Journal of Constructional Steel Research 62 (2006) 739–746www.elsevier.com/locate/jcsr

dy

pplicationmembers.omeys channelthe need tolines. Thens for the

ic tests andnnectors ofRBS). It is

RHS beam-to-column connection with web opening—experimental stuand finite element modelling

S.R. Satish Kumar∗, D.V. Prasada Rao

Department of Civil Engineering, IIT Madras, Chennai 600 036, India

Received 10 August 2005; accepted 30 November 2005

Abstract

Rectangular Hollow Sections (RHS) have superior structural performance compared to conventional steel sections. However, the aof RHS in structural steel framework is limited because suitable connection configurations have not been developed between suchAlso adequate information on the moment–rotation characteristics of connections between RHS members is not available for design. To overcthese problems, a new and efficient connection is proposed which is easy to fabricate and convenient for erection. The connection emploconnectors welded to the column flange and bolted to the beam to transfer beam flange forces into the column webs thereby avoidingprovide internal diaphragms in the column. An opening in the web facilitates the installation of bolts and can be used to pass servicebolts are loaded in shear, so as to obtain improved performance of the connection under cyclic loading. By choosing suitable dimensiochannel connectors, the strength and stiffness of the connection can be varied. The behaviour of the connection is evaluated by cyclnon-linear finite element analysis. Test results are presented in the form of hysteretic curves and failure modes. In the case of channel cohigh strength, failure occurs at the beam net section away from the face of the column, similar to beams with Reduced Beam Sections (shown that the connection has adequate ductility and hystereticenergy dissipation capacity. The moment–rotation characteristics of the connectioncan be expressed in terms of the three-parameter power model, for semi-rigid frame analysis.c© 2006 Published by Elsevier Ltd

Keywords: Rectangular hollow sections; Semi-rigid connections; Moment–rotation curves; Web opening; Cyclic tests; Non-linear analysis

in

iDlogyf

tn

dhio

.

seed

nshasRHSionideesf

dercessidems,

to

ldedighors

1. Introduction

Rectangular hollow sections (RHS) are increasingly beused in steel framed structures due totheir higher efficiency inresisting compression, bending, and torsion in comparison wconventional sections and also for architectural reasons.to the absence of outstands they are less susceptible tobuckling and so exhibit higher ductility and hysteretic enerdissipation capacities. This makes them particularly suitablemoment resisting frames designed to resist seismic loads. Theyalso possess high strength to weight ratios resulting in lighstructures. RHS columns filled with concrete, popularly knowas concrete filled tubes (CFTs), have also been widely usehigh rise buildings. The frames with CFTs are light in weigand have high ductility and stiffness. However, the applicat

∗ Corresponding author. Tel.: +91 44 2257 4287; fax: +91 44 2257 6287E-mail addresses: kim@iitm.ac.in(S.R. Satish Kumar),

dvpr@rediffmail.com(D.V. Prasada Rao).

0143-974X/$ - see front matterc© 2006 Published by Elsevier Ltddoi:10.1016/j.jcsr.2005.11.016

g

thuecal

or

er

intn

of RHS in structural steel frameworks is limited becausuitable connection configurations have not been developbetween such members.

Although sufficient information is available on connectiobetween I-beams and I-columns, very little research workbeen done on connections between RHS beams andcolumns. The current practice is to go for full penetratwelds and for larger sections, diaphragms are inserted insthe column at beam flange levels. However, this involvconsiderable fabrication and cost and has the disadvantage othe field welds introducing potential zones for fracture unseismic loading. The use of bolts gives considerable toleranin fabrication but conventional bolts require access on the into facilitate tightening of nuts. To overcome these problea new and efficient connection is proposed which is easyfabricate and convenient for erection.

