INFLUENCE OF COLUMN STRAIGHTENING PROTOCOL ON CONNECTION PERFORMANCE

Reidar Bjorhovde The Bjorhovde Group, Tucson, Arizona, U S A

ABSTRACT Some of the major findings of a study into the performance of certain types of beam-to-column connections are presented. 17 full-scale specimens used common beam and column sizes as well as welded flanges and cover plates and bolted webs; they were designed in accordance with the AISC Seismic Specification. Some connections had rotary straightened columns, others used gag straightened or unstraightened members. Loading was applied quasistatically or dynamically. The results show that the form of column straightening has no effect on connection performance. INTRODUCTION Steel has been the construction material for a large number of buildings, bridges and other structures for more than 100 years. Its elastic and inelastic responses to loads and load effects make it a material with predictable and reliable behavior under a wide range of service conditions. The ease and speed of fabrication yield significant economies of construction. The most complex elements of a structure are the connections. For buildings this is especially true for the beam-to-column connections, where details and fabrication processes combine to produce three-dimensional conditions. For reasons of economy and fabrication, over the years certain connection types became common. They were proven through uses in many structures, designers were confident about design methods and details, and fabrication was of high quality. In particular, connections with welded beam to column flange joints and a bolted web were used extensively. Tests and analyses showed that these connections were capable of developing appropriate moment and shear capacities, and the deformation characteristics were very good. Use of steel structures in high seismicity areas was considered especially advantageous, due to the inherent inelastic deformation capacity of the material, and the structures performed well in a number of minor and major earthquakes. However, the understanding of seismic effects and structural behavior has advanced significantly over the past ten years, and tools such as computers have facilitated increasingly fine-tuned designs. Structural systems have also changed, to satisfy space demands of architects and owners. As a result, structures in some ways have become simpler, with fewer load-carrying elements, but at the cost of reduced redundancy. The effects of earthquakes, for example, therefore have had to be accommodated by fewer members and especially fewer connections. The 1994 Northridge earthquake had a major influence on state-of-the-art thinking about ductile structural response. Cracks were found in the connections of many steel-framed structures, and it was perceived that the earthquake was the primary reason. It is now clear that many of

403

the cracks had occurred before the earthquake, and that the same type of cracking had taken place in structures in the absence of seismicity. However, it was also evident that accepted material and structural behavior and design and fabrication approaches needed re-examination. As an example of a non-seismic cracking incidence, the fabricator for a large California project experienced cracking in the column during the shop fabrication of beam-to-column assemblies. The connections had welded flanges and bolted webs, as well as welded beam flange cover plates and column continuity plates (stiffeners). Figure 1 shows the connection that was used. Although the member sizes shown here are not identical to those of the actual project, they are representative of what was utilized.

Figure 1 Connection with As-Built Details The cracks in the as-built connection were found in the web of the column, in the region of the cross section commonly known as the k-area. This is a small area of the web surrounding the location where the transition fillet from the flange enters the web. The k-dimension measures the distance from the outside of the flange to the end of the fillet. Originally the occurrence of the earthquake- as well as the fabrication-related cracks was attributed to inadequate material properties of the steel. Much discussion took place about yield stresses that were significantly higher than the specified minimum values, and some investigators opined that this was a result of faulty steel mill practices. Additional problems evolved as subsequent examinations found that the k-area tended to exhibit high strength and hardness, but low ductility and fracture toughness. These characteristics are related to the fact that many sizes of wide-flange shapes are rotary straightened in the steel mill as a matter of course, to meet the straightness requirements of ASTM (1). Although the straightening is common practice for steel mills all over the world, the properties and response of the k-area material is nevertheless a phenomenon that merited study. These issues are the background for the study that is presented here. The complete test data are given in the report by Bjorhovde et al. (2).

404

SCOPE OF TESTING PROGRAM Table 1 gives the details of the 17 tests. All specimens used W14x176 columns and W21x122 beams; the steel grade was ASTM A572 Grade 50. Table 1 Major Features of Connection Testing Program Number Of Tests 8 6 Conn. Type As-Built Connection Description* CJP welds for beam to column and continuity plates to column; 1-5/8 cover plates; 10 1-1/8 A325 bolts for beam web Straight. Protocol** 5R 2G 1U Load Prot*** 4Q 4D 4Q 2D 1D 2D

Revised As As-Built, but with 1 inch cover plates and fillet 5R Type 1 welds for the continuity plates 1U 1 Revised As As-Built, but with a inch fillet weld transition 1R Type 2 from the beam to the column 2 Revised As Revised Type 1, plus transition fillet weld 2R Type 3 and repositioned continuity plates * CJP = complete joint penetration ** Straightening protocols: R = rotary; G = gag; U = unstraightened *** Loading protocols: Q = slow cyclic (quasi-static); D = dynamic (1 Hz frequency)

Figure 2

Revised Type 1 Connection

Figure 3

Revised Type 2 Connection

The eight as-built connections reflected current practice. The six Revised Type 1 specimens used thinner cover plates, and in a major departure from current practice it was decided to use the more economical fillet welds for the continuity plates. The single Revised Type 2 specimen had the slight change of the inch transition fillet weld

405

between the beam (cover plate) and the column flange, in an attempt to provide a better force transfer path. Figures 2 and 3 show the Revised Type 1 and 2 Connections. The Revised Type 3 connections had the same details as the Revised Type 1, plus the inch transition fillet, and, in what turned out to be an important modification, the continuity plates were repositioned to allow for a better load transfer and fracture path for the connection. In traditional usage the continuity plates would be placed with their mid-thickness line at the same level as the interface between the beam flange and the cover plate. For the Type 3 connections it was decided to place the continuity plates with the outside edge in line with the beam flange to cover plate interface. Figure 4 shows the details of the Revised Type 3 connection.

Figure 4 Revised Type 3 Connection COLUMN STRAIGHTENING PROTOCOLS One of the issues that prompted the decision to perform the series of tests was the performance of the steel itself in the web of the column. Specifically, the steel was called into question because of the cracks that had developed in the k-area of some columns during welding, and because of the cracks that were found in some structures after the Northridge earthquake. It was also noted that the k-area is deformed heavily because of the nature of the rotary straightening that is used by steel mills to meet the straightness requirements of the materials delivery standard (1). Since this form of straightening is applied continuously, the localized areas of changed material characteristics appear along the full length of a member. Another factor in the rotary straightening process is the initial out-of-straightness of the shape as it comes from the cooling bed in the steel mill. It is possible to have shapes that are very straight after rolling; these require less straightening effort to meet the delivery criteria. On the other hand, some shape lengths may be fairly out-of-straight; these require larger straightening loads. For ease of production, current US steel mill practice dictates that all shapes be straightened.

406

Heavier shapes are straightened by the application of concentrated loads at discrete points; this process is termed gag straightening. This procedure does not introduce continuous k-areas of high strength and low ductility in the manner of the rotary straightening because of the way the load is applied. Unstraightened shapes are not commercially available. However, for the purposes of the testing program it was possible to obtain two unstraightened members, to establish the response of a connection assembly in the absence of the straightening effect. Overall, as Table 1 shows, of the total of 17 tests, 13 specimens had rotary straightened columns, 2 used gag straightened members and 2 utilized unstraightened columns. It was felt that this testing array would establish conclusively the influence, if any, of the straightening protocol for the columns. LOADING PROTOCOLS A number of beam-to-column connections have been tested in past research projects. Many of these used slowly increasing or effectively static loads. Recognizing the importance of dynamic and especially seismic response characteristics, the US Applied Technology Council (ATC) has developed criteria that are based on slow cyclic loads (3). These are often referred to as quasistatic loads, since it is not attempted to model actual earthquake loading. Rather, using a displacement control approach, the load is applied in alternate directions, with increasing amplitudes of the load application point. This is the loading protocol that was used for eight of the connection tests; they are identified by the letter Q in Table 1. More recent studies have emphasized the need to have the test loading simulate seismic conditions as closely as possible. This led to the development of criteria that focused on loads being applied at certain loading or strain rates, to mimic the earthquake response of the structure. Although the discussion continues as to whether it is more realistic to use dynamic loads, it is all but certain that these will impose more severe demands on the structure and its connections. It was therefore decided to test the second group of nine specimens dynamically, using a frequency of 1 Hz. These tests are identified by the letter D in the Loading Protocol column in Table 1. Three complete cycles were used for each beam tip deflection, with a maximum displacement of 10 inches. CONNECTION DESIGN The connections were designed in accordance with the criteria of the AISC LRFD Specification, including the 1997 and 1999 seismic requirements (4, 5). In order to achieve optimal seismic performance of structures and their elements, current US principles utilize the strong column, weak beam concept, whereby plastic hinges will form in the beams at the ultimate limit state. This leads to improved redundancy and ductile failure modes for the structure as a whole. However, for the tests of this project it was decided to impose the most demanding conditions possible on the column material, primarily since the cracking that had been observed took place in the columns. It was felt that this would represent a worst case scenario. The test specimens therefore reflect assemblies with strong beams and weak columns, where plastic hinges will form in the column panel zones.

407

DISPLACEMENTS AND ROTATIONS Measure of Performance The primary measure of the performance of a connection is its plastic rotation angle (capacity), identified by the symbol p . This expresses the ability of a connection to sustain plastic deformations prior to failure, and is therefore regarded as a criterion by which the connection can be evaluated for seismic performance and suitability. Assumptions If the test frame is infinitely stiff, the true beam tip displacement would be defined as the vertical deflection at the beam end, measured from its original, undeformed position. However, the frame is not rigid, and deformations of some magnitude will take place. Therefore, if the displacements were measured in relation to the test frame, the frame deformations would be added to the true specimen displacements, yielding incorrect data. To account for these effects during the testing, vertical and horizontal displacements were measured at the top and bottom column pin supports, at the center of the pin at the beam end, and at the center of the column panel zone. These data were then used to determine the actual beam tip deflections and to eliminate the influence of the test frame flexibility. In addition to the vertical translations of the column end supports, horizontal movements will occur at both column ends as well as at the panel zone center. The sense of these displacements will indicate whether the entire test specimen rotates as a rigid body around a point on the column close to the bottom pin. Testing of a separate specimen (not one of the 17 listed in Table 1) showed that the top column pin support moved horizontally, in direct proportion to the beam tip deflection, with magnitudes up to 0.3 inches. The lower column pin also moved, but in the opposite direction; these displacements were never larger than 0.04 inches. It was decided to treat this deformation as insignificant, concluding that the entire specimen rotated as a rigid body about the bottom column pin support. Plastic Rotations The deformation demands of an earthquake are partly accommodated by the elastic displacements of the frame. Additional deformations have to be provided by the structure in the form of plastic hinge rotations in the beams and by plastic deformations in the column panel zones. The FEMA Interim Guidelines (6) that have been developed over the past several years recommend that new steel-framed construction should be able to accommodate plastic rotations of at least 0.030 radians in the connection regions. The requirement is reduced to 0.025 radians for retrofitted structures. Further, these rotations must be sustained for at least one full cycle of loading. Cumulative Plastic Rotations as a Measure of Energy Absorption The most common presentation of connection test results has been in the form of plastic rotations, the number of cycles to failure, and whether the specimens met the FEMA criteria. Hysteresis loops and failure modes and details are also given, but no data are provided on the

408

overall energy absorption capacity of the connection, which is a central measure of suitability for seismic conditions. For the past several years, Japanese reports have included data on cumulative and normalized cumulative plastic rotations (7). The former is the sum of the plastic rotations associated with each cycle of loading; the latter is a relative measure of the same. It is evident that the cumulative plastic rotation of a connection reflects its energy absorption capacity, and therefore provides key information on its performance ability. The survival of a connection for one cycle, as specified by FEMA (6), says little about its potential response under sustained seismic activity. It is understood that updated Japanese seismic criteria now utilize a cumulative plastic rotation capacity of 0.3 radians as the measure of acceptable performance for a connection. The cumulative plastic rotation p = p is defined as the sum of the individual plastic rotations that occur during each complete half cycle of the connection test. The quantity also includes the excursion amount that occurs at failure. PERFORMANCE OF MATERIALS AND CONNECTIONS Materials Testing Detailed and extensive testing was conducted for the material in the columns of the test specimens, since the cracking and eventual connection failure was expected to occur in these members. The tests included the usual uniaxial tension tests, chemical analyses, Charpy VNotch fracture toughness tests, and Rockwell B hardness tests. In addition to the ASTMrequired tension (1) specimen location in the flanges of the wide-flange shapes, tensile tests were also conducted for the material in the k-area, web and core. The fracture toughness samples were taken from the same areas in order to map the variability of the CVN values of the material; similar mapping was used for the hardness tests. As expected, the tensile properties of the web and flange steel met and reasonably exceeded the minimum requirements of the ASTM A572 (50) standard. The k-areas of the rotary straightened columns had higher yield and tensile strengths and lower ductility, also as anticipated. The gag straightened and unstraightened members exhibited nearly uniform strength and ductility at all locations. The flange and web locations of the rotary straightened columns had excellent CVN toughness; the core and especially the k-area values were much lower. The gag straightened and unstraightened columns did not display reduced toughness in the k-area, other than as part of the typical pattern of variability. The single lowest CVN value for any of the test specimens was 5.0 ft-lbs; it was found in the k-area of one of the rotary straightened columns. The hardness tests showed the same tendencies as the CVN values. These results also delineated the extension of the high hardness area, showing that it typically extends approximately inch into the web, beyond the end of the transition fillet. Connection Tests Figures 1 through 4 show the connections that were tested, and Table 1 gives the structural and fabrication details. It is emphasized that for all but the Revised Type 3 connections, the continuity plates were located with their mid-thickness line at the same level as the interface

409

between the beam flange and the beam flange cover plate. The Type 3 connections had the outside edges of the continuity plates in line with the beam flange-to-cover plate interface. Figure 5 shows the column face moment vs. plastic rotation hysteresis loops for Specimen A4. This was an as-built connection with a rotary straightened column; it was tested dynamically. The somewhat ragged nature of the curves is a result of the dynamic testing and the high rate of data recording. Failure was initiated through a crack at the toe of the bottom cover plate to column weld. Full stiffness was maintained through one complete cycle of 0.028 radians.

Figure 5 Moment vs. Plastic Rotation Curves for As-Built Connection A4 Figure 6 shows the hysteresis loops for Specimen A7. This was an as-built connection with a gag straightened column; the testing protocol was dynamic. The failure was identical to that of A4, except that the initiating crack took place at the toe of the top cover plate to column weld. The full stiffness was maintained through one complete cycle of 0.028 radians.

Figure 6 Moment vs. Plastic Rotation Curves for As-Built Connection A7

410

Figure 7 shows the hysteresis loops for Specimen A8. It had an unstraightened column, and the test was run quasi-statically (this explains the smooth appearance of the curves). As for tests A4 and A7, the failure was initiated by a crack at the toe of one (top) of the cover plate to column welds, although a crack also developed at the edge of the cover plate, close to the edge of the column flange. The full stiffness was maintained through one complete cycle of 0.028 radians.

Figure 7 Moment vs. Plastic Rotation Curves for As-Built Connection A8 Influence of Column Straightening Protocol It is essential to note that the response characteristics, failure modes and rotation capacities for the three specimens A4, A7 and A8 were identical for all practical purposes. The same was observed for the other as-built connections that had rotary straightened or gag straightened columns. Similar behavior in terms of the influence of the straightening protocol was found for the 6 Revised Type 1 connections that were tested, one of which used an unstraightened column and the other five had rotary straightened members. It can therefore be concluded that the column straightening protocol has no influence on the response of beam-to-column connections. These findings have since been further confirmed by research conducted at the University of California at San Diego (8). Brief Comments on Other Connection Tests in Research Program The performance of the Revised Type 1 and especially the Revised Type 2 connections was generally very good, and all but three connections met the FEMA single cycle rotation requirement (6). Subsequent fractographic analysis showed that one of these had a significant weld flaw, which contributed to its early failure. The other two non-FEMA acceptable tests would meet the Japanese cumulative plastic rotation requirement of 0.3 radians; this criterion was well satisfied for all of the other connections. The single Revised Type 2 test maintained its full stiffness through two complete cycles of 0.038 radians. It is recalled that Type 2 was identical to the as-built connections but for the presence of a inch transition fillet weld between the cover plate and the column. The two Revised Type 3 connections performed extremely well in all respects. As an example, Figure 8 shows the moment vs. plastic rotation hysteresis loops for Test R3-2; the data for the other Type 3 connection are very similar. The test continued through 26-1/2 cycles, including three at 0.038 radians. By the end of the test the cumulative plastic rotation was 1.840

411

radians; this compares to the values 0.500 to 0.700 radians, which were typically achieved for the as-built, and the Revised Type 1 and Type 2 connections. Clearly the modified placement of the continuity plates had a very major effect on the response of the connections.

Figure 8 Moment vs. Plastic Rotation Curves for Revised Type 3 Specimen R3-2 CONCLUSIONS AND RECOMMENDATIONS 17 full-scale beam-to-column connections were tested by quasi-static or dynamic loading protocols. The effects of steel mill straightening practices were evaluated through the use of rotary straightened, gag straightened and unstraightened columns. The results show that the straightening protocols have no influence on the performance of the connections. The results also indicate no significant differences in strength, failure modes and rotation capacities between specimens that were tested quasi-statically and dynamically. The dynamic loading protocol is felt to be more severe, but its complexity and hydraulic testing demands make this approach impractical and very expensive for most testing laboratories. The quasistatic testing remains suitable. It is recommended that cumulative plastic rotations be used for connection performance assessment rather than the single cycle rotation measurement that is spelled out in current guidelines. The cumulative plastic rotations are more representative of the energy absorption capacity of the connections. Using the updated Japanese criterion of 0.3 radians, all of the connections but one would be satisfactory, and the one that fails the criterion had a significant weld flaw. The tests of the Type 3 connections demonstrated excellent plastic rotation and energy absorption capacities. It was also found that although cracks developed and eventually propagated through the column material, the propagation was slow and stable, with numerous crack arrests during the testing. Such was also the case for the cracks that propagated into the column k-area, demonstrating that a crack in this region will propagate in stable fashion, given appropriate connection details and fracture paths. Further, these connections used thinner cover plates and fillet welded and repositioned continuity plates. Finally, the cropping of the

412

continuity plates is important, to the effect that the ends of the welds need to be kept away from the k-area, but this observation applies to all kinds of welds and connections. In brief, fabrication and construction economies will be obtained with the Revised Type 3 connection. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. American Society for Testing and Materials (ASTM) (1998), Specification for General Requirements for Rolled Structural Steel Bars, Plates, Shapes and Sheet Piling, ASTM Standard A6. ASTM, Conshohocken, PA. Bjorhovde, Reidar, Goland, L. J., and Benac, D. J. (1999), Tests of Full-Scale Beam-to-Column Connections, Technical Report, Nucor-Yamato Steel Company, Blytheville, AR, August. Applied Technology Council (ATC) (1992), Guidelines for Cyclic Seismic Testing of Components for Steel Structures, ATC Guideline No. ATC-24. ATC, Redwood City, CA. American Institute of Steel Construction (AISC) (1993), Specification for the Load and Resistance Factor Design of Steel Building Structures, 2nd Edition. AISC, Chicago, IL, December 1. American Institute of Steel Construction (AISC), Seismic Provisions for Structural Steel Buildings (1997) and Supplement 1 to the 1997 Seismic Provisions (1999), AISC, Chicago, IL. Federal Emergency Management Agency (FEMA) (1995), Interim Guidelines: Evaluation, Repair, Modification and Design of Steel Moment Frames, Report No. FEMA-267, FEMA, Washington, DC. Nakashima, M., Suita, K., Morisako, K., and Maruoka, Y. (1998), Tests of Welded Beam-Column Assemblies. I: Global Behavior, Journal of Structural Engineering, ASCE, Vol. 124, No. 11, November (pp. 1236-1244). Uang, C.-M . (2000), Private Communication, San Diego, CA.

413