Professional Documents
Culture Documents
The Effect of Base Connection Strength and Ductility On The 1 Seismic Performance of Steel Moment Resisting Frames
The Effect of Base Connection Strength and Ductility On The 1 Seismic Performance of Steel Moment Resisting Frames
The Effect of Base Connection Strength and Ductility On The 1 Seismic Performance of Steel Moment Resisting Frames
net/publication/339830174
CITATIONS READS
5 224
4 authors, including:
Farzin Zareian
University of California, Irvine
88 PUBLICATIONS 1,760 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Statistical Frameworks for Improved Seismic Demand Analysis of Bridge Structures View project
All content following this page was uploaded by Pablo Torres-Rodas on 10 March 2020.
1
Assistant Professor, Faculty of Civil and Env. Eng., Gdansk University of Technology, 80-233 Gdansk, Poland.
2
Assistant Professor, College of Science and Engineering., Universidad San Francisco de Quito. Ecuador. 170157.
3
Associate Professor, Department of Civil and Env. Eng., University of California, Irvine, CA, 92697, USA.
Corresponding author.
4
Professor, Department of Civil and Environmental Engineering, University of California, Davis, CA, 95616, USA.
1
28 INTRODUCTION AND MOTIVATION
29 Steel moment frames in seismically active regions, such as the West Coast of the United States,
30 are typically designed to concentrate yielding in plastic hinges at the ends of the beams while the
31 remainder of the structure remains predominantly elastic (AISC, 2010). The intent of this practice
32 is to fully leverage ductile and dissipative response of beams over the entire height of the building,
33 while discouraging soft-stories that limit overall deformation capacity. To achieve such response,
34 other elements of the structure are either designed for the overstrength seismic loading (denoted
35 as ), or are capacity protected (e.g., columns are designed using the Strong-Column-Weak-Girder
36 – SCWG check). Despite these measures, some plastic hinging in the columns is unavoidable; this
37 arises due to either: i) the kinematic necessity of plastic rotations at the column base to mobilize a
38 full-building mechanism – Figure 1a, or ii) higher mode effects that contravene the first-mode
39 response inherent in the SCWG check – Figure 1b. In the first story of the building, plastic hinges
40 may be accommodated either in the lower region of the column section itself (Figure 1c), or in the
41 base connection or foundation (Figure 1d). However, ductile response of the connection is required
42 to obtain the latter response. Specifically, AISC 341-16 (AISC, 2016) Section D2.6c allows use of
43 the overstrength seismic load to design base connections provided that “a ductile limit state in
44 either the column base or the foundation controls the design.” Connection qualification data for
45 column bases is relatively sparse, and moreover, the required ductility capacity (to resist the
46 demands corresponding to this situation) is not quantified explicitly either in the AISC 341-16 or
47 investigated in prior research. As a result, the dominant practice is to accommodate the plastic
48 hinge at the column base within the lower region of the column section itself (Figure 1c), rather
49 than within the base connection (Figure 1d). This is because rolled sections are presumed to be
50 ductile, whereas connections are presumed to be brittle. The condition shown in Figure 1c (which
2
51 is consistent with the current “Strong-Base” design practice) is achieved by capacity protecting the
52 connection, i.e., designing it to remain elastic by being stronger than the fully-yielded and strain
54
55 Designing the connection to be stronger than the column usually implies the use of thick base
56 plates and multiple large anchor rods in exposed type base connections (Fisher and Kloiber, 2006).
57 When these become unfeasible, embedded type connections (Grilli and Kanvinde, 2017) are
58 specified. These connections usually require multiple concrete pours. In either case, constructing
59 a strong base entails significant expense and logistical challenges. However, several recent
60 experiments (Barnwell, 2015; Gomez et al., 2010; Grilli et al., 2017; Trautner et al., 2017, 2016;
61 Tryon, 2016) indicate that even when not explicitly detailed for ductility, column base connections
62 (of various configurations) may be highly ductile with desirable hysteretic and dissipative
63 characteristics; this has been observed for various types of base connections, including those with
64 significant axial load. This suggests that the current practice of designing bases to remain elastic
65 may be unwarranted given their expense and potential ductile performance. Moreover, recent
66 studies (Elkady and Lignos, 2018, 2016; Newell and Uang, 2008) have revealed that column
67 sections themselves may have limited ductility due to local and lateral instability, especially in the
68 presence of axial load. When considered along with the recently observed ductile response of
69 column base connections, this practice is potentially counterproductive (protecting the more
70 ductile element), as well as expensive. Despite these strong mechanistic justifications for weak-
71 base design, practical considerations prevent its widespread adoption. First, although the observed
73 evidenced by premature fractures in the field (base plate fracture in the Oviatt Library during the
3
74 1994 Northridge Earthquake, Gioncu and Mazzolani, 2013), as well as in laboratory specimens
75 (Petrone et al. 2016). This may be attributable to the absence of prescriptive guidance (e.g.,
76 analogous to AISC 358, 2016 for beam-column connections). Nonetheless, the high ductility of
77 many other connections is encouraging from the standpoint of developing such guidance. Second,
78 although design codes notionally permit weak base design, the professional practice is reluctant to
79 implement it in the absence of quantitative relationships between base connection ductility and
81
82 Addressing these issues requires coordinated efforts in the following areas: i) understanding the
83 relationships between base connection stiffness & strength, ductility, and structural performance,
84 ii) developing connection details and design methodologies that satisfy these relationships for
85 acceptable performance, and iii) refining design considerations for the entire frame incorporating
86 these interactions between the base connection and frame properties. Against this backdrop, this
87 study parametrically examines the effect of base connection strength and ductility on the seismic
88 performance of steel moment resisting frames, with the objective of establishing a quantitative
89 context within which design/detailing strategies for connections as well as moment frames may be
90 pursued. The main scientific basis of this study is Nonlinear Time History (NTH) analysis of four
91 moment-resisting frames (4-, 8-, 12-, and 20-story) designed as per current design standards
92 (AISC, 2005a; ASCE, 2005). The NTH analyses are supplemented by Nonlinear Static Pushover
93 (NSP) analysis to provide generic insights into modes of response. The NTH analysis and
94 performance assessment is conducted within the framework of FEMA P-695 (2009) to examine
95 the effect of various base connection conditions on the collapse resilience of the frames. Two sets
96 of analysis are conducted. In the first set, the strength of base connections is parametrically varied.
4
97 These strengths range from a low value wherein the connections are designed for a reduced seismic
98 load, i.e., R = 8 (implying highly inelastic response as in the case of beam-column connections),
99 to a high value wherein the base connections are capacity designed for 1.1RyMP of the attached
100 column, implying a strong-base condition, and elastic response of the connection. For each of
101 these, various aspects of frame and connection response are studied. The primary concerns are: i)
102 to examine the effect weakening the base on the various aspects of structural response, and perhaps
103 more importantly, ii) to examine the interrelationships between the strength and the required
104 ductility to achieve acceptable performance in the context of collapse safety. A second set of
105 analyses examines a collateral implication of weak-base design, recognizing that a weaker base is
106 likely to be more flexible as well. Previous research (Zareian and Kanvinde, 2013) has shown that
107 base rotational flexibility has the potential to alter frame response, by lowering the inflection point
108 in the first story column, possibly triggering a soft story. As a result, examination of this collateral
109 implication is important from a practical standpoint when evaluating prospective design
110 methodologies that utilize weak-bases. Each set of analyses uses the most current research for
111 modeling base connection properties, including strength, stiffness, and hysteretic response.
112
113 The paper begins by outlining the methodological components of this research. This includes the
114 building designs themselves, followed by the manner in which they are modeled for the NTH and
116 framework, which follows procedures mandated by FEMA P-695 (2009). The subsequent section
117 presents the two analysis sets outlined above, conducted within this framework. Results of these
118 simulations are discussed with an emphasis on their implications for design of the base
5
119 connections, as well as the moment frame itself. The paper concludes by summarizing limitations
120 of the work, while summarizing future work to overcome these limitations.
121
122 METHODOLOGY
123 Figure 2 schematically illustrates the methodology for parametric simulation and performance
124 assessment used in this study. Referring to the figure, the components of this methodology include
125 the archetype moment frames, structural models to represent the response of these frames through
126 Nonlinear Static Pushover (NSP) as well as Nonlinear Time History (NTH) analysis, the
127 probabilistic framework within which these analyses are conducted, and performance indicators
128 for assessment of collapse and other aspects of response. These methodological components are
130
131
133 Four moment frames (4-, 8- , 12-, and 20-story) were considered in this study. Figure 3 shows the
134 geometry of these frames along with member sizes and the building floor plan. The approach of
135 using “archetype” frames allows the inference of generic insights about system response and its
136 sensitivity to various parameters. While numerous combinations of base connection parameters
137 (strength and stiffness) are considered, the number of stories (and consequently, building height)
138 is selected as the sole parameter to reflect variations between the buildings themselves, since it is
139 anticipated based on previous work (e.g. Medina and Krawinkler, 2005; Zareian and Kanvinde,
140 2013; Zareian and Krawinkler, 2012) to be the most influential. The frame heights range from
141 16.5m for the 4-story frame, to 80.0m for the 20-story frame. Each frame has 3 bays, and is located
6
142 at the perimeter of the building plan shown in Figure 3. Referring to the figure, the orthogonal
143 frames do not share any columns, such that biaxial bending of the columns is not a concern; each
144 frame is assumed to resist forces only in the in-plane direction. The key aspects of frame design
146 Loadings and design checks corresponding to ASCE 7-05 (ASCE, 2005) as well as AISC 341-
147 10 (AISC, 2010) are followed. Additional details of these frames (including connection details)
149 The beams all have Reduced Beam Section (RBS) connections, in which beam inelastic
151 A value of R = Cd = 8 is assumed for design of the beam members, along with site-class D
152 conditions under the seismic design category D. Seismic hazard and site conditions are
153 consistent with a non-near fault location in the Los Angeles Basin area.
154 Four distinct designs for column bases are considered; these are combinations of (R, = (8,1),
155 (3, 8/3), (8,3); and a case in which column base strength is equal to 1.1RyMP. These are denoted,
157 The code-based fundamental periods of the 4-, 8-, 12-, and 20-story frames are 0.95, 1.64, 2.25,
158 and 3.37 seconds respectively. These correspond to design (i.e., 10% of exceedance in 50
159 years) spectral acceleration values of 0.63g, 0.37g, 0.27g, and 0.18g for the four frames.
160 Corresponding MCE (i.e., 2/50 exceedance) values are 0.95g, 0.55g, 0.40g, and 0.27g
161 respectively.
162 Following current practice, each frame is designed using the fixed base assumption; this
163 information is important for appropriately interpreting the results discussed later. The main
164 implications of this are that: i) the effect of base flexibility on building period is not
7
165 incorporated, and ii) strong column weak girder for the first story columns design assumes a
167
169 Figure 4 schematically illustrates key aspects of the frame models constructed in OpenSees
170 (Opensees, 2017) for the NSP as well as the NTH analysis. Referring to the figure, these aspects
172 All members (beams and columns) are simulated as elastic beam-column elements with
174 In the beams, a rotational springs is placed at the location of each RBS where yielding is
175 anticipated (i.e. at the two ends). The rotational response of these springs is represented through
176 the Ibarra-Medina-Krawinkler (IMK) model (Ibarra et al., 2005). This model includes a
177 bilinear hysteretic response, with a “cap” to simulate strength loss, in addition to several forms
178 of strength and stiffness deterioration. The model requires the estimation of at least 14
179 parameters; these are calibrated based on the cross-sectional properties of the beam and RBS,
181 In the columns, rotational springs are placed at each end. These springs are similar to the
182 springs in the beam, i.e., the IMK model is used to represent their constitutive response.
183 However, as discussed in the next point, the model is calibrated to reflect the presence of axial
185 A representative axial force was estimated from pushover analyses to approximately account
186 for the effect of axial force on column bending strength; bending strength was reduced using
187 this axial force and the AISC interaction equation (AISC, 2005b). The representative axial load
8
188 is equal to the summation of column axial load due to gravity load and half of the maximum
190 Finite joint sizes are simulated along with inelasticity in the joint panel zones. Panel zones are
192 representing the panel zones as an assemblage of pin-connected rigid links, in which one of
193 the connections also features a rotational spring. The properties of this spring are calibrated to
194 reflect expected panel zone properties through a bilinear hysteretic response, as per the practice
196 As shown in Figure 4, a leaning column (with the appropriate gravity loads), along with large-
197 displacement geometric nonlinearity simulates destabilizing P- effects that are responsible
199 Special attention was afforded to column bases, since methods to simulate their inelastic moment-
200 rotation response are not as well-established as for other connections. As shown in Figure 4, the base
201 connections were simulated as rotational springs connected to the base of the columns, with their
202 translation tied to the column bases. The constitutive response of these springs was represented by
203 the IMK (Ibarra et al., 2005) model as well. This follows recent research (Torres-Rodas et al., 2018,
204 2016), which indicates that the IMK model is able to represent the important aspects of base
205 connection hysteretic response, based on validation against experimental data by Grilli et al., (2017),
206 and Gomez et al., (2010). Figure 5a shows moment-rotation response from one experiment on
207 embedded column bases (conducted by Grilli et al., 2017) alongside a photograph (Figure 5b) of the
208 corresponding experiment. Figures 5c and 5d shows the counterpart response as determined from
209 the calibrated IMK model. Specifically, Figure 5c illustrates the monotonic backbone curve of the
210 IMK model, whereas Figure 5d illustrates cyclic evolution derived from this backbone curve.
9
211 Referring to these figures, the IMK model is able to functionally reproduce important aspects of base
212 connection response including yielding, hardening, strength and stiffness degradation, as well as
213 pinching. At least 12 parameters are required to define this hysteretic response. Two of these
214 parameters, i.e., the strength (denoted 𝑀 – see Figure 5c, with the peak strength defined such that
215 𝑀 = 0.7𝑀 ) and the stiffness (denoted 𝐾 – also see Figure 5c) are considered to be
216 “core,” for this study since i) they dominate overall response, ii) are directly pertinent to the goals of
217 this study, which focuses on weakened bases, and iii) they are strongly correlated with connection
218 configuration, such that they may be controlled through design and detailing. As a result, these
219 parameters are controlled as variables in this study. The remaining parameters (termed “ancillary”)
220 control other aspects of hysteretic response such as degradation and pinching, and cannot be
221 conveniently correlated with connection configuration (Torres-Rodas et al., 2018, 2016). Values
222 prescribed by Torres-Rodas et al., (2018, 2016) are used for the ancillary parameters in all
223 simulations.
224
225 It is relevant to note three points here: (1) the strength of the base connection referred to hereafter
226 denotes the yield, i.e., 𝑀 = 𝑀 (rather than ultimate, i.e., 𝑀 ) strength, since this is
227 consistent with how connections are designed, (2) the yield and ultimate strengths, and
228 correspondingly the backbone curve is symmetric in both directions, and (3) the backbone curve
229 of the hysteretic model for the bases (Figure 5c) does not have a descending (failure) branch or
230 rotation capacity. This is appropriate because in the parametric simulations discussed later, the
231 rotation capacity is varied in a notional manner such that exceeding this notional capacity is
10
233 assumption, it is necessary to simulate structural response only up to (but not beyond) the point of
235
237 Once constructed as described in the previous subsection, the frame models provide the opportunity
238 to interrogate various parameters for performance assessment using NSP and NTH analysis. Table
239 1 shows a simulation matrix summarizing all parameter sets for which performance assessment is
240 conducted. Referring to the Table, for each frame, the following parameters are varied:
241 Base connection strength: For each frame, four levels of base connection strength are
242 examined. For each of these levels, the interior and exterior base connections have the
243 following flexural strengths (1) 𝑀 =𝑀 /𝜙 in which 𝜙 = 0.9, and 𝑀 is the moment
244 induced by the base shear corresponding to R = 8, i.e., Reduced Seismic Load (2) 𝑀 =
245 𝑀 /𝜙, in which 𝑀 corresponds to a design moment for R = 3, which is prescribed by the
246 Seismic Provisions (AISC 341-16) for systems not explicitly detailed for ductility, (3) 𝑀 =
247 Ω𝑀 /𝜙 , which corresponds to the overstrength seismic load, and finally (4) 𝑀 =
248 1.1𝑅 𝑀 /𝜙, wherein the bases are capacity designed for the attached column. As such, these
249 levels span a range of base connection strengths and provide examination of plausible
250 prospective design alternatives. In the cases that utilize the base shear (rather than column
251 capacity), the base connections (for both interior and exterior columns) are design based on an
252 equivalent lateral load analysis assuming a fixed base – which is consistent with current
253 practice for estimating earthquake loads in other elements of the moment frame.
254 Base connection flexibility: Conventionally, base connections are assumed fixed in the design
255 and performance assessment process. The implicit assumption is that if designed to be stronger
11
256 than the column (as per current practice), the connections have sufficient rotational fixity such
257 that they may be assumed fixed. However, research by Zareian and Kanvinde (2013) indicates
258 that: i) even if designed as fixed, column bases may show significant flexibility, ii) the degree
259 of this flexibility depends on the actual connection detail including whether the connection is
260 exposed or embedded, the size of embedded plate, anchor rod size and pattern, the level of
261 axial force and iii) structural response is highly sensitive to base flexibility. To quantify the
262 effects of this sensitivity, each simulation is conducted with two assumed values of base
263 flexibility for each base connection (i.e., interior and exterior) in the frame. One of these
264 (denoted = ∞) corresponds to a fixed base, whereas the other corresponds to an estimated
265 flexibility that represents the expected rotational flexibility of each base connection (denoted
266 = 1). To characterize the expected rotational flexibility of each connection, the following
267 process is adopted. First, for each level of base connection strength summarized in the previous
268 point, the connection is designed using approaches outlined by Grilli & Kanvinde (2017), as
269 well as the AISC Seismic Design Manual (2015). For each of the base connections, it is
270 determined that embedded type base connections are more appropriate, given the magnitude
271 of the forces that must be resisted. The design results in estimates of embedment depth which
272 are also summarized in Table 1. Second, base connection rotational stiffness is determined for
273 each of the connections based on the designed configurations, using approaches developed by
274 Torres-Rodas et al. (2017) and Kanvinde et al. (2012) that have been validated by test data.
275 For the exposed type connections, the flexibility depends on the axial compression in the
276 column which delays uplift of the base plate, and this is incorporated into the model by
277 Kanvinde et al. (2012). As done for the column rotational hinges, this axial load corresponds
278 to the gravity load and half of the maximum axial force due to lateral loading. These rotational
12
279 stiffnesses are assigned to each of the hysteretic springs (see Figure 5) as a core parameter.
280 This exercise is repeated for all base connections (interior as well as exterior). The resulting
282 Once the rotational flexibility and strengths are determined as above, 8 parametric combinations
283 (4 base connection strengths × 2 rotational flexibilities) are generated for each of the four frames,
284 resulting in a total of 32 models – as indicated in Table 1. For each of these 32 models, the base
285 rotation capacity is monitored during analyses as an additional parameter, to examine tradeoffs
286 between base rotation capacity and the other parameters, to obtain acceptable frame response. Five
288 i.e., ∞ essentially induces sidesway collapse without failure of the base). As discussed in the
289 previous section, this parameter is examined in a “post-processing” sense, meaning that: i)
290 different frame models are not considered for each value of this parameter, and ii) the variation of
291 this parameter is conducted in a notional sense, such that exceeding this value in each of the
293 This results in a total of 160 parametric combinations as summarized in Table 1. Each of these
294 parametric combinations is subjected to Nonlinear Static Pushover and Nonlinear Time History
295 analysis within the overall scientific framework as illustrated previously in Figure 2. These
297
299 For each of the 160 parametric combinations, Nonlinear Static Pushover (NSP) analysis are
300 conducted along with Nonlinear Time History (NTH) simulations. The objective of the former
301 (NSP) is to obtain general insights regarding frame response modes (i.e., collapse mechanisms,
13
302 ductility with respect to P- instability) that are otherwise difficult to obtain from suites of NTH
303 simulations. On the other hand, the NTH simulations provide quantitative estimates of
304 failure/collapse probabilities under seismic excitation. The NTH-based performance assessment is
305 carried out as per the procedure outlined in FEMA P695. The FEMA P695 process provides a
306 rational basis for evaluating various design options (i.e., combinations of base strength, stiffness,
307 and deformation capacities), by examining their effect on the probability of unacceptable response
308 (e.g., collapse or exceedance of other critical limit states). The procedure is summarized briefly
309 here, especially since it was slightly adapted to suit the objectives of this study.
310 Figure 6 shows response spectra of unscaled 44 far-field ground motions as per FEMA P695.
311 Also shown in Figure 6 are the median intensities (spectral accelerations) of the entire suite of
,
312 motions at the fundamental periods of the frames; these intensities are denoted as 𝑆 , ,
, , ,
313 𝑆 , ,𝑆 , , and 𝑆 , .
314 For each of the frames, a scale factor is determined, which when applied to the entire suite of
315 motions results in exactly 50% of the motions (i.e., 22 motions) inducing unacceptable
316 response (or failure), as defined later in this section. For example, if scaling up the entire suite
,
317 of motions by a factor 𝑆𝐹 % results in failure (of the 4-story building) for 22 motions,
318 then the median intensity corresponding to 50% motions inducing failure may be determined
, , , ,
319 as 𝑆 , % = 𝑆𝐹 % ×𝑆 , .
320 As per the processes outlined in FEMA P695, this median intensity is compared with the site-
,
321 specific hazard (corresponding to a 2% in 50-year probability of exceedance 𝑆 ) to
322 determine an Adjusted Collapse Margin Ratio (ACMR), with the adaptation that the collapse
323 limit state is substituted by a failure limit state as discussed later. Subsequently, the probability
14
324 of failure is determined under a 2/50 hazard. This probability may assessed relative to
326 As discussed above, it is important to define unacceptable response or failure in the context of
327 frame behavior. Conventionally (i.e., as per FEMA P695), the sole criterion for unacceptable
328 response is collapse of the frame, which typically manifests as excessive lateral deformations in a
329 sidesway mode due to hinging at the ends of the beams or columns. However, in the context of the
330 current study, exceeding the deformation capacity of the base connection is also considered
331 unacceptable response equivalent to collapse. Referring to prior discussion, this is a conservative
332 assumption. Nonetheless, it is judicious for two reasons: i) limited information is available
333 regarding the post-failure response of base connections, and its impact on system performance,
334 and ii) the assumption allows for estimation of deformation capacities that are necessary to meet
335 acceptance criteria for various designs. As a result, unacceptable response or failure of the frame
336 (during any ground motion) is defined as the attainment of either of the following two limit states:
337 1. Sidesway collapse of the structure: This determined to occur when interstory drift in any story
338 attains a value of 10%. This is a well-accepted indicator of collapse, since it does not
339 necessitate computationally intensive simulation of the structure “down to the ground,” while
340 also providing a significant drift threshold from which recovery is extremely unlikely
342 2. Exceedance of rotation capacity of base connection: As discussed previously, the rotation
343 capacity of the base connection is varied in a parametric manner, and five values (i.e., 0.01,
344 0.02, 0.04, 0.05, and ∞) radians are selected as notional capacities. Results are generated and
15
346 Once failure is defined in this manner, the FEMA P695 methodology may be applied to determine
,
347 𝑆 , % for each frame, by following the steps outlined above. Using this intensity, as well as
348 the site-specific hazard, the probability of failure corresponding to each notional value of rotational
349 capacity is computed for all parametric combinations. Additionally, for each frame model and
350 notional rotational capacity combination, information is retained regarding the number of motions
351 (out of the 22 motions inducing failure) that induce failure due to sidesway collapse versus
352 exceedance of rotational capacity. For the cases with sidesway collapse, information is also
353 retained regarding the stories participating in the sidesway collapse mechanism. The next section
354 examines the effect of various parameters listed in Table 1 (i.e., frame height, base connection
355 strength, base connection stiffness, and base connection rotation capacity) on the observed modes
357
359 Figures 7a and 7b show a representative pushover curve (shown here for the case of 8-story with
360 R= 3 and = 1, along with a corresponding collapse mechanism for the purposes of defining
361 performance indicators. As discussed earlier, the main objective of the NSP simulations is to obtain
362 generic insights regarding various aspects of frame response. Two aspects are quantitatively
364 The deformation or ductility corresponding to failure (i.e., unacceptable response). As in the
365 case of the NTH simulations, this deformation may be defined as the lesser of the following
366 two: i) the lateral deformation at which the P- moments overwhelm the strength of the
367 structure, leading to sidesway collapse (Kanvinde, 2004), or ii) the lateral deformation at which
368 the notional rotation capacity of the base connection is exceeded. Similar to the NTH
16
369 simulations, the notional capacity is parametrically varied to take the values of 0.01, 0.02, 0.04,
370 0.05, and ∞ (of these, the results corresponding to rotational capacity equal to ∞ essentially
371 correspond to failure due to sidesway collapse). More specifically, the deformation
372 corresponding to failure is formalized as the Period Based Ductility determined from the
373 pushover curve. The calculation of Period Based Ductility (which is defined as per FEMA
375 The system overstrength factor (denoted as referred to herein as system overstrength) is an
376 indicator of the ability of the frame to redistribute forces as well as the overstrength demands
379 In addition, the collapse mechanism is also monitored for each of the parametric runs, to
380 investigate causal relationships between physical response and the quantitative performance
381 indicators summarized above, for example to examine whether soft-story response is more likely
382 for any of the parameter sets. Figure 7b illustrates a representative collapse mechanism
384
385 Figures 8a-d show results of the NSP analysis for each of the four frames for the fixed base ( =∞)
386 condition. Within each figure, the histogram bars indicate Period Based Ductility (determined as
387 shown in Figure 7a) for each of the base rotation capacities plotted against the base connection
388 strength. Figures 9a-d show similar results for the flexible base ( = 1) condition. Figures 10a and
389 10b plot the system overstrength of the frames (for = ∞ and 1, respectively) against the base
390 connection strength. The system overstrength (being controlled by the peak base shear) is
391 insensitive to the base connection strength. Consequently, Figures 10a and 10b include single
17
392 system overstrength values from each of the frames. Referring to Figures 8-10, the following
394 The period based ductility (which provides generic sense of the deformation capacity of the
395 system) for each of the buildings increases with respect to the base strength, as well as with
396 respect to the base rotation capacity. These trends are virtually identical for both values of base
397 flexibility.
398 Within the above observation, the effect of base rotation capacity on period based ductility is
399 most dominant in all structures with R = 8 (Figure 8a). This is not unsurprising, because the
400 base connections yield earlier in these situations, thereby accumulating larger rotations.
401 Following from the above point, as the base strength is increased (especially beyond R=3), the
402 period based ductility tends to saturate with respect to the base rotation capacity. Most notably,
403 for the base strengths corresponding to = 3, and 1.1RyMp, the period based ductility for each
404 of the cases (except one situation – the 8-story frame for = 3) for each of the rotation
405 capacities equals the period based ductility for a base rotation capacity of infinity. This
406 indicates that in these cases, the structural deformation capacity is controlled by P- instability
408 The base strengths corresponding to = 3 and 1.1RyMP result in similar performance (as
409 implied by the period based ductility). This observation is particularly interesting when noted
410 against the actual values of base strength corresponding to these two cases as indicated in Table
411 1. Referring to the Table, the base strengths corresponding to = 3 are on average 70% lower
412 than those corresponding to 1.1RyMP. (i.e. 95%, 56%, 64%, and 64% lower in 4-, 8-, 12-, and
413 20-story frames respectively). These differences may represent a significant cost differential
18
414 between the connections designed for the two cases (= 3 and 1.1RyMP), which produce very
416 Referring to Figures 10a and 10b, the system overstrength of the structures follows a trend
417 similar to their ductility, such that increasing the base strength result in an increase in system
418 overstrength. Similar to the period based ductility, this effect saturates with respect to the base
419 rotation capacity at higher levels of base connection strength, indicating a transition to P-
420 instability driven collapse. Moreover, the system overstrength is only mildly sensitive to base
421 flexibility (i.e., the difference between Figures 10a and 10b is modest). With the exception of
422 the 8-story frame, the system overstrength of the structure (for all design cases, including when
423 the column base is designed for 1.1RyMP) is lower than 3, which is the system overstrength as
424 implied by AISC 341-16 (AISC, 2016). The system overstrength is virtually insensitive to the
425 base rotation capacity; consequently this is not indicated as a variable on these figures.
426 It is important to emphasize here that although the above observations (being based on static
427 pushover simulations) do not reflect effects of higher mode effects, or phenomena such as cyclic
428 deterioration of various components. As such, they are regarded as broad insights that help
429 contextualize results from the NTH simulations that are discussed next.
430
431 Results of the NTH simulations are plotted in Figure 11 (for the =∞) and Figure 12 (for = 1
432 cases). Specifically, Figures 11 plot the probability of failure at 2/50 hazard, against the base
433 connection strength levels. The probability of failure is determined as defined earlier, and
434 corresponds to the attainment of either one of the two limit states (i.e., sidesway collapse or
435 exceedance of base rotation capacity). Each histogram group in Figures 11 and 12 correspond to
436 one building, whereas the bars within the histograms represent probabilities of failure
19
437 corresponding to various values of base rotation capacity. Additionally, each group also includes
438 a bar that represents the probability of failure when the base rotation capacity is infinity, implying
439 sidesway collapse. All the figures include a dashed line corresponding to the 10% probability of
440 failure. This is consistent with FEMA P695 acceptance criteria for collapse. Numerous
441 observations may be made referring to Figures 11 and 12; the overall trends are first discussed,
442 followed by more specific observations with possible implications for design.
443 As a general observation, the probabilities of failure represented in Figure 11 are relatively
444 similar (in most cases identical) to those shown in Figure 12. At first glance, this suggests that
445 base flexibility has a modest effect on failure probability; this mirrors the trends noted in
446 Figures 8 and 9 previously with regards to period based ductility determined from monotonic
447 pushover analysis. Nonetheless, for NTH simulations, this trend is somewhat surprising since
448 it may be anticipated that more flexible bases lower the point of inflection in the first story
449 column, possibly triggering soft-story response. A closer inspection of these trends suggests
450 why this is not borne out. For the cases with higher base strength (i.e., = 3, and 1.1RyMP),
451 the base connections (owing to their design for moment demands and associated detailing, e.g.,
452 embedment depth), have high stiffness – such that in effect, the = 1 cases are functionally
453 equivalent to a large value ( = ∞ cases), insofar as building response is concerned. At the
454 other end of the spectrum (i.e., the R = 8 case), there is a significant difference between the
455 initial stiffnesses associated with the = 1 and ∞ cases. However, in these cases, the base
456 yields early, such that the post-yield stiffness (which is active during most of the strong
457 shaking) is quite similar between the two, resulting in similar performance. Based on this
458 observation, the subsequent points do not distinguish between the = 1 and infinity cases,
459 focusing only on the other parameters, i.e., strength and rotation capacity.
20
460 A second general observation is that (across all frames), the probability of failure for all values
461 of base rotation capacity decreases with an increase in base strength. For each of these base
462 strengths (e.g., within Figure 11a or 12a), the probability of failure is strongly dependent on
463 base rotation capacity, such that increasing the base rotation capacity results in a decreased
465 Specifically, it appears as though for the base strengths corresponding to R = 8, and R = 3, the
466 probabilities of failure are unacceptable (i.e., significantly higher than the 10% acceptance
467 criterion) for all the frames (with the exception of the 4-story frame), and for all base rotation
468 capacities, including the base rotation capacity of 4% which is on the order of the expected
470 Figures 11d and 12d represent current design practice, i.e., designing the base connections for
471 1.1RyMP (i.e., capacity design). Referring to these figures, the probabilities of failure are very
472 close (i.e., in the range of 10-12%) to the FEMA P695 acceptance criterion (~10%) in most of
473 the situations, barring a few. An exception is the 20-story building, in which the probabilities
474 of failure are in the range of (15-20%) for all base rotation capacities, and controlled by
475 sidesway collapse. A closer inspection of the column base time histories reveals that the
476 column base moments exceed 1.1RyMP due to strain hardening of the column section, thereby
477 forcing yielding in the base. This is not uncommon – see Inamasu et al. (2018), and Elkady
478 and Lignos (2015). Both these observations indicate that even current design practice may not
480 Considering the above, it appears that only the case with = 3 (i.e., base connection designed
481 based on overstrength seismic loads) is prospectively viable, from the standpoint of providing
482 acceptable response. Consequently, this case (the data in Figures 11c and 11d) requires closer
21
483 examination. As an additional point of reference, Table 2 summarizes the ratios between the design
484 strengths associated with the 1.1RyMP case and the =3 case. Based on these figures and Table 2
485 the following points (with implications for design) are evident:
486 At the outset, Table 2 establishes that the moment capacities associated with the 1.1RyMP case
487 are significantly larger than those associated with the =3 case. The values range from 1.00 to
488 2.24 (mean = 1.71 for interior columns, and 1.32 for the exterior columns). This indicates that
489 switching from the capacity based design to that based on overstrength offers significant
491 Referring to Figures 11c and 12c, it is apparent that for the 4-story frame, design based on the
492 overstrength provides acceptable response (comparable to that of frames with capacity-
493 designed bases – see Figure 12d) regardless of base rotation capacity. This is because for the
494 4-story building, failure is controlled entirely by sidesway collapse without the development
495 of significant inelastic rotation in the base connection (see far right column of Table 1). From
496 a design standpoint, this has two implications. First, there is no negative consequence of
497 designing the column base for the low-rise frame based on the strength corresponding to =
498 3 condition. Second, the connections designed for these reduced forces do not require special
499 detailing since the rotational capacity does not appear to influence the probability of failure.
500 Nonetheless, it is important to acknowledge that of all the cases considered (see Table 2), the
501 difference between the overstrength and capacity design approaches is modest for the 4-story
502 building. While this results in relatively modest savings by using the lower design moment, it
503 also explains the relatively similar performance between the two values of design moment –
22
505 For the 12-, and 20-story frames, the failure probability decreases as the base rotation capacity
506 is increased. In each case, the base rotation capacity values of 1% and 2% result in grossly
507 unacceptable probability of failure. In each case, a base rotation capacity value of 4% (which
508 is compatible with the mandated standard for beam-column connections as per AISC 358)
509 results in probabilities of failure that are close to acceptable, i.e., in the 14-18% range.
510 The 8-story frame shows somewhat higher probabilities of failure even for a rotation capacity
511 of 4% (18% probability of failure for the ∞ case, and 25% probability of failure for the
512 1 case). However, this is not significantly higher than the probability of failure
513 corresponding to infinite base rotation capacity. A closer examination of the results –
514 especially the far-right column of Table 1 suggests the underlying reason for this. Specifically,
515 for the 8-story with Ω=3, building a large majority of collapse-causing motions induce
516 sidesway collapse. Mechanistically, this response is similar to that observed for the 4-story
517 frame (i.e., collapses dominated by sidesway), but quantitatively, these collapses tend to occur
518 at lower shaking intensities for the 8-story frame, resulting in higher probabilities of failure.
519 Nonetheless, a closer examination of time history simulations suggests that these sidesway
520 mechanisms occur in the mid- to upper stories, excited by higher mode effects. The implication
521 is that this does not indicate an issue with the weak-base design, but rather the frame design
522 itself.
523 Providing a base rotation capacity of 5% results in failure probabilities that are comparable to
524 current design practice (i.e., based on capacity design), except for the 8-story frame (which is
525 largely controlled by sidesway collapse). Although a base rotation capacity of 5% appears to
526 be high, previous research by the authors (Gomez et al., 2010; Grilli and Kanvinde, 2017), and
527 others (Barnwell, 2015; Trautner et al., 2017, 2016) indicate that achieving this level of base
23
528 rotation capacity is not unrealistic, and may well be available even in connections constructed
529 as per standard detailing practices for both exposed and embedded type connections.
530 The above observations illustrate limitations of current design practice, as well as tradeoffs
531 between various base connection properties, suggesting possible paths for refinement of the current
532 practice. The next section outlines such prospective refinements, as well as research that may be
533 required to develop connection details that are required for successful realization of these
534 refinements.
535
537 This section interprets the observations of the preceding discussion in the context of their
538 implications for system or frame design considerations, as well as opportunities for development of
539 connections within this overall context. A shortcoming of the current design practice is the high
540 estimates of moment demands for base connection design, and the associated high cost; this was a
541 major motivating factor for this study. Additionally, the results of the NTH simulations indicate that
542 even designing base connections based on column capacity does not guarantee acceptable
543 performance as per current criteria (FEMA P695). In interpreting this, it is important to note two
544 factors: i) the current design methodology for steel moment frames in the United States (i.e., using
545 ASCE 7-16 in conjunction with the performance modification factors and detailing guidelines of
546 AISC 341-16 as was done for the frames in this study) predates the evaluation procedure outlined in
547 FEMA P695, and ii) the probabilities of failure, in most cases, are only marginally higher than the
548 acceptance criteria. From a mechanistic standpoint, the collapses in these cases (where the
549 probabilities of failure are higher than those mandated by FEMA P695) are all the result of sidesway
550 instability (i.e., P- effects), and related to frame (rather than connection) design. Consequently, they
24
551 are outside the scope of this paper. Nonetheless, the performance of structures designed as per current
552 practice serve as a useful benchmark to assess performance of frames designed with weaker bases.
553
554 From the viewpoint of prospective design, the major finding is that designing base connections for
555 the overstrength seismic load, rather than based on capacity design offers the most promise in
556 terms of performance (comparable to capacity based design) as well as economy (designing the
557 bases for significantly lower moments; see Table 2). For the 4-story frame, the difference between
558 capacity based design and overstrength based design is modest, such that the base connections are
559 not subjected to inelastic rotation. This suggests the possibility of designing bases in low-rise
560 structures (similar to the one investigated herein) without explicit consideration of base
562
563 On the other hand, performance of the mid- to high-rise frames (8-, 12- and 20-story) with base
564 connections designed for overstrength loads shows a dependence on base rotation capacity. For
565 these, performance comparable to frames with capacity-designed bases is achieved only when the
566 base connection has rotation capacity is on the order of 5% (i.e., 0.05 radians). While this appears
567 to be a large rotation capacity, numerous experimental studies conducted recently have
568 demonstrated that column base connections designed per current construction practice provide
569 deformation capacities that are in this range or higher. For example, in the Gomez et al. (2016)
570 and the Trautner et al. (2016, 2017) studies, exposed base connections showed rotation capacities
571 in the range of 0.06-0.10 radians. In the Grilli and Kanvinde (2017) study, embedded base
572 connections were determined to have rotation capacities in the range of 0.03-0.06 radians. These
573 details were not detailed explicitly for ductility, and utilized standard materials and welding
25
574 practices consistent with AISC 341-16, and base connection details consistent with Design Guide
575 One (Fisher and Kloiber, 2006), and the Seismic Design Manual (AISC, 2015). Other recent
576 research (Trautner et al., 2016), indicate that if explicitly designed for ductility (e.g., by providing
577 unbonded anchor rods with a large yield length), base connections are able to provide rotation
578 capacities well in excess of those required to achieve acceptable system performance (these
579 specimens showed rotation capacity in the range of 0.06-0.12 radians). In summary, requiring
580 ductility in base connections (e.g., through prequalification testing), and development of details
581 that meet such prequalification standards may be well within reach, given the inherent ductility of
582 these connections. Given the associated benefit of designing these connections for a lower strength
584
586 Column base connections in steel moment resisting frames are typically designed to be stronger
587 than the adjoining column, with the intent of forcing a plastic hinge into the column section itself
588 instead of in the connection. This practice is based on the presumption that column bases are less
589 ductile as compared to the member. This practice necessitates expensive base details that require
590 heavy base plates, multiple large anchor rods, or deeply embedded columns in concrete footings.
591 However, recent research has indicated that column base connections may have significant
592 deformation capacity, which is unutilized in the current design practice, possibly at a high cost.
593 Against this backdrop, this study seeks to i) quantify the interactive effects of base connection
594 strength, ductility, and flexibility on structural performance, and ii) based on this understanding,
595 to provide guidance for refining design considerations for the moment frames as well as the base
596 connections.
26
597
598 To achieve this, main scientific basis of the study is a series of parametric simulations on four (4-
599 , 8-, 12-, and 20-story) archetype moment frames designed as per current design standards in a
600 high seismic region (the Los Angeles Basin). Given the aims of the study, the base connections
601 are designed for four levels of seismic load; these levels include i) a reduced seismic load,
603 systems not explicitly designed for ductility, iii) the overstrength seismic load, corresponding to
604 = 3, and finally iv) the current design practice, in which base connections are capacity-designed,
605 to develop a moment corresponding to 1.1RyMP of the adjoining column. Frame models for these
606 buildings are constructed based on best simulation practices and member/component models that
607 are validated against experimental data. Parameters interrogated through these frame models
608 include four levels of base connection strength, five levels of base rotation capacity, and two levels
609 of base flexibility (corresponding to a fixed base, which is commonly assumed in design and
610 analysis, as well as a flexible base, based on estimated rotational flexibilities for each of the base
611 connections). In total, this results in 160 parametric combinations. Each of these parametric
612 combinations is evaluated through Nonlinear Static Pushover (NSP) and Nonlinear Time History
613 (NTH) simulations. The former (NSP) are conducted to obtain basic broad behavioral insights
614 regarding system overstrength, modes of failure, and period-based ductility. The latter (NTH) are
615 conducted within the framework of FEMA P695 to rigorously assess the performance of the
616 parameter combinations. Specifically, this performance assessment involves conducting suites of
617 NTH runs with 44 ground motions scaled to various intensities, to determine the probabilities of
618 collapse (or failure) at the intensity corresponding to a 2% probability of exceedance in 50 years
619 (i.e., MCE level shaking). Failure is assumed to occur due to one of two mechanisms: sidesway
27
620 collapse of the frame, or exceedance of the base connection rotation capacity in any of the base
621 connections. The main findings and design implications of the NSP as well as the NTH simulations
623 All aspects of response (i.e., period based ductility and system overstrength in the NSP
624 simulations, and failure probabilities in the NTH simulations) are only mildly sensitive to base
626 In the NSP simulations, both the system overstrength and the period-based ductility follow a
627 similar trend, wherein increasing the base connection strength or rotation capacity results in
628 higher system overstrength or period based ductility. However, this effect saturates with
629 respect to the base rotation capacity at higher levels of base connection strength, indicating a
631 The NTH simulations indicate that the frames (except for the 4-story frame), with base
632 connections designed for the R = 3 and R = 8 conditions do not achieve probabilities of failure
633 comparable to those for frames with capacity designed bases for any level of base rotation
634 capacity and thereby offer least promise in terms of a prospective design method. On the other
635 hand, the frames with base connections designed for overstrength seismic load ( = 3) show
636 the greatest promise in terms offering acceptable performance, while significantly reducing
637 design moments for the bases (by up to 120%). To achieve this performance, the base
638 connections must possess rotational capacity of approximately 0.05 radians. Based on recent
639 experimental research, the development and qualification of connection details with these
641 In conclusion, results of the study encourage examining whether it is feasible to design base
642 connections for lower strength (i.e., overstrength seismic as compared to capacity design), while
28
643 developing connection details with sufficient deformation capacity to achieve satisfactory
644 performance. Notwithstanding this positive outcome, this study has several limitations, which must
645 be considered in its interpretation in support of design development. Most importantly, the paper
646 utilizes limited number of frames with a limited set of parameters. Although these frames are useful
647 for obtaining broad insights about structural systems, the response of specific structural designs may
648 deviate significantly from those used in this study. This is an important consideration in generalizing
649 results of this study. Multiple design solutions are often possible for a given set of design constraints,
650 and this paper examines only one design outcome. Other factors that may affect the findings include
651 biases due to ground motion effects (including the effects of vertical acceleration). Owing to these
652 limitations, it is acknowledged that generalization of these findings and adoption into code based
653 design guidelines (e.g., AISC 341) is not likely without further research, albeit the study certainly
654 provides impetus for such research. Until then, these findings encourage performance based design
656
657 Although all relevant aspects of nonlinear structural response were simulated with high fidelity,
658 errors or bias in the simulation cannot be discounted. These include issues with the modeling
659 framework itself, e.g., the use of frame members with concentrated plasticity; this implies axial
660 force-moment interactions are accounted for in an indirect way, without consideration of a time-
661 varying axial force. Such a framework also cannot effectively model three dimensional modes of
662 response such as lateral-torsional buckling. Within the framework, the selection of hysteretic
663 model functional forms and parameters, as well as the lack of consideration of Soil Structure
664 Interaction effects may introduce additional error. Finally, the study only focuses on refinements
665 to design practice in the context of base connections and their interaction with the systems. As
29
666 discussed in the paper, even outside of base connection design, the current design practice is
667 subject to refinements; these are outside the scope of this study. Notwithstanding these limitations,
668 the study provides broad insights into the response of steel moment frames with weak bases,
669 demonstrates that weak-base design is a prospectively feasible design strategy, and provides a
671
672 REFERENCES
673 AISC, 2016. Seismic Provisions for Structural Steel Buildings (ANSI/AISC 341-16),
674 ANSI/AISC 341-16. ed. American Institute of Steel Construction, Inc., Chicago, IL.
675 AISC, 2015. AISC Seismic Design Manual. American Institute of Steel Construction, Inc.,
677 AISC, 2010. Seismic Provisions for Structural Steel Buildings (ANSI/AISC 341-10),
678 ANSI/AISC 341-10. ed. American Institute of Steel Construction, Inc., Chicago, IL.
679 AISC, 2005a. Seismic Provisions for Structural Steel Buildings (ANSI/AISC 341-05),
680 ANSI/AISC 341-05. ed. American Institute of Steel Construction, Inc., Chicago, IL.
681 AISC, 2005b. Steel Construction Manual 13th Edition. American Institute of Steel Construction,
683 AISC, 2005c. Prequalified Connections for Special and Intermediate Steel Moment Frames for
684 Seismic Applications, ANSI/AISC 341-05. ed. American Institute of Steel Construction,
686 Appiled Technology Council, 2009. Quantification of Building Seismic Performance Factors
30
688 ASCE, 2005. Minimum Design Loads for Buildings and Other Structures (ASCE 7-05),
689 ASCE/SEI 7-05. ed. American Society of Civil Engineers, Reston, VA.
690 Barnwell, N., 2015. Experimental Testing of Shallow Embedded Connections Between Steel
692 Elkady, A., Lignos, D.G., 2018. Full-Scale Testing of Deep Wide-Flange Steel Columns under
693 Multiaxis Cyclic Loading: Loading Sequence, Boundary Effects, and Lateral Stability
695 https://doi.org/10.1061/(ASCE)ST.1943-541X.0001937
696 Elkady, A., Lignos, D.G., 2016. Dynamic stability of deep and slender wide-flange steel columns
697 - Full scale experiments, in: Structural Stability Research Council Annual Stability
698 Conference 2016, SSRC 2016, April 12, 2016 - April 15, 2016, Structural Stability
699 Research Council Annual Stability Conference 2016, SSRC 2016. Structural Stability
701 Elkady, A., Lignos, D.G., 2015. Analytical investigation of the cyclic behavior and plastic hinge
702 formation in deep wide-flange steel beam-columns. Bull. Earthq. Eng. 13, 1097–1118.
703 https://doi.org/10.1007/s10518-014-9640-y
704 Fisher, J., Kloiber, L., 2006. Design Guide 1: Base Plate and Anchor Rod Design (Second
705 Edition).
706 Gioncu, V., Mazzolani, F., 2013. Seismic Design of Steel Structures. CRC Press. ISBN
708 Gomez, I., Deierlein, G., Kanvinde, A., 2010. Exposed Column Base Connections Subjected to
31
710 Grilli, D., Jones, R., Kanvinde, A., 2017. Seismic Performance of Embedded Column Base
711 Connections Subjected to Axial and Lateral Loads. J. Struct. Eng. U. S. 143.
712 https://doi.org/10.1061/(ASCE)ST.1943-541X.0001741
713 Grilli, D.A., Kanvinde, A.M., 2017. Embedded column base connections subjected to seismic
715 https://doi.org/10.1016/j.jcsr.2016.11.014
716 Ibarra, L.F., Medina, R.A., Krawinkler, H., 2005. Hysteretic models that incorporate strength and
718 Inamasu, H., Lignos, D., Kanvinde, A., 2018. Influence of embedded steel column base strength
722 Kanvinde, A.M., 2004. Micromechanical simulation of earthquake -induced fracture in steel
724 Kanvinde, A.M., Grilli, D.A., Zareian, F., 2012. Rotational stiffness of exposed column base
725 connