Engineering Journal American Institute of Steel Construction 11 Experimental Investigation of Steel Joist Design for Ductile Strength Limit State 21 Experimental Verification of Spliced Buckling Restrained Braces 43 Local Stabllty of Double-Coped Beams 53. ErrataENGINEERING JOURNAL AMERICAN INSTITUTE OF STEEL CONSTRUCTION Dedicated to the development and improvement of steel construction, through the interchange of ideas, esperiences and data Editorial Staff Editor: Kerri A, Grvm, S.E., PE. Research Editor: REIDAR BIORHOVDE, PH.D. Production Editor: ARETI CARTER Officers {TerFREY E. Dave, PE, Chairman Dave Steel Company, inc, Asheville, NC JAMES G. THOMPSON, Vice Chairman Palmer Stee! Supplies, Inc., MeAlten, TX Rocien E, Frc, RE., President American Institute of Steet Construction, Chicago Davin B. RArTERMAN, Secretary & General Counsel American Institute of Steel Consiuetion, Chicago ‘CHARLES J. CARTER, S.E,, PE. PLD. Vice President and Chief Sructural Engineer American Institute of Steel Consiuetion, Chicago IaCQUES CATTAN, Vice President American Institute of Steel Construction, Chicago JOWN P. CROss, PE, Vice President American Institute f Stee! Construction, Chicago Scorr L. Matick, Vice President American Institute of Steel Construction, Chicago ‘The amicls contained herein are not inended to represent official stud, recommendations or policies ofthe Institute. The Ista not responsible for any Statemens made or pins expressed by contributor otis our The opinions of the authors herein do not represent an oficial position of Tnsints, and in every ease the offically adopted publications of the Tat will cont and supersede any suggenions or modifications consid in any tices herein. “The information presented herein is ose on recognized engncerng principles and is for general information only. While itis belcved to be accurate, this infermaton should not be applied to any specific application without competent Professional examination and verification by a ticesed professional engine. ‘Anyone making wc ofthis information sume ll habit rising from sch se. "Manuscript are welcomed, ht pubiaton canna guaranteed, All manuscripts should be submited in duplicate, Authors do no recive a remuneration. A *Cuide for Authors" sprinted on the inside ack cover ENGINEERING JOURNAL (ISSN. 0013-8029) is published quarterly Subscriptions: Members: one subscription, S40) per year. included in dues Aktional Member Subscriptions: $40 per year Non-Members US. S160 per yer. Foreign (Canad and Mexico: Meme 0 per year, Non-Member 6) per ye. Pablished by the American Insite of Stel Constastion a One Fast Wacker Drive, Suite 7H), Chicago I 660, Peviegicals postage paid at Chicago, TL and additonal mailing offies Postmaster: Send adress changos to ENGINEERING JOURNAL in cate of the American Insti of Stel Construction, One Fast Wacker Drive, Suite 700, (Chicago,IL coed. Copyright 2014 by the American Insitute of Stel Consmistion, All rights reserved. No prt ofthis publication may be repealed without writen permission. “The AISC logo isa repistered trademark of AISC, Subscribe to Engineering Journal by visting our website walsecarwle) or by calling 112600.8848 Copies of current and past Bnsincering Journal antics tug avilable ie to members online at wrwalscorgels Non-members may prchase Englcering Journal ance download atte ATSC Bocksore a wirwaiscant for S10 cachExperimental Investigation of Steel Joist Design for Ductile Strength Limit State JOSEPH ROBERT YOST, TIMOTHY J. HARRINGTON, JOSEPILJ. POTE, SHAWN P. GROSS and DAVID W. DINEHART ABSTRACT Open web steel joists are prefabricated truss assemblies designed in accordance with specifications set forth by the Stee! Joist Institute (6d). Curronty, the Sl design raquirement is based on capacity, with no consideration for the governing-member strength limit stato. Tha ‘purpose of ths research sto Investigate a ducte design methodology tor Stoo Jolsts where the primary strength limit stato 's characterized by tension-member yielding and large inelastic deformation, followed by a secondary strength mit state of compression-member bucking To achieve ductile behavior, a series of experimental joists were designed and manufactured using controled overstrength ratios of rela- tive momioer strengths so that tension-member yielding precluded compression-momiber bucking. The consequence of adjusting member strengths to induce ductile failure a slight inereaso in oist weight. The experimental matrix consisted of 18 jist samples: six identical 93-"- Jong K-seres joists, sx identical 33-1-long LH-seresjolsts and six identical 32-f-long rod web joists Al jasts were tested o colapee under simaly supported uniform load conditions. Experimenta results show that a ductile design is achievable because all 18 joists demonstrated tension-member yielding with significant deformation prior to a secondary limt state of compression-mmamber bucking Keywords: ste! joists, strength design yielding, ductile iit state. INTRODUCTION ppen web sicel joists are prefabricated truss assemblies typically used in supporting roof and floor systems ‘The current specification for stcel joist design is published, by the Stee! Joist Institute (SJ1) and titled Standard Specifi cations, Load Tables and Weight Tables for Steel Joists and Joist Girders, 43rd edition (SHI, 2010). This specification is, bbased on the 2005 AISC Specification for Structural Steel Buildings, ANSUAISC 360-05, and has a dual format allow- ing either allowable stress design (ASD) or load and resis- tance factor design (LRFD). The strength limit state for steel joist members occurs by tension yield, compression buck- ling or interaction between axial force and bending moment. osspn Rober Yost. Areociate Prctssor, Cepartment af Ch anc Enaron= ‘mortal Engneeng Vilnova Univer, Vilarove PA cotespondig). Emal Fesepr yoatavianeva ck Timoty J Hannon, Grasute Resoarch Asitirt, Department of Ci 4nd Envronmantl Engnssing, Vilanora Unversty anova, PA, Ema thaetOvilanovaads Joseph J. Pole, Diector of Recexrch and Developmen, New Milenium Balding Systoms, LLC, Hope, AF, Era Joe Potenenml.com ‘Shawn P Goss, Associate Professor, Deparment of Gil are Envrenmen= {al Engneedng. Vilineve Unverty, Vibnova, PA. Ema shaw gone ‘avis W, Dine, Professor, Deparment of Civ and Environmental Engl neonng, Vanova Urverty, Viana, PA Emit david. dnehanilanora Presently, the SJT acknowledges four main joist types: K-series joists, longspan joists (LH-series), deep longspan joists (DLH-series) and joist girders. This paper explores the LRFD design methodology as related to design and behavior ‘of K-series and LH-series joists. In accordance with the current SIT design specification individual joist members are designed to meet strength requirements for a given design load combination. Using the LRFD design methodology, the member-strength desi requirement may be expressed as follows tele foe=defero © InEquation 1, SR ismember specific and defined us the stress ratio, f; is the required strength or member stress resulting from external factored loads and OF, is the design strength at the ultimate limit state, Both f, and QF, are defined for K series joists and LHL-series joists in Sections 4.2 and 103.2 of the SUI design specification, respectively (SII, 2010). In effect, the SR isa measure of member efficiency, where SR= 1.0 indicates a member at its design capacity and SR < 10 indicates reserve strength. ‘Typically, joists are designed for economy so that individ- ual design siress ratios of multiple tension and compression ‘members are all simultaneously at or near 1.0. Therefore, ‘no consideration is given by the SJI standard (2010) to con- trolling the member-strength capacity limit state, and duc- tile tensile yielding (of the bottom chord or tension web) is given no specified preference over sudden compressive ENGINEERING JOURNAL / FIRST QUARTER / 2014/1‘Table 1. Design Conditions and Ductile Design Limits denen aces" Ro ShergPenorre | Sendomens Un ae a ee Boom chord and fatal tenion 1.00 240 soo eg tn | tanto ue bbuckling (of the top chord or compression web). As well, in truss design for non-seismic-load cases, the 2005 ATSC Specification does not make specific reference to a preferred, member-strength limit state, Importantly, the discussion hore neglects connection-related limit states and is restricted, to the memberlevel joist design, where the strength limit state will occur by tension-member yield or compression- member buckling, At the member level, research has shown that buckling of a steel joist compression member results, in an instantaneous and significant loss in load bearing capacity (Yost et al, 2004, 2006). The authors suggest that, given this behavior, a joist designed for a ductile tensile- member yield limit state is desired over one with a sudden ‘compression-member buckling limit state. In the event of severe overloads, the gradual yielding and deformation of ‘2 ductile clement provides visual warning, load sharing 10 neighboring members and time for evacuation. Additionally, there is inherently less strength variance in components that ‘are controlled by tensile yield limit state than those con- trolled by compression limit states, where force eccentric- ity, variability in end fixity and other variables contribute to less predictable buckling strengths. This characteristic was noted by Engelhardt et al. (2000) as related to strength and failure mode of experimental open web steel joists and by Rao et al. (2011) as related to strength and failure of latice- type transmission towers “The objective of this research study is experimental inves- tigation of a stee! joist design methodology where ductile tensile yielding is the intended primary-strength limit state As the controlling tension member(s) yield, the load-bearing capacity of the joist remains intact. Ultimately, with suf ficient inelastic deformation and in the absence of tensile fracture, the yield limit state is followed by a secondary strength limit state of compression-member buckling. At ‘compression-memiber buckling, the joist strength is drast cally reduced and ultimate collapse occurs. This paper out Tines a design philosophy investigated with experimental 2./ ENGINEERING JOURNAL / FIRST QUARTER / 2014 testing of K-series and LH-series joists that have been designed for a ductile tensile yielding limit state. The focus of the study is exploratory, where experimental results are compared to predicted behavior in terms of load capacity and strength limit state mechanisms. From this comparison, conclusions are established regarding further pursuit of the proposed ductile methodology as related to steel joist design. EXISTING SJI DESIGN REQUIRE} AND METHODOLOGY NTS Open-web steel joists are designed in accordance with SJI 2010), which has both LRFD and ASD methodologies. Design considerations not explicitly covered by SIT fol low the AISC Specification for Structural Steel Buildings (AISC, 2005) or the AISI (American Iron and Steel Insti tute) North American Specification for the Design of Cold- Formed Steel Structural Members (AISI, 2007). LRED is a probability-based philosophy that implements both load and resistance factors to ensue a minimal chance of inadequate capacity due to overload and/or understrength. This paper will focus solely on the LRFD methodology. For K-series and LH-series joists, bottom chord and web members are designed for axial force only. Also, interior top chord panels of K-series joists that are less than 24 in, in length are designed for axial compression only. All other K-series and LH series top-chord panels must consider axial compression and bending interaction. Table 1 summarizes these design conditions for all members. The SJT (2010) LRFD design requirements for axial tension, axial compres- sion and interaction are given in Equations 2 through 6 as follows: PulAg OF, Axial tension: <1Axial compression = St % Interaction a pane point +e ® Interaction at mid-panct 0 6 Sa fos CoS for Sat cqg9 Su 4|__Gom lev 6) OF, Beker | fox 9g, ory ‘The numerator terms of Equations 2 through 6 are LRFD. required strength terms defined as: factored axial force M, = factored bending moment A, =member cross-sectional area S* =minimum section modulus about axis of bending Jou factored axial compression stress = P, /Ay fos = factored bending stress =M,/S moment factor taken as —0.3fu /-F. for end panels = 1 = 04fau/QeFe for interior panels ‘ler backling stress =n°£/(KL. fr)? F, _=yiold sress assumed for design a5 50 ksi maximum member slenderness Jastc modulus = 29.000 ksi ‘The denominator terms of Equations 2 through 6 are LRFD_ design strength terms and, for those not yet defined, are given as follows: fy =0[ 0659" ]p or ters 4.