Experimental Analysis of Beam-To-Column Connection in Steel Storage Racks Using Cantilever Test and Portal Test Method PDF

Preview

Summary

EXPERIMENTAL ANALYSIS OF BEAM-TO-COLUMN CONNECTION IN STEEL STORAGE RACKS USING Title CANTILEVER TEST AND PORTAL TEST METHOD Author(s) ASAWASONGKRAM, N.; CHOMCHUEN, P.; PREMTHAMKORN, P. Proceedings of the Thirteenth East Asia-Pacific Conference on Structural Engineering and Construction (EASEC-13), ...

Description

Title

Author(s)

Citation

Issue Date

Doc URL

EXPERIMENTAL ANALYSIS OF BEAM-TO-COLUMN CONNECTION IN STEEL STORAGE RACKS USING CANTILEVER TEST AND PORTAL TEST METHOD

ASAWASONGKRAM, N.; CHOMCHUEN, P.; PREMTHAMKORN, P.

Proceedings of the Thirteenth East Asia-Pacific Conference on Structural Engineering and Construction (EASEC-13), September 11-13, 2013, Sapporo, Japan, I-1-3., I-1-3

2013-09-13

http://hdl.handle.net/2115/54471

Type

proceedings

Note

The Thirteenth East Asia-Pacific Conference on Structural Engineering and Construction (EASEC-13), September 1113, 2013, Sapporo, Japan.

File Information

easec13-I-1-3.pdf

Instructions for use

Hokkaido University Collection of Scholarly and Academic Papers : HUSCAP

EXPERIMENTAL ANALYSIS OF BEAM-TO-COLUMN CONNECTION IN STEEL STORAGE RACKS USING CANTILEVER TEST AND PORTAL TEST METHOD N. ASAWASONGKRAM1*†, P. CHOMCHUEN1, and P. PREMTHAMKORN1 1

Faculty of Engineering, Mahanakorn University of Technology, Thailand

ABSTRACT This paper presents experimental results of a typical steel storage rack beam-to-column connection, which steel tabs welded at the end of the beam are engaged into the perforated column. The beam-to column connections are tested using the cantilever test and the portal test method according to the international racking design specification. Four different levels of vertical service loadings are conducted for the portal test. The results from a pushover analysis of a single story frame using a moment-rotation relation of beam-to-column connections obtained from cantilever tests are compared with the experimental results from portal tests. The analytical results show a good correlation with the experimental results from the portal test for no vertical loadings and medium vertical loadings. For high vertical loadings and very high vertical loadings, the analytical results do not produce a good correlation with the experimental results from the portal test. The cantilever test cannot represent the effect of the presence of the vertical service loading in actual field. These effects increase the initial stiffness and decrease the strength of the connection comparing with the results from the cantilever test and lead to a non-correlation of the analytical results compared to the test results from portal tests for a high and very high vertical loading. Keywords: Cantilever tests, Connection, Moment-rotation, Portal tests, Steel storage racks. 1.

INTRODUCTION

In recent years, steel storage racks are extensively used in industry and large public warehouse for storing products. This type of structure has become a common feature in several countries. The main components of the structure are box beams, beam end connectors, and perforated thin wall cold-formed open sections columns. The beam connects to the column by the steel tabs inserted into the perforated column. According to the previous experimental study, this type of connections can be classified as a semi-rigid connection with a low to moderate strength and stiffness (Bernuzzi and Castiglioni 2001, Bajoria and Talikoti 2006, Prabha et al. 2010). Generally the storage racks are not braced in the longitudinal due to the continuously accessible to the product in service. Such a

*

Corresponding author: Email: [email protected]

†

Presenter: Email: [email protected]