In the proposed connection, channel connectors are weto the column and connected to the beam flanges by hstrength friction grip (HSFG) bolts. The channel connect

740 S.R. Satish Kumar, D.V. Prasada Rao / Journal of Constructional Steel Research 62 (2006) 739–746

mnnd

shedetedt isgth

tions

f

heoutonucss

nsf I-sinattioviomn

idthnendith

eyth

iveeoulfero

als

nsonsithop

ybectiornaluredsngheywa

nce.o

nsendeenk ofTheundr toess,

fionedor

d an

ess,nelngesingseened andctediredhe

andttainthennel

ithr

thee

transfer the beam flange forcesinto the column webs therebyavoiding the need to provide internal diaphragms in the coluAn opening in the web facilitates the installation of bolts acan be used to pass service lines. The bolts are loaded insoas to obtain improved performance of the connection uncyclic loading. The behaviour of the connection is evaluaby cyclic tests and non-linear finite element analysis. Ishown that the connection has adequate stiffness, strenductility and energy dissipation capacity. The moment–rotacharacteristics of the connection can be expressed in termthe three-parameter power model [1,2]. A parametric study oconnection behaviour is presented in a companion paper [3] tofacilitate the use of the proposed connection.

2. Review of literature

In view of the several advantages of RHS, some researcinitiated the use of RHS in columns only and carriedthe experimental and analytical investigations on connectibetween I-beam and box-column sections. However, sinformation is still limited and in particular does not addrethe problem of connecting two RHS members.

White and Fang [4] carried out experimental investigatioon the behaviour and performance of different types obeam to box-column top and seat angle connections, ua combination of welding and bolting. They observed thalthough web cleat connection suffered significant deformaof column flange, the top and seat angle connection behadid not depend on the width-to-thickness ratio of the coluflange.

Ting et al. [5] presented the results of finite elementanalysis of I-beam to box-column connections with beam wless than the column width. The connections were stiffeexternally with triangular plate stiffeners, angle stiffeners aT-stiffeners. The results indicated that the connections wexternal T-stiffeners were the most efficient form as thredistribute the stresses more effectively and enhancestiffness of theconnection. They concluded that for effecttransfer of stresses to the side walls of the column, thcombined width of beam flange and stiffeners at the joint shbe equal to the width of the column to allow the transof stress to the column side walls. The optimum lengththe stiffener which results in minimum stress levels wasderived.

Shanmugam et al. [6] carried out experimental investigatioon the cyclic behaviour of I-beam to box-column connectistiffened internally using continuity plates and externally wT- or angle stiffeners. The load–displacement hysteretic lowere found to be stable and connections exhibited ductilitvalues ranging from 5.1 to 10.7, which were considered toadequate. They concluded that performance of the connewith T-stiffeners was better than the connections with intecontinuity plates and external angle stiffeners. The faimodes of the tested specimens were fracture of the weljoints, excessive yielding of stiffeners followed by buckliof either the column web or beam flange at the joint. Talso pointed out that plate slenderness of column webs

.

ar,r

,

of

rs

sh

g

nur

d

e

d

fo

s

nl

at

s

Fig. 1. Proposed RHS beam-to-column connection.

an important parameter affecting connection performaAnother study [7] carried out on interior connection alsyielded similar results.

Korol et al. [8] carried out experimental investigatioon the performance and failure modes of extendedplate, high-strength, blind-bolted moment connections betwI-beams and box-columns, to overcome the problem of lacaccess inside the hollow column section to tighten the nut.specimens were tested under monotonic loading. It was fothat the behaviour of the proposed connection was similathe connection using A325 bolts and possesses good stiffnmoment capacity and ductility.

Wheeler et al. [9,10] reported test results and analysis oconnections between RHS using an end plate and pretensbolts. The connections failed by tensile bolt fractureexcessive deformation of the end plate.

3. Proposed connection and connection performance

The proposed connection between an RHS Beam anRHS Column is as shown inFig. 1. The essential features of theconnection are two channel connectors, of uniform thicknwelded to the column in the fabrication shop. These chanconnectors transmit the stress resultants of the beam flato the column webs by shear-lag action. Rectangular openwere provided in the beam webs so as to allow bolting betwthe beam flange and the channel connector. Adequate tolerancand end clearance can be provided between the beam enthe column face to facilitate erection. The beam can be erebetween two such connections by lowering it to the requlevel and rotating in plan, to get the required alignment. Tmain considerations in deciding the size of the openingthe channel connectors are that the connection should aan ultimate moment close to the full-plastic moment ofbeam and the failure modes like shear failure, of the chaconnector or of bolts, should be avoided. Connections wHSFG bolts subjected to shear only areknown to perform betteunder cyclic loading.

