K2.1 Scope
The development of testing requirements for beam-to-column moment connections was motivated by the widespread occurrence of fractures in such connections in the 1994 Northridge earthquake. To improve performance of connections in future earthquakes, laboratory testing is required to identify potential problems in the design, detailing, materials, or construction methods to be used for the connection. The requirement for testing reflects the view that the behavior of connections under severe cyclic loading cannot be reliably predicted by analytical means only.
It is recognized that testing of connections can be costly and time consuming. Consequently, this section has been written with the intent of providing the simplest testing requirements possible, while still providing reasonable assurance that connections tested in accordance with these Provisions will perform satisfactorily in an earthquake. Where conditions in the actual building differ significantly from the test conditions specified in this section, additional testing beyond the requirements herein may be needed to ensure satisfactory connection performance. Many of the factors affecting connection performance under earthquake loading are not completely understood. Consequently, testing under conditions that are as close as possible to those found in the actual building will provide for the best representation of expected connection performance.
It is not the intent of these Provisions that project-specific connection tests be conducted on a routine basis for building construction projects. Rather, it is anticipated that most projects would use connection details that have been previously prequalified in accordance with Section K1. If connections are being used that have not been prequalified, then connection performance must be verified by testing in accordance with Section K2. However, even in such cases, tests reported in the literature can be used to demonstrate that a connection satisfies the strength and rotation requirements of the Provisions, as long as the reported tests satisfy the requirements of this section. Consequently, it is expected that project-specific connection tests would be conducted for only a very small number of construction projects.
Although the provisions in this section predominantly address the testing of beam-to-column connections in moment frames, they also apply to qualifying cyclic tests of link-to-column connections in eccentrically braced frames (EBF). While there are no reports of failures of link-to-column connections in the Northridge earthquake, it cannot be concluded that these similar connections are satisfactory for severe earthquake loading as it appears that few EBF with a link-to-column configuration were subjected to strong ground motion in that earthquake. Many of the conditions that contributed to poor performance of moment connections in the Northridge earthquake can also occur in link-to-column connections in EBF. Further, research on link-to-column connections (Okazaki, 2004; Okazaki et al., 2004b) has demonstrated that such connections, designed and constructed using pre-Northridge practices, show poor performance in laboratory testing. Consequently, in these Provisions, the same testing requirements are applied to both moment connections and to link-to-column connections. In the 2016 edition of the Provisions, requirements were added for testing beam-to-column connections in composite special moment frames (C-SMF) and composite intermediate moment frames (C-IMF).
When developing a test program, the designer should be aware that the authority having jurisdiction may impose additional testing and reporting requirements not covered in this section. Examples of testing guidelines or requirements developed by other organizations or agencies include those published by SAC (FEMA, 2000a; SAC, 1997), the ICC Evaluation Service (ICC, 2008), and the County of Los Angeles (County of Los Angeles Department of Public Works, 1996). Prior to developing a test program, the appropriate authority having jurisdiction should be consulted to ensure the test program meets all applicable requirements. Even when not required, the designer may find the information contained in the foregoing references to be useful resources in developing a test program.
K2.2 Test Subassemblage Requirements
A variety of different types of subassemblages and test specimens have been used for testing moment connections. A typical subassemblage is planar and consists of a single column with a beam attached on one or both sides of the column. The specimen can be loaded by displacing either the end of the beam(s) or the end of the column. Examples of typical subassemblages for moment connections can be found in the literature—for example, in SAC (1996) and Popov et al. (1996).
In the Provisions, test specimens generally need not include a composite slab or the application of axial load to the column. However, such effects may have an influence on connection performance, and their inclusion in a test program should be considered as a means to obtain more realistic test conditions. An example of test subassemblages that include composite floor slabs and/or the application of column axial loads can be found in Popov et al. (1996); Leon et al. (1997); and Tremblay et al. (1997). A variety of other types of subassemblages may be appropriate to simulate specific project conditions, such as a specimen with beams attached in orthogonal directions to a column. A planar bare steel specimen with a single column and a single beam represents the minimum acceptable subassemblage for a moment connection test. However, more extensive and realistic subassemblages that better match actual project conditions should be considered where appropriate and practical, in order to obtain more reliable test results.
