplacement cycle at each excursion level was repeated forthree times to get the repetitive behavior of the specimensat a particular level.

4. Experimental results

The cantilever test specimens were subjected to bothaxial compressive load of 450 kN (32% of axial yield load)and cyclic lateral load at free end. Experimental resultspresented in the following sections illustrate how lateralload–lateral displacement response, lateral stiffness, mo-

ment-curvature response and energy dissipation capacity ofbuilt-up battened cantilever specimens of equal slendernessratio was influenced by different configurations of battensnear the fixed end.

4.1. Lateral load–lateral displacement response

Fig. 9 shows the lateral load–lateral displacementresponse of the test specimens. The elastic behavior of

ARTICLE IN PRESS

Specimen DC1C

Specimen DC1MR Specimen DC2MR

Specimen DC1MB Specimen DC2MB

-120

-90

-60

-30

0

30

60

90

120

-4 -3 -2 -1 0

La

tera

l L

oa

d (

kN

)

Drift (%)

YLB

GLB

120

-120

-90

-60

-30

0

30

60

90

-4 -3 -2 -1 1

1

2 3 4 1 2 3 4

1 2 3 4

0

Drift (%)

GLB

LB

Y

La

tera

l L

oa

d (

kN

)

-120

-90

-60

-30

0

30

60

90

120

-4 -3 -2 -1 0

Drift (%)

La

tera

l L

oa

d (

kN

)

GLB

Y LB

-120

-90

-60

-30

0

30

60

90

120

-4 -3 -2 -1 0 2 3 4

Drift (%)

La

tera

l L

oa

d (

kN

)

GLB

Y

LB

Drift (%)

-4 -3 -2 -1

-120

-90

-60

-30

0

30

60

90

0

La

tera

l L

oa

d (

kN

)

120

LBY

GLB

1 2 3 4

Fig. 9. Lateral load–lateral displacement response of specimens.

D.R. Sahoo, D.C. Rai / Thin-Walled Structures 45 (2007) 552–562558

ARTICLE IN PRESS

80

100

120

)

D.R. Sahoo, D.C. Rai / Thin-Walled Structures 45 (2007) 552–562 559

the specimens was observed up to a drift ratio of 1.25%,which is approximately linear in the hysteretic curve.Yielding (Y) may be defined as an event when significantdeparture from linear behavior of specimen was observedin the load–displacement curve. Yielding of the specimenswas observed in the form of cracks in the white wash at adrift ratio of 1.5%, which corresponds to a lateral load of48.5 kN. Local buckling of components of the specimenDC1C was observed in the panel near fixed end at a driftratio of 1.8% and the global lateral buckling of thespecimen was observed at a drift ratio of 3.1% with closingof the gap between the main components in the panel nearfixed end. The maximum lateral load carried by thespecimen DC1C was 68.5 kN, which was about 70% ofits expected flexural strength. The maximum lateral loadscarried by specimen DC1MR and DC2MR were 87.6 kNand 71.9 kN at drift ratio of 3.2% and 2.5%, respectively.Local buckling of components of specimens DC1MB andDC2MB was observed in the region just after the ‘box’configuration at a drift ratio of 3.3%. The maximumlateral loads carried by the specimens were 99.0 kN and94.4 kN, respectively, at drift ratio of 3.4% and 3.2%,respectively. The details peak lateral loads and peak lateraldisplacements of test specimens are given in Table 5.

The hysteretic curves showed the higher lateral loadcarried by the specimens in negative lateral displacementexcursion level. The reason may be due to additional workdone by the specimens against the force of gravity, whenthe specimen was displaced in upward direction. Therefore,the recorded lateral load value was higher in one directionthan the other direction. Hence, the lateral strengths of thetest specimens were obtained considering the average oflateral load in both directions so that the effect of gravityload was nullified.

4.2. Lateral stiffness

The lateral stiffness of specimen was calculated as theratio of lateral load required to produce unit lateraldisplacement. The initial stiffness of specimen DC1C wasfound to be 4.5 kN/mm, whereas that of specimen DC2MBwas found to be 6.5 kN/mm. The increase in initial stiffnessof DC1MR as compared with specimen DC1C was 20%.Similarly, the increase in initial stiffness of specimenDC2MB was 32%. As the spacing of battens in the expected

Table 5

Peak lateral loads and lateral displacements of specimens

Specimen Displacement (mm) Drift (%) Lateral load (kN)

+ve �ve +ve �ve +ve �ve Average

DC1C 38.4 35.2 3.2 2.9 47.8 79.2 68.5

DC1MR 42.9 34.3 3.6 2.8 77.8 97.3 87.6

DC2MR 30.9 28.6 2.6 2.4 60.4 83.4 71.9

DC1MB 44.7 37.6 3.7 3.1 82.6 115.4 99.0

DC2MB 42.6 32.8 3.6 2.8 83.0 105.7 94.4

plastic hinge region decreased, the lateral stiffness ofbattened built-up beam columns increased and specimenswith ‘box’ configuration in the expected plastic hinge regionshowed higher lateral stiffness. The lateral stiffness ofspecimens was decreased with increase of cyclic excursionlevel and the stiffness at the failure was about 40% of thecorresponding initial lateral stiff ness of specimens.