(E/0R,* =0.877F. for kL fr > 4.71(E/OF, \” 2 =reduetion factor for slender compression clements k ——=effective length factor Lr ‘member slenderness 4 O. 05 =resistance factors for tension, compression and bending, respectively =090 ‘The effective length factor () is specified by the SII specif cation ($31, 2010) based on the joist series and member type. In addition to the desiga strength requirement identified in Equations 2 through 6, SIT also limits the maximum slender- ness ratio (L /») for the various member types. A summary of existing SJT design requirements as related to design con dition and maximum slenderness is given in Table 1. DUCTILE DESIGN PARAM AND METHODOLOGY ER To design an open-web steel joist for a controlling, ductile tensile-yielding, strength limit state, the relative strengths of the individual members must be considered so that ten- sion yielding precedes compression-member buckling. Thus, using the ductile design philosophy, the maximum stress ratio must be controlled by a tension member, and compression-memiber stress ratios must be sufficiently less to ensure yield before bucking. Accordingly, the relative strength factor (p) has been implemented to require mini- ‘mum over-strength for all compression members (top chord and interior webs), as a function of the maximum member stress ratio, The relative strength relationship is given as | o In Equation 7, pi the member relative strength factor, (SR), is the corresponding member stress ratio and (SR) is the ‘maximum stress ratio for all members. Again, for a ductile design, (SR)ngx Will be controlled by a tension member, ‘The stress ratios (SR) are as defined in Equation 1 and calculated using the procedures of Equations 2 through 6. Accordingly, as a structural system, the primary strength Timit state of the joist will depend on the maximum rela- tive strength factor forthe tension and compression member groups. This relationship is shown in Table 2, where it is noted that to achieve a ductile limit state, p< 1,00 for all ‘compression members and p = 1,00 for the maximum tension, ‘member, It is understood that these p limits are theoretical values, and the maximum for all compression members ‘will need to be less than 1,00 by a sufficient amount so that a tension yielding strength limit state is statistically probable, For this paper. the p factor is used in member selection and design of experimental joists so that ductile tensile yielding ‘of end web or bottom chord is the primary strength limit state. With regard to compression member aver-strength, the aforementioned relative strength factor (p defined in Equa tion 7) is the primary design variable governing member selection and achievement of a ductile limit state. A duc- tile design is theoretically achieved by setting the rel tive strength factor for the compression members to some value less than 1,00, as noted in Table 2. The lover relative strength factor encourages a ductile limit state by providing additional strength to compression members. Statistically, ENGINEERING JOURNAL / FIRST QUARTER / 2014 /3Table 2. Relative Strength Factor and Limit States Member Group Relative Strength Factor [Compression | Tension _| Primary Strength Limit State =1.00 00 Tensile yield z Simuitaneous tensile yield Fee cree re and compression buckling =1.00 = 1.00 ‘Compression buckling the lower p factor on the compression members decreases, their probability of failure. The maximum member relative strength factors used for design of experimental K-series, rod-web K-series and LH-series joists tested inthis research, defined as ductile designs, are as follows: + Bottom chord and end web members Pyuc= 1.00 + Interior tension webs Pu = 0.95 + Compression web members Pax = 0:80 + Top chord Pn = 0.90 “The use ofthese values results inthe prediction of bottom= chord of end.web tension yielding as the primary strength limit state, followed by top-chord buckling as the second- ary strength limit state. Again, the relative strength factors are determined using Equation 7, with the stress ratios as, given in Equation I, and the selected members satisfy the proposed maximum relative strength limits. As part of the ductile design methodology, the maximum slenderness limit, ‘on end-wweb and bottom-chord tension members is increased from 240 to 300. The slenderness limit of 300 is consistent th the recommended maximum slenderness ratio for ten- sion members in Section DI of the 2005 AISC Specifica- ton, The ductile design parameter limits are summarized, in Table 1 From the maximum p factors noted, the compression web has the most reserve strength, With Pag: = 0:80. The {op chord, in comparison, has a smaller margin of relative foverstrength, With Pras = 0290. In selecting these values, it was considered that compression-web buckling is more variable than top-chord buckling, justifying the lower rela- tive strength factor. The web-strength variability is due 10 the unsupported condition of the member length, variation in end fixity and eccentricity of axial load resulting from Weld location and member alignment, In comparison, the top chord is typically continuously braced by the support- ing deck, resulting in more predictable behavior, Also, the top chord is @ continuous double-angle member, and size selection is generally controlled by the middle-panel stress ratio, For economy of the joist, top-chord selection is rit cal and excess over-strength is to be avoided. As well, the top-chord compression strength is known to be higher than 4 ENGINEERING JOURNAL / FIRST QUARTER / 2014 that predicted using SII procedures from research done by Taboni et al. (2007) and Cianei et al. 2009). These results ‘were all considered in setting the compression-web and top- chord Pye. at 0.80 and 0.90, respectively, for design of the experimental joists, Future research may justify different Prax Vllues for the compression members. Tn conclusion, the proposed relative strength factors pro- mote design of a joist with a high probability of tension member yielding and associated ductile Timit state, Tntro- ducing a design parameter that regulates the respective strengths of compression and tension members resulted in the joist designs tested in this experimental study. Impor- tantly, the duetile design procedure is offered as an explor- atory exercise to investigate feasibility of the ductile design philosophy. SAMPLE DETAILS AND EXPERIMENTAL PROGRAM Experimental investigation included testing modified ver sions of three series of joists: K-series, LH-series and rod- web K-series, each designed per the ductile joist design procedure described earlier (and summarized in Table 1 For statistical validation, six identical samples were tested foreach joist series. Figure | provides experimental load and support information, and Figure 2 provides sample details related to pane! point layout and member sizes. The test ing apparatus for all 18 joists was designed to simulate a uniformly distributed load on a simply supported. span Referring to Figure 1, the joists are loaded via four hydrauli cylinders, spaced 8 ft apart. The hydraulic load was manu- ally pumped into the system and equally distributed to each of the four cylinders. As shown in detail A of Figure 1, each cylinder contacts a built-up load distribution unit that further distributes the load into eight point loads on the top chord at 1-fspacing. The system applies 32 equal point loads, spaced at 1 ft on center, along the 32- or 33-ft length of the jos. ‘Accordingly, the distributed force pattern, w, applied to the top chord is calculated a8 w = Pyyi/32 ft. The top chord ‘was laterally braced at 2-F intervals to prevent out-of-plane buckling. This combines with the multiple-point loads to simulate a uniformly loaded, continuously braced top-chord condition typical for joists. Additionally, the bottom chordAaa A AS Ren To Conenaed aa pois sce at 2° om LS Thee Dette Diao Unit Fig. 1. Experimenial load and support detail. 10 gato, oO Fig. 2. Sample details: (a) ypical K-seres details (samples JI-1 through J1-6}:() typical LH series details (samples J2-J through 12-6: (c) typical rod-web K-series detail samples J3-1 through J3-6). ENGINEERING JOURNAL / FIRST QUARTER / 2014 /5Te 5 Sample Design Paramaors and Faure Sequence aries L-Series oa Web Predicted Predicted Sirona strength Sone ten | SRandp | usc Siet | candy | SHER” | emanty | unites Member oO a) ‘Sequence o Sequence oO Sequence Tm crow SPIE] ogy [858 269 as ia Pal 67 —[ Secondary [ 0:8 —| sosordary [087 | Soooary Gatton Grd vw B88 | Prmary | — 088 ‘ot | Paar res [etoeeea| sane a8 7 on PS 0.98 0.40 0.54 0.36 Bimry | —~p5——| ~ag0 | 088 oF O75 er | 095] a 08 028 re [080] naa 059 sia | 080 or See te peer eae ene ey can be ese ee bridging applied to joist systems. The bottom-chord center ppancls and both end webs were painted with lime wash to identify yielding during testing. During testing, load was applied at an approximate rate of 1,000 Ib/min. Fifty-kip capacity load cells were placed at each support (Figure 1) 10 capture the total load applied to the system. A linear va able differential transducer (LVDT) recorded the deflection of the bottom chord at mid-span (Figure 1). Load and di placement were recorded ata sampling rate of 10 Hz using a 16-bit data acquisition system, ‘Sample details are shown in Figure 2. The K-series sam- ples (Figure 2a) featured continuously crimped (CO), single~ angle web members, which ate characterized by bending. the outermost region of each leg to fit and align the angle ‘centroid in the same plane as the centroid of the chords. The larger LH-series samples (Figure 2b) featured a combination ‘of continuously crimped, single-angle and double-angle webs members, The rod-web K-series samples (Figure 2c) con- sisted of a continuous round bar bent at the panel points 10 form the web members. Also, the P2 and P4 compression ‘webs were fabricated as rods reinforced with single angles, ‘and the corresponding design strength for these reinforced, members was used in determining the relative strength fac tors of Table 3, All experimental joist samples were designed for a duc- tile limit state using an assumed yield strength (F,) of ‘50 ksi and by the limiting relative strength factors described earlier (.e., compression-web Pygx = 0:80, top-chord Pyar 0.90, interior-tension Web Pyaas = 095, bottom-chord and 6/ ENGINEERING JOURNAL / FIRST QUARTER / 2014 strenath factors (Ppyax)- member stress ratios (SR) andl mem- ber relative strength factors (p) for the three joist series fab- ricated and tested in this study. For K-series samples, tensile yielding of both the bottom chord and end web is predicted as the relative strength factor for each at 1.00. This will be followed by top-chord buckling (p = 0.87). The controlling ccompression-web relative strength factor is 0,64 for member 4, so that web buckling is unlikely. For LH-series, ductile yielding ofthe end web is predicted (p = 1.00). Bottom-chord yielding (p =0.95) could occur before achieving a secondary limit state of top-chord buckling (p = 0.88). Compression- web buckling is controlled by member P6 with p = 0.77 and is unlikely to accur before top-chord buckling. For the rod-web K-series joists, bottom-chord yielding is predicted (p = 1.00), followed by top-chord buckling (p = 087). Compression-web strength is controlled by P12 with p = 0.77. The overstrength on the compression web predicts top-chord buckling as the secondary limit state for the rod web joists. ‘The SJT-factored LRFD design loads determined by Equations 2 through 6 for the ductile joists detailed in Fig- ure 2 are 418 lbift for the 20-in-deep ductile K-series, 1303 It for the 28-in-deep ductile LH-series and 420 Ibvft for the I6-in.