Examples of subassemblages used to test link-to-column connections can be found in Hjelmstad and Popov (1983), Kasai and Popov (1986c), Ricles and Popov (1987b), Engelhardt and Popov (1989a), Dusicka and Itani (2002), McDaniel et al. (2002), Arce (2002), and Okazaki et al. (2004b).
K2.3 Essential Test Variables
K2.3a Sources of Inelastic Rotation
This section is intended so that the inelastic rotation in the test specimen is developed in the same members and connection elements as anticipated in the prototype. For example, if the prototype moment connection is designed so that essentially all of the inelastic rotation is developed by yielding of the beam, then the test specimen should be designed and perform in the same way. A test specimen that develops nearly all of its inelastic rotation through yielding of the column panel zone would not be acceptable to qualify a prototype connection wherein flexural yielding of the beam is expected to be the predominant inelastic action.
Because of normal variations in material properties, the actual location of inelastic action may vary somewhat from that intended in either the test specimen or in the prototype. An allowance is made for such variations by permitting a 25% variation in the percentage of the total inelastic rotation supplied by a member or connecting element in a test specimen as compared with the design intent of the prototype. Thus, for the example above where 100% of the inelastic rotation in the prototype is expected to be developed by flexural yielding of the beam, at least 75% of the total inelastic rotation of the test specimen is required to be developed by flexural yielding of the beam in order to qualify this connection.
For link-to-column connections in EBF, the type of yielding (shear yielding, flexural yielding, or a combination of shear and flexural yielding) expected in the test specimen link should be substantially the same as for the prototype link. For example, a link-to-column connection detail that performs satisfactorily for a shear-yielding link, , may not necessarily perform well for a flexural-yielding link, . The load and deformation demands at the link-to-column connection will differ significantly for these cases.
Satisfying the requirements of this section will require the designer to have a clear understanding of the manner in which inelastic rotation is developed in the prototype and in the test specimen.
One of the key parameters measured in a connection test is the inelastic rotation that can be developed in the specimen. The acceptance criterion in the Provisions is based on story drift angle, which includes both elastic and inelastic rotations. However, inelastic rotation provides an important indication of connection performance in earthquakes and should still be measured and reported in connection tests. Researchers have used a variety of different definitions for inelastic rotation of moment connection test specimens in the past, making comparison among tests difficult. In order to promote consistency in how test results are reported, these Provisions require that inelastic rotation for moment connection test specimens be determined based on the assumption that all inelastic deformation of a test specimen is concentrated at a single point at the intersection of the centerline of the beam with the centerline of the column. With this definition, inelastic rotation is equal to the inelastic portion of the story drift angle. Previously, the Provisions defined inelastic rotation of moment connection specimens with respect to the face of the column. The definition has been changed to the centerline of the column to be consistent with the SAC recommendations (SAC, 1997; FEMA, 2000a).
For tests of link-to-column connections, the key acceptance parameter is the link inelastic rotation, also referred to in these Provisions as the link rotation angle. The link rotation angle is determined based upon an analysis of test specimen deformations and can normally be determined as the inelastic portion of the relative end displacement between the ends of the link divided by the link length. Examples of such calculations can be found in Kasai and Popov (1986a), Ricles and Popov (1987a), Engelhardt and Popov (1989a), and Arce (2002).
K2.3b Members
The intent of this section is that the member sizes used in a test specimen should be, as nearly as practical, a full-scale representation of the member sizes used in the prototype. The purpose of this requirement is to ensure that any potentially adverse scale effects are adequately represented in the test specimen. As beams become deeper and heavier, their ability to develop inelastic rotation may be somewhat diminished (Roeder and Foutch, 1996; Blodgett, 2001). Although such scale effects are not yet completely understood, at least two possible detrimental scale effects have been identified. First, as a beam gets deeper, larger inelastic strains are generally required in order to develop the same level of inelastic rotation. Second, the inherent restraint associated with joining thicker materials can affect joint and connection performance. Because of such potentially adverse scale effects, the beam sizes used in test specimens are required to adhere to the limits given in this section. For C-SMF and C-IMF systems, the weight per foot of the structural steel member that forms part of the test beam must adhere to the specified limits. However, there is no limit on the total weight per foot of the beam in the test specimen.
This section only specifies restrictions on the degree to which test results can be scaled up to deeper or heavier members. There are no restrictions on the degree to which test results can be scaled down to shallower or lighter members. No such restrictions have been imposed in order to avoid excessive testing requirements and because currently available evidence suggests that adverse scale effects are more likely to occur when scaling up test results rather than when scaling down. Nonetheless, caution is advised when using test results on very deep or heavy members to qualify connections for much smaller or lighter members. It is preferable to obtain test results using member sizes that are a realistic representation of the prototype member sizes.
As an example of applying the requirements of this section, consider a moment connection test specimen constructed with a W36x150 (W920x233) beam. This specimen could be used to qualify any beam with a depth up to 40 in. (1 000 mm) and a weight up to 200 lb/ft (2 900 N/m). The limits specified in this section have been chosen somewhat arbitrarily based on judgment as no quantitative research results are available on scale effects.
When choosing a beam size for a test specimen, several other factors should be considered in addition to the depth and weight of the section. One of these factors is the width-to-thickness ratio, , of the beam flange and web. The ratios of the beam may have an important influence on the performance of specimens that develop plastic rotation by flexural yielding of the beam. Beams with high ratios develop local buckling at lower inelastic rotation levels than beams with low ratios. This local buckling causes strength degradation in the beam and may therefore reduce the load demands on the connection. A beam with very low ratios may experience little if any local buckling and will therefore subject the connection to higher moments. On the other hand, the beam with high ratios will experience highly localized deformations at locations of flange and web buckling, which may in turn initiate a fracture. Consequently, it is desirable to test beams over a range of ratios in order to evaluate these effects. For C-SMF and C-IMF systems, ratios are pertinent to steel members that form part of the composite system. For some composite systems, local buckling of steel members may be restrained by concrete elements. For example, filling a steel tube with concrete or encasing a steel member in concrete may delay the onset and reduce the severity of local buckling. These effects should be considered when designing a test specimen and when considering how test results can be extrapolated to the prototype.
These Provisions also require that the depth of the test column be at least 90% of the depth of the prototype column. Tests conducted as part of the FEMA/SAC program indicated that performance of connections with deep columns may differ from the performance with W12 and W14 (W310 and W360) columns (Chi and Uang, 2002). Additional research on moment connections with deep columns is reported by Ricles et al. (2004b). For C-SMF and C-IMF systems, this limitation only applies to the depth of the structural steel member that forms part of a composite column, not to the overall depth of the composite column.
In addition to adhering separately to the size restrictions for beams and to the size restrictions for columns, the combination of beam and column sizes used in a test specimen should reasonably reflect the pairing of beam and column sizes used in the prototype. For example, consider a building design that calls for the use of a W36 (W920) beam attached to a W36 (W920) column. For the connection type proposed for this building, successful tests have been run on specimens using a W36 (W920) beam attached to a W14 (W360) column and on other specimens using a W24 (W610) beam attached to a W36 (920) column. Thus, test data are available for this connection on specimens meeting the beam size limitations of Section K2.3b and separately on specimens meeting the column size restrictions of Section K2.3b. Nonetheless, these tests would not be suitable for qualifying this connection for the case of a W36 (W920) beam attached to a W36 (W920) column because the combination of beam and column sizes used in the test specimens does not match the combination of beam and column sizes in the prototype, within the limits of Section K2.3b.
K2.3f Steel Strength for Steel Members and Connection Elements
The actual yield stress of structural steel can be considerably greater than its specified minimum value. Higher levels of actual yield stress in members that supply inelastic rotation by yielding can be detrimental to connection performance by developing larger forces at the connection prior to yielding. For example, consider a moment connection design in which inelastic rotation is developed by yielding of the beam and where the beam has been specified to be of ASTM A36/A36M steel. If the beam has an actual yield stress of 55 ksi (380 MPa), the connection is required to resist a moment that is 50% higher than if the beam had an actual yield stress of 36 ksi (250 MPa). Consequently, this section requires that the materials used for the test specimen represent this possible overstrength condition as this will provide for the most severe test of the connection.