4. Cyclic tests on connections

The primary objective of the tests was to obtainmoment–rotation characteristics and also to validate the finit

S.R. Satish Kumar, D.V. Prasada Rao / Journal of Constructional Steel Research 62 (2006) 739–746 741

le).

ilityrtae

tae

naH

ines

ed

rs-turineth

ingt

ertitha

ThrnThnndinm

e bo.

lledting

dtestsl

onentdintscemn,end,

ethe

f akNsi-

enttude

thewasndad

fromoff

n.mn

berins at

asing

pec-The

arent

noding

is-ackace-

neteec-

re-f the

Fig. 2. Typical test specimen (all dimensions in mm—drawing not to sca

element model of the connection. In seismic design, ducthysteretic energy dissipation and mode of failure are impoand so the tests were carriedout under incremental amplitudcyclic loading.

4.1. Test specimens

Four beam-to-column connection specimens, whose deare given inTable 1, were tested in the laboratory. The samRHS was used for both the beam and the column. The origiobjective was to develop a connection between hot-rolled Rbut only small size sections in limited range are availableIndia. In order to extend the use of RHS to multi-storied framfabricated sections can be used. Therefore, it was decidfabricate equivalent RHS by welding.

The RHS were fabricated by two methods. In the fimethod, two ISMC 200 channel sections were welded toetoe (Fig. 2) and abox beam was fabricated by welding foplates using full-penetration groove welds. The RHS obtaby welding two ISMC 200 channel sections was used for allcolumns.

The channel connectors were fabricated by welding two ISA75 × 75 × 10 equal angle sections toe-to-toe or by weld8 mm thick plates. Two channel connectors were weldedthe column by full-penetration butt welds, at an appropriatdistance from each other so as to facilitate the smooth inseof the beam between them. To prevent shear yielding ofcolumn web in the joint region, a 6 mm thick doubler plate wwelded to it using plug welds and also at its outer periphery.rectangular openings were cut in the beam web and the cowere rounded by grinding to prevent stress concentration.holes for HSFG bolts, in the beam flange and in the chaconnectors, were also match drilled to ensure uniform loaof all the bolts. The HSFG bolts were of diameter 24 mM10.9 grade. Hardened washers were used both under thhead and under the nuts. The bolts were pre-tensioned trequired amount using calibrated manual torque wrenches

,nt

ils

lS,

,to

to-

de

o

one

se

erseelg,olt

the

Tensile test coupons were extracted from the hot-rochannel, angle sections and plates, used for fabricathe beam–column test specimens. Average values of yieland tensile stress obtained from the three couponare 260 N/mm2 and 475 N/mm2 respectively for channeconnectors and plates.

4.2. Test setup and test procedure

The test setup used for simulating the seismic loadingan exterior beam-to-column moment connection in a momresisting frame is as shown inFig. 3. The column was supporteon steel rollers spaced 1000 mm apart, to simulate poof inflexion. It should be noted that the shorter distanensured that the bending moments developed in the coluat either beam face, were less than that at the beamthereby aiding the development of the plastic hinge in thbeam. Each specimen was loaded at the free end ofcantilever beam of clear span 1750 mm by means ocomputer-controlled hydraulic actuator of capacity 1000and stroke±125 mm. The tests were carried out quastatically by applying cycles of incremental displacemamplitudes and two cycles at each increment. The amplifor each successive step was increased in multiples ofyield displacement of the specimen. The test specimencompletely arrested from rigid body translation by using eblocks, which were anchored to the strong floor. The loand displacement of the actuator head were obtainedthe internal transducers of the actuator. The total rotationthe connection was determinedfrom the average reading otwo dial gauges installed at thetips of the channel connectorsdivided by the distance of the tip from the axis of the columTwo LVDTs were placed vertically at a distance of 150 mfrom the center line of the beamso as to measure the rotatioof the joint due to the elastic bending of the column. A numof post-yield strain gauges were also pasted to get the stravarious locations. The rotation of the connection is definedthe total rotation minus the rotation of the joint due to bendof the column.

