Effect of holt preloading on joint bebavioue 1. Jaspart MSN, University of Liege Belgium R Maquoi MSN. University of Lidge, Belgium ABSTRACT: In Furocode 3 Part 1-1 as well asin most ofthe existing models for prediction ofthe ess and strength properties of structural bolted joints, the effec of bolt preloading is disregarded. Inthe present Paper, the influence of this Tact om the characteristic joint deforma iy curves is pointed out and related formulae for stiffness and stength prediction are proposed | INTRODUCTION Structural unstifened bolted joints -baam-o-eolumn Joints, beam-to-beam joints, beam splices. ~ are very "economical foe fabrication and erection Generally, such joints have a semiigid behaviour snd are parially resistant. The prediction of thie rotational response, in terms af moment M relative rotation @, is therefor of particular importance in view of the study of their influence on the global structural response of the building frames Inthe most recent analytical procedures for the prediction of the joint response Caspart 1991 Revised EC3 Amnex 1 1994), the so-called component method (Jaspart etal 1994) is used. The ‘ofthis metho isto consider any joint not 5 a whole but as a set of windividyal basic componentsy. In the particular caso of an unstffened beam:to-clunn joint with extended end plate connection subjected to bending, the televant components are the following: = compression zone + column web in compression; + beam flange in compression: + column wob in tension; * column flange in bending ; bolts in tension 5 end plate in bending 5 + beam web in tension = in shear zane + column web pana in shear Fach of these basic components possesses it fown lovel of strength and siffaess in tension, compression oF shear and may therefore be characterized by a F-A curve where F is the applied force and A the cated deformability. The coexistence of several components within the same joint element ~ for instance, the column seb which is simultncously subjected 19 compression (or tension) and shear = can obviously lead 10 stress interactions that are likely to decrease oth strength ‘and stiffness of each individual basic component; ‘this imeration affects the shape ofthe deformability ‘curve ofthe related components but dees not eal the principles of the component method in question component mathod requires the following steps 8) listng ofthe components that are “activated in the joint 1) evalusion of the stiffness andlor strength characteristics of each individual base component (specific. characteristics ~ inital stifness, design strength... ~ or whole deformability curve); ©) assembly» of the components in view of the ‘evaluation of the sfiTaese andlor strength sharactersties of the whole joint (specific Characteristics - initial stiffness, design resistance. - or whole deformability curve) ‘The «assembly» is based on a disibution of the intoral frees within th joint. AS a matior of fact ‘he extemal loads applied to the joi distribute, at ech loading step, between the individualcomponentsthey do that in accordance with the instantaneous stiffness of each component The application of the component method requires ‘obviously a suficien knowledge of the behaviour of the basic component ‘The framework of the component method is suiicionly general 10 allow the use of various feehnigues of component characterization and joint ‘essen, Tn particular, the stifess and strength characteristics of the components may result fom cexpariments in laboratory, mumerial simulations by means of finite element programs or analytical ‘mavels hase on theory. In Lidge, experimentations ‘nd numerical simulations have been performed and ‘used references when developing and validating lytical models Jspart 1991), ‘The analytical modals may be developed with different levels of sophistication according to the parsons 10 whom they are devoted: the expressions presented in (sspart 1991) aro able to cover the influence of all the parameters Which affect significantly the component behaviour (strain hardening, bolt head and net ‘dimensions, bolt presresing,..) since the beginning of the loading up to collapse and ft therefore well with a scientific publication + the rules which ave been introduced,” for instancs, in Annex J of Eurocode 3 (1993) — nex devoted to joint desipn- and in its revised version (1994) to which the University of Lidge has largely contributed with the University of ‘Aachen (D) and TNO Delft (NL) ate fat more simple and are therefore more suitable for practical use, imilar loves of sophistication exist also for what