As an example of applying these Provisions, consider again a test specimen in which inelastic rotation is intended to be developed by yielding of the beam. In order to qualify this connection for ASTM A992/A992M beams, the test beam is required to have a yield stress of at least equal to 47 ksi (324 MPa) for ASTM A992/ A992M. This minimum yield stress is required to be exhibited by both the web and flanges of the test beam.
The requirements of this section are applicable only to members or connecting elements of the test specimen that are intended to contribute to the inelastic rotation of the specimen through yielding. The requirements of this section are not applicable to members or connecting elements that are intended to remain essentially elastic.
K2.3i Welded Joints
The intent of the Provisions is that the welds on the test specimen replicate the welds on the prototype as closely as practicable. Accordingly, it is required that the welding variables, such as current and voltage, be within the range established by the weld metal manufacturer. Other essential variables, such as steel grade, type of joint, root opening, included angle, and preheat level, are required to be in accordance with
AWS D1.1/D1.1M. It is not the intent of this section that the electrodes used to make welds in a test specimen must necessarily be the same AWS classification, diameter, or brand as the electrodes to be used on the prototype.
K2.3j Bolted Joints
The intent of the Provisions is that the bolts and bolt holes on the test specimen replicate the corresponding bolts and bolt holes in the prototype. The Provisions require that the test specimen and prototype have the same bolt grade, type and orientation of bolt holes, installation procedure, and faying surface preparation. Different methods are available to produce bolt holes and, when inelastic rotation within the connection by slip or yielding is anticipated, it is the intent of the Provisions that bolt hole fabrication methods result in similar levels of bolt hole quality, resulting in similar behavior with respect to the relevant behavioral characteristics of the connection. In some cases, this might require that the test specimen and prototype employ the same bolt hole fabrication method, while in other cases, the desired connection behavior may be obtained using different fabrication methods. Determination of what constitutes similar levels of quality would be subject to approval by the authority for prequalification.
K2.4 Loading History
For biaxial loading of columns, the intent is to require that both axes are loaded using a pseudo-statically applied load (variable load) as specified in Section K2.4b. The option to apply simultaneous varying loads using Section K2.4b is not prohibited, although the coordination of the two loading sequences would require judgment, presumably supplied by the connection prequalification review panel (CPRP). It does not appear reasonable to try to explain how the loads would be coordinated in the Provisions because different connections might suggest different phasing of the loads. Proponents and reviewers are reminded that coordination of loading must be considered. Although not stated explicitly, biaxially symmetric columns would not require duplicate testing about both axes.
The Provisions require that testing include the most demanding combination of beams for which prequalification is sought. For some systems, particularly composite systems, the “largest beam” might not always represent the most demanding situation.
The Provisions provide an option to apply a variable load about at least one axis while a constant (static) load, equal to the expected demand from the beam in the orthogonal direction, may be applied about the orthogonal axis. The use of a static load, equal to the expected strength of the orthogonal beam, is intended to address the lack of test data demonstrating how and at what magnitude simultaneously variable loads should be applied. The Provisions allow for other loading sequences should alternate loading be deemed more appropriate by the proponent and reviewers.
The loading sequence prescribed in Section K2.4b for beam-to-column moment connections is taken from Protocol for Fabrication, Inspection, Testing, and Documentation of Beam Column Connection Tests and Other Experimental Specimens
(SAC/BD) (SAC, 1997). This document should be consulted for further details of the loading sequence, as well as for further useful information on testing procedures. The prescribed loading sequence is not intended to represent the demands presented by a particular earthquake ground motion. This loading sequence was developed based on a series of nonlinear time history analyses of steel moment frame structures subjected to a range of seismic inputs. The maximum deformation, as well as the cumulative deformation and dissipated energy sustained by beam-to-column connections in these analyses, were considered when establishing the prescribed loading sequence and the connection acceptance criteria. If a designer conducts a nonlinear time history analysis of a moment frame structure in order to evaluate demands on the beam-to-column connections, considerable judgment will be needed when comparing the demands on the connection predicted by the analysis with the demands placed on a connection test specimen using the prescribed loading sequence. In general, however, a connection can be expected to provide satisfactory performance if the cumulative plastic deformation and the total dissipated energy sustained by the test specimen prior to failure are equal to or greater than the same quantities predicted by a nonlinear time history analysis. When evaluating the cumulative plastic deformation, both total rotation (elastic plus inelastic) as well as inelastic rotation at the connection should be considered. SAC/BD-00/10 (SAC, 2000) can be consulted for further information on this topic.
Section K2.4c specifies the loading sequence for qualifying tests on link-to-column connections and is based on work by Richards and Uang (2003) and Richards (2004).
The loading sequence specified in Guidelines for Cyclic Seismic Testing of Components of Steel Structures (ATC-24) (ATC, 1992) is considered as an acceptable alternative to those prescribed in Sections K2.4b and K2.4c. Further, any other loading sequence may be used for beam-to-column moment connections or link-to-column connections, as long as the loading sequence is equivalent to or more severe than those prescribed in Sections K2.4b and K2.4c. To be considered as equivalent or more severe, alternative loading sequences should meet the following requirements: (a) the number of inelastic loading cycles should be at least as large as the number of inelastic loading cycles resulting from the prescribed loading sequence, and (b) the cumulative plastic deformation should be at least as large as the cumulative plastic deformation resulting from the prescribed loading sequence.
Dynamically applied loads are not required by the Provisions. Slowly applied cyclic loads, as typically reported in the literature for connection tests, are acceptable for the purposes of the Provisions. It is recognized that dynamic loading can considerably increase the cost of testing and that few laboratory facilities have the capability to dynamically load large-scale test specimens. Furthermore, the available research on dynamic loading effects on steel connections has not demonstrated a compelling need for dynamic testing. Nonetheless, applying the required loading sequence dynamically, using loading rates typical of actual earthquake loading, will likely provide a better indication of the expected performance of the connection and should be considered where practical.
K2.6 Testing Requirements for Material Specimens
Tension testing is required for steel members and connection elements of the test specimen that contribute to the inelastic rotation of the specimen by yielding. These tests are required to demonstrate conformance with the requirements of Section K2.3f and to permit proper analysis of test specimen response. Tension test results reported on certified mill test reports are not permitted to be used for this purpose. Yield stress values reported on a certified mill test report may not adequately represent the actual yield strength of the test specimen members. Variations are possible due to material sampling locations and tension test methods used for certified mill test reports.
ASTM standards for tension testing permit the reporting of the upper yield point. Yield strength may be reported using either the 0.2% offset or 0.5% elongation under load. For steel members subjected to large cyclic inelastic strains, the upper yield point can provide a misleading representation of the actual material behavior. Thus, while an upper yield point is permitted by ASTM, it is not permitted for the purposes of this section. Determination of yield stress using the 0.2% strain offset method based on independent testing using a common specimen size for all members is required in this section. This follows the protocol used during the FEMA/SAC investigation.
This tension testing utilizes potentially different specimen geometry, testing protocol, and specimen location; therefore, differences from the material test report are to be expected. Appendix X2 of ASTM A6/A6M discusses the variation of tensile properties within a heat of steel for a variety of reasons. Based on previous work, this appendix reports the value of one standard deviation of this variance to be 8% of the yield strength using ASTM standards.
This special testing is not required for project materials as the strength ratios in Table A3.2 were developed using standard producer material test report data. Therefore, supplemental testing of project material should only be required if the identity of the material is in question prior to fabrication.
Only tension tests for steel members and connection elements are required in this section. Additional materials testing, however, can sometimes be a valuable aid for interpreting and extrapolating test results. Examples of additional tests, which may be useful in certain cases, include Charpy V-notch tests, hardness tests, chemical analysis, and others. Consideration should be given to additional materials testing, where appropriate.
For C-SMF and C-IMF specimens, material testing is also required for reinforcing steel and concrete. Because of potentially significant differences in specified concrete compressive strength compared to the actual compressive strength, limits are placed on the degree to which the actual tested compressive strength of concrete in a specimen is allowed to differ from the specified value. An exception to these limits is provided if it can be demonstrated that differences in concrete beyond these limits will not result in unacceptable differences in connection performance between the test specimen and the prototype.