5. Test results

The depth and thickness of the channel connector in simens 1 and 2 were 75 mm and 10 mm respectively.load–displacement hysteretic loops for the specimensshown in Fig. 4. The reproducibility of load–displacemecurves in the second cycle of each amplitude step, withstrength or stiffness degradation, can be observed. The yielof the beam net section is shown inFig. 5(a). The connectionwas able to carry increased load with increase in the tip dplacement and the trend continued until the initiation of crat the beam net section. With further increase in the tip displment the crack extended leading to the rupture of the beamsection, as shown inFig. 5(b). The failure is away from the facof the column and similar to beams with Reduced Beam Stions (RBS) [11,12]. As the connection was subjected to pdominant bending and due to the presence of four webs o

742 S.R. Satish Kumar, D.V. Prasada Rao / Journal of Constructional Steel Research 62 (2006) 739–746

000

Table 1

Summary of fabricated RHS beam, RHS column and channel connectors (all dimensions are in mm)*

Specimen no Beam Column Channel con-nector

Db Bb tb f tbw Dc Bc tc f tcw dch tch

1 200 150 11.4 6.1 200 150 11.4 6.1 75 12 200 150 11.4 6.1 200 150 11.4 6.1 75 13 200 150 11.4 6.1 200 150 11.4 6.1 38 14 200 150 8 6 200 150 11.4 6.1 50 8

∗ For notation refer toFig. 1.

Fig. 3. Test setup.

(a) Specimen 1. (b) Specimen 2.

Fig. 4. Tip load–displacement hysteretic loops.

ibiotio

ilaan

veyinp

eleen. Thehen asnelcantion.ts

channel connector the shear effects were found to be negligin the connection region. The hysteretic-loops of the connectare free from the pinching effect and hence these connecwere able to dissipate large amounts of hysteretic energy.

The behaviour of specimens 3 and 4 was found to be simFig. 6 shows the hysteretic loops for specimens 3 and 4. It cbe observed that the reproducibility of the loops was essentiallythe same andno strength or stiffness degradation was obserin the second cycle of each amplitude step. The load carrcapacity of the connection increased with increase in the ti

lenns

r.

dg

displacement until the occurrence of a crack in the channconnector. However, the stiffness of the connection has breduced due to the reduced depth of the channel connectorfailure was initiated by cracking of the extreme fiber of tchannel connector section close to the face of the columshown inFig. 7 and then leading to the fracture of the chanconnector section. The slight pinching of the hysteretic loopbe attributed to the cracking of the channel connector secThe yield and ultimate loads and corresponding displacemenfor all the specimens are summarized inTable 2.

S.R. Satish Kumar, D.V. Prasada Rao / Journal of Constructional Steel Research 62 (2006) 739–746 743

(a) Initiation of yielding. (b) Rupture of the beam net section.

Fig. 5. Failure mode of specimens 1 and 2.

(a) Specimen 3. (b) Specimen 4.

Fig. 6. Tip load–displacement hysteretic loops.

Table 2Results for the tested specimens

Specimen Yield Yield Ultimate Ultimate Initialload displacement load displacement stiffnessPy δy Pu δu Rki (kN/m)

(kN) (mm) (kN) (mm)

1 54 15.1 92 118 152.2e–32 52 15.0 91 115 154.7e–33 35.5 12.2 68 95 32.85e–34 33.6 10.0 70 90 30.2e–3

theonla

serelate

o getm

6. Finite element analysis of the proposed connection

Due to symmetry of the structure and loading aboutvertical plane passing through the beam and column axes,one-half of the beam-to-column connection sub-assembwas modeled, as shown inFig. 8, using the finite element

lyge

analysis software MSC/NASTRAN [13]. The component plateof the beam, column, and the channel connectors wdiscretised using isoparametric four-noded quadrilateral pelements. Finer mesh was used in the connection region tthe magnitude of the stress concentration accurately. The beaweb was not connected to the column. The effect of HSFG

744 S.R. Satish Kumar, D.V. Prasada Rao / Journal of Constructional Steel Research 62 (2006) 739–746

ctinn

smdoatiodin

domoin

ra

dtipsrthore

el

thetithewa2

anloloaan

f t

e

age.

c-rms.ed.ishi-am-ota-

ageery

ionThetethen.

