THE EFFECT OF BEAM TO COLUMN CONNECTON N ARC PORTAL FRAME Asko Keronen Rakenteden Mekankka, Vol. 26 No 2 1993, ss. 35-5 SUMMARY A full scale rc (renforced concrete) portal frame has been bult n order to study the effect of beam to column connecton on deflecton and supportng moment of a column n practce. Fve dfferent connecton types were used n tests: the am was to fnd out the dfference between a hnged connecton and a connecton wth a bearng pad. Another am was to fnd out a connecton type whch s more rgd than a connecton wth a full sze bearng pad: two examples of new connectons are shown. The results of hnge- and bearng pad-connectons are estmated wth a calculaton method, whch has been proposed and used n references (Lndberg 1987, Keranen 1984, Keranen 1991). Ths method takes nto account the effect of a bearng pad n a beam to column connecton on the deflecton of a frame and the moment surface of a column. The results show that beam to column connecton type has a sgnfcant effect on deflecton and supportng moment of a column. The hnge-connecton caused clearly - eaven 4.3 tmes - greater deflecton and supportng moment than a connecton wth a full sze bearng pad. However, the least deflectons were found for new connecton types. On grounds of ths research t s mportant to take nto account the beam to column connecton type n constructng columns n a rc portal frame. Also t s mportant - as the fnal am of ths research s - to develope the rules for effectve length factor of ths knd of columns. 35
NTRODUCTON n Fnland one-stored ndustral and commercal buldngs and storehouses are often constructed of prefabrcated rc portal frame structures. Stablty s usually acheved by cantlever acton of columns. At least one column n a frame s cantlevered; others can be pn ended. n beam to column connecton there s an elastomerc bearng pad n order to prevent peak stresses on the contact surfaces and cracks due to the dfferent rotaton of a column and a beam. t s generally assumed that the beam to column connecton behaves as an deal hnge. However, a bearng pad allows a movement of the reacton of the beam as a result of dfferent rotatons between the beam and the column. The movement of the acton pont causes a moment that may be opposte sgn compared to common practce. The acton pont of the reacton of the beam vares due to the horzontal load. PROPERTES OF THE BEARNG PAD The movement of the reacton of the beam depends on the sze and the elastcty of the bearng pad. Also the smoothness of the.contact surfaces, tme and temperature have an effect on the movement. Equal shape factors ( = the relatonshp between loaded and unloaded surface) do not necessarly quarantee that the compresson propertes for bearng pads are smlar. Thnner bearng pad usually behaves stffer. n compresson tests (Keronen 1984) bearng pads were loaded between two concrete blocks of sze 2 2 2 mm 3 (smooth steel mold). The compressve stran of a bearng pad was measured usng Demec -pns that were attaced n the concrete cubes. Two pars of pns were used n order to elmnate the compresson stran of concrete. Elastc propertes were examned for bearng pads 15 15 mm 2 (7 RH). The thcknesses were 3, 4, 5, 6 and 8 mm (shape factor k = 12.5... 4. 7) and the 36
compressve stress.4... 12. MPa (wth loadng speed.4 MPamn). Every pad was loaded quckly up to 12-13 MPa and unloaded several tmes before tests. For the thckness of 8 mm, the modulus of elastcty was 11... 22 MPa. For other thcknesses ths value vared between 1... 