4.3. Moment–curvature response

The curvature at a section of specimen was calculatedfrom the strains recorded by the strain gauges. The strainscorresponding to the peak lateral displacement of anexcursion level were used to determine the curvature of thespecimen. The ideal moment–curvature response of dou-ble-channel sections was obtained assuming the material tobe elasto-plastic with a yield stress of 250MPa. Fig. 10shows the actual moment–curvature response of specimensalong with the ideal moment–curvature response ofdouble-channel section. Specimen DC1C showed smallcurvature of 1.5� 10

�5

with small moment capacity. Themaximum curvature-attained specimens DC1MR andDC2MR was about 3.3� 10

�5

, which was about twice ofspecimen DC1C. However, the specimens failed to reach itsexpected plastic moment-carrying capacity. In contrast,specimens DC1MB and DC2MB reached the curvature of5.3� 10

�5

, which was about 3.5 times of specimen DC1C.In addition, the specimens DC1MB and DC2MB reachedexpected plastic moment capacity of the sections. The slopeof the ideal moment–curvature curve was higher than theslope of moment–curvature response of test specimens.This was due to the effect of shear flexibility of battens,which was not considered for ideal moment–curvaturecurve.

4.4. Energy dissipation capacity

Fig. 11 shows the cumulative energy dissipated by thespecimens at different cyclic excursion levels. The energy

0

20

40

60

0.0 0.1 0.2 0. 3 0.4 0.5 0.6

Curvature (x10-6)

Mom

ent

(kN

-m

Ideal

DC1C

DC1MR

DC2MR

DC1MB

DC2MB

Fig. 10. Comparison of moment–curvature response of specimens.

ARTICLE IN PRESS

0

0.2

0.4

0.6

0.8

1.0

0 5 10 15 20 25 30 35 40 45

Excursion Level (mm)

Norm

aliz

ed E

nerg

y

DC1C

DC1MR

DC1MB

DC2MR

DC2MB

Fig. 11. Energy dissipation capacity of specimens.

D.R. Sahoo, D.C. Rai / Thin-Walled Structures 45 (2007) 552–562560

dissipated at an excursion level was normalized to themaximum energy dissipated due to a maximum lateraldisplacement of 40mm and flexural strength of 100 kN.The energy dissipated in an excursion level was calculatedby taking average of energy dissipated in three cycles of thedisplacement level. Specimen DC1C showed the leastenergy dissipation capacity in each cyclic excursion levelas compared to other specimens and could reach about65% of the ideal energy dissipation. The energy dissipatedby the specimens DC1MR and DC2MR was more thanspecimen DC1C and about 80% of the ideal energydissipation. However, specimens DC1MB and DC2MBshowed higher energy dissipation capacity as comparedwith other specimens.

Fig. 12. Comparison of lateral load–displacement response.

5. Comparison of experimental and FE results

The experimental results are compared with the FEanalysis results. Three FE models, namely, FEM1, FEM2and FEM3 similar to the specimens DCC, DCMR andDCMB, respectively, were analyzed. Four-noded shellelements were used to model main components as well asbattens of built-up members. Axial compressive load of450 kN (same value as applied in experiment) was appliedat the free end, distributed equally at eight pointssymmetric to the cross-section that would eliminate theeffect of eccentric loading on the member. Static analysiswas carried out for the axial compressive load and the stateof the models at the end of analysis was considered asinitial conditions for the gradually increased lateraldisplacement. Nonlinear static collapse analysis wascarried out in the present study to predict maximumload-carrying capacity as well as the post-bucklingbehavior of battened members.

5.1. Load–displacement response

Fig. 12(a) compares the experimental load–displacementresponse of specimens designed as per current practice with

the FE study. The maximum lateral load carried by thebattened member was about 83.0 kN in the FE analysis,which was about 20% higher than the experimental value.Both experimental and FE studies showed exactly same

ARTICLE IN PRESSD.R. Sahoo, D.C. Rai / Thin-Walled Structures 45 (2007) 552–562 561

lateral load and lateral stiffness for small lateral displace-ments. For the higher lateral displacements, the FEanalysis predicted the higher lateral loads than theexperiment. This may be due to the presence of residualstresses in the components developed by welding, absenceof absolute fixity of the fixed end and unintentionaleccentricity of loading in the test specimen, which wasnot considered in the FE analysis.

