Journal of Constructional Steel Research 95 (2014) 242–255

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Fuse performance on bracing of concentrically steel braced frames undercyclic loading

Frederic Legeron a, Etienne Desjardins b, Ehsan Ahmed c,⁎a Department of Civil Engineering, University of Sherbrooke, Quebec J1K 2R1, Canadab AECOM, Montréal, Québec H3B 0A1, Canadac Department of Civil Engineering, University of Sherbrooke, J1K 2R1, Canada

⁎ Corresponding author.E-mail addresses: Frederic.Legeron@Usherbrooke.ca (F

etienne.desjardins@aecom.com (E. Desjardins), A.Ehsan@

0143-974X/$ – see front matter © 2014 Elsevier Ltd. All rhttp://dx.doi.org/10.1016/j.jcsr.2013.12.010

a b s t r a c t

a r t i c l e i n f o

Article history:Received 21 February 2013Accepted 21 December 2013Available online 25 January 2014

Keywords:DuctilityConcentrically braced framesFuse designConnection capacitySeismic demandCyclic loading

This paper describes the development of ductile fuse system for steel angle bracings to reduce the seismic de-mand to the connections of concentrically braced frames. In such type of structure, the connections often needreinforcement to resist the tensile capacity of the bracing to comply with the capacity design procedure recom-mended by most design codes. In this research, various models of ductile fuses consisting of smoothly reducedangle section are placed on the X-bracing of a braced frame. The fuses are designed to reduce the tensile capacityof bracedmembers to the capacity of the bolted connections. To evaluate the performance of the fuse system, twodifferent thicknesses of single angle members with leg width of 64 mm part of an X-bracing are tested cyclicallyin a 1.75-m high frame representing a building story. It was observed that the braced frame with fuse could beused to reduce seismic load demand to the connections sufficiently to avoid connection strengthening thatwould result from the application of the capacity design principles. It was observed that properly designedfuse system in braced frame showed stable hysteretic response under cyclic loading and maintained adequateductility with a reasonable compromise on the compressive strength of braced members. Finally, based on thestudy results, most efficient fuse patterns are identified for practical design applications.

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Concentrically braced frames are one of the most common lateralload resisting systems for steel buildings and have been studied exten-sively for performance evaluation under seismic loading. Results fromprevious research investigations [1–3] and damage to concentricallybraced frame during past earthquake show that severe local bucklingand premature net fracture of the brace in the connection areas signifi-cantly decrease the effectiveness of concentrically braced frame withinits inelastic range of behavior. The response of concentrically bracedframe in resisting seismic loads is governed by the performance ofbraces and connections subjected to reverse cyclic loading. To obtainthe desirable performance from concentrically brace frame, the bracesmust fail first by showing acceptable ductility after several cycles of in-elastic deformation including stretching in tension and buckling incompression. Past research [4–6] shows that concentrically bracedframes can provide good seismic performance if premature fracture ortearing of the brace connection is avoided. Hence, guidelines havebeen produced in different codes of practice for the design of the bracesand connections to give a desired capacity under seismic events. These

. Legeron),Usherbrooke.ca (E. Ahmed).

ights reserved.

codes [7–9] require that the connection is ‘stronger’ than the brace;therefore, the brace will fail before the connection. In other words, thefactored resistance of the bracing connections must exceed the axialtensile strength of the bracings members, AgRyFy, where Ag is the crosssection area of the brace, Fy is the yield strength of the brace and RyFyis the probable yield strength of the brace accounting for variation inyield strength of actual members. In the case of brace made withangle members, unless the steel ultimate strength (Fu) is considerablylarger than the yield strength (Fy), the effective net area, An, wouldhave to be greater than the gross area to respect this requirement. Forexample, with typical values of An/Ag = 0.8 and shear lag effect to re-duce connection strength by 20%, and Fy = 300 MPa, according toS16-09, RyFy = 385 MPa, and the required Fu would be 601 MPa forthe failure to happen in the brace instead of the connection, whereastypical values of Fu that are used in design is 450 MPa. In many in-stances, this requirement implies the need for reinforcing the brace atthe connections. The drawback of such reinforcement is that it adds sig-nificant cost and complexity to the connections.

The design of a tension–compression braced frame is usually con-trolled by the compressive resistance of the brace member, which ismuch less than the tensile resistance. The tensile resistance of brace istypically larger than what is required by calculation. The strengtheningof the connection is not due to computed loads but instead to the appli-cation of the capacity design method and the actual excess of tensile

Table 1Full-scale test specimens.

Test ID 64 × 64 × 13 Ag = 1443,rz = 12.16 (Group 1)

Remarks

R13 Without any fuse(reference Test)

F13-1 Fuse at ends, on both legs(2 fuse system)

F13-2 Fuses at ends, on boltedleg (2 fuse system)

F13-3 Fuse at ends, on un-boltedleg (2 fuse system)

64 × 64X × 9.5 Ag = 1110,rz = 12.15 (Group 2)

R9.5-1 Without any fuse(reference test)

R9.5-2 Without any fuse(reference test)

F9.5-1 Inner and outer fuses onboth legs (4 fuse system)

243F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

capacity in the case the member is designed for its compression capac-ity. Reducing the cross section of the member on sufficiently smalllength in order to limit reduction of compression strength would resultin a reduced tensile strength and therefore limit the need for connectionstrengthening. The concept of incorporating a fuse in bracing to reducetensile strength to the level strictly required by calculation is thereforeinteresting and has been investigated in the past [10, 11] on HSSbrace. It was shown that the ductility demandwith the reduced sectionbrace was localized around the hole of the brace. This resulted in veryhigh local material straining, but given the short brace length that thisplastic straining occurred over it didn't result in a system ductility(drift of entire bracing system) that was sufficient in comparison withwhat was expected given the R values used in design.