-deep ductile rod-web K-series. In comparison to standard joists of equal span that are designed for the same factored loads but with no preference for controlling the strength limit state, the ductile joists weigh about 8% more. ant amount of this weight increase is related to the top-chord size, where over-sttength related to the designlimit of p $0.90 requires a larger section. Itshould be under- stood, however, that the 8% weight increase noted is spe~ cific othe joists tested inthis study and that, in general, the ‘weight increase associated with the ductile design methodol- ogy will vary with many factors, such as span length, joist type, material availability and manufacturer The actual yield strength of the bottom-chord mat rial was experimentally measured using the procedures of ASTM E8-04b, Standard Test Methods jor Tension Testing of Metallic Materials (ASTM, 2004), For each ofthe 18 joists, tested, a coupon was removed from the bottom-chord end panel. This location (bottom-chord end panel) was selected because of the low stress in this length, From test results, the average yield strength for the J1, 12 and J3 samples was, found to be 60.3, 60.6 and 61.5 ksi respectively. Thus, the ‘measured yield strength is about 20% higher than was used for design. Significantly, unusually high yield strength may. defeat the onset of tensile yield and ductile behavior. How- cever, in the event of yield strength high enough to preclude the desired tensile yielding limit state, the use of relative ‘over strength factor, p. on the compression members would, censure strength in excess of a joist designed in accordance ‘with SIT 2010), 700 Yield in Be 00 400 200 sy tintosded to adjust ‘TEST RESULTS In preparation for testing to collapse, all joists were first pre- loaded to a nominal displacement of 1 in., or about 40% of the LRFD factored design load. This was done to ensure that all data acquisition was functioning properly and to seat the test sample in the loading frame, thereby removing any ‘gap deformation among the loading apparatus, joist sample and supports. Upon release of the prefoad, the data acquisi- tion system was zeroed and testing to failure commenced. Results for the K-series, LH-series and rod-web K-series are shown in Figure 3, where itis noted that the horizon- tal axis is center span deflection and the vertical axis is the ‘equivalent uniformly distributed load (w) determined as the total hydraulic force (Piya) divided by 32.1 (= Phy /32 6 Also, the load axis in Figure 3 includes the dead weight of the testing apparatus and self-weight of the joist, which are simply added to the force applied by the hydraulic eylinders. ‘The total dead load was determined by weighing all compo- nents of the system in the absence of hydraulic force. ‘As cam be seen from Figure 3, initial response forall joists is elastic with a Tinear Joad-deflection response, indicating all member stresses below yield. For each of the three joist Strain Hardening ——n3 test appara. wa 1s —18 6 7 8 8 wou ow “Midspan Displacement (ia) (a) KeSeries Fig. 3. Load: displacement results: (a) K-series (continued next page) ENGINEERING JOURNAL / FIRST QUARTER / 2014 /7Strain Hardening = 1000 + iS 7505 Unloaded —n3 500 gj apparatus aA —ns 20 | DL=77 IWR 16 ° mo aeepeee ata statement Peep etig ttteeeaa Midspan Displacement (in) wo 30 > annem strain hacdening 700 Yield of BC on 00 2 44-16 f/f yf lies. 3 10 Design Capacity ~ $20 Ink —ni —n2 ae Unloaded to adjust TB 7 test apparatus —— 1 —ns 100 —n6 o + o 1 2 3 4 5 6 7 mo ou RB -Midsan Displacement in.) fc) Fig. 3 continued. Load-displacemens results: (b) LH-series; (c)rod-we K-seres 8/ ENGINEERING JOURNAL / FIRST QUARTER / 2014series, ahigh degree of repeatability is evident among the six identical samples. As the extreme fiber stress of the critical tension member reaches yield, the behavior transitions from, clastic to inclastie, as is evident in the accelerated deflec- tion rate and curvature of the load-deflection response, For all joist specimens, first yield occurred at the bottom-chord, mid-span location and was visually identified by flaking of lime wash at that location. Typical bottom-chord yielding ‘and lime-wash flaking can be seen in Figure 4a, Eventu- ally, the critical tension member fully yields, and a plastic, load plateau is achieved on the load-deflection response. For K-series and rod-web K-serics, yielding of the end web was also detected, as shown in Figure 4b, No end-web yield was, detected in the LH-series samples. Following the fully plastic condition, all joists were loaded to approximately 6 in. of deflection and subsequently unloaded. The residval deflection is noted in Figure 3 and, ranges from about 2.5 to 3.5 in, Unloading at 6 in, of defor- ‘mation was necessary to reset the LVDIT to a lower position ‘and capture the full deformation of the joist, which would, ‘otherwise exceed the LVDT stroke capacity, The joists were then reloaded, resuming the load plateau and continued plastic deformation, Stiffness of the reloading branch was. neatly identical to stiffness of the initial elastic response, Al joists were loaded to collapse, which oceurred by ‘compression-memiber buckling af either the top chord or a ‘web member. Table 4 summarizes test results in terms of load and displacement at yield (Y), plastic (P) and ultimate (U), as well as the secondary strength limit state and vari- ‘ous ratios. In Table 4, yield (Y) is taken at intial departure. from linear-elastic behavior; plastic (P) is defined as the load plateau, which is taken at 4 in, of joist displacement; and ultimate (U) is the secondary strength limit state of ccompression-member buckling. Figure 5 presents results graphically in terms of ultimate-displacement-to-yield- displacement or ductility ratio (Figure Sa) and average load. ratios at yield, plastic and ultimate relative to the ductile design capacity (Figure Sb). Individual series results are dis- ‘cussed in the following sections K-Series Discussion (Samples J1-1 through J1-6) ‘The predicted failure mode for these joists is given in Table 3 and expected to he simultaneous yielding of the fend web and bottom chord (9 = 0.99 and 1.00), followed by top-chord buckling (p = 0.87). The failure behavior observed during testing is consistent with this prediction. ‘As mentioned, flaking of lime wash, indicating yielding, ‘was observed on the bottom-chord and end-web members. Bottom-chord yielding was observed to initiate adjacent 10 the panel points of the bottom-chord mid-span segment (B4), As inelastic displacement increased, the bottom chord, yielded region spread down the length of panel Bd (Figure 4d). At ultimate load capacity, the complete cross-section ‘of the bottom-chord B4 panel for the entire member lensth showed evidence of yielding: however, no yield was detected in the neighboring bottom-chord panels (B3 and BS). End- web yielding generally occurred after significant bottom- ‘chord yielding and was evident over approximately the ‘middle half of both end-weby members. ‘fier significant plastic deformation, each sample expe- rienced top chord buckling. In general, top-chord buckling ‘occurred between brace points and about the y-axis, or out of the plane of the joist, characterized by an unrecoverable «drop in load-carrying capacity. A typical top-chord buck- Jing condition is shown in Figure 4c. For sample JI-2, the Joad distribution unit was rotated out-of-plane, inducing an ‘eccentric load on the top chord and promoting premature buckling. Thus, strength and deformation at ultimate ofthis sample are not considered in Table 4. For samples JI-1, 11-4 and J1-6, there appears to be a region of strain hardening after approximately 7 in, of deflection (Figure 3a). This is evident in the slight increase in load after the plastic load plateau. Excluding sample J1-2, deformation at ultimate ranged from 7.37 in. (Sample J1-1) © 10.8 in. (sample 1-4). In terms ‘of span length, this represents L/54 to L/37, a tremendous amount of deformation. Ductility results in Table 4 and Fig- ure Sa show a range from 2.47 (sample J1-5) to 3.55 (sample Ti-4) with a series average of 2.83, demonsrating a signifi cant amount of energy dissipation in the form of inelastic deformation. Average strength ratios in Figure Sb show pro- gressively inereasing capacity at yield, plastic and ultimate relative to design, and strength ratio standard deviations are all very low (Table 4), Specifically, the yield-strength-to- design-strength ratio (Y/D in Table 4 and Figure 5b) ranges from 1.23 to 1.34, with aseries average of 1.29, and ultimate- strength-to-design-strength ratio (U/D in Table 4 and Figure $b) ranges from 1.45 to 1.52. with a series average of 1.49. ‘The strength ratios and associated standard deviations indi- cate predictable behavior with low variability, conservative design relative to the primary strength limit state of tensile Yield and substantial reserve strength re: ary strength limit state of compression buckling. ive to the second- LHL-Series Discussion (Samples J2-1 through J2-6) The predicted strength limit state for these joists is given in Table 3 and expected to be end-web yielding (p = 1,00) followed by secondary buckling of the top chord (p = 0.88). The bottom chord has a p of 0.95 and may yield prior to a ccompression-member buckling. The observed behavior was not consistent with this prediction. Rather, in all six cases, yielding occurred on the bottom chord only, initiating on either side of the mid-span panel point, as shown in Fig- ure 4e. Bottom chord yielding was observed simultaneously in both members adjacent to the mid-span panel point (B3 and B4). As deformation increased, bottom-chord yielding ENGINEERING JOURNAL / FIRST QUARTER / 2014 /9{yea ze = eps npn 4 COE = Sape5-H "Ya BHP = Sef) HT USSG OIE OAL « Eumang om aed = ane Bun era GoM FS = BPS Buena pow-dek = EOL woHROHELUDS SNE SEC = OVT + 760 PIS, ws | ore | zee | ebeieny gos | see | ees | oer zs [ere [ves | Ser | cougy eos | pre | eis | rer | 9"-Pou wes | ore | ais | eer eas | aoe | ws | eer zas | ave | eze [rer er | soo | oe | %aPS zouet| pre | 1901 | ebeveny sser| eve | aio | oer goer wwe | over | Ser wer] ove [er | rer jet eee] se | veer | eer woot] we | wor | eer eer] eve | ov | rer i perorjree | 0 nis 65 | oe | org | abeiony zs | oe | oss | our wes | eve [uss | our eos | woe fos | mir | S™ ra ws | ose | eos | eur ost | zez | oa | sor | oor | eas | woz | sis o;ay;alealea ©] cay | wan [eves [tun | twa) | Cu) | Gua seues "yu tuens ain | aa | aia | peor | “daa | peor | “dsia | “asia | P07 | copuocss | “5d | Pee | “asia | peOT ewnin, (a)onseig PIA ‘Sunsay 1sa1 jo Mewuing “py a1geL qenewpeotusisea | asnonew | Ala neu 10/ ENGINEERING JOURNAL / FIRST QUARTER / 2014spread down the length of the B3 and B4 panels. However, no end-web yielding occurred in any of the samples. The LH-series end webs are double-angle members with battens, ‘welded to the third points, The end webs did not yield as pre- dicted, possibly due to a higher yield strength than the bot- tom chord. At approximately 7.0 in, of deformation, a region of strain hardening occurred (Figure 3b) where the load capacity gradually increased until ultimate collapse under the secondary strength limit state of compression-member buckling. The secondary strength limit state for samples J2-2 and 12-5 was a combination of buckling of the $4 web mem- ber followed by buckling of the top chord at center span. ‘The S4 web buckling was not a sudden condition, rather a deformed bent shape of the member was evident as the joist approached failure by top-chord buckling, as is shown in Figure 6a, The deformed shape of the $4 web only occurred, at very high deflection, in excess of approximately 7 in. At ultimate collapse, top-chord buckling was out ofthe plane of vein) Centerline a the joist for sample 12-2, For sample J2-S, top-chord buck- Jing occurred in the plane ofthe joist between the two pane! points located 28 in, on either side of center span, Simul- taneous buckling of the S4 member was also observed, which resulted in the loss of in-plane bracing by the $4 web at center span on the top chord, This behavior is shown in Figure 6b, For the remaining four LH-series samples, some evidence of $4 buckling or bending was observed; however, ultimate collapse occurred by top-chord buckling out of the plane of the joist at an interior panel. ‘The $4 member is a secondary vertical web member and is not predicted by clastic analysis to be highly stressed by direct application of externally applied loads. The clastic analysis is based on relatively small deformations where the S4 member force is largely based on tributary loading, as is shown in Figure 7a. Secondary web members are gener- ally not important force-resisting memibers, but rather pri- marily provide in-plane bracing for the top chord. However, considering its location at the highly stressed top-chord o Jeistcaneine @ Fig. 4. Member yielding and buckling: (a) Bottom-chord yielding: (0) end-web yielding: (e)top-chord buckling: (d) bottom-chond ylelding in K-series samples; (e)bottom-chord yielding in LH-series samples; (f) botiom-chord and end-web yielding in vod-web K-series samples. ENGINEERING JOURNAL / FIRST QUARTER / 2014/11mid-panels, test results show that the $4 member appears to become highly stressed under extreme inclastc joist dis placement, This condition likely results from the very large vertical deformation and corresponding development of & vertical force component in the top-chord axial force that, by joint equilibrium, acts on the vertical web, as shown in Figure 7b. This results in stresses well beyond those caleu- lated for clastic behavior and small deformations. Currently, the SIT requires only that vertical web members be designed, for gravity load plus 0.5% of the top-chord axial force (SII, 2010). This requirement should be reexamined with co cration of ensuring that vertical-web members possess the necessary strength required to resist axial forces associated ‘with large deformation behavior. It should be noted that this, behavior was only evident after the test joists had success- fully demonstrated the intended goal of extreme inelastic pa BK-Series LH-Series "= Rod-Web-Series 10 | | 12 3 45 Average Sample fa) ductile deformation while retaining full load-bearing capac- ity and that the vertical web buckling was not sudden in nature Duciility results of Table 4 and Figure Sa show that the LH-series achieved the highest performance of the three series tested. Ductility ranged between 3.26 (sample J2-4) and 4.08 (sample 32-5), with a series average of 379. The high ductility ratios appear to result in a strain-hardening region that begins at approximately 7 in. of deformation, approximately three times the yield displacement, At these