ards the joint «assembly». In the present paper, attention is paid 10 the tension zone ofthe bolted joints and more especially to the plate components subjected 10 transverse forces such end plates or columa flange, including bolts in tension “Two general appronches termed respectively plate ‘models and «T-stub model» have been proposed to analyse these components. In the eplte model, the ‘components are considered as plates in bending and studied accordingly while the Testub idealization ceonssts in reducing. the components to T-stub sections of appropriate length, that are connected by their lange onto a presumably infinitely rigid foundation and ate subjected 10 a uniformly isubuted force acting inthe web plane (igure 1) ‘Tho T-stb idealization allows to darive relatively simple prédiction rules for stiffness and strength hich can similar be applied to end plates, column Fanges or cleats; forall these reasons, it has been ‘uted by diferent authors and is also referred 1 in ECS Annex J (1998; 1994) Figure |= Testub In the present paper, the effeet of she bolt ‘resresing on the tensile eesponse of the T-stu is ‘contomplated. Tt influences the stiffness and strength properties of the Testubs but is still disreparded in the revised FC3 Annex J (1994). Indeed same design rules are given for joints with preloaded and non- preloaded bolt. 2. EXPERIMENTAL AND. DESIGN T-STUB DEFORMABILITY CURVES ‘A Testub tested in Inboratory exhibits a non linear behaviour, as shown in figure 2. The corresponding F-A curve may be characterized by 4 main Parameters the initia resistance F, the stain ‘he ulti theoretically defined as the maximum tension force developed by « T-stub made of a material which ‘exhibits an elastc-perfecly plastic sieststeainFigure 3 - Collapse modes ina T-stub, diagram; thus sean-bardening would be fully disregarded, To identify such a eapacity on an experimentally recorded F-A curves neither obvious nor easy because of unavoidable st coffee. Hovr 40 define the experimental «plasticn fonacity is diseussed elsewhere (Inpart 1991); ‘sun present paper, itis given as shown in figure 2 In BC3 Annex J, design rules are provided to derive a so-called design F-A curve. When applied to an actual Tstub tested in laboratory, Le. on the basis of the measured geometrical properties, ‘the measured mechanical properties of ste) + partial safety factors equal 10 1.0, ‘the BCH design curve should idemiy itself v9 tho “olastc-perfcly plastic” curve dofind in figure 2; indeed strain-hardening is disregarded ‘The BC3 design curve is fully derived from the sole knowledge of two main characteristics: the initial sifness K, andthe design resistance Fye Pa, which is supposed to be equal to the "plastc’ F +m aforementioned conditions apply, is depend ‘onthe collapse mode of the T-stub, ‘The later ean be due to (figure 3) 8) bot fracture with no prying forces, as a result of a very large stiffness of the T-stub ange, oF by onset of yisd liner machsniem in the T-tab before the seength of the bolts be oxhausted, or ©) mixed collapse involving yield lines atthe toe of the fillets in the T-stub and exhwostion of the bolt strength. ‘The related Fa values are expressed as follows (4 bolts in tension): ‘Mode a: bolt fracture (igure 3) Fay 4 By o Mode b: plastic mechanism (Bigue 30) Pygy tee ® ‘Mode e: mixed collapse (igure 3e) 2b, m, +4 By Fg ee 8 Bm eo where om, oe 5 (= ange thickness; f, = stool yield stress; Y= partial safer Factor) and a! 11,25 m, Byqy is the design resistance of the bots and by ~ see figure I i defined in accordance ‘withe ECS, ‘The design strength Fy of tho T-stub is Gerived asthe smallest value got from expressions orton 3. EFFECT OF BOLT PRELOADING ON T- STUB RESPONSE. “The effect of bolt preloading on the F-A response of| ‘8 T-stb is pointed out in figure 4 which relates to test cently performed in Lidge on quite similar T= stubs connected respectively by means of preloaded ‘and non-prelonded bots. Tests shove that Tus sigiicative inucace stiffness and on the "plat resistance of the connection and; non significative influence on tho ultimate stato due tothe completo lost of preloading between the connected flanges at that load level ‘The effect of bolt preloading is disregarded in ECS; the use of ECS design rules fr joins with non preloaded bolts can therefore lead to Undorestimations of the initial stiffness and of the design strength In (Jaspart 1991; Jaspart etal 1991), numerous comparisons between experimental results on‘extended end plate connections and theortical Py values computed according 1 the 1993 draft of EC ‘Annex allow to conelude tha 4 good agreement is found only when the tension capacity af the T- is governed by collapse mode n or ¢ (figure 3). When a plastic mechanism forms in the Testud (collapse mode b), ECS is found much 100 safe Similar conclusions can he drawn from the tests on fush end plate connections (Moore 1988) as well as from the tess of figure 4 where mode b is relevant ti + Now PRELOADED =H PRELOADED S010 150 700. Almmi Figure 4 Effect of bolt preloading on the T-stub stesponse Inthe two following. sections, the necessary modifications 10 bring 1 the existing prediction design rules For the initial stiffness K, sn the design resistance F, are presented 4. EFFECT OF BOLT PRELOADING ON THE DESIGN RESISTANCE [Refinements can be brought 10 the T-stub model of ECS with the result hat the amended model provides 4 higher resistance for collapse mode b without affecting significant the accuracy repatding both collapse modes a and e. ‘An attempt in tis respect has been made by the junior author Gaspart 1991). In most ofthe existing ‘models, the Fores inthe bols ae alvays idealized fs point loads. Thus, it is never explicitely accounted for the actual sizes of bolts and washers, fon the one hand, and on the degree of bolt preloading, on the other hand. Care taken of both Aspects should influence firs, the location of some of the yield lines forming the plastic mechanism, and Second, the contebution of the external loads tothe virtua! 0+ olive to this mechanism, In the T-stub model, the plastic mechanism is composed of patalla straight yield lines which ‘lvelop in the flange of the T section. Two af them fre alvays located atthe toe of the fillets. The two ‘thers are loeted inthe vicinity ofthe bolt rows ithe they coincide with the axes of the bolt rows (Zoctomeijr 1974) - what is done in ECS -; that ‘means the bolt size is fully disregarded and the load is applied in the ais of the bolts (ig. 5.8). Or bolts and washers are assumed so sti thatthe yield lines fave forced 10 develop atthe inner extremity of the boltivasher diameter (Kishi ct al 1988), where the bolt foad is also assumed to bo applied (ig. 5.0). [None of thase models is in very far agreement with ‘experimental observations. Indeed the lines of maxim curvatare are tually not straight but slighty cuted and their pattern inthe close vicinity of the bolts is found to depend on the stifess of bots and onthe degree of bolt preloading (fig. 5.) For prctes purpose, ane cannot imagine to account for such a complex actual pattern, It must be notice tha, for well proportioned connections, the Yield lines’ are not far from complying with Zostemeijer’s assumption (1974), it is therofore justified to refer tothe latter for what regards the location of the yield lines However account will be tnken of the bolt size; it wll be assumed that the bolt load exerted onto the T-stub Mange is uniformly distributed over a certain. length D located symmetrically with respect othe bolt axis (fig. 5b): D meas the diameter of the bolt head/screw of ‘washer, Of course the locaton of the yield lines does no more coincide necessarily with the section of fraximam bending monvent and results in a non- ‘compliance withthe Fundamental theorems of plastic ‘design; the authors are of the opinion thatthe eror emalins sufficiently small to. be acceptable. “Accordingly half ofthe force inthe bolts devalops a negative external work when the forms with the result of an expected. higher connection eapacty compared to EC3 model, For sake of simplicity the resultant bolt loads 2B is substitated by two equal statically equivalent load ‘components B acting aa distance + e= 0.25 D from the bolt axis (ig. 6) ‘Applying the principle of virtual work to above plastic mechanism, on the one hand, and the ‘equations of equilibrium, on the other hand, provides the limiting force Fyyy associated to collapse by onset of aplastic mechanism: yy = C! = 26), m/l2 mn! ~ em #04) A)©. Figure 6 - Forces ating on the T-tub vith © n= min fu 125m) o m= 0.25 o bin aecordance with ECS, OF course, eq(4) confines iselt to Zoctemeijers (1974) and BC3 (1993) formulae when distance is Vanishing, Formula (4) has been included in the revised Annex J of EC3 (1994) a a less conservative altemative to formula (3) ‘What is sid above is not explicitly influenced by the degree of bolt preloading. Actually the free in preloaded bolts ofan elementary assemblage subject tan increasing external load N, (parallel tothe bolt =) evolves according 1 figure 7, First, che bolt tension increase is ® reduced proportion of the extemal load because of the compensating effect of the reduction in plate compression 2C. When the Figure 7 ~ Bolt foree evolution ter becomes equal and opposite to the inital preloading force ofthe plate (Ny = Nao. the plates start to separate and the system becomes statically eterminste. At any higher load, the bolts ‘experience the whole Toad N,, Separation occurs at Ny *29/K* o where § is the preloading force per bolt and K" = (UCL + 16). The factor & = AVA, is the ratio between the axial stiffaess of the effective plate compression area A, and the resisting bolt cross- sectional area Ay; is taken as 3 as an average value {Agerskov 1978), Proceeding as before with attention duly paid to the effect of bolt preloding yields theamended expression ofthe limiting force yy iy (Un! ~201 = Ke} hy m, + eS Iam! ~e(l = Ko +89) nr the reservation that i) 2B (yn! +2b, ma! 2) Ny) because equ(8) is valid in the range prior to plate Separation. Should condition (9) not be fulfilled, then reference is made to (8) where K° = 0, what resultsin Flas = Fig, With Fy, given by equ), ‘The limiting force according to (8), subordinated to the addtional check (9), constitutes a refinement of the relevant EC3 design rule. Tt is exsy 10 demonstrate Jaspart 1991) that above bolt eects do 1 at all alter the strength capacities associated respectively tothe two ather collapse modes ofthe Test 5, BFPECT OF BOLT PRESTRESSING ON THE INITIAL STIFFNESS ‘The initial stiffuess is rolated to the elastic deformation of the T-stub flange in bending and the bolls in tension. When to T-subs are connected together ~ a, for instance, the column flange and endplate idealized T-stubs in an endplate beam 0 column connection - the initial stiffness of each T stub can not be evaluated independently. As & matter of fact, the bolts in tension belong 10 the 10 Tetubs and, through their deformabily, the compatibility of the respective deformabilties of the Testub flanges is ensured, In (Jaspart 199), expressions providing the elastic inital stiffiess of each Testub have been proposed; they allow the coupling effect between the T-stubs to be taken into consideration. When non preloaded bots are use, the elongation A, (gute 8) of the bolts simply results from the elongation of the bolt shank ‘subjected to tension, On the eontray, when bolts are preloaded, tansion frees result moro in a decrease of the compression forces between the connected plates than in an increase of the bot tension Free, a stated in section 4, The "decompression" stiffness of the plates is much higher than that of the bolts in tension, ‘The relative stiffaess between the plates and the bolts has been expressed in section 4 in the foomat of a & factor, Preloading therefore results in ‘higher global sifness oF the T.stub. Formulae for ‘he B -A, deformability a he bot level have been proposed by Agerskov (1976): ~ for non prcloaded boli 8 2k, A ra ) (10) foe proton bts But) 1 EA, OD where (igure 9) oh 143), +07, ky =], #143), #0911, + O41, Figure & -Testub deformability In the revised EC3 Annex J (1994), the laste sihesses of the rlovant T-stubs and bolts to expressed in arder to avoid complexity due to coupling effects, Expression (10) has been approximated by the following 4B as «2 with Lt Ha thet According to EC3, this expressionhas also to be used for joins with preloaded bolts, even if it onlytension. Based on the Agerskov's proposal (1976) to approximate the value of the plate-o-boltstifness ratio to 5, it may be demonstrated that the following simple expression may bo substituted to ay: ayy ‘Through this expression, the influence of the bolt preloading on the T-stub inital stffaess could be simply taken in consideration in the revised! EC3 Annes 1 (1994), 6. NUMERICAL APPLICATION In figure 11, the rovised ECS Annex J (1994) has ‘heen sed in combination with above formulae (8), () and (13) to derive the deformability moment rotation curve of « heam-to-column jint with Mange cleats (igute 10) and praloaded bolts, Ths joint has "tested some years ago at Sheffield University (Davison eta 1987), Besides the experiment thre analytical curves afe reported in figu “They correspond respectively 0: + the revised EC3 Annex J (1994) where the fores in he bots are idealized as point loads - formulae (1)@2) and (3) for strength evaluation = + the revised EC3 Annex 1 (1994) where the forces in the bolts are assumed to be uniformly distibuted under the bolt head/nuvwasher ~ formula (1),(3) and (4) for strength evaluation -; + ease similar to the previous one but with bolt preloading taken into consideration through Formulae (8) for strength and (13) for stiffness, ‘The effect of the bolt preloading on the joint rengh is found to be quite significant in such ‘ase where the cleat in tension isthe weaker part of ‘he joint and fils through Mode 1. The increase of i Figure 10 Joint with Mange cleats ‘he ini stiffness is less marked, This is clearly justified by the relatively small contribution of the bolt deformation (or plate “decompression") to the total deformability of the joint the later results ‘nainly from the clest and column flange deformation ‘The good agreement between the experimental ‘curve andthe analytical one taking) the bolt preloading into consideration has 10 be highlighted Many other comparisons with results of experimental tests performed in various european research contos ‘are reported in (Jaspart 1991). 7. EFFECT OF THE BOLT PRELOADING ON ‘THE SHAPE OF THE ANALYTICAL JOINT DEFORMABILITY CURVES In the revised annex J (1994), the secant titness of joint momentsoation curve corresponding, 10 the design moment resistance My, is defined as 8,4/M ‘where §,.,i8the joint initial stiffness and 1] = 3 {Tor welded joints and joints with endplates) of 3,3 (or joints with flange cleats). The value 1) = 3,5 in ECS 's valid for Mange elated joins where bolts are non preloaded or where bolts are preloaded in such away that no slip curs between the cleats and the beam Manges before Myyis reached, nthe first ease, the stiffness calenlation inclndes tha bearing eformabilty of tho cats and the beam flanges a5 ‘well asthe shear deformability of the related bolts ‘These deformability sources ate disteparded in the second case, In the particular case of figure 11, the bolts are pralosdod bu he slip resistance ofthe joint is lower than May When slip occurs, the beating deformability of the plates and’ the shear deformability of the bolts have to be added 10 the ‘other Joint deformability sources, then the secant joint stfitess. corresponding 10. Myy is found ‘Sghificantly lower than S,,/3,5. In (aspart 1991), more refined values of 1 have been proposed toFigure 11 = Comparison ‘ote sich eases a3 shay et ella +14 sf 02 sa, <1 In (14), @ is the rato between the slip moment resistance Of the joint and May. For eases where 0, F 1, ECS suggests a value of = 3,5 while a value of 1) = 3 is obtained through formula (14). This is justified in spat 1991) based on comparisons with ‘numerous test results Fore valies lower than 0,2 ‘he bolts may be considered as non preloaded Formula (14) has been used to derive the analytical curves of figure 11 a4 F CONCLUSIONS In this paper, the effect of bolt preloading on the itness and strength properties of structural joints is Pointed out. Simple formulae are suggested which col be used co implement the possibilities offered bby the revised Annex I of Eurocode 3. which disregards the inflence ofthe bolt prstressing, An application of EC3 in combination with the hein proposed formulae is performed in order to highlight the possible benefit to be drawn fiom. the consideration of bolt preloading. Lastly itis refereed to another document where the new proposed formulae have been valideted through comparisons with numerous test results than experimental test REFERENCES AAguskoy, H. 1976, High treagih bolted connections ‘subject o prying ASCE Stet Division, vo. 102, nO STH, pp. 161-175 Davison, HB, Kiroy PA. and Nethereot DA. 1987 Rotational stiffness characterises of stel beamto- ‘column connections. Jl of Consntctonal Stet Reserch, v8 pp. 17S, rocode 3 1993: Design of Stel Structures, Pat 1 (General Roles and Roles for Buildings, European Prestindard - ENV 1993-1 Jaspac, P. 1991, rade de In semisigitité dex ‘ssomblages et de son iuence sur la éitnce et la Stable des ossatues en acer. Ph.D. Thesi, University of Lige, 1991 Saspart, HP. and Maguci, R. 1991, Pate capacity of endplate and flange leted connections. Second Intemational Worshop on Connections, Pesburgh, US.A, 10-12 Api pp. 225.238 Jaspat, Pad Maguol, R. 1994, Prection ofthe semi- fig and partal-stength popetes of structural jos. ‘SSR Anmul Techical Serton att Mesting, high USA. Kish, Na Chen, W.R, Matsuoka, K.G, and Momachi, ‘SG. 1988, Momentotaion rlatonsip of top-end seat angle wth double web-angle connections, Elsevier ‘Applied Se. Publ pp. 121-134, ‘Moat, DB. 1988, The design of endplate comectons. TRL Wood Memadal Conference, Garston, Watford Revised Annex Jof Euroende 3 1994, Document voted ‘by CEN and being now edited. ‘oetennijer, P. 1974. A design method for the tension ‘side of ‘stically loaded bolted. bean-o-columa ‘connections. Heron, Deft Univ, vol 20, NO 1