In this paper, a simplified fuse system is developed for single anglemembers. The designed fuses are capable of reducing strength demandson connections while maintaining the load carrying capacity and ade-quate ductility in the braced frame system. The performance of fivefull cross single angle bracing with fuses is compared to the behaviorof three cross angle bracings without any fuse. All X-braces are subject-ed to cyclic loading up to failure. Recommendations for design applica-tions are presented based on the experimental study.

F9.5-2 Inner and outer fuses on legsintersection (4 fuse system)

2. Experimental program

The experimental program consists of eight (8) tests on single storyconcentrically braced frame subjected to the effect of cyclic loading. Theload is applied at the top right corner on the frame of the bracing sys-tem. The cross bracings of the braced frame are single angle memberswhich are connected to each other at mid-point and at the ends to theframe by gusset plates assembly. Two different sizes of angle membersare used in the experimental program. In each test similar sizes ofangle members are used as cross bracings. In the fuse equipped tests,the fuses are placed on the anglemembers near the connections. The ef-fect of designed fuses on the performance of concentrically bracedframe is evaluated from the experimental study. Fig. 1 shows the di-mension of test frame assembly with crossed angle members andTable 1 tabulates the test programs. The tests R13, R9.5-1 and R9.5-2are considered as the reference tests (tests without any fuse system).In other fuse equipped tests, the locations of fuses were empirically

Fig. 1. Dimension of the fr

chosen (see Table 1) with the objective to limit reduction of compres-sion strength while achieving the goal of connection protection.

2.1. Design of specimens

Two different angles 64 × 64 × 9.5 and 64 × 64 × 13 of normalgrade steel (350 W) were used for the preparation of specimens. It isobserved from Fig. 1 that the length covered by the angle member inthe test frame is 2657 mm (center to center of frame). The radius of gy-ration for the angle members about the weak principal axis, r, is12.15 mm. As the two cross bracings are connected at their mid point;assuming an effective length factor, K of 0.5, the slenderness ratio ofthe bracing KL/r is109. The calculated plate slenderness ratio of anglemembers, b/t, where b is the width of leg and t is its thickness, were al-ways below 145/Fy0.5 in order to comply with Canadian standard [7] for

ame with X-bracings.

244 F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

the chosen ratio of KL/r. Any thickness below 9.5 mm would not havemet this criterion. It is worth noting that the AISC [8] imposes a slightlyhigher limit of 170/Fy0.5 for this limit.

The cross bracings were connected to the gusset plate through three3/4″ A490 bolted connections. The gusset plate assembly in the framewas designed to allow the out of plane rotation according to Canadianstandard S16-09 [7] and AISC 2010 [8], by providing clear distance offree rotation of 13 mm which is equal to two times the gusset platethickness. All connections plates were sand blasted to reduce connec-tion slippage during cyclic loading. The sand blasting is also requiredby code CSA S16-09 [7]. The objective was to control the quality of thesurfaces of connected plates so that slips of plates can be minimizedduring cyclic loading.

The tensile capacity of the bracedmember is governed by the tensilerupture in the net section and accordingly, the following equation isused to calculate the un-factored tensile strength [7]:

Tr¼ 0:85Ane Fu ð1Þ

where, Ane = AnU, An is angle's net area and U is the shear lag factor forconnections to tension member.

The code CSA S16-09 [7] recommends the use of a shear lag factor of0.60 for single anglemember connectedwith one row of three bolts. Forfused equipped anglemembers, the large reduction of section due to theuse of shear lag factor of 0.60 may result early buckling in the fuse. Con-sidering this, the following equation [9] was used for the calculation ofthe effective net area, Ane:

Ane ¼ An 1− el

� �: ð2Þ

In the above equation, l, is the length of connectionmeasured centerto center of first to last bolt of the connection, An is the angle's net area;and e is the connection eccentricity measured perpendicular fromconnection plane to the centroid of the member section resisting theconnection force, as defined inAISC 2005 [9]. By using the above expres-sion, the value of Ane is about 85% of An instead of 60%, thus an increaseof 40% in strength is obtained.

The compressive resistance of the bracedmember was computed byfollowing the expression given by Eq. (3), as used in Canadian Standard[7].

Cr ¼Ag Fy

1þ λ2n� �1=n ð3Þ

where n = 1.34 and λ ¼ KLr

ffiffiffiffiffiffiFyπ2E

q.

The compressive resistance calculated using the expression of AISC2005 [9] gives slightly higher value and thus, Eq. (3) provides a conser-vative estimation of compressive resistance of the brace andwas used inthe analysis.

2.2. Design of fuse system

The fuse system is composed of a reduced section of sufficient lengthand it is positioned on a part of the angle members. Once installed onthe brace, the connection does not require any kind of strengthening.