Fig. 7. Failure mode of specimens 3 and 4.

bolts was represented by the use of rigid elements connethe corresponding nodes on the beam flange and the chaconnector, so as to simulate no-slip condition and to tranthe beam flange force to the channel connector. As the testeRHS connections were designed not to slip at ultimate lstage thebolt slip has not been modeled. The bearing acof the RHS beam on the channel connectors during benof the beam was simulated using spring elements of highcompressive stiffness. Symmetric constraints were applieall nodes lying in the plane of symmetry. The top and bottends of the column stub were hinged to represent the pof inflexion.

The material was modeled as an elastic–plastic sthardening (EPH) material with yield stress of 265 N/mm2

and ultimate stress of 475 N/mm2. The analysis was repeatewith different values of strain-hardening slopes until thedisplacement curve obtained from the finite element analysimatched perfectly with that obtained from the test on similabeam-to-column sub-assemblages. Finally it was foundthe strain-hardening slope of E/80 was found to be mappropriate to account for cyclic strain-hardening effects.

The downward load was uniformly distributed to the nodof the web at the beam tip section. Non-linear incrementaload analysis was carried out, in increments of 8 kN tillfailure of the connection to obtain the tip load–displacemenrelationship. Initially convergence studies were carried out wdifferent mesh sizes in the beam-to-column sub-assemblagmodel. Using the results of the convergence studies itdecided to carry out the analysis with 4575 elements and 4nodes.

The failure criterion is based on the normal stress atnode reaching the tensile strength of the material and thecorresponding to that load step is taken as the ultimatecarrying capacity of the connection. Comparison of testanalytical results, shown inFig. 9, indicates that the finiteelement analysis can be used to predict the behaviour oconnection with sufficient accuracy.

7. Analytical model for moment–rotation relationship

An analytical model of the connection moment(M)–rotation(θ) characteristics will be of great use in the analysis of fram

ngelit

dng

to

ts

in

at

s

s50

yaddd

he

s

Fig. 8. Finite element model of the beam-to-column sub-assembl(a) Global view. (b) Local view.

to study the realistic behaviour. The moment–rotation charateristics of the connection can be represented in various foBut the three-parameter power model is the most widely usTherefore, the three-parameter power model proposed by Kand Chen [2], shown in Fig. 10, has been chosen to represent the connection behaviour. As per this model, the pareters required to be determined are the initial tangent rtional stiffness(Rki ), theultimate moment capacity(Muc), andthe shape factor(n), which determines the curvature of themoment–rotation curve.

The model is given by the equation

M = Rki × θ

(1 +

(θθ0

)n) 1n

(1)

in which θ0 is a reference plastic rotation= MucRki

.The initial stiffness of the connection(Rki ) can be obtained

as the slope of the moment–rotation curve at the initial stof the loading, i.e., when the connection was subjected to vlow levels of loads. In the present study the initial connectstif fness was determined corresponding to the load of 8 kN.ultimate moment of the connection is the product of ultimaload carrying capacity of the connection and the distance ofpoint of action of the load from the center line of the colum

S.R. Satish Kumar, D.V. Prasada Rao / Journal of Constructional Steel Research 62 (2006) 739–746 745

(a) Specimens 1 and 2. (b) Specimens 3 and 4.

Fig. 9. Comparison between test and finite element analysis results (tip load–displacement curves).

te

th

esr

d

anie

ionathesctoom

three-

esandreticctiontyn

ctivetion. Themender

se

ns

Fig. 10. Three-parameter power model.

Table 3Three parameters for the tested specimens

Specimen Ultimate Reference plastic Shape paramemoment(Muc) rotation (θ0) (n)

(kN m)

1 and 2 168.4 3.8e–3 1.353 124.5 1.1e–3 2.04 135.8 1.52e–3 1.78

The moment–rotation characteristics of the connection andleast-square technique can be used to determineθo andn forthe connection. The comparison of the moment–rotation curvobtained from the test and the analytical three-parameter powemodel is shown inFig. 11. The three parameters for the testeconnections are given inTable 3.

8. Summary and conclusions

A new connection configuration between an RHS beaman RHS column was presented. The connection is convenand has several advantages in terms of structural and functperformance. The details of test specimens, test setuptesting procedure were also described. Depending uponstrength of the channel connector, two types of failurmodes were observed. In the case of channel conneof high strength, the beam net-section failure is away fr

r

e

dntal

nde

rs

Fig. 11. Comparison between moment–rotation curves of the test andparameter power model.

the face of the column. Non-linear finite element analysof RHS beam-to-column connections was carried outvalidated with test results. The load–displacement hysteloops of the tested specimens indicate that the connepossesses considerable hysteretic energy dissipation capaciand also sufficient ductility. Hence, the proposed connectiocan be recommended for use in frames in seismically aregions. The moment–rotation relationship of the conneccan be modeled using the three-parameter power modelconnection modeling can be incorporated in semi-rigid fraanalysis programs to get the performance of the frame uvarious design loads.

References

[1] Chen WF, Kishi N. Semi-rigid beam-to-column connections: data baand modelling. J Struct Eng ASCE 1989;115(1):105–19.

[2] Kishi N, Chen WF. Moment–rotation relations of semi-rigid connectiowith angles. J Struct Div, ASCE 1990;116(7):1813–34.

746 S.R. Satish Kumar, D.V. Prasada Rao / Journal of Constructional Steel Research 62 (2006) 739–746

ntru

ing

ions

mns.

am

SS

ions.

.s for

es7;

les

[3] Prasada Rao DV, Satish Kumar SR. RHS Beam-to-column connectiowith web opening — parametric study and design guidelines. J ConsSteel Res 2005.doi:10.1016/j.jcsr.2005.11.015.

[4] White RN, Fang PJ. Framing connections for square structural tubJ Struct Div, ASCE 1966;92(ST2):175–94.

[5] Ting LC, Shanmugam NE, Lee SL. Box-column to I-beam connectwith external stiffeners. J Construct Steel Res 1991;18(3):209–26.

[6] Shanmugam NE, Ting LC, Lee SL. Behavior of I-beam to box-coluconnections stiffened externallyand subjected to fluctuating loadJ Construct Steel Res 1991;20(2):129–48.

[7] Shanmugam NE, Ting LC. Welded interior box-column to I-beconnections. J Struct Div, ASCE 1995;121(5):824–30.

[8] Korol RM, Ghobaran A, Mourad S. Blind bolted W-shaped beam to H

ct

.

columns. J Struct Div, ASCE 1993;119(12):3463–81.[9] Wheeler AT, Clarke MJ, Hancock GJ, Murray TM. Design model for

bolted moment end plate connections joining rectangular hollow sectJ Struct Div, ASCE 1998;124(2):164–73.

[10] Wheeler AT, Clarke MJ, Hancock GJ. FE modeling of four-bolt tubularmoment end-plate connections. J Struct Div, ASCE 2000;126(7):816–22

[11] Chen S, Yeh CH, Chu JM. Ductile steel beam-to-column connectionseismic resistance. J Struct Div, ASCE 1996;122(11):1292–9.

[12] Chen S, Yeh CH, Chu JM, Chou ZL. Dynamic behaviour of steel framwith beam flanges shaved around connection. J Construct Steel Res 19942:49–70.

[13] MSC/NASTRAN Reference Manual, Volumes 1, 2 and 3. Los Ange(California): MacNeal-Schwendler Corporation; 2000.

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具