high deflections, strain hardening in the yielding tension ‘member is achieved, resulting in the tangent stiffness appar- cent in the test results. This strain hardening behavior is further supported by the load-strain results for sample 12-6 shown in Figure 8, where strain data were collected at the ‘middle length of both end webs, as well asthe bottom chord 7 YD =P =UID 163 Kon LHD) Joist Series 0) Rod Web (3) Fig. 5. Ratio results (a) ductility ratio: (b) average strength ratios, @ Fig. 6. Secondary limi state filure for J2 and J3 series: (a) sample J2-2;(b) sample J2-5: (c) sample 53-1; (d) sample J3-S. 12/ ENGINEERING JOURNAL / FIRST QUARTER / 2014at center span. Figure 8 shows thatthe end webs remained clastic (SGI, SG2, SGS, $G6), but there is severe yielding in the bottom chord (SG3, SG4) The sain at yield is mea- sured to be about 2,100 microstrain, after which there is a sharp transition to a perfectly plastic condition. At about 16,000 microsiain, strain hardening initiates and contin- ues to about 19,500 microstrain, at which point secondary compression-member buckling occurs. After testing, the bottom chords were all visually inspected and measured ‘witha digital caliper, and there was no visible or measurable evidence of a reduced cross-section or necking. Average strength ratios in Figure Sb show progressively increasing capacity at yield, plastic and ultimate relative to design and also indicate strength ratio standard deviations are all very low (Table 4), The yicldstrength-to-design- strength ratio (Y/D in Table 4 and Figure Sb) ranges from 1:24 to 1.32, with a series average of 1.28, and ulkimate- strength-o-design-srength ratio (U/D in Table and tre Sb) ranges from 1.43 to 1.56, with a series average of 1.52. As with the K-series joists, strength ratios and associ ated standard deviations indicate predictable behavior with low variability conservative design relative to the primary width to $4 Note bonding in the top chord not shown for clarity Fewity (a Cy resultant top chord axial force Choy = Vemticalcomponcat of Cy. 4 axial force Note = ending inthe op chord not shown for clarity Faw(L)+2C ye, o Fig. 2. 84 member axial fore: (a) small deformation; () large deformation. ENGINEERING JOURNAL / FIRST QUARTER / 2014/12strength limit state of tensile yield and substantial reserve strength relative to the secondary limit state, Rod-Web K-Series Discussion (Samples J3-1 ‘through J3-6) ‘The predicted strength limit state sequence for these joists as provided in Table 3 is bottom-chord yielding (p = 1.00) followed by top-chord buckling (p = 0.87). Yielding of the end web before top-chord buckling is possible with p = 092. ‘The load-deflection results for the rod-web K-series joists are shown in Figure 3e, where extreme ductile behavior is, noted. For all samples, response is initially elastic followed by bottom-chord yielding, The observed yielding result is consistent with predicted behavior. In all cases, first yield- ing occurred in the bottom-chord mid-span panel (B7). This ‘mechanism was observed to start adjacent to the panel points fon each side of the B7 panel, as is shown in Figure 4f. As. inelastic displacement increased, yielding spread down the length of the B7 panel as well as ito adjacent bottom-chord panels (B6 and B8). At very high displacements, Iime-wash flaking was also observed in the BS and B9 panels. After the onset of bottom-chord yielding, end-web yielding was also observed, In some instances, the end-web yielding was, observed over the entire cross-section and length of the end bars (Figures 4f and 60). After full, bottom-chord yielding, the behavior appears to directly enter a strain-hardening region (Figure 3c). This, is evident in the absence of a horizontal load plateau fol- lowing full yielding of the bottom chord. Rather, there is an immediate resumption of load increase, albeit at a much reduced rate, The load versus displacement plot of Figure 3c shows an inclined yield plateau, indicating an i strength as deflections increase, This behavior was attrib- uted to strain hardening after about 7 in. of deflection in the Keseries and LH-series joists (Figures 3a and 3b); however, for the rod-web K-series, its onset is immediate and con- tinuous throughout the loading from first yield to ultimate collapse. This appears to be the result of more numerous rod-web members relative to the geometric layout of deeper joists. With the spacing between panel points minimized, the stress redistribution can occur more continuously as the bottom-chord member is yielding, ‘At secondary limit state, only two of the six joists (13-2 and J3-4) experienced the top-chord buckling anticipated by the theoretical failure sequence. The top-chord buckling locations were confined to the central panels (T7 through TIO). Three specimens (J3-1, J3-3, 13-6) collapsed in buck- ling ofthe first interior compression web member (P2/P27), which is a %-in-round bar reinforced with a 1 in, x 1 in. x 0.109 in. angle. In each of these cases, the buckled web ‘member was adjacent to an end web member that demon- strated substantial yielding prior to buckling. These types of failure are shown in Figure 6c, The J3-5 specimen had a relatively unique secondary limit state of buckling in the P8/P21 and P6/ P23 compression web members, ay shown in Figure 6d. While not necessarily expected, their ultimate buckling is understandable in that they are the first unrein- forced compression bars from the end of the joist and have design stress ratios of 0.59 and 0.75, respectively. el Seo ae = aoe ie Jasna ial Fig. 8, Tension-membersirain results for sample J2-6. 14/ ENGINEERING JOURNAL / FIRST QUARTER / 2014Ductility ratios from Figure Sa range from 2.53 for sample t0 3.53 for sample 13-4, with a series average of 3.15. As ith JT and J2 samples, average strength ratios in Figure Sb show progressively increasing capacity at yield, plastic and ultimate relative to design, and strength ratio standard, deviations are all very low (Table 4). The yield-strength- to-design-strength ratio (Y/D in Table 4 and Figure 5b) ranges from 1,22 to 1.29, with a series average of 1.26, and. ultimate-strength-to-design-strength ratio (U/D in Table 4 and Figure Sb) ranges from 1,52 to 1.69, with a series average of 1.63. The strength ratio U/D is the highest of the three Joist series tested, and represents additional strength devel- ‘oped from the post-yield inclined load plateau (Figure 3c) As with the K-series and LH-series joists, strength ratios and, ‘associated standard deviations indicate predictable behav- ior with low variability, conservative design relative to the primary strength limit state of tensile yield and substantial reserve strength relative to the secondary limit state DISCUSSION OF TEST RESULTS AND RELABILITY INVESTGATION ‘The ductile design methodology related to controlling mem- ber strength by implementation ofthe relative strength factor 1p in member selection (all as summarized in Table 1) is fully, ‘supported by 18 experimental data points. For each joist tested, the intended ductile behavior was achieved and characterized by tension yield, followed by significant plastic deformation and ultimate failure by com- pression buckling. Thus, the p factors selected for design provided sufficient compression-member over-strength, allowing tensile yield as the initial strength limit state and, importantly, yield and ultimate collapse at strengths well in excess of the predicted design capacity. As well, the experi- mental test results are uniquely significant when measured relative (© the design limit state (D) at each the yield (¥) (or primary strength limit state, plastic limit state (P) and ultimate (U) or secondary strength limit state conditions. ‘That is, experimental behavior is characterized by three ‘unique limit states that occurred sequentially with increas ing strength capacity in all 18 samples (Figure 5b). Ult mately, these three limit states (Y, Pand U) are a predictable sequence, where the plastic Limit state represents an impot- tant transition from the primary to the secondary strength Timit states. Strength ratios and corresponding, statistical data are important in this discussion as well. As is seen in ‘Table 4 and Figure 5b, the strength ratios at each ofthe yield, plastic and ultimate conditions relative to LRFD ductile design capscity are progressively increasing, with each ratio, having a significant factor of safety and low standard devia- tion. This ensures a conservative design with the primary strength limit state corresponding to the yield Timit condi- tion, significant residual strength at the secondary limit state ‘and low variability for all three strength ratios. Although an in-depth reliability study is beyond the scope of this paper, the reduced variance in the strength of mem bers in tension relative to the strength of slender members in compression results in improved reliability and is worth exploring as an extension of the test results presented. If ‘materials purchased for the bottom chord and end webs are limited to 2 specific stel alloy supplied by a specific mill, further reductions in variance can be found, resulting in ‘more reliable strength and limit state control, Accordingly, an exploratory reliability investigation has been performed based om the tested joist plastic strengths (P), together with ‘merchant bar steel mill test data, and employing the crite- ria and assumptions used to develop the 2005 AISC LRFD design approach. Merchant bar steel mill test data have been furnished by Steel Dynamics Roanoke Bar Division, for ASTM A529-50 steel, covering a time frame from May 2008 to October 2012 (Steel Dynamics, 2012). In all, the data included 11,546 test samples representing 4,337 batches of steel. The yield stress population distribution and statis- tical data for these 11,546 samples are provided in Figure 9, For joist test strength ratios, the 18 tested joist plastic strengths have been divided by the joist experimental desiga strength, which is defined as the nominal strength times the ratio of member tested F, to specified minimum F, (50 ksi). ‘A summary of these loads and ratios is provided in Table 5. Lacking data for broad comparisons of actual section prop- erties t0 nominal section properties, the industry standard data published in Table FI of AISI 2007 (AISI, 2007) have been used for these ratios. ‘The LRED design approach used by both the 2005 AISC Specification and the 2007 AISI design specification fo- lows the equations and procedures presented in a series of eight articles in the September 1978 issue of the Journal of the Structural Division. However, the two differ in the ratio of Tive-to-dead loads used for calibration of LRFD to the historieal ASD design method. For LRFD calibration, the 2005 AISC Specification uses a live-to-dead load ratio of 3, whereas AIST uses a Tive-to-dead load ratio of 5. From AISI Chapter F, Tests for Special Cases, and Commentary fon Chapter A, General Provisions, the relative reliability index is calculated as: tn [SearaFara] ¢ Wi VE +CpvE +VB ® where B. =relative reliability index 1 =resistance factor =0.90 Cy =calibration coefficient, which may be shown [using procedures demonstrated in ATST (2007) Commen- tary on Chapter A, General Provisions] t equal ABI for LRFD with live-to-dead load ratio of 3 ENGINEERING JOURNAL / FIRST QUARTER / 2014/15‘mean value of material factor, M = 1.135 from Steel Dynamics Roanoke Bar Division data for ASTM A529-50 merchant bar (sce Figure 9) ‘mean value of fabrication factor, F 1.0 from AISI (2007) Table FI P,, = mean value of professional factor, P, for tested ‘component = 1033 from test-average-plastic-to-experimental- design ratio (see Table 5) ‘Vy = coefficient of variation of material factor 0.0602 from Steel Dynamics Roanoke Bar Division data for ASTM A529-50 merchant bar (Figure 9) coefficient of variation of fabrication factor = 0.05 from AISI 2007 Table FL -orrection factor 1.196 for 18 test samples ‘efficient of variation of jist test results = 0.029 from test results data (see Table 5) ‘efficient of variation of load effect, which may. be shown (using procedures demonstrated in AIST (2007) Commentary on Chapter A, General Provi- sions) to equal 0.187 for LRFD with livesto-dead. load ratio of 3 Number of Samples Substitution of these values into Equation 8 yields an approximate plastic strength B = 32. This is an improved reliability as compared tothe approximate f= 2.6 for mem- bers, reported in the 2005 AISC Specification and reflec- tive of expectations of a joist designed in accordance with the SIT standard (2010). Tt should be noted that the caleu- lated approximate plastic strength B = 3.2 is based on the joist tested plastic limit state (Tables 4 and 5), above which the joists demonstrated consistent reserve eapacity before attaining ultimate maximum load capacity. In summary, the ductile design methodology employed in this experimental program produced the predicted behav- jor related to achieving ductile failure, resulting in a slow collapse mechanism characterized by large inelastic defor mation and improved reliability. ‘The loss in economy acknowledged as a consequence of adjusting-member strengths, CONCLUSIONS AND FINDINGS ‘The research presented in this study experimentally explores, a design methodology for open-web steel joists, where the primary strength limit state is ductile tensile yielding of the bottom chord or end web, which, after significant inelastic Panamese | Y#lSte5 | Sr a0 (3) | ksimin Average | Set64 | 1.383 ‘Misimum | 30,000 | 1.0000 Maximum | 76,570 | _1S3I4 SwDev | 3016 | 0.0083 cov | 00a | 0.0002 Dhases ELL Stress (psi) Fig, 9. Yield stress population distribution and statistical dara 16/ ENGINEERING JOURNAL / FIRST QUARTER / 2014deformation, is followed by a secondary strength limit state of buckling of the top chord or compression web. Impor- tantly, the study scope is restricted to gravity loading of simply supported K-series and LH-scries joists, where the bottom chord and end webs are in tension andthe top chord is in compression. For other load and support conditions, where the force sense in these members is different, the stated findings may not appl Ductile behavior was achieved by adjusting the relative design strengths of the individual tension and compression members so that tension yielding precedes compression- member buckling. Adjusting the individual member ten- sion and compression strengths to the appropriate relat strength factors results in a predictable failure sequence characterized by ductile behavior and sufficient capacity to support SIT LRFD design loads. The proposed ductile design methodology was experimentally investigated in the n, manufacturing and testing of modified K: LH¥-series and rod-web Kcseries joists. For each joist series ‘Table 5, Statistical Data for Joist Tests ‘SJILRFD Fy Experimental Ratio Plastio/ Design Load | Experimental | Design Load Plastic Exp Design Series Sample ) (ksi) (bit) | Strength (ib/tt) | Load (-) aw 565, 1.01 we 574 1.02, ws 567 1.01 K-Seri a ‘Series ha 8 603 560 Se0 tos we 592 1.06) we 582 1.04) wea 1878 1.07, rz 1882 1.07, 23 1886 1.07, L-Series 1303 608 1785 2-4 1852 1.08) 25 1868 1.08) w8 1855 1.06) 34 582 1.01 2 588 1.03) Rod-web 33) fe sie a 587 0.99 K-Series 4 Es 0.38 135 572 1.00) 136 508 0.39 ‘Average 1.0330 Std. Dev. 0.0302 All ‘cov 0.0293, Quantity” 8 six identical joists were tested, for a total of 18 tests. All joists were simply supported with a uniformly distributed Toading pattern applied to the top chord, The lengths were either 32 or 33 ft, and the top chord was laterally braced st 2-ft intervals. The following conclusions are derived from the test results: + AILI8 joists behaved in a ductile fashion, as predicted, with tension yielding as the primary strength Limit state followed by compression-member buckling as the sec- ‘ondary strength limit state. For K-series and rod-web K-series joists, both the bottom chord and end web expe- rienced tension yielding, For the LH-series joists, only the bottom chord experienced yielding. + For the joists tested in this research, implementation of the ductile design relative strength factors as the basis for member selection resulted in an 8% increase in ‘weight as compared to a conventionally designed joist of ‘equal span and capacity. In general, the weight increase ENGINEERING JOURNAL / FIRST QUARTER / 2014/17associated with implementation of the ductile design ‘methodology will vary with many factors such as span length, joist type, material availability and manufacturer. ‘+ ‘The yield strength forall 18 samples exceeded the LRED. design strength by a significant amount, For K-series, Lit-series and rod-web K-series joists, the six sample average yield loads were 1.29, 1.28 and 1.26 times the LRED design load, respectively. Reference for this con- clusion is made to Table 4 and Figure Sb. This indicates a conservative design relative to the primary yield strength limit state. ‘+ The series average ultimate strength ratio, which is defined as the load at ultimate divided by the LRED. factored design load, is 1.49, 1.52 and 1.63 for K-series, LH-series and rod-web K-series joists, respectively. Ref= erence for this conclusion is made to Table 4 and Figure ‘sb. This indicates substantial reserve strength relative 10 the secondary strength limit state +The average ductility ratio for six identical samples, which is defined as the deflection at ultimate divided by the deflection at yield, was 2.83, 3.79 and 3.15 for K-series, LH-series and rod-web K-series joists, respec= tively. For all 18 samples tested, this ratio ranged from 247 to 408. Reference for this conclusion is made 10 Figure Sa and Table 4. +The relative reliability factor calculated using joist test results and statistical data from 11,546 merchant bar test samples was 3.2, an increase of 23% over the 2.6 used. by current SJT LRFD methodology. Reference for this cconelusion is made to Figure 9 and Table 5. + For the K-series and LH-series joists, the yield limit state was followed by a horizontal load plateau. After significant deformation, the load plateau terminated, and these joists experienced a gradusll increase in load 18/ ENGINEERING JOURNAL / FIRST QUARTER / 2014 capacity that is associated with strain hardening in the jelded tension member. Reference for this conclusion is made to Figures 3 and 8, Ultimate collapse occurred asa secondary limit state of buckling of the top chord or ‘compression web. + For rod-web K-series joists, the yield limit state was fol- lowed by an immediate resumption of increasing load- bearing capacity, The post-yield behavior was an inclined linear increase in loading until secondary compression failure, Ultimate collapse occurred as a secondary limit state of top-chord buekling or compression-web buck- Ting. Reference for this conclusion is made to Figure 3c + In soveral of the LH-series joists, bending and buckling ‘of the secondary $4 web at mid-span was observed. ‘This occurred after very high inelastic deformation and is attributed to development of a vertical component to the resultant chord axial force that delivers a substantial ‘compression force on the web. Reference for this conclu- sion is made to Figures 6a, 6b and 7. In conclusion, the ductile design philosophy was success- fully implemented using the relative strength factor (p) as the basis for member selection, ensuring sufficient compres- sion member over-strength relative to tension-member yield strength, ACKNOWLEDGMENTS ‘The authors ate grateful to Commercial Metals Company Ine. for providing financial support and materials for thi project and to Steel Dynamics Roanoke Bar Division for sharing statistical data from mill tests of ASTM A529.50 ‘merchant bar, The authors also wish to thank the Villanova University Office of Research and Sponsored Projects for providing graduate tuition support and the Villanova Uni- versity Center for Undergraduate Research and Fellowships for supporting an undergraduate researcher on this project.SYMBOLS. Ae = gross cross-sectional area Cw = moment factor Cres Crey = top-chord resultant force and vertical force component, respectively CoCo = correction factor and calibration coefficient, respectively E = elastic modulus. ou = factored axial stress ve = factored bending stress he = required member stress F = axial force in $4 web member Fer critical buckling stress Fe = Euler buckling stress Fy = nominal member stress at ultimate = yield stress = mean value of fabrication factor, material factor and professional factor, respectively = effective length factor L member length = tributary length to top-chord panel point My = factored bending moment r, = factored axial force = local buckling reduction factor r = radius of gyration s = seetion modulus SR = stress ratio Vis Vex Vo. Vo. = coefficients of variation for material, fabrication, joist test results and load effect, respectively w = applied distributed force pattern B = relative reliability index = strength reduction factor = relative strength factor REFERENCES AISC (2005), Specifications for Structural Steet Buildings, 13th ed., American Institute of Steel Construction, Chi cago, IL. AISI (2007), North American Specification for the Design of Cold-Formed Steel Structural Members, American Iron and Steet Institute, Washington, DC. ASTM (2004), “ES-04b Standard Test Methods of Tension ‘Testing of Metallic Materials,” ASTM E8-04b, American Society of Testing and Materials, West Conshohocken, PA Gianci, PA.. Yost, IR., Gross, S.P. and Dinchart, DW. (2009), “Design and Behavior of Ductile Open Web Steel Joists: Phase MI,” Research Report, Commercial Metal Corporation, Hope, AR Engelhardt, M.D., Kates, 2, Beck, H, and Stasney, B. 2000), “Experiments on the Effects of Power Actuated Fasteners ‘on the Strength of Open Web Steel Joists.” AISC, Eng neering Journal, Fourth Quarter. Taboni, NJ, Yost, IR., Gross, SP. and Dinchart, DW. (2007), "Design and Behavior of Ductile Open Web Steel Joists,” Research Report, Commercial Metal Corporation, Hope. AR Rao, N., Knight, $., Seetharaman, S., Lakshmanan, N. and Iyer, N. Q011), “Failure Analysis of Transmission Line “Towers.” Journal of Performance of Constructed Facili- ties, May/June SH (2010), Standard Specifications, Load Tables, and Weight Tables for Steel Joists and Joist Girders, 43rd ed., Steel Joist Institute, Myrtle Beach, SC. Steel Dynamics 2012), “Internal Quality Control Report.” ‘Steel Dynamics Corporation, Roanoke, VA. Yost, JR., Dinehart, DW. Gross, S.P, Pot, J. nd Gargan, B, 2004) “Strength and Design of Open Web Steel Joists with Crimped-End Web Members,” Journal of Structural Engineering, Vol. 130, No. 5. Yost, JR., Dinzhart, DW., Gross, S.P, Pote, J. and Deeney, J. (2006), “Buckling Strength of Single Angle Compres- sion Members in K-Series Joists.” Engineering Journal, second quarter ENGINEERING JOURNAL / FIRST QUARTER / 2014/19)20 / ENGINEERING JOURNAL / FIRST QUARTER / 2014Experimental Verification of Spliced Buckling Restrained Braces KENNETH T. TAM, RONALD L. MAYES, DAVID L. MCCORMICK, ANINDYA DUTTA and CRAIG B. GOINGS ABSTRACT ‘A citical facility constructed inthe 1980s was housed in a building whose original seismic force-resisting system included chevion-braced frames and pre-Northrdge earthquake moment frames. inthe mic-19006, a code-basod seiamc retrofit was designed, Ths phased retraft was only partaly completed. & more recent seismic evaluation of the building in ts parvally retrofitted condlton revoaled malor structural oficiencies. Another retrofit was dasigned using feld-spiced buckling restrained braces (BABS) that allowed the bulaing to remain opera- tional during construction. The paper summarizes the analyses performed fr the sesmic retrofit and the development, prototype testing and ‘rection ofthe spiced BABS. Tho project testing protoccl was compared with current AISO Seismic Provisions testing protocol for BABS (AISC 341-10), which shows that it propery Gotated and fabricated, BRB used for thie retro project can have poak and cumulative cuetiity Wallin excess of the currant AISC testing err, ‘Keywords: seismic retro, bucking restrained braces, fld-splce Geta INTRODUCTION ‘A critical facility constructed in the 1980s was housed in ‘a building whose original, seismic force-resisting system, includes chevron-braced frames in the long dircetion and pre-Northridge earthquake moment frames in the short direction. In the mid-1990s, a code-hased seismic retro- fit was designed. This retrofit, which was scheduled to be ‘completed in phases, was only partially completed after approximately 12 years, A seismic evaluation ofthe building, in 2008 in its partially retrofited condition revealed major structural deficiencies. Consequently, a seismic retrofit that consisted of a combi- nation of exterior buttresses with buckling restrained braces (BRBs) and BRB frames at some interior locations was, developed. One of the major design constraints was that the facility needed to remain operational during the construc- tion with minimal disruption to existing process utilities. During the design phase, concern was expressed about the maneuverability of the BRBs in tight spaces inside the Kenneth I. Tam, SE. Senior Project Manager, Simpson Gumpenz & Hager, Ine San Franco, GA coveepondieg), Era ktamogh com Ronald. Mayes, PRD, Sta Conautat, Sinpson Gumparts & Hager, ne. ‘San Fancieco, CA. Eat: rayee@ghcom Dav L. MeCormik S.., Snir Prhcipal, Smpeon Gumpert & Hager ne ‘San Frencieco, CA. Ema dmccornsck@egh com ‘Anindya Duta PRD. SE. Senior Project Manage, Sepeon Gumpert & ‘Hepes ine. San Francico, GA Ema [email protected] Craig & Goirgs, SE, RE. Assocate Principal Simpson Gumpers & Hage, In, San Franco, CA, Eat ebgomngaBagh com building, This resulted in the development of a field-splice detail for the BRBs, which permitted erection of the braces in half-seements, Because splicing of BRBs had not been performed previously, the supplier was required to demon- strate similar performance to the un-spliced BRBs through prototype testing. After the testing, the spliced BRBs were successfully installed and the project was