The objectives of the fuse design are to:

– Reduce the load carried by the bracing in tension to the theoreticalcapacity of the connection and hence avoid connection failure.

– Limit reduction of compression capacity as so the theoretical capac-ity of story bracing is affected only marginally and total story capac-ity is sufficient to carry the design loads.

– Achieve an acceptable ductility of about 3 to 4 for moderate ductilitysystem and 2 to 3 for limited ductility system, as consistent withmoderate or limited ductility [12].

Unlike the previous researchwork [10], the fuses in this study are notdesigned to reduce capacity of the bracemember in tension to the capac-ity in compression because preliminary studies showed that a fuse de-signed to reduce capacity in tension to the capacity in compressioncauses a serious reduction in cross section of the angle member. Largeplastic rotation would result from early buckling of the fuse whichwould lead to failure at low ductility and this method was abandonedearly on in the project. In this study, the fuse is designed to have a capacityequal to that of the bolted connection. Thus, the fuse area,Afuse, is equal to:

Afuse ¼Trmin

φFu: ð4Þ

whereϕ is the resistance factor and it is 0.9 for structural steel and Fu rep-resents the ultimate strength of steel. Tr_min is the minimum connectionresistance calculated from:

i) the member effective net area capacity,ii) shear capacity of connection boltsiii) bearing capacity of connection, andiv) tear-off capacity of connection.

Other failuremodeswere also analyzed, but theywere found not crit-ical. The strength of the gusset plate is also important, but itwas designedto have larger strength than the resistance of the bolted connection.

The fuse length is estimated so as a ductility of 3 is reached for thebrace system with a strain in the fuse exceeding 5%. The objective ofductility of 3 is consistent with a moderately ductile system as definedin CSA S16 [7]. Assuming plasticity is spread over the whole length ofthe fuse, Lfuse, the required length Lfuse to achieve the ductility of 3 istherefore:

Lfuse ¼Δall−εyLbrace

ε f−εy: ð5Þ

where εy ¼ FyE , εf = 5% and Δall = 3εyLbrace.

E is the modulus of elasticity; and Lbrace is the length of the bracemember (2036 mm). The value of εf is chosen as 5% based on earlier ex-perience of cyclic tests on steel members [13]. It is a conservative valuesignificantly lower than the ultimate strain of 10–15%, which is general-ly obtained on a coupon tested in tension only [13]. The total fuse lengththus calculated is divided by 2 or 4 to get the individual length of fuse ina 2 or 4 fuse-equipped brace member, respectively. The depth requiredto cut for the fuse element is calculated from the area of fuse and thetransition zones are composed of milling machined circular cuts. Fig. 2shows the dimension and location details of the fuses on the anglebrace members in the fuse equipped tests.

2.3. Specimen properties

Table 2 presents the properties of the test specimens as the averageof the measured yield and ultimate strength on two coupons cut out ofthe member made with the same material and tested according toASTM-E8 testing protocol [14]; and the calculated nominal resistanceof specimens in both tension and compression accounting for measuredmechanical properties of steel. It is to be noted that the calculated nom-inal resistance is not taking into account the effect of fuses on capacity.The table also shows the calculated story strength for frame in tension/compression, Sr(T/C), and in tension only design, Sr(T). The story strengthis calculated by using the following expressions, where θ is the angle ofinclination of bracing with the horizontal (41° in all tests).

Sr T=Cð Þ ¼ 2Cr cos θ ð6Þ

and,

Sr Tð Þ ¼ Tr cos θ: ð7Þ

(F13-1)

(F13-2)

(F13-3)

(F9.5-1)

(F9.5-2)

Fig. 2. Details of fuses on angle members (All dimensions are in mm).

245F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

It is observed that the calculated Tr values were always about 2.3times higher than the corresponding Cr values. A tension only designwould then be interesting to reduce member size; however it would

reduce the story overstrength and tension only brace design is not per-mitted by code AISC 2010 for such special concentrically braced frames.It is observed from Table 2 that Tests F13-1, F9.5-1 and F9.5-2 used fuse

Table 2Properties of test specimens.

Test ID Section size Fuse geometry Mechanical properties Theoretical resistance Nominal story strength

Area (mm2) Fuse rz(mm) Fy (MPa) Fu (MPa) Cr (kN) Tr (kN) Sr(T/C) (kN) Sr(T) (kN)

R13 64 × 64 × 13a 1443 12.16 375 543 198 457 298 344F13-1 857 7.91 372 542 202 457 304 344F13-2 847 3.96 365 539 190 454 286 342F13-3 847 3.96 369 542 200 457 301 344R9.5-1 64 × 64 × 9.5b 1110 12.15 334 526 137 346 206 261R9.5-2 1110 12.15 319 495 130 326 196 245F9.5-1 673 7.72 387 543 151 357 227 269F9.5-2 647 7.51 371 543 142 357 214 269

a 64 × 64 × 13 : Ag = 1443 mm2 An = 1176 mm2 Ane = 991 mm2.b 64 × 64 × 9.5 :Ag = 1110 mm2 An = 910 mm2 Ane = 774 mm2.