28 MPa beng at ts most for 5 ja 6 mm pads by compressve stress 1 MPa. The bearng pads (also 7 RH) used n ths research were tested. The modulus of elastcty was 37-82 MPa (under stress 1.8-7.1 MPa) for sze 15 15 8 mm 3 (hole <> 34 n the mddle and a cuttng: k = 3.3) and 5-126 MPa (4.1-16.4 MPa) for sze 65 15 8 mm 3 (k = 2.8); correspondng loadng speeds were 6 and 2 MPamn. These values were measured durng 5. compresson cycle ( -> max. load - > ) between smooth concrete blocks 2 2 2 mm 3 n frst loadng tests the pads were between the blocks all the tme from the precdng fast loadng-unloadngs to the end of the test. The am was to represent the effect of wnd load. n second tests the am was to smulate the portal frame n loadng tests: n these tests the pads were released from the connecton between every vertcal loadng. There were no precdng loadngs before measurements. So n bearng pad loadng tests there were any precdng loadngs before measured loadng cycles. Ths means that the regon of low modulus of elastcty n the begnnng of the tests s ncluded n second loadng measurements, but s pressed away n frst tests. Ths s the obvous reason - n addton to the dfferent shape factors - for dfferent test results n these two tests. The modulus of elastcty n computer calculatons was 6 MPa, as n earler studes. n fgure 1 the rotaton dfference w between beam end and column top s shown as a functon of the eccentrcty ed. The reacton of the beam s N = 5 kn, the area of the bearng pad s 35 35 mm 2, the modulus of elastcty E = 6 MPa and the thckness t 5 and 1 mm. Values are typcal for the calculatons concernng arc portal frame. One can see that the eccentrcty ea grows ntensvely wth small values of the rotaton w. Ths leads to a quck ncrease of the moments n beam to column connecton. 37
. o, 14,12 o, 1 o, oa rotat1on (rad ) 1,6 o, 4 o, 2!!! 1 ------- t =, ""' 1! [-t=1 mo!! ' '!! J :1! v!! l '! l!: l~l_---- 1 '----------------o,-os---o-,1 -----~-~-~:_-_--_~~-:--~-- o~ ~_:_ -_-o_-~:_--_-_- ~._~' ed o_. o._" o,s_j, Fg. 1. Relatonshp between the eccentrcty ed and the rotaton dfference w. STRUCTURE OF THE TEST FRAME A full scale precast concrete portal frame was constructed for loadng tests. The frame conssts of precast columns and a beam. The columns are supported rgdly by a steel beam HE 32 A that s fastened to the floor (fg. 2). The sze of the columns s 18 18 mm 2 and length 33 mm. So, the slenderness 1.. s 14 (L = 2. 2 L). There s one renforcementbar of <> 12 mm n each corner of the column cross-secton. The tes are <1>6 k 15. The rc beam s very rgd compared wth the columns: cross-secton s 28 6 mm 2 and length 13 mm. Strength of concrete was 45 MPa, strength grade of renforcement A5HW and steel at least Fe 51 C. 38
hnge jont ma de by usng two roller bea rngs renforced co ncrete beam 28 x 6 mm 2 renforced co ncrete columns 18 x 18 mm 2 X n 4h ' ' ' ' ' ' ' ' ' ' ' : ~ " " " 13 horzontaoadng (both drectons): ~ ' mechancal weghts :5 6 kn ',-, J(j \.t~-~:' '! ~1 Llf-1 L,,,,,, "',,,,,,,,, < =------~:: _-, L ': " ' vertcaoadn g: hydraulc jack s 4 kn + 4 kn q HE3QA " ::.: ~ ==~~7..) fr:--~fc HE 1 A ~ "' tenson bars C!l32mm hnge jont m ade by usng two roller be "'"'' ~ supportng b eam ~ lw l HE 32A % ' rgd support Ah '<W rfh lrfh ~} :Q - lw lp 8"",r 1 - ;:r, :,,-- ',,,'$,~ ~~~ '.!---"..-=- ~ 3 w Fg. 2. Rc portal frame. 39
BEAM TO COLUMN CONNECTONS n the tests there were fve connecton types: 1) a bearng pad whch covers whole supportng area (15 15 8 mm 3 ), 2) a bearng pad whch covers only half of the supportng area (65 15 8 mm 3 ), 3) a hnge, 4) a steel component and 5) a steel component wth a pretensoned bar (fg. 3). n every connecton the transfer of the lateral force was ensured by a renforcement bar 4> 16. Connecton type 1 corresponds a stuaton where the column top supports only one beam end. n second type the column top s prepared to support two beam ends; pads are placed symmetrcally n the frame. These connectons are common n portal frame structures. Connecton 3 corresponds common dealsaton n calculatons. Type 4 s a new dea how to make a connecton more rgd. n last type ths connecton s pretensoned wth a renforcement bar <!> 16 (stress 3 MPa). The maxmum exentrcty n steel component connectons s 6 mm... - l T - - - - - - - - - - -1 ~!! :! - - - - 1.- - - - - - 1.- - - - - - -,.r- - - - - - - - - - -,! j ( :.> - - - J.L._._. - - -, r- - - - - - - - - - -,. _(.) ( ) - - _J.L._._.. Fg. 3. Beam to column connectons. 4