completed with ‘minimal discuption to the operations of the facility. ‘Assummary is presented of the analyses performed for the scismie retrofit and the development, prototype testing and erection of the spliced BRB, The project testing protocol ‘was compared with current AISC testing protocol for BRBs (AISC 341-10; ATSC, 2010), which shows that, if properly detailed and fabricated, BRBs used for this retrofit project cean have peak and cumulative ductility well in excess of the current AISC testing requirement, ‘The subject building is Tocated in South San Francisco, California, and was built circa 1980, Tt is two stories with a rectangular plan (100 ft by 364 {1 The original building" scismie force-resisting system includes braced frames in the long direction and pre-Northridge earthquake moment frames in the short direction. Figure | shows an aerial view of the building, In the mid-1990s, a seismic retrofit was designed to the requirements of the 1991 Uniform Building Code (UBC) with an importance factor of = 2.5. The retro- fit primarily involved the addition of braced frames and aug- ‘mentation of existing moment frame connections atthe roof, Tevel with haunches, The retrofit, which was progressing in phases during the yearly shutdowns, was only partially (about 60%) completed in 2008 when a new scismic evalu- ation of the building was commissioned, Major deficiencies in the structure in its partially retrofited condition were identified. Consequently, a retrofit design was undertaken, ENGINEERING JOURNAL / FIRST QUARTER / 2014/21Many of the spaces are extremely congested with process piping, mechanical ducts and other equipment. Most of the processes could not be altered or taken out of service, even temporarily. Furthermore, two buildings exist adjacent 10 the north elevation with minimal seismic separation, and a large multilevel pipe rack exists along most of the length ‘at the south elevation, Given the site constraints, a retrofit, scheme was developed that utilized exterior steel buttresses, (on the south elevation, connected to the building with hori- zontal BRBs at the floor and roof levels, and added BRBS ‘within some of the frame lines that were accessible. SEISMIC DEFICIENCIES ‘The partially retrofitted building was analyzed, and the fol- lowing deficiencies in the seismic load resisting system were identified. Pre-Northridge Moment Frames ‘The original lateral system in the transverse direction of the building comprises of W24 (roof) and W30 (second, floor) wide flange beams connected to WI columns with typical pre-Northridge moment connections (WUF) that are susceptible to brittle failure. The frames also have weak panel zones that have capacities well below that required 10 develop yielding of the beam. Chevron-Braced Bays All ofthe original braced frames and a majority of the ret- rofitted braced frames have a chevron configuration with beams that were not sized to earry the unbalanced force. ‘This is worsened by the fact thatthe original braced frames, consist of double-tiered chevron frames (two levels of chev- ron frames per story) where the intermediate beam between story levels consists of W10 wind girts that were oriented with the strong axis inthe horizontal position Inadequate Brace Connections All the braces in the original building, as well as a major ity of the retrofitted braces and their connections, were not detailed to develop the strength of the braces. Close Proximity of Adjacent Buildings and Propensity for Pounding ‘The subject building is located in a congested building com- plex with at least two buildings spaced closely enough to cause a pounding hazard (see Figure 2), Other Issues ‘The building also has a relatively weak, bare-metal roof diaphragm, as well as inadequate collectors and collector connections, in part due to the fact that the retrofit from the 1990s was not completed, Fig. 1. Aerial view of the facility showing the subject building. 22/ ENGINEERING JOURNAL / FIRST QUARTER / 2014RETROFIT ANALYSIS ‘A code-type analysis using an R factor was not adequate 10 account for the differing behavior ofthe various lateral sys- tems that were builtin different periods of time, Therefore, 2 performance-based design approach outlined in ASCE. 41-06, Standard for Seismic Rehabilitation of Existing Buildings (2006), was followed. The performance objective is w exceed life safety under the BSE-1 level earthquake and collapse prevention under the BSE-2 level earthquake. A nonlinear dynamic analysis using ASCE 41-06 was performed. The analysis was iterative im that it was fist required to conceive retrofit concepts and introduce them. into the 3D model and then validate through the dynamic analysis that they alter the behavior of the structure in a. ‘way that was anticipated. The owner next provides input regarding acceptability with regards to operations and cost. ‘The scheme was then incrementally modified until a cost effective scheme with litle or no temporary or permanent effect on operations was developed. Original Moment Frame —New BRB $ Spliced RB Sti tesl eat Subject Building aetna cea hace Retrofit concepts including dampers and conventional braced frames were considered, but it was soon determined that BRBs provided the best protection for the existing braced frames and their connections and limited the drifts in the direction of the moment frames to protect the weak connections and to minimize pounding. Through strategic placement of BRBs—which required replacement of exist- ing chevron braces at some locations—diaphragm strength- ening was limited to local areas, and only a limited amount of collector strengthening was required. As the design progressed, the contractor expressed con- cerns about maneuverability of the BRBs (approximately 440 ft long) inside the congested building. A field-splice detail for the BRBs, allowing transport ofthe braces in half- segments, was developed. Because of concerns regarding alignment of the spliced segments, the vendor was required to demonstrate through prototype testing that the spliced braces would perform as well as the non-spliced BRBs. The BRBs were tested at the NCREE facility in Taiwan and the results compared to the analytical model. == New Exterior Buttress connected to building with horizontal BRB Fig, 2. Plan showing proximity of adjacent buildings. ENGINEERING JOURNAL / FIRST QUARTER / 2014 / 23,Ground Motion for Nonlinear Analysis A suite of seven ground motions was used for the nonlinear time-history analyses. These ground motions were scaled to a site-specific spectra generated using the next-genera attenuation (NGA) relationships. As the soi profi dramatically under the building, the worst-case ground motions associated with a firm and soft sol site were used, Nonlinear Analysis Model A nonlinear finite element model of the lateral-load systems Of the building was created using the analysis program C’ Perform. Existing double-angle and tube steel braces were modeled using backbone parameters from ASCE 41-06. One ‘of the most important criteria for modeling the braces was 10 capture the behavior of weak connections. Where strength ‘of the connection governed, a drastie drop in strength was, incorporated into the model. The brace was considered inef- fective after the connection fails. Similar modeling was per- formed to recognize substantial loss of strength when the braces buekled. The moment frames were modeled with Table 1. Acceptance Criteria for BRB ‘Average of ‘Maximum of ‘Shaking Level Performance Goal Time Histories 7 Time Histories BSE-t Life safety Boy 12a) BSE-2 Collapse prevention 12 18 ‘nonlinear panel zone spring to capture the effects of the weak panel zone. ‘The effects of foundation flexibility were included into the model through the use of nonlinear elastie springs repre- senting the passive and frictional resistance of the footings ‘against the supporting soil. Nonlinear gap elements with compression only stiffness were used to capture foundation uplift. The upper- and lower-bound analysis of the structure ng 150% and 67% of the soil stiffness to capture the vari ability in the behavior were used in accordance with the requirements of ASCE 41-06. ‘The BRBs were modeled with a component consisting of Milf end zones and a yielding core. The backbone param- eters are shown in Figure 3. These parameters were based on prior testing performed by the BRB manufactures. Because there are no criteria for checking BRBs in ASCE 41-06, strains in the individual BRBs were checked to the crite- ria listed in Table 1. These criteria were derived based on a review of existing test data from the BRB manufacturer. It can be seen from Tuble 1 that separate criteria for the average of the seven time histories and maximum from any single event were used. Both criteria needed to be satisfied ‘Compression Fig. 3. Backbone properties of buckling restrained braces. 24 / ENGINEERING JOURNAL / FIRST QUARTER / 2014DETAILS OF THE SPLICED BRB As indicated previously, a spliced BRB was developed 10 help with the maneuverability of the BRB members within congested areas. The spliced BRB consists of two BRB seg- ments; each segment alone is similar to a standard BRB con- sisting of a yielding core, transition zones and nonyielding tend zones (see Figure 4), The concept behind the spliced BRB is that the splice zone is designed 10 remain elastic, (similar to typical end zones). The core atthe splice zone has, 2a cross-sectional area much larger than the yielding core, and the splice of the casing was designed to have flexural stiffness exceeding the casing along the remaining length of the BRB, Several conceptual splices were assessed, includ- ing welded and bolted options (ee Figure 5). The welded option was chosen based on the stability of the cruciform shape as well as the constructablity of the detail. The final detail consists of cruciform cores at the splice zone, which are connected with full-penetration ‘welds. The outer casing was connected with plates welded, to each other and to the casing with fillet welds, The final detail of the splice is shown in Figure 6. / Saif t i Fig. 5. Conceptual studies of splice options ENGINEERING JOURNAL / FIRST QUARTER / 2014/25Note that the yielding core of the spliced BRB is shorter than a standard BRB of the same length because there are essentially four end and transition zones, doubling that of standard BRB. The analysis model included this to capture the correct load distribution within the structure as well as, the ductility demand on the cores. Locations of the spliced [BRB within the structure were closely coordinated with the ‘contractor during the design phase, Many site visits and some inspection openings were made during the design phase to ‘work out the locations of the spliced and nonspliced BRB. ‘Subsequently, many iterations of the model were created. BRB PROTOTYPE TESTING Because spliced BRBs have not been implemented before, prototype specimens were fabricated for testing. The objec- tives of the testing program were as follows: ‘+ To validate that the spliced BRB is feasible and the bbchavior is as anticipated, + To determine any potential obstacles related t0 the ‘erection of the spliced BRBs. ‘+ To ensure that the two halves ofthe spliced BRB can be properly aligned given the congested field conditions. 1 RACCESS HOLE IN TRANSVERSE STIFFENERS. (CENTERED ON LEG OF MAIN ‘CORE PL BEVEL, TYP LUNE oF ‘CASING, TYP 2. CENTER PLSINCASING ‘he MAX TOLERANCE 1 Pte, TYP Fig. 6, Final details of splice 26 / ENGINEERING JOURNAL / FIRST QUARTER / 2014+ Todevelop quality assurance requirements for splicing. BRBS in the field. ‘Additionally, testing on three unspliced BRBs was per formed to study the effects of the following: + Performance of pin-ended BRBS with casing fab- ricated with circular HSS section, The pin-ended BRB was developed to be a high-performance brace that includes some of the special detailing not used in previous versions. Thus, testing was performed to validate the performance. + Performance of relatively long and slender BRBs, [Note that this was also intended to be a control, or ‘comparison, tothe spliced version. DESCRIPTION OF PROTOTYPE SPECIMENS A total of six specimens were tested. The yielding cores of the BRB were fabricated with ASTM A36 plate stock with the following properties: F, = 41.9 ksi and Fy = 71.6 ksi based on average of two coupon tests using the 0.02% off- ‘set method The gusset plates were made with ASTM A572, Grade 50 steel. Three specimens were spliced. Two of the spliced spe ‘mens were 21 ft long and the third was 31 ft long. Three specimens were not spliced: two 12-ft-long pin-ended speci- mens, and one 31-ft-long bolted-end specimen. Properties of the test specimens are summarized in Table 2 DESCRIPTION OF FABRICATION ‘The prototypes were fabricated at the manufacturing facility in Utah, The construction of the spliced specimens (speci: mens 3,4 and 5) was observed to develop procedures to per- form the splices and to identify potential problems during, construction, Table 2. Summary of BRB Test Specimens Length, Ls | Yield Length, L, End Pree Number | ID (ty (in) Casing Size | Connections | _(kips) | Spliced 1 | tater | 12 86.2 HSS 10.6 Pin-end 250 No 2 | se01az | 12 86.2 HSS 10.06 Pin-end 250) No. 3 | toot | at 1256 HSS10x100% | Bolted-end 230) Yes. 4 | soa | 21 1256 HSS10x10.% | Bolted-end 230) Yes. 5 | 903 | ot 238.7 HSS10x10.1 | Bolted-end 250 Yes 6 | ia | st 268.5 HSS10<100 | Bolted-end 250. No Method of Alignment One of the objectives of the prototype testing was to deter mine whether the two halves could be properly aligned in the field, Alignment plates to align the cruciform cores were proposed (see Figures 7 and 8). The alignment plates were removed in stages as the cores were welded together. ‘To align the casing, two pairs of steel angles were clamped ‘onto the steel casings. A small amount of welding was used tw connect the angles with the casing (see Figure 9), The alignment angles were removed in stages as the casing splice plates wore welded. Figure 10 shows the Finished spliced condition ‘The original intent was to make the splice in the hori- zontal position (on the ground) prior to lifting the BRB into the final locations. However, it was determined that this required substantially more floor area than is available in the building. During one of our visits to the site during the design phase, the contractor requested the option of lifting, ‘each half of the BRBs into the inclined position (final posi tion of the brace) prior to performing the splice. In order to replicate the actual condition of the site, one of the speci- ‘mens was spliced in the inclined location (ee Figure 11) As shown in Figuee 11, the prototype was supported at the splice point by a crane. In the field, this would be accom plished by a combination of chains and come-alongs. The alignment, welding of the cores and welding of the casing plates of the specimen were performed in this position Effects of Welding of the Core ‘The cores were spliced with prequalified, full-penetration welds. The weld of the core splice was treated as a demand- critical weld with associated quality assurance (ultrasonic testing, magnetic particle testing) even though no yielding is expected to occur at this location. The maximum thick- ness of the core plates was 1 in, Some amount of distortion was observed, primarily due to the heat generated during the welding process. AS a result, some out-of-straightness ENGINEERING JOURNAL / FIRST QUARTER / 2014/27‘was measured after the cores were welded. It was noted that, the sequence of laying down the weld beads was an impor tant aspect of controlling the amount of distortion (ie, the ‘welder needed to alternate the weld placed on each leg of the cruciform core). Additionally, in order to prevent initial stressing of the core due to heating and subsequent cooling, of the core, tightening of bolts at the end connections was, permitted only after the splice was complete and adequate time passed for cooling of the welded one. Tolerance As discussed earlier, the cores were aligned using align- ‘ment plates. Following this procedure, the outof-plane offset tolerance of the cores was limited to Yis in, The out-of-straightness tolerance (both in-plane of the core and out-of-plane of the core, see Figure 12) was initially set to be Vis in, over the length of the completed brace. This was measured prior to welding of the cores. As part of the field duality assurance, a second measurement was required after the cores were welded. For this measurement, the tolerance wwas relaxed to 4 in. over the length of the brace. It was noted that during the welding of the splice of specimen 5, the outof-straightness tolerance was exceeded in the out. of-plane direction. The measured out-of-straightness in the specimen was 4 in. over the length of the brace. This was likely due to sequence of laying down the weld beads. In lew of attempting to straighten this specimen, it was decided to test the specimen to study the effect on performance of some Fig, 9 Alignment aid for casing. 28 / ENGINEERING JOURNAL / FIRST QUARTER / 2014 Fig, 10. Finished splice.Fig. 11. Prototype spliced in inclined position. ENGINEERING JOURNAL / FIRST QUARTER / 2014 / 29‘Table 8. Measured Alignment on Test Specimens with Spliced BABS ‘Alignment before Welding | Alignment after Welding Core + Stiffener Core + Stiffener (in.) but before Welding Casing Plates (in.) Number 'o In-Plane _ | Out-of-Plane In-Plane Out-of-Plane 3 190241 He te ie ay 4 190242 ete ote <é 02d, the following equations, which are based on a plate-buckling model (Muir and Thornton, 2004), are applicable, AQ 6 ‘The reduction factor for plate buckling is when 2<07 Lo (a) when 0.7 <2. 1Al 34-0486) (6b) when > LA (60) ‘The slenderness parameter is, ig ae ie may ue vse") Figure 2 shows a plot of the critical stress, Fy, versus Inj for e/d = 1 and 1, = 03 in. The eritial stress for le eral torsional buckling is calculated with Equations 3 and 4 without the yield stess limit. The critical stress for local p24 ~ Face of Cope Setback, | Out Plane Restraint Ry + 4x Fig. 1. Double-coped beam. 44 / ENGINEERING JOURNAL / FIRST QUARTER / 2014 buckling is calculated with Equations 5, 6 and 7. It can be seen that the two curves are trending in opposite directions, with F,, increasing with h/d for lateral torsional buckling and decreasing with hy /d for local buckling. This indicates the need for a single. continuous function that covers the entire range of applicability AISC Specification Section FLL Because the Manual equations developed by Cheng et al (1984) were based on a lateral torsional buckling model, AISC Specification (AISC, 2010) Section FIT will be reviewed here. Section FI] provides design information for the flexural strength and stability of rectangular members bent about their major axis, For yielding, M, = M,=F,ZS1.6M, 0. For inclastic lateral torsional buckling, waofis-om(! nan, @ z For last lateral torsional buekling, £84 5 LE mare My = FS 5M, co) ‘The critical stress is ay \ © \ g 0 ao 2 Re 20 - —Latera-Torsional Buckling Local Buckling ° 00 02 08 06 08 10 nda Fig, 2. Critical stress versuslateral torsional buckling modification factor stance between brace points, in ‘Mg = nominal moment, kip-in Id moment, kip-in, lastie moment, kip-in. lastc section modulus, in.” laste modulus, in." yeam width, in M, My Ss Zz ' Equation 11 is the theoretical solution for lateral torsional buckling (Timoshenko and Gere, 1961) multiplied by C, and, simplified by substituting the properties for a rectangular cross-section. It can be shown that Equation 3 is equal to Equation 11 by substituting + 19 Ly =e and Ch =f into Equation 11. Therefore, fy is simply a lateral torsional ‘buckling modification factor applied to the theoretical equa- tion for the critical moment of a rectangular beam, FINITE ELEMENT MODELS AISC Specification Section F2 equations for lateral torsional buckling of wide flange beams are based on the theoreti cal solution (Timoshenko and Gere, 1961), with C, factors developed primarily using elastic finite element models. The inelastic portion of the buckling curve was developed by ‘mapping, based on limited testing and finite clement results, inthe inelastic zone, Because much of the inelastic research ‘was based on a constant moment along the beam length (Cy = 1), the full beam length was inelastic. Therefore, the Duckling curves are conservative for C, > 1 because they {don’t account for partial inelasticity along the beam. This, ‘same procedure was used in this research to develop equi tions for the local stability of coped beams. The finite element program was designed to address three issues related to the local stability of double-coped beams: 1. Cope depths greater than 20% of the beam depth. 2. Unequal cope depths at the top and bottom. 3. Unequal cope lengths at the top and bottom. ‘A parametric study consisting of 54 elastic, finite element ‘models was used to determine the effect of each variable on the critical load. Using the variables shown in Figure 3, the ‘program consisted of 30 models with ¢,=¢,, 12 models with, ¢,> ¢, and 12 models with c; < cy, The details are listed in Appendix A, Tables A1, A2 and A3, respectively. ‘All models were built with the nominal dimensions of a W16x26. Models for additional beam sizes are not required, Decause the critical moment is proportional to 1° for beams, ‘with identical proportions on the cope geometry, Following, the modeling techniques of Cheng etal. (1984), BASP finite clement software was used to determine the critical loads, assuming the flanges were laterally braced atthe face of the cope. There was no setback dimension in the models; there- fore, ¢,=¢, and ¢ =e. RESULTS Accuracy of Manual Equations For the models with equal cope lengths atthe top and bot tom flanges, Table Al in Appendix A compares the finite element results to the current design procedure in the AISC Manual (AISC, 2011). Cohuma 6 lists te critical reactions from the finite element models, Rj, and column 7 Tists the critical reactions from AISC Manual equations, Rx. Models 6,11, 12, 18 and 16 had d, $0.24; therefore, Re was calculated with Equations 1 through 4, For the remaining specimens, Reg was calculated with Equations 5, 6 and 7. ‘The average Rje/R ato, listed in column 10, is 1.54 with a standard deviation of 0.496. Design Model All of the finite clement models buckled in a similar man- ner, as shown in Figure 4. Confirming the results of Cheng et al. (1984), the tension edge of the coped cross-section experienced lateral translation, and the shear center expo- rienced lateral translation and twisting. The compression edge of the coped section buckled in the shape of a half sine wave, which extended partially into the uncoped portion of the beam due to lateral translation atthe reentrant corner of the cope, ‘To form a design model, the buckling mode must be iden- tified. The buckled shapes have the appearance of several independent modes, including local buckling, lateral tor- sional buckling, shear buckling, and distortional buckling. -— ce Sh e Fig, 3. Different cope si sat the top and bottom flanges. ENGINEERING JOURNAL / FIRST QUARTER / 2014 / 45‘The dominant buckling mode is dependent on the cope ‘geometry, Short copes are controlled by shear buckling, and, Tong copes are controlled by lateral torsional buckling, with, some aspects of local buckling and distortional buckling, present in all cope geometries. Because the buckled shapes most closely resemble lateral torsional buckling over the critical variable range, the design model is based on Equi tion 11, with the buckling modification factor, C,, account- ing for contributions from the other buckling modes. Factor, , was determined by curve-fitting the finite element d Curve-Fit Equations ‘The required flexural strength at the face of the cope is M,= Rea «2 ‘The nominal moment is calculated with Equations 10 and 11 with = fy and d = hp, The equation for Cis dependent on the ¢,/¢y ratio. For beams with c, =), Cy is calculated using, Equation 13 with Ly = cj = cj. For beams with ¢,< cy, Cis calculated using Equation 13 with L, = 0.9¢;+0.1ep, cla Fig, 4, Buckled shapes: (a) ¢,= 6; (0) & > &y (0) 6, < 46 / ENGINEERING JOURNAL / FIRST QUARTER / 2014 For beams with c; > cy» Cp is calculated using Equation 14 with La = Ct 69/2 oe fomnfEoe 63] where ¢4 = length of bottom cope, in, & =length of top cope, in dy = depth of bottom cope, in. da; = depth of top cope, in. ¢4.= distance from the face of the bottom cope to the end reaction, in ¢ =distance from the face of the top cope to the end reaction, in nig = minimum of e, and ey ‘The results for all models are listed in Tables Al, A2 and A3 in Appendix A. For beams with ¢, = cy, the aver age, finite eloment-to-caleulated load ratio is 1.1 and the standard deviation is 0.0535. For ¢, < cy, the average load ato is 1.02, and the standard deviation is 0.0902. For c, > «em the average load rato is 1.06, andthe standard deviation is 00752, Equation 13 is plotted in Figures 5 and 6 with the finite element results for c, =). Figure 5 shows Cy, versus ¢;/d for four values of ded. Figure 6 shows Cy versus day/d forfour values of ey/d. DESIGN Moment-Shear Interaction ‘The design procedure in the AISC Manual (AISC, 2011) uses beam theory as the basis for calculation of the flexural stresses. Because the maximum normal and shear stresses occur at different locations on the cross-section, combin- ing these stresses is not required. However, the design pro- cedure proposed in this paper utilizes the plastic flexural strength. Because the plastic stress distribution requites the ‘maximum shear and normal stresses to act atthe same loca- tion on the cross-section, the flexural strength is reduced in the presence of shear loading. For short cope lengths, the requited shear load can be close to the shear yield strenath. To account for the interaction between the flexural and shear loads, a reduction factor can he applied to the plastic ‘moment capacity, M,, Neal (1961) developed Equation 15 for the plastic capacity of a rectangular member subjected to ‘moment about one axis, axial load and shear, as)where P= required axial load, kips xia yield load, kips quired shear load, kips hear yield strength, = 0.6F at. ips ‘The plastic moment strength, reduced to aecount for the required shear load is Because M,, must be based on the available shear strength rather than the nominal value, Equations 17a and 17b should be used in design: (16) oo 01 02 03 ad LRFD My (7a) ASD My, «iT its of Applica Because the curve-fit equations were derived using finite ‘element models, the range of applicability should be based ‘on the cope geometries studied. The geometry of the ten- sion flange cope has a limited influence on the buckling Toad; therefore, no limits are required on dey orcs. The finite clement models were limited to a maximum cope length of + FE (6ctld = 0.109) = FE (dctd= 0.207) + FE (ded = 0.04) + FE (dctld = 0.402) —Equation (devd = 0.109) — Equation (devd = 0.207) — Equation (devs = 0.304) = FE(Cud= 0.489) = FE (Cud=0978) + FE (CUd= 147) o4 08, Fig. 8. Cy versus dy / ENGINEERING JOURNAL / FIRST QUARTER / 2014/47twice the beam depth. Observation of Figure 5 shows that C), increases slowly for ¢,/d greater than 2: therefore, in the rare ‘case that c; > 2d, it is recommended that cy=2d is used in the calculations, Because the depth of the compression flange ‘cope was limited to 40% of the beam depth, the equations are valid only for des $O.4d. When ci/d is less than 0.5, the curve-fit equations can produce unrealistically low values for C). This is because shear buckling dominates the behavior for short copes. ‘To eliminate erroneous calculations, a lower limit ean be applied to Cy. According to equations developed by Dow- swell (2004), Cy = 184 for a rectangular cantilever beam loaded at the shear center with bracing at each end and a concentrated load at the tip. Therefore, 2 minimum value of C,= 184 is recommended, Design Proposal ‘To account for inelastic action, AISC Specification Section FIL can be used with ¢= 1, and d= hg, The following design procedure is suggested: For yielding, 2-< My = My, as) For inelastic lateral-torsional buckling, 2y <2 <2 «9 For clastic lateral-torsional buckling, 4 > hy: My= FS. 5 Mp 2, “The eritical stress is 19EC, f= eu Z x ) where a 3) 196 he 4 Fr ‘) 48 / ENGINEERING JOURNAL / FIRST QUARTER / 2014 Simplified Equations Simplified versions of Equations 13 and 14 can be used for {design purposes. For beams with ¢,= cp and beams with ¢,< ep, Ly =, and Cy is calculated with Equation 25. cfeltps © (¢ + ep)/2 and Cy is calculated with Equation 26. Abels ‘The simplified equations are compared with the Finite ele- ment models in Appendix A, Tables Al, A2 and A3. For beams with ¢)= cy, the average finite element-to-calculated load ratio is 1.18, and the standard deviation is 0.139. For ¢, < the average load ratio is 1.05, and the standard deviation 4s 0.0736, For ¢, > cy, the average load ratio is 1.19, and the standaed deviation is 0.0949, For beams with ¢,> ey, Ly CONCLUSIONS ‘This paper used the results of a parametric study to formulate 4 design procedure for the local strength of double-coped beams. Lateral torsional buckling modification factors, for use with AISC Specification Section FI, were formulated by curve-fiting the results of 54 finite element models. The proposed solution is based on a larger database of finite ele- ‘ment models than the Manual procedure, and the limits of applicability have been extended, In contrast to the Manual procedure, which has curves trending in opposite directions, the proposed solution provides a single, continuous equation cover the entire range of applicability. ‘The proposed design procedure was shown to be more accurate than the current Manwal procedure. For beams with equal cope Iengths at both flanges, the Manual proce- dure has an average finite element-to-caleulated load ratio of 1.54 with a standard deviation of 0.496, The curve-fit ‘equation developed in this paper (Equation 13) produced an average finite element—to-calculated Toad ratio of LOL with a standard deviation of 0.0535. For the simplified equation (Equation 25), the average finite clement-to-calculated load ratio is 1.18 with a standard deviation of 0.139 Using all $4 of the finite element results, calculations using the curve-ft equations (Equations 13 and 14) produced an average finite element-to-calculated load ratio of 1.02 and a standard deviation of 0.0665. The simplified design equations (Equations 25 and 26) had an average finite cle- ‘ment-to-calculated Toad ratio of 1.15 and a standard devia- tion of 0.115ACKNOWLEDGMENTS ‘The authors would like to thank Professor Joseph Yura at the University of Texas at Austin for providing the finite ele- ment program, BASP, used in this study. SYMBOLS Cy, = lateral torsional buckling modification factor ksi E_ = modulus of elast F, critical stess, ksi F, = specified minimum yield stress, ksi 1, = distance between brace points in My = nominal moment, kip-in My = yield moment, kip-in My, = plastic moment, kip-i. ‘Mpr= plastic moment, reduced to account for the required shear load, kip-i, M, = required moment, kip-in. P, = required axial load, kips P, = axial yield load, kips Q_ = reduction factor for plate buckling Ru = critical reaction with C; caleulated with the simplified design equation Rj. = critical reaetion from finite element model R, = required end reaction, kips Ry. = ctitical reaction with C, calculated with the original regression equation Sue) = elastic section modulus of the coped section, in. S, = clastic section modulus, in. Vy, = shear yield strength, kips Z_ = plastic modulus, in.? € = cope length, in. 4. = length of bottom cope, in a length of top cope, in. beam depth, in dy, = depth of bottom cope, in. de, = depth of top cope, in € = distance from the face of the cope to the end reaction, in. * tance from the face of the botiom cope to the end reaction, in ¢, = distance from the face of the top cope to the end reaction, in, nig= minimum of e, and ey fy = adjustment factor fh, = reduced depth of web, in. 1 = beam width in. fy = web thickness, in Q, = safety factor for shear 6, = resistance factor for shear = slendemess parameter iy. = limiting slenderness for the limit state of yielding 2, limiting slenderness forthe limit state of inelastic lateral torsional buckling REFERENCES: AISC (2010), Specification for Structural Steel Buildings, ‘American Institute of Stee! Construction, Chicago, IL. AISC 011), Steel Construction Manual, Ith ed,, Ameri can Institute of Steel Construction, Chicago, TL, Akay, H.UL, Johnson, C.P. and Will, KM. (1977), “Lateral and Local Buckling of Beams and Frames.” Journal of the Structural Division, ASCE, Vol. 103, No. STS, Sep- tember, pp. 1821-1832. Cheng, 1., Yura, J.A. and Johnson, C.P, (1984), “Design ‘and Behavior of Coped Beam,” Ferguson Lab Report, The University of Texas at Austin, July Dowswell, B. (2004), “Lateral-Torsional Buckling of Wide Flange Cantilever Beams,” Engineering Journal, AISC, Second Quarter, Vol. 41, No. 2 Muir, L.S. and Thornton, W.A. 2004), “A Direct Method for Obtaining the Plate Buckling Coefficient for Double- Coped Beams.” Engineering Journal, AISC, Third Quar- ter, Chicago, IL. Neal, B.G. (1961), “The Effect of Shear and Normal Forces ‘on the Fully Plastic Moment of a Beam of Rectangular Cross Section,” Journal of Applied Mechanics, Vol. 28, pp. 269-274 Timoshenko, S. P. and Gere, J. M. (1961), Theory of Elastic Stability, 2nd ed., MeGraw-Hill, New York. ENGINEERING JOURNAL / FIRST QUARTER / 2014 / 49,APPENDIX A. TABLES 2, =e! eacton fom A'SC Mans! equations [R,~erieal ection with C, caleutod ws th avn reptecslon equation ‘i, =a feacton wih G, clcuated wie te sme design equation ‘80 / ENGINEERING JOURNAL / FIRST QUARTER / 2014 dered ecto eet) ce seg ec ae ae | pce (pt | a | Number | (in.) (in) fin.) Gin.) | (kips) | (kips) | (kips) | (kips) | Ree Bre wo psa [tea [trp ear | tea | 710 | 142 | er ae | 1 a8 ap wor [sor [tmp in} sox] te | eae] 027] 1108 | o9a8 | 0970 wf sor [sor | aze pari] 5x6] sz | 520] 0] 158 | nee | 108 wear [sor [are pani | 496 | aa | 420 367 | oar | 108 [48 efor [sor [est pari | 30s | a6 [37 P26 [1.05 [108 138 we_[-so7 | sor [a7] sar | sz0| aso | eco] ea0 | 116 | 0920 | 066i we_[ wor | sor | ape | sat | aoe] 260 | eae] sto | 155 | 0970 | 108 a7 [-s07 | sor | are | sat | oer] aus | aso] one | 100 | 402 [iar we_[ or | sor | a7] ave | eat | sas | ast] 470 | ceo | oon6 | oon? so sor [aor [ave [are | 203] 2a6 | 200 | 2a a8 [016 Ress 154 [11 [178 Standard | 5.496 | o0sss | 0.199Table A2, Finite Element Results with ¢; > ce tesa |e cep ce |etae | ogi |e time | time |e Number | __(in.) (in) (in) (in) (kips) (kips) {kips) re o2_ [sor | isa | aze | am | er | s72 | sas | 120 | 109 oa | sor_| 154 | 1m | 171 | ear | aza [771 | 100 | 107 36 | sor_| 154 | 17 | am | 70a | 721 | 674 | vor | 1.414 3a [sor | 154 | a24 | 171 | 7a1_| er | 600 | 111 | 129 40 [154 | 708 | a2 | am | m5 | v92 | 14 | tor | 125 | 1sa_| res | 1m | 171 | 79 | ore | os | tor] 1.14 “| tsa | 7es | 171 | am | as7 | 242 | ata | 108 | 120 ae | 15a | 7ea | a2e | 171 | aa | 204 | 191 | o9s | 1.17 as | or | rea | sas | am | 794 | ea | sea | 119 | 126 so_|s07_| ree | 1m | 471 | 910 | 957 | 075 | 95a | 1.06 se_|so7_| ree | 17 | az | evo | ear | 705 | 105 | 1.35 ss_[ sor | ree | o2e [471 | 025 | 775 | oe: | 108 | 121 ‘Average | 1.08 | 1.19 ‘Standard Deviation | 0.0752 | 0.0949 FR, ~eitcaleacton fom i A, ~ertcal ection wth, cal A, meres waction wih C clea jevent nodal ith th onal rgrcon equation th the implied deign extn ‘Table AS. Finite Element Results with o, < Gy Model Cs ce dee es Re Pre Ree Boe Bee Number | _(in.) (in) fin) fin) | (kips)_|_tkips)_| (kis) Roe 31 154 | 307 3.24 24 | 141 14.7 13.6 0959 | 1.04 33. 154 | 307 171 a7_| 218 214 203 or | 1.05 35, 154 | 307 171 aaa | 185 185 17.8 100 | 1.04 37, 154 | 307 324 171 | 165 174 158 0982 | 1.04 39 768 | 154 324 a24 | 428 45.1 a7 sae | 1.03 a 768 | 154 171 171 | 726 65.0 624 112 | 1.16 a 768 | 154 171 324 | 644 569 546 113 [1.18 45 7.68 | 154 3.24 471 | 487 52.6 406 0.926 | 1.00 47 7.68 | 307 3.24 324 | 403 a7 a7 0.967 _| 0.967 49 7.68 [307 171 171 | 683 60.0 62.4 44 | 1.09 51 7.68 | 307 171 a2za_| 613 525 546 447 | 142 53 7.68 [307 3.24 17 | 455 | 486 48.6 0.936 | 0.936 ‘Average | 1.02 | 1.05 ‘Standard Deviation| 0.0902 | 0.0736 Fa, ~eitealeacton for fi, ~ertcal ection wth, cal fa. =etea reacton wth, alot jzvent modal ith th onal rgrcon equation it the simi dos etn ENGINEERING JOURNAL / FIRST QUARTER / 2014/51'52/ ENGINEERING JOURNAL / FIRST QUARTER / 2014ERRATA The Behavior of Steel Perimeter Columns ina High-Rise Building under Fire Paper by MARIA M. GARLOCK and SPENCER E. QUIEL (4th Quarter, 2007) Figure 11 should be replaced with the following figure 10 09 os o7 08 fos oa 03 02 ot 00 o 2 04 08 o8 1 0 02 04 06 08 1 “ime (hs) “rane (rs) @ Fig, 11, Computational ($APIR) results for (a) axial load ratio and () plastic momens rato for the column just below Floor 23 ENGINEERING JOURNAL / FIRST QUARTER / 2014/53‘84 / ENGINEERING JOURNAL / FIRST QUARTER / 2014ERRATA Flange Bending in Single Curvature* Paper by BO DOWSWELL (nd Quarer, 2013) Equations 33 and 38 should be replaced with the following: Cp = 181 ~ LISA + 1.0667 a3) Cy =2.23 1492-4 1.3989 38) + Another erat for this paper appears in 3d Quarter 2013 ENGINEERING JOURNAL / FIRST QUARTER / 2014/55SOOPE: GENERAL: MANUSCRIPT PREPARATION: GUIDE FOR AUTHORS ‘The ENGINEERING JOURNAL. 8 dees 1 the imyoverent sd advancement of sel consttion. Ils pages ae open tall who wish to report on now development of tchnigus i stzel design, researeh, the {design andor consution of new peojs ie ibicaion methods, o© new products of significance tthe uses of sce in consttion. Only ‘sina papers shoud e sbi Papers tended for publication may be submited by mail to the Editon Keith Grubb, ENGINEERING JOURNAL. AMERICAN INSTITUTE OF STEEL CONSTRUCTION, One East Wacker Drive, Site 700, Chicago, TL, 6060 by ema to srubbeaisc one ‘The antlespubishod inthe Bnginering Journal undergo peer review before publication for (1) origina of sation: (2) tecnial vale to the see! construction community: (3) proper ero others orking in ‘he same area: 4) por publication of the material and (8) ustification of the caclison bese om the repo Alpers thine cope otned abo willbe rviewod by engineers seteted from among AISC, indasy, design fim, and universies The standard review proces inciodes outside review by an average of tts reviewers, who ae experts in their rspoctive tachnial rea, and ‘volunters in the progam, Papers not accepted wil ot be feted othe stor. Published papers bocome the property of the American Insti of Sel Construction and are poteted by appropriate copyrights. No proofs vill be sent to author. Each author receives thc copics of the ise in ‘shish his contribution appears “Manuscripts must be provided in Microsoft Word format Include a PDF ‘wth your submit. View our complete author guidelines at ww. ore‘There's always a solution in steel. ENGINEERING JOURNAL ‘American Institute of Steel onstruction ‘One East Wacker Drive, Sulte 700 Chicago, IL 60601 312.670.2400 wwwaise.org