1

configuratiion makes the longitudinal directtion particu ularly criticaal under a lateral exciitation. Thee lateral behhavior of thhe storage rack in thhe longitudiinal direction dependss significan ntly on thee behavior oof the beam m-to-column n connectionns. Thereforre, it is imp portant to hhave properr method off determininng it. For beam-to-col b lumn conneections, theere are a great g differeence exist in memberr geometry aand connecttion system ms (Markazi et al. 1997), all international stanndards for stteel storagee rack (AS 1993; RM MI 2008; FEM F 1998)) require specific s tesst to deterrmine the reliable off moment-rootation relattion of beaam-to-colum mn connectiion. Two teests to deteermine the connectionn behaviors aare: the canntilever test and the porrtal test. In the canttilever test a short lengtth of beam is connecteed to a colum mn and thenn load to faiilure. Manyy researcherss have perrformed the cantileveer test metthod (Mark kazi et al. 1997; Bernuzzi andd Castiglionii 2001; Bajooria and Talikoti 2006)). Although h the cantilev ver test prov ovides a sim mple methodd of determinning the connnection strrength and sstiffness, the disadvanttage of this m method is the t effect off the verticaal service loading l can nnot represeent. In ordeer to betterr represent the effect of verticall service loaading in a beeam-to-colu umn connecction, the po ortal test meethod has beeen consideered. In thiss testing meethod, the beam b is con nnected to ttwo column ns to form a portal fraame. Both ends e of thee column shaall be pinneed at its basse. The verttical load sh hall be applied to the ppallet beam to simulatee the usual sservice loadding. Then a lateral looad is appliied on the column c at tthe level off the beam.. Although tthe portal teest can repreesent the beehavior of raacks in actu ual conditionn, quite a feew numberss of portal teest have beeen reported (Krawinkleer et al. 197 79; Harris 2006) due too it is ratherr difficult too perform. This paperr presents exxperimental results of the cantilev ver tests an nd the portal al tests on th he behaviorr of beam-too-column coonnection of o steel storrage rack. The T portal tests are peerformed with w variouss vertical looading leveels. The efffects of vvertical loaading on the t momennt-rotation relation off beam-to-coolumn connnection are describbed. An analysis a off a singlee story frame fr withh moment-rootation of connections c obtained bby a cantileever test is compared with the ex xperimentall results from m portal tesst. The correelations betw ween such the t two tests are discusssed. 2.

EXPE ERIMENTA AL INVESTIGATION NS

The test sppecimens are a selected from a coommercial Thailand T manufacture. The geom metry of thee specimens is depictedd in Figuree 1. These values aree given as an averagee value of the severall measuremeents from eaach series.

Figure 1: Compon nent of the tested speccimens (dim mensions inn (mm)). 2

2.1. Canttilever test The cantileever test provides a sim mple methood of determ mining the connection c strength an nd stiffness.. The cantileever test seetup consistss of a shortt cantilever of pallet beam conneccted to the center of a short lengtth of a collumn. Both h ends of thhe column are rigidly y supportedd. The load d is appliedd monotoniccally by a hyydraulic jacck placed onn a load celll. The free end e of the bbeam is con nstrained byy a vertical gguide to prevent an un ndesirable oout-of-planee movemen nt of the beaam. The ap pplied loadss are recordded at eachh incremen nt of loadiing until the t failure is occurreed at the connection. c . Displacem ment transducers are mo ounted to meeasure beam m and colum mn deflectioons. The rottation of thee connectionn is calculatted from thee deflectionns for each load step. Figure F 2 shoows the gen neral layoutt of the expeerimental seetups and th he arrangem ment of the transducerss. Full detaiils of the ex xperimentall results are available inn an earlier paper by thhe authors (A Asawasongk kram 2012)).

(a) Traansducer arrangemen nt.

(b) Actual test seetup.

Figure 2: C Cantilever test setup. The cantileever tests are a conductted of six cconnection samples wh hich consistt of three for f hoggingg moment teesting and three for sagging s mooment testin ng. The sam mples of exxperimentall results off moment-rootation curvve are show wn in Figuree 3. The exp perimental results show w that, the connectionn exhibited llooseness at a the initiall stage bec ause the stteel tabs aree not fit inn the colum mn slot. Thee looseness iis overcomee when the rotation reaached an ap pproximately 0.006 radd. The initiaal loosenesss of the connnection is also a found in the prevvious of exp perimental studies s (Berrnuzzi and Castiglionii 2001; Bajooria and Tallikoti 2006;; Prabha et al. 2010). The T strength h and stiffnness of the connections c s are differeent under hogging h an nd sagging moments due to thee asymmetrric of the connectionn geometry. The failure took place when the taab in the ten nsion side was w cut by thhe column perforation. p . The resultss also show w that the av verage of pplastic rotatiion capacity y of the connnection is about 0.088 rad. The cconnection stiffness s an nd the ultim mate momen nt capacity are tabulatted in Tablee 1. In thiss study, the connection stiffness iss determineed by the co oncept sugg gested by FE EM (FEM 1998) afterr the initiall loosenesss is termin nated. Thee FEM usses an iterrative grapphical proccedure thatt approximaately balancees the areass below the actual and ideal i curvess up to the ffailure pointt.

3

(a) Hogging mom ment.

(b) Sag gging mom ment.

Figuree 3: Examp ple of momeent-rotation n curve fro om cantilevver test. Table 1:: Beam-to-ccolumn con nnections teest results from f cantillever test Test nno.

Type of loading

Stiffness (kN-m/rad)

1 2 3 4 5 6

Hogginng moment Hogginng moment Hogginng moment Sagginng moment Sagginng moment Sagginng moment

51.42 51.54 45.52 45.33 48.54 40.75

ge Averag stiffnesss (kN-m/rrad) 49.50 0

44.87 7

mate moment Ultim capacity (kN-m) ( 1.44 1.34 1.32 1.36 1.59 1.63

Average ultimate moment caapacity (kN-m m) 1.37

1.53

2.2. Porttal test Under the portal test, the connecction is subbjected to sh hear force, bending m moment and axial forcee thus repressenting the actual field d conditionss. A portal test setup is i shown inn Figure 4. Two portall frames weere mountedd on hingess supports which are clamped to o the strongg floor. Thrree bracingg element arre connectedd between two t portal fframes by channel c secttions to prevvent any diisplacementt in the transsverse direcction. The vertical v disttance betweeen the center of the hiinges and th he center off the beam is equal to 700 7 mm. Th he horizontaal distance between b thee column ceenterlines iss 2500 mm.. The distannce betweenn two portal frames is 11000 mm. A horizontall rigid transsfer beam bolted to thee two colum mns is used to distribu ute the laterral loads eq qually to th he two fram mes. Lateraal loads aree applied by a hydraulicc jack placed on a load cell attacheed to a very y rigid steel support and d connectedd to the stronng floor. Veertical loadss are simulat ated by sand d bags restin ng on standaard wood paallet.

((a) Transdu ucer arrang gement.

(b) Actual test seetup.

Figure 44: Portal teest setup.

4

Four typess of content weight are used for thhe test seriess to investig gate the effeects of vertiical loadingg level on sttrength and stiffness of beam-to-ccolumn con nnection. Th he 1st case is no verticcal loading.. The 2nd is “medium” loading l level which haas a total weeight of 1 to on (2 kN/m m/portal fram me). The 3rdd is “high” lloading leveel which haas a total w weight of 2 ton (4 kN/m m/portal fraame). The 4th is “veryy high” loadding level which w has a total weiight of 2.5 ton (5 kN N/m/portal fframe). Theese verticall loading levvels corresppond to 0%,, 40%, 80% % and 100% of the allow wable mom ment of the portal p beam m respectivelly. Four dissplacement transducers t s, LVDT1-L LVDT4, aree placed on column at the t level off centerline of the porttal beam to measure thhe lateral displacemen d nt of the poortal frame. Additionall displacemeent transduccers, LVDT T5-LVDT8, are placed at a the base of o the colum mn to check k the slidingg of the coluumn. The lateral l load ds and laterral displaceements are recorded aat each incrremental off loading unntil the failurre is occurred. The expperimental moment-rot m ation curvees are given in Figure 5. 5 The main characteristtics of the moment-rot m tation curvees are show wn in Table 2. The mom ment at thee connectionn are expressed as (Kraawinkler et aal. 1979)

H h  P 2

M 

(11)

The averagge rotation of o the beam m-to-columnn connection n is given by y



 Mh ML L    h  3 EI c 6 EII b



  

(22)

where H is the laterral load app plied to onee portal fram me, h is thee vertical ddistance from m center off hinge suppport to the center of porrtal beam, P is the ax xial force in n the columnn due to verrtical loads,,  is the lateral displacement at the center oof the portaal beam.  is taken ass the averagge recordedd displacemeent obtain from fr LVDT1-LVDT4 aas shown in Figure 4(a)).

ment-rotatiion curve ffor portal test with fou ur vertical loading lev vels. Figgure 5: Mom Table 2: Beam-to o-column cconnectionss test resultts from porrtal test Type of vertical loading No veertical loadingg 1 toon (2 kN/m) 2 toon (4 kN/m) 2.5 tton (5 kN/m)

Moment at the initial sstage, Mi (kN-m) 0.081 0.184 0.286 0.538 5

Ultimate mom ment capacity,, Mu (k kN-m) 1.612 1 1.473 1 1.439 1 1.152 1

Mi/Mu (%) 5.0 5 12.5 19.9 46.7 4

From the pportal testinng, the follo owing obserrvation can be made. The T structurre with high her verticall loading levvel has a loower connecction capaccity than thee structure with w lower vertical loaading level.. For exampple, the ultimate momeent capacityy of the very high loaading is 70% % comparin ng with thee ultimate m moment capaacity of the no vertical loading. In n the higher vertical loaading levels, both sidess of the beam m-to-colum mn connectio ons have a hhigher initial hogging moment annd a higher shear forcee acting on a positive diirection. Wh hen the lateeral load is subsequentl s y applied, th the left side connectionn would be ssagging mooment and a shear forcce acting on n a negativee direction whereas the right sidee connectionn would be a hogging moment annd a shear force acting g on a posiitive direction. For thee right side cconnection, the hogging moment aand shear fo orce due to the lateral lload will co ombine withh a high iniitial hogginng momentt and a hi gh initial shear forcee due to thhe vertical load. Thee combinatioons of bendding momen nt and sheaar force pro oduce the right side coonnection reeaches theirr ultimate m moment capaacity prior to o the other llower verticcal loading levels. The portal frame defllected shapee under the vertical an nd lateral lo oad are show wn in Figurre 6. At thee initial, the angle betw ween the beaam and the column deccrease both on the leftt and right connections c s due to the applicationn of verticall loads. Whhen the structure subjeccted to an iincreasing lateral l load,, an angle oon the left side s connecction starts to increasee and the an ngle on thee right side connectionn continue too decrease. The failure occurred onn the right side s connecction at the ssteel tab plaaced fartherr above the nneutral axiss. Another obbservation is i that the portal p test eexhibits a hiigh initial stiffness s at tthe initial stage s due too the tightenning of the vertical v load d. In this stuudy, the efffect of the vertical loadd on the initial stiffnesss is represennted by thee ratio of th he momentt at the iniitial stage to t the ultim mate moment capacityy (Mi/Mu). T The effect off the verticaal load effeect is low, which w Mi/M Mu equal to 55% and 12.5%, for thee no verticall loading annd the mediium verticaal loading. The T vertical load has a certain efffect, whichh Mi/Mu equual to 19.9% %, for the high verticall loading. The T effect of o the verticcal load is rather r high,, which Mi/M Mu equal too 46.7%, forr the very hhigh verticaal loading. At A the end oof the initiaal stage, thee tightening of the connnection is released r byy the moment due to th he lateral looad, then th he stiffnesss decrease. T The momennt-rotation curve show w the “slip”” at the en nd of the innitial stage for the noo vertical loading and the medium m vertical lloading. Th he slip reprresent by a short flat line in thee moment-rootation curvve. This slip come from m the steel taabs are not perfectly fitt with the column c slot.. However tthere is no slip s for the high and veery high verrtical loadin ng due to a large friction betweenn the steel taaps and coluumn slot from a high vaalue of vertiical loading gs.

pplication of o vertical load. (a) Ap

(b) ( Applica ation of lateeral load.

Figu ure 6: Portaal frame deeflected sha apes. 6

3.

STRU UCTURAL ANALYSIIS

The correlaation of thee experimen ntal test resuults obtained d from canttilever tests and portal tests is onee of the maiin objectivees of this study. s The cantilever tests distin nguish betw ween the ho ogging andd sagging m moment-rotaation of thee connectionns whereass the portal tests give a mean value of thee moment-rootation of thhe connectiions. To coompare the experimenttal results inndirectly, a non-linearr pushover aanalysis of a single sto ory frame iss performed d by means of a compuuter program m SAP20000 (CSI 2009). The resuults between n the portal test and th he numericaal analysis uusing beam m-to-columnn connectionn obtained from the cantileverr test are investigateed. Non-linnear behav vior of thee beam-to-coolumn connnection is modeled m by a non-lineaar rotationaal link elem ment. The liink elementt connected to the end of beam an nd the end oof column as shown in n Figure 7. The characcteristics off link elemeents are obttained from m the cantillever test. The T effects of geomettric nonlineearities andd material nnonlinearitiees are also taken innto accoun nt in the numerical model. A nonlinearr moment-rootation relattion of a con nnection is modeled by y a tri-lineaar model to account forr the effectss of the initial loosenesss. The tri-lin near model is illustrateed in Figuree 8. The vallue of k1 is the t averagee stiffness inn the initial looseness stage. s The vvalue of k2 is the average connecction stiffness obtainedd by the conncept suggessted in FEM M (FEM 19998). The paarameters of o the tri-linnear model (k1, k2,  i ,

 y , Mi, Mu) obtained from experrimental resuults of the cantilever c teest. The pusshover anallysis resultss compare w with the expeerimental reesults from the portal teest as illustrrated in Figuure 9. The analyttical resultss using beaam-to-colum mn connecttions from the cantilevver test show a goodd correlationn with the experimenta e al results froom the porttal test for no vertical loadings an nd medium m vertical loaadings. For high and very v high v ertical load dings, the an nalytical ressults do nott produce a good correelation withh the experim mental resuults from the portal tests. The expperimental results r from m portal testss produce a higher stifffness approxximately tw wo times hig gher than thhe analytical results forr the high annd very higgh vertical loadings. T The portal tests t also produce a loower streng gth than thee analytical results for very high vertical loaadings. Theese results show a lim mit of a can ntilever testt method forr high and very v high veertical loadiings. As exp plained in section 2.2, tthe cantilev ver tests cann not represeent the effeect of a tigh htening at thhe connectiion and a combinationn of shear and bendingg moment duue to the veertical servicce loading. These effeccts lead to a non-correllation of the analyticall results com mpared to thhe test resultts from porttal tests for high and veery high verrtical loadin ng level. Mu Mi

i Figure 7: Configurattion of the case study frame and d beam-to-ccolumn join nt modelin ng.
<...