LOADNG SYSTEM The vertcal load was arranged wth two hydraulc jacks of 4 kn. The jacks were located between two short HE-beams n the mddle of the frame. The beams were connected to the frame by tenson bars of <!>32 mm. The frame was loaded by vertcal loadngs N = 37, 7, 12, 135, 167, 265, 292 kn for one column. These loadngs corresponds mean compresson 1.1-9. MPa. Horzontal loadng was arranged wth mechancal weghts. Tenson wre was fastened to the end of the beam. Two wres were used: the frame was loaded symmetrcally on both drectons to mnmze the permanent deflecton. There were 1-4 cycles for one vertcal loadng. n tests horzontal loads were H ~ 6 kn for whole frame. The maxmum value depends on the vertcal loadng and the beam to column connecton type. The horzontal effect from vertcal loadng system s subtracted from fnal horzontal loadng. Proposed loadngs are from frst cycle and the horzontal loadng s always to same drecton. TEST RESULTS Deflecton The deflecton of the frame depends strongly on the connecton type between the beam and the column. The full sze bearng pad has same knd of effect on the rgdty of the frame as n tests wth a steelframe (Keronen 1984): a frame wth full sze bearng pads s clearly more rgd than a frame wth hnge connectons; and the more vertcal loadng the larger s ths dfference (fg. 4-7). 41
l H[l<N] 3 6,---------,--------, ---------,--------,-------~---------, 5 1 15 v(mm) 2 25 3 Fg. 4. The deflecton of the frame, when N connecton type. 37 kn. Numbers refer to the 5 4 5 1 15 v(mm] 2 25 3 Fg. 5. The deflecton of the frame, when N = 12 kncolumn. 42
5 H[kN] 3, r 5 4 l 1 ~ ~ ~~ l-- 2 3 1 ~ ~ ' 5 1 15 v[rnm] 2 25 3 Fg. 6. The deflecton of the frame, when N = 167 kncolumn. H[kN] 3 ' 4 j' 5 1 1 ~~ 1 v -~ l------------ 3 5 1 15 2 25 3 v[rnm] Fg. 7. The deflecton of the frame, when N = 265 kncolumn. 43
For example, when vertcal loadng N s 12 kn and horzontal loadng Hs 2 kn, the relatve deflecton wth hnged connecton s 2.6, and when N s 167 kn respectvely 2.8 compared wth the deflecton wth bearng pad connecton. Under vertcal loadng 267 kn and horzontal loadng 1 kn the relaton s 4.3. Also the half sze bearng pad connecton effects smaller deflecton than the hnged one although the dfference s now slghter: the correspondng relatve deflectons are now 1.4 (N = 12 kn, H = 2 kn) and 1.5 (N = 167 kn). The deflectons are n these cases 1.8-1.9 tmes larger compared wth deflectons wth full sze pads. The effect of the locaton of the hnge was also studed. There were no dfference between the deflectons ether the hnge was stuated centrcly or eccentrcly (e = 42 mm symmetrcally). The frame wth connectons 4 or 5 s most rgd under every loadng stuaton. Pretensoned connecton effects clearly smallest deflecton when pretenson force s at least.5 N. For smaller values ths effect becomes nsgnfcant. The mnmum pretenson force was. the same for a steel component connecton wth eccentrcty e = 45 mm nstead of 6 mm. The steel component connecton reduces deflectons 49-57 % (N = 12-167 kn, H = 2 kn) compared wth deflectons wth full sze bearng pad and respectvely 8-85 % compared wth hnged connecton. For shorter eccentrcty (e = 45 mm) respectve fgures are 43-46 % and 76-81%. Wth every connecton type the deflecton of the frame was the less the more the vertcal loadng was. Vertcal loadng effects most to deflectons wth steel components and least to deflectons wth hnges. Typcal deflecton fgures after loadng cycles show (fg. 8 and 9), that there s clearly more dsspated energy n tests wth bearng pad than hnge connectons. These fgures located symmetrcally n regard to orgo; deflecton wthout horzontal loadng was normally 1-4 mm. 44
11 N = 12 kn H{kll) -3 3 v ( mm ] -6 Fg. 8. Typcal loadng cycles for connecton 1. H { kll) -3 3-4 -6 Fg. 9. Typcal loadng cycles for connecton 3. 4~
Supportng moment Also the supportng moment of the column depends on the connecton type. n fgure 1 supportng moments are counted up(= supportng moment of the frame). The results are smlar to deflecton fgures: the relatve supportng moment wth hnged connecton s 2. (N = 12 kn, H s 2 kn) and 2.1 (N = 167 kn) tmes greater than wth full sze bearng pad connecton. Relatons are here smaller than wth deflectons because the effect of vertcal loadng to supportng moment s slghter compared wth horzonta11oadng. The respectve relatve supportng moment relaton between hnged and half sze bearng pad connecton s here 1. 3 (N = 12 kn and 167 kn, H = 2 kn). The supportng moments are n these cases 1.6 tmes greater compared wth full sze pads. The steel component connecton reduces supportng moments 31-38 % (N = 12-167 kn, H = 2 kn) compared wth full sze bearng pad frame and respectvely 65-71 % compared wth hnged connecton. For shorter eccentrcty respectve values are 27-29 % and 63-67%. Fg. 1. The sum of the supportng moments of the columns n a frame, when N = 12 kncolumn. 46
- 18-16 - 14-1 2-1 M[kNm] - 8-6 - 4-2 (' 1 2 4 ~ s!'7 7 ~~,~""" --- -- - v -!---- --- ---- -~ ~ - v-------- #1...--------- F------ - 1 5 H[kN] 6 Fg. 11. The sum of the supportng moments of the columns n a frame, when N = 167 kncolumn. Theoretcal calculatons The deflectons accordng to the analysng method are represented n fgure 12 and the supportng moments n fgure 13. Before these loadngtests the columns were cracked as result of shrnkage and earler loadngs. Therefore the rgdty of the column s taken here from the tests. As can be seen, the effect of connecton to the deflecton and supportng moment s smlar also n theoretcal calculatons. 47
r- 5 H[KN) J +-----f------+---~+-----+-------1----~ 5 1 15 v(mm) 2 25 JO Fg. 12. Deflecton n tests (sold lne) and accordng to the analysng method (dashed lne). -18-16 -14-12 -1 H[KNm) -8-6 -4-2 Fg. 13. 48 l,j.' v >;.(; l.. v 2 '..'.... E. v J.. 1 ~ :' p 5 H[KN) Supportng moment n tests (sold lne) and accordng to the analysng method (dashed lne).! 6
CONCLUSONS Tests wth the rc portal frame show that the effect of the connecton between a beam and a column s sgnfcant to the rgdty and the supportng moment of the frame. The full sze bearng pad connecton reduces deflectons over 6 % and supportng moment about 5 % compared wth respectve values wth hnged connecton (N = 12-167 kn, H = 2 kn). Also tests show that the bearng pad should be full sze type: the type where the column top s prepared to support two beam ends reduces respectve deflectons about 3 % and supportng moment over 2 %. New steel component connectons proved to be worth of developement: deflectons reduced over 8 % and supportng moment over 6 % compared wth hnged connecton. The result does not seem to be senstve for eccentrcty. t s qute obvous that the connecton type should be taken nto account n dmensonng a cantlever column. 49
REFERENCES Keronen, A. 1984, RC Column n a Portal Frame. (n Fnnsh). Tampere Unversty of Technology, Department of Cvl Engneerng. Dvson of Structural Engneerng. 119 p. Keronen, A. 1991, RC Portal Frame wth ntegral Footng. (n Fnnsh). Tampere Unversty of Technology, Department of Cvl Engneerng. Dvson of Structural Engneerng, Report 45. 156 p. Lndberg, R. 1987, Actual Behavour of a Beam to Column Connecton n a Renforced Concrete Portal Frame. Tampere Unversty of Technology, Publcatons 46. 82 p. Asko Keronen, M.Sc. (Eng), Assstant Dvson of Structural Engneerng Department of Cvl Engneerng Tampere Unversty of Technology 5