246 F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

patterns that have a minimum radius of gyration, rz, only 30% smallerthan that the radius of gyration about the weak axis of the initial mem-ber. The fuse patterns used in tests F13-2 and F13-3 have aminimumra-dius of gyration about 60% smaller than the radius of gyration of theinitial member.

2.4. Test set-up and instrumentation

Fig. 3 shows the typical test set-up and instrumentation used for theexperimentations. It also shows the actuator location and the supportconditions of the frame. Two angle members of similar geometry wereused as X-bracings in a steel frame for the eight full scale tests. Thestory load was applied by a 500 kN actuator at the top right of bracingsystem. In the test set-up, “Brace A” refers to the member in tension inthe first cycle (see Fig. 1) and “Brace B” refers to the member undercompression under the first cycle. Strain gauges are placed on the tophorizontal member of the frame and also installed on the left andright vertical double angles. The strain readings were used to calculatethe force distributions in the braces. Displacement transducers (LVDTsand string potentiometers) were placed at various locations to measurehorizontal, vertical and differential movements of the test specimen(see Figs. 1 and 3). All instrument data channels were connected to acentral data acquisition system, which was in turn connected to a com-puter for data collection. The data acquisition system was high speed

Fig. 3. Test set-up and

portable data acquisition system with a capacity of 1 Mhz. The rate ofdata acquisition was 2 hz.

It is to be noted that the experimental setup was designedwith a re-movable 6.35 mm gusset plate bolted in between two 9.5 mm platebolted at the connection of the frame in order to reduce operation of re-placement of gusset and braces,with the exception of test R9.5-2,whereconventional gusset plates were used instead of removable ones. As theresults of reference test R9.5-2were comparable to that of reference testR9.5-1 with removable plate, it was concluded that removable gussetplate did not alter the results and hence, found suitable for use in allthe remaining tests. In the tests equipped with 4 fuses (inner andouter fuses) F9.5-1 and F9.5-2, small projected bars are placed insidethe fuses, to estimate manually the deformation of each fuse duringthe test.

2.5. Loading protocol

Fig. 4 shows the loading protocol used for the tests. It is a commonlyused loading protocol [15] and the selected loading pattern is intendedto cause forces that simulate high levels of inelastic deformationsthat may be experienced by the frame during a severe earthquake. A±150 kN test cycle was conducted at the beginning of every test toevaluate the bolt displacement in the frame and also to verify the properbehavior of the system. In the following cycle, the bracingwas loaded upto first buckling occurring in Brace B. This displacement at this first

instrumentation.

Fig. 4. Reverse cyclic loading for full scale test.

247F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

buckle is considered as the yield displacement,Δy, and thedisplacementimposed at each cycle, Δ, is based upon this yield displacement.However, this yield displacement considers the frame displacement,due to bolt displacement and slipping, that occurs during every loading.The frame displacement was about 15 mm in most of the tests andtherefore, the actual yield displacement, Δy, was reduced by 15 mm.Thus, the actual displacement imposed in cycle of ductility μ isΔ = μΔy + 15. The bracingswere cycled at 2, 3, 4, 5 times the yield dis-placement up to failure. The speed of loading is indicated on Fig. 4. Eachcycle was repeated once, so two full cycles were applied to each bracingat each level of displacement.

3. Experimental results

3.1. Test results and general behavior

Fig. 5 shows the normalized hysteretic behavior of all eight (8) testsconducted in the laboratory. The normalized load value is the storyforce (SF) divided by the nominal story strength in tension/compression,Sr(T/C), according to the simplified analysis. Results of test R9.5-2 was acalibration test similar to test R9.5-1 that did not provide any additionalinformation, except that the removable gusset plates did not modify thebracing system behavior. In Fig. 5, each graph also displays the designvalues for frame in tension/compression and in tension only. All graphsshow the average maximum force achieved during the reference test(R13 or R9.5-1). Test F13-3 shows very high pinching effect and didnot reach tension only story strength. It is observed that unlike testsF9.5-1 and F9.5-2, test F13-1 and test F13-2 endured many inelasticcycles. Although all the fuse patterns managed to meet the T/C require-ment, it took more cycles for test F13-1 and F13-2, unlike test F9.5-1and F9.5-2 which met it during the first cycle. Furthermore, test F13-1also reached the tension only requirement.

Table 3 summarizes the full scale test results under cyclic loading. Itshows the cycle of failure, the failure mode, the maximum and mini-mum load deployed by the frame during the test, the load required bythe story to buckle chord B for the first time and the ratio of maximumto theoretical story strength in tension compression design for all theeight tests. The ‘rupture’ column of Table 3 shows the loading cycle at

which the system has failed. The first term inside the parenthesis indi-cates the displacement ductility level and the second shows in whatcycle it failed (first or second cycle at that ductility level). Fig. 6 depictsthe failure mode of specimens in tests.