The built-up battened members with reduced battenspacing in the panel near fixed end carried a lateral load of91.5 kN in FE analysis, which was about 12% more thanthe both experimental value. Fig. 12(b) shows thecomparison of experimental and FE load–displacementresponse of the battened members having reduced battenspacing. The load–displacement bahaviour of both the testspecimens was comparable with the analytical study.

Fig. 12(c) shows the comparison of experimentalload–displacement response of battened members withthe box configuration in the panel near fixed with the FEstudy. The average maximum load carried by the speci-mens was about 96.7 kN as compared with 104 kN in theFE study. Both the experimental and FE load–displace-ment results matched very well. The difference in the lateral

Fig. 13. Comparison of mode of

loads carried by the specimens in the experimental study aswell as the FE study was about 7%.

5.2. Mode of failure

The mode of failure of battened member as observed inexperimental and FE studies is shown in Fig. 13. The localbuckling in form of depression in the webs and bulging offlanges of battened member was observed in the panel nearthe fixed end. Thus, the initial gap between the maincomponents was reduced as shown in Fig. 13(i). Thelocal buckling in form of outward bulging of websof main components of battened members with reducedspacing of battens was observed in the panel betweenthe batten near fixed end and the additional batten of thespecimen. However, the initial gap between the maincomponents of the member was maintained after thefailure. The plastic hinge of battened member with ‘box’configuration shifted away from the fixed end. The localinstability of components was not observed within the‘box’ region. The mode of failure of battened membersobserved in the experimental study matched very well withthe FE study.

failure of battened members.

ARTICLE IN PRESSD.R. Sahoo, D.C. Rai / Thin-Walled Structures 45 (2007) 552–562562

6. Conclusion

The following conclusions can be drawn from thepresent study:

(i)

The battened double-channel beam columns designedas per current code provisions could not reach theirexpected flexural strength due to local buckling ofcomponents. The uniform spacing of battens ofbattened members resulted in low plastic rotation,energy dissipation, and ductility. The built-up memberswith reduced spacing of battens in the expected plastichinge region performed better in terms of flexuralstrength, lateral stiffness, plastic rotation and energydissipation. However, the members with ‘box’ config-uration in the expected plastic hinge region, using verywide battens, was able to reach their expected flexuralstrength and exhibited excellent performance in termsof plastic rotation, and energy dissipation capacity.

(ii)

The design of battens should be based on the moment-shear interaction of the member. However, designing allthe battens for the shear demand due to moment capacityrequires higher depth of battens and the battens remainineffective as the distance from the fixed end increases.Hence, designing only the battens near the fixed for ashear demand due to moment capacity of member andother battens for a shear demand of 2% of axial load as

per current practice is sufficient to reach the expectedplastic moment-carrying capacity of the member.

References

[1] Bleich F. Buckling strength of metal structures. 2nd ed. New York:

McGraw-Hill Book Company; 1952.

[2] Aslani F, Goel SC. Analytical criterion for buckling strength of built-

up compression members. Eng J AISC 1991;28(4):159–68.

[3] Lin FJ, Glauser EC, Johnston BG. Behavior of laced and battened

structural member. J Struct Div, Proc Am Soc Civil Eng 1970;96(7):

1377–401.

[4] Duan L, Reno M, Uang C. Effect of compound buckling on

compression strength of built-up members. Eng J AISC 2002;39(1):30–7.

[5] Galambos TV. Guide to stability design criteria for metal structures.

5th ed. New York: Wiley; 1998.

[6] Aslani F, Goel SC. Stitch spacing and local buckling in seismic-

resistant double-angle braces. J Struct Eng ASCE 1991;117(8):2442–63.

[7] Astaneh-Asl A, Goel SC, Hanson RD. Cyclic out-of-plane buckling

of double angle bracing. J Struct Eng 1985;111(5):1135–53.

[8] BIS-800. Code of practice for general construction in steel.

New Delhi: Bureau of Indian Standards; 1984.

[9] AISC. Load and resistance factor design specification for structural

steel buildings. Chicago, IL: American Institute of Steel Construc-

tion; 2001.

[10] Hibbit HD, Karlsson BI, Sorensen P. ABAQUS/Standard User’s

Manual (v6.4-1). Pawtucket, RI: Hibbit, Karlsson & Sorensen, Inc.;

2003.

[11] ATC-24. Guidelines for cyclic seismic testing of components of steel

structures. Redwood City, CA: Applied Technology Council; 1992.