As expected, the failure of all the reference tests occurred due to thefailure at the connection but offer a significant ductility which was notexpected as it is considered in codes as brittle failure. This might bedue to high ratio of ultimate to yield strength Fu/Fy in the anglemember.This ratio has a major impact on the achievement of member gross areastrength or connection failure. It is noted that the failure of fuseequipped test F13-3, was also due to failure of connection at net section(refer to Fig. 6). Failure of connection in this test may be attributed tothe orientation of the fuse that may have induced local flexure in theconnection and may have overstressed the connection. The other twotests (F13-1 and F13-2) of this group had a fuse fracture failure. It wasalso observed that although in test F13-2 the fuse was able to protectthe connection, its performance was relatively weak in compressionand the theoretical story strengthwas barely achieved. This ismainly at-tributed to the reduction of area due to the use of relatively large fuse inthat test. In the second group, the test F9.5-1 and test F9.5-2 show fusefracture failure as F13-1 and F13-2.

Table 4 summarizes the buckling, maximum and minimum stressobtained in both braces for each of the tests conducted. The bucklingstrengths of braces equipped with fuses are significantly lower tothose of respective reference tests (R13, R9.5-1 and R9.5-2).

In the reference test, the brace strength is limited by the connectioncapacity. Ideally, the fuse strength would be just below this value. How-ever, the maximum stress in Table 4 shows that the braces with fusereach similar stress in the brace as the reference test. However, it isslightly lower inmost cases, with a reduction ranging from1% to 10% ex-cept for test F13-3 where the failure was initiated in the connection, notin the fuse. The strength reduction can be explained by the variability inthe steel ultimate strength, as well as plastic rotation of the fuse that re-duces the strength of the fuse, but it also provides some conservatism inorder to make sure the failure is in the fuse, not in the brace connection.

Table 4 also reports the fuse stress (Fbrace/Afuse). It is observed that itis well beyond yielding and around fracture, which was the designobjective.

Fig. 5. Hysteretic normalized behavior.

248 F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

3.2. Dissipation of energy

Fig. 7 presents the dissipated energy as a function of the cyclereached for all tests. The dissipated energy is calculated from the area

of force displacement graph as obtained in the cyclic tests. The valuesof the last cycles have a softer slope as the last cycles were not com-pleted, one of the member having fractured in the course of loading. Itis observed that all tests sustained a loading cycle of at least (3Δ + 1),

Table 3Full scale test results of frame under cyclic loading.

Type Test Rupture Rupture type Max. story load, SF(max)

(kN)Min. story load, SF(min)

(kN)First buckling story load(kN)

RatioSF(max)/Sr(T/C)

64 × 64 × 13 R13 (3Δ + 1) Connection 402 −389 400 1.34F13-1 (5Δ + 1) Fuse fracture 351 −330 212 1.15F13-2 (3Δ + 2) Fuse fracture 292 −322 118 1.02F13-3 (5Δ + 1) Connection 302 −305 50 0.99

64 × 64 × 9.5 R9.5-1 (5Δ + 1) Connection 318 −308 296 1.54R9.5-2 (4Δ + 1) Connection 306 −309 230 1.56F9.5-1 (3Δ + 2) Fuse fracture 307 −255 307 1.35F9.5-2 (3Δ + 1) Fuse fracture 283 −262 283 1.32

249F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

and R9.5-2 and F9.5-1 had the largest energy dissipation at that loadingcycle. The rate of energy dissipation for most tests did not vary muchand remain nearly the same, even for those tests sustained highercycle of loading. Test F13-3 lies definitely outside the range. This maybe due to very low capacity in compression of this bracing system thatreduces the reloading branch of the hysteretic curve.

3.3. Distribution of plasticity

Fig. 8 presents the strain obtained in the fuses during tests F9.5-1and F9.5-2. Unlike other tests, F9.5-1 and F9.5-2 were the only teststhat were equippedwith four fuses, hence considered critical to presentthe results of distribution of plasticity.

In test F9.5-1, the strain of fuses in members A and B did not varymuch especially for the external fuses and the upper external fusesshow more strain compared to the lower ones. It is also observed thatthe external fuse strain was relatively high compared to that of internalfuses at higher cycle of loading.

During test F9.5-2, the fuses of member ‘A’ deformed more thanthose of member ‘B’ (in particular the external fuses). This differencecan be explained partly by the position of the members to the cyclesof loading and also by the properties of steel employed. It should benoted that member A was always the first being subjected to tensiondue to the loading protocol. The force required to achieve a positive(tensile) displacement is typically larger than the force required forthe negative displacement. In addition, test member used as chord Ahad a relatively lower yield stress compared to its counterpart B(371 MPa against 384 MPa). This slight difference in yield stress per-haps foster a faster yielding of member ‘A’ which is accentuated withthe loading cycles. Test F9.5-2 finished during loading towards cycle(3Δ + 1) with a force lower than the cycles 2Δ.

3.4. Distribution of forces in bracing

Fig. 9 shows the distribution of forces in braces A and B for all thetests. The distribution is represented by the ratio of force carried by amember to the total story force taken by both members in a phase ofgiven cycle. In the graph, the positive and negative values representthe pushing and pulling phases respectively. When the assembly ispushed, brace A is in tension and B is in compression, the opposite phe-nomena results when the load is reversed. It should be noted that thepresented distribution is for the force in the jack either reaching a max-imum or minimum for a given cycle. In Fig. 9, two reference lines areused to show equal distribution (0.5/0.5) and singular distribution(1.0/0).

It is observed from Fig. 9 that equal distribution of forces never oc-curred in reference tests and in the tests equippedwith fuses. The refer-ence test R9.5-1 shows the distribution of forces in braces A and B as0.65/0.35 in the first loading and of 0.40/0.60 after reverse loading inthe first cycle. The forces turned quickly from the buckled situation totension in brace ‘A’ due to second cycle lateral pressure. With theincreasing cycle of loading, the braces have the tendency to become

tension only member. However, this tendency is faster in fuse equippedassembly than the reference tests without any fuse.

Fig. 10 shows the stress and strain in brace A and brace B due to ap-plied lateral loading. It is observed from Fig. 10 that in some cases, thereis an increase in compression load between cycles. This may be due totoo conservative estimate of the first buckling deflection applied duringthe first cycle (1Δ + 1). It is possible that the test in the first cycle wasreversed before the full buckling was reached and only the first sign ofbuckling was taken as full buckling. However, there is no loss of com-pressive capacity and it remains close to the value reached during thefirst cycle (1Δ + 1).

It is interesting to note the de-synchronization of forces occurredduring the test cycles (Fig. 10). That explains the difference observedin Figs. 9 and10. In thefirst cycle, the lateral force is taken by the tensionand compression member at the same time. However, with increaseddisplacement, the members in tension take an increasing part of thestory force. In fact, the maximum compressive strength in the braceoften occurs shortly after that the same framehas been subjected to ten-sile stress.When themember is fully-extended for a tensile force, all de-formations caused by previous compression cycle disappear, resultingin a re-alignment of the chord in the plane of bracing. Once, the braceis re-aligned, it is ready to take adequate compressive loads. The differ-ent responses of the members in the compression and in tension to theside loading have not decreased themaximum capacity of the system orthe individual response of each chord in maximum tension.

4. Analysis of the results and discussion of fuse performance

In this section the experimental results are analyzed to address theobjectives of the fuse design. The performance of the designed fuses inconcentrically braced frames is discussed and their effectiveness is eval-uated based on three design criteria. Thewell designed fuse includes theselection of the location of the fuse, the total number and the calculationprocess to reach appropriate strength and ductility.

4.1. Ductility achieved

It is observed from the tests (refer to Fig. 5 and Table 3) that all thefuse-equipped tests showed appreciable degree of ductility before failure.In these tests, the braces in tension yielded and improved the overallductility of the system.However, in all reference tests the failure occurredfrom the brittle failure of the connection. Although reference tests R9.5-1and R9.5-2 showed unexpected ductility, the braces in these tests did notyield before reaching the connection capacity, and ductility camemainlyfrom severe inelastic deformation at the connections.

It is observed from Fig. 7 that in the first group of tests, the un-notched test R13 dissipates the least energy. The energy dissipation oftest F13-1 was very high compared to other two tests in this group(F13-2 and F13-3). In test F13-2, the failure of fuse occurred from localbuckling of braces without the development of significant energy dissi-pation. Test F13-3 lies outside the range and this fuse configuration isnot recommended for use in practice. In the second group, referencetest R9.5-1 dissipates the most energy. The energy dissipation of tests

Fig. 6. Failure mode of specimens in tests.

250 F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

Table 4Brace test results.

Type Test Stress computation First buckling stress (MPa) Maximum stress (MPa) Minimum stress (MPa)

Brace A Brace B Brace A Brace B Brace A Brace B

64 × 64 × 13 R13 Fbrace/Ag −120 −129 330 298 −139 −129F13-1 Fbrace/Ag −65 −78 313 309 −98 −80

Fbrace/Afuse −110 −131 527 521 −165 −135F13-2 Fbrace/Ag −34 −35 269 298 −48 −73

Fbrace/Afuse −58 −60 459 507 −83 −125F13-3 Fbrace/Ag −34 −46 271 277 −54 −46

Fbrace/Afuse −59 −78 462 472 −92 −7864 × 64 × 9.5 R9.5-1 Fbrace/Ag −137 −129 362 364 −173 −151

R9.5-2 Fbrace/Ag −139 −115 327 325 −166 −124F9.5-1 Fbrace/Ag −53 −101 327 307 −111 −108

Fbrace/Afuse −88 −166 539 507 −182 −179F9.5-2 Fbrace/Ag −82 −52 320 311 −87 −66

Fbrace/Afuse −140 −89 550 533 −149 −114

251F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

F9.5-1 and F9.5-2 that used four fuseswere lower than the energy of ref-erence test but theywerewithin the range. Thus, it is concluded that theenergy dissipations of these tests were within the range of limited duc-tility as defined in the code CSA S16.

Based on results shown in Fig. 8, the distribution of plasticity for testF9.5-1 and F9.5-2; in test F9.5-1, the upper external fuses show morestrain compared to the lower ones. The upper external fuses haveexceeded the strain 5% up to about 14%, while the lower externalfuses strain were below 2%. The internal fuses of this test have less de-formation, but it was more uniform throughout the test cycle. As fortest F9.5-2, the interior fuses of brace A deformedmore than their coun-terparts of the brace B.

It is concluded that although some ductility was achieved in all fuseequipped tests, tests with two fuses showed better ductility than thetests equipped with four fuse system because there is no plasticity con-centration. Thus two different level of ductility is reached for the designof fuse equipped braced frames. With two fuses, a moderate ductility isreached whereas X-brace reached only limited ductility when providedwith four fuses.

4.2. Story strength achieved

The story strength is the maximum horizontal force applied to thebrace systemduring the tests. The theoretical story strength is calculatedwith Eq. (6) and (7) assuming Tension/Compression (T/C) and Tensiononly (T) bracing system respectively. It is observed from Fig. 5 and alsofrom Table 3 that the theoretical T/C story strength was achieved in alltests. In test F13-2 and F13-3, the theoretical story strength (T/C) wasbarely reached. In all other fuse equipped tests (F13-1, F9.5-1 and

Fig. 7. Dissipation of energy.

F9.5-2), the actual story strength exceeds the theoretical T story strength(see Fig. 5).

In general, test equippedwith only two fuses (F13-1 and F13-2) offercurves that are flatter than the tests equipped with four fuses (F9.5-1and F9.5-2). However, cracks appeared earlier in tests equipped withfour fuses than for the tests equippedwith two fuses due to yielding lo-calized in one fuse instead of distributed over the 4 fuses (see Fig. 8).This resulted in superior story strength for two fuse systems as com-pared to the four fuse systems. Tests equipped with two fuses offer aconstant growth of normalized load unlike tests on braces with fourfuses. This growth is also more pronounced than the reference tests.Both the reference tests R13 and R9.5-1 exceeded both tension/compression and tension only values (refer to Fig. 5). Thus, the devia-tion of graph peak from the line representing the corresponding refer-ence test story strength gives an estimation of the strength loss due tothe implementation of fuses. Even though all the braces equippedwith fuses lost a significant amount of strength, they all remained strongenough to withstand the tension/compression requirements expectedfrom cross type braced frame. This achievement is mainly due to theload redistribution between the two braces during the loading. Thefuse system also provided an appreciable degree of ductility mainlydue to the brace in tension that would yield and improve the overallductility of the system instead of brittle failure of frame in one of itsconnections.

It is observed from Fig. 10 that for the bracing members equippedwith fuses, the actual compression capacity is well below the estimatedtheoretical capacity whereas for the reference tests it is closer to the cal-culated capacity. In tests equipped with fuses F13-1, F9.5-1 and F9.5-2,the reduction of compression capacities as compared to the referencetest R13 are 27.5%, 28.8% and 44.2% respectively, that did not affectthe story strength significantly. However, in tests 13-2 and 13-3, thecompressive strength reduction was significant and these are 47.1%and 60.1% respectively. This over reduction of compression capacity intests F13-2 and F13-3 reduced the story strength. Thus it is concludedthat in all the tests with well designed fuse system, the story strengthwas achieved (refer to Fig. 5) and the reduction of the compression ca-pacity of braces in these tests did not affect significantly the storystrength. Based on the tests, a reduction of the radius of gyration of30% seems acceptable which also corresponds to the ratio of slender-ness KL/r of fuse to slenderness of brace member of about 0.14.

4.3. Protection of connections

It is observed from Fig. 10 that in all fuse equipped tests, the stress inboth the braces crossed the yield line of the fuse. The crossing of this lineindicates the yielding of fuses. Fig. 10 also shows the theoretical ulti-mate capacity line of braces in both tension and compression. The cross-ing of ultimate capacity line in the positive field indicates the failurewilloccur at connection.

Fig. 8. Strain of fuses in tests F9.5-1 and F9.5-2.

252 F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

In the reference test R13, the yielding of the bracing did not occurand it reached the connection capacity at higher displacement. Accord-ingly, this test failed at connection before the yielding of the bracing. Inthe fuse equipped tests F13-1 and F13-2, the fuses yielded and afteryielding the fuses were able to take more loads at higher displacementbut did not reach the ultimate capacity of the fuse (around 1.45 timesthe yield strength). In these tests, the member stress to the connectionwas well below the estimated capacity line. In test F13-3, although thefuse yielded at lower cycle, it did not perform satisfactorily due to im-proper orientation of fuses. Although, the bracing stress did not reachthe capacity line, the failure of this test at connection occurred due tothe rotational influence of brace member. It supports the previous con-clusion that this orientation of fuse failed to protect the connection andshould be avoided in the practice for fuse design.

In the reference test R9.5-1 or R9.5-2, the bracing stress crossed thecapacity line before reaching the yield stress. Thus in these tests failureof connection occurred, although it showed unexpected ductility byyielding of the fuse at higher cyclic loading which might be attributedto steel characteristics. In fuse equipped tests F9.5-1 and F9.5-2, thestress in fuses not only reached the yield line but also reached its ulti-mate capacity. Thus in these tests, after yielding, the fuses were capableof taking more load while reaching its ultimate capacity. In these tests,higher forces in the bracings were observed at relatively lower cycles.

It is concluded from the above observations that in all fuse equippedtests, the fuses yielded first before reaching the connection capacity andwere capable to protect the connection. The only exception is the testF13-3, where although the fuse yielded, the failure at connection oc-curred due to the orientation of the fuses.

5. Conclusions and recommendations

In this paper, the experimental performances of various fuses inangle brace parts of concentric frames were evaluated. The fuses havebeen designed with a tensile capacity equal to the capacity of boltedconnections in order to avoid a connection failure. Based on experimen-tal investigation, the following conclusions are made:

• In most fuse configurations, the load at connection was lower than itscomputed resistance while providing a full ductile behavior. The fullscale tests also evaluated the performance of various fuse designs,fuses positions and distributions. In general, the fuse pattern usedfor tests F13-1, F9.5-1 and F9.5-2 were the most efficient, unlike testF13-2 and especially test F13-3which behaved poorlymainly becauseof their inadequate fuse geometry.

• Braces equipped with two fuses placed at both ends of the brace (testF13-1)met the tension criteria and endured asmany cycles as the ref-erence test and exhibited moderate ductility. These braces also dem-onstrated constant strain hardening during the loading but showeda significant loss in compression capacity. Braces equipped with fourfuses placed at four places (tests F9.5-1 and F9.5-2), did not lose asmuch compression resistance, but reached their ultimate loads soonerin the test loading and then failed at the level of limited ductility. Thisbehavior could be attributed to the concentration of plasticity in exte-rior fuses and lower dissipation in internal fuses. Even though thebraces equipped with fuses lost a certain amount of strength com-pared to the reference tests, they all met the requirement of atension/compression design. The load redistribution between the

Fig. 9. Distribution of forces in members.

253F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

Fig. 10. Stress and strain in bracings A and B.

254 F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

255F. Legeron et al. / Journal of Constructional Steel Research 95 (2014) 242–255

two braces managed to meet the requirement set by the tension/compression design which was based upon a 50/50 load distribution.

• A braced member with fuse system reduced its tensile capacity to alevel closer to its compression strength. It is thus possible to includebolted connections without any reinforcement for net effective areatensile failure mode because the fuses were designed to have thesame capacity as the bolted connection. The decrease in capacitydoes not affect the energy dissipation and avoid the over sizing ofother transmitting elements of concentrically braced frame system.Design with fuses can therefore comply with the capacity design re-quirements while avoiding the disadvantages of the requirement ofmaintaining the integrity of the capacity protected elements in con-centrically steel braced system.

• It is realized that the effective slenderness ratio is the single most im-portant parameter that affect the compression capacity and theshape of the hysteretic loop of bracing. The effectiveness of designedfuse systems needs to be evaluated for different level of slendernessratio. Furthermore, better estimate for the loss of compression strengthis needed for the successful design of fuses.

• The effect of fuse system on the over strength reduction factor, R0,

should be evaluated in further study so a complete design methodwould be available for the practicing engineers.

References

[1] Astaneh-Asl A, Goel SC, Hanson RD. Cyclic behavior of double angle bracingmemberswith end gusset plates. Research report UMEE 82R7. Ann Arbor, MI: Department ofCivil Engineering, University of Michigan; 1982.

[2] KimHI, Goel SC. Seismic evaluation and upgrading of braced frame structures for po-tential local failures. Report no. UMCEE 92–24. Ann Arbor, Michigan: Department ofCivil Engineering, University of Michigan; 1992.

[3] Tremblay R, Archambault M, Filiatrault A. Seismic behavior of ductile concentricallysteel X-bracings. Proceedings 7th Canadian conference on earthquake engineering,Montreal, Canada; 1995. p. 549–56.

[4] Kahn LF, Hanson RD. Inelastic cycles of axially loaded steel members. Journal ofStructural Division, ASCE 1976;102:947–59 [No. ST5].

[5] Foutch DA, Goel SC, Roeder CW. Seismic testing of a full scale steel building— part I.No. ST11 Journal of Structural Division, ASCE 1987;113:2111–29 [New York].

[6] Aslani F, Goel SC. Report no. UMCE 89–5 , Experimental and analytical study of theinelastic behavior of double angle bracing members under cyclic loading. Ann Arbor,Michigan: Department of Civil Engineering, University of Michigan; 1989.

[7] CAN/CSA S16-09 Design of steel structures. Mississauga, ON, Canada: CanadianStandards Association; 2009.

[8] ANSI/ AISC 341–10. seismic provisions for structural steel buildings. Chicago, IL:American Institute of Steel Construction; 2010.

[9] AISC specification for structural steel buildings. American Institute of Steel Construction;March 9, 2005.

[10] Kassis D. Ajout de fusibles aux diagonales de contreventements pour la conceptionsismique de charpentes métalliques à un étage. Montreal QC: École polytechniquede Montréal; 2008 (251 pp.).

[11] Rezai M, Prion H, Tremblay R, Bouatay N, Timler P. Seismic performance of bracefuse elements for concentrically steel braced frames. Proc. of STESSA Conference,Montréal, QC; 2000. p. 39–46.

[12] Newmark NM, Hall WJ. Earthquake spectra and design. Berkeley, California: Earth-quake Engineering Research Institute; 1982.

[13] Desjardins E, Legeron F, Ahmed E. Performances of ductile fuses in reducing seismicdemand on connections of concentrically steel braced frames. 15th World confer-ence on Earthquake Engineering (15th WCEE), 24–28 September, Lisbon, Portugal;2012.

[14] ASTM E8/E8M—13a standard test methods for tension testing of metallic materials.ASTM Standards 03.01, 2013, 28.

[15] Paultre Patrick, Eid Rami, Ita Robles Hugo, Bouaanani Najib. Seismic performanceof circular high-strength concrete columns. ACI Structural Journal 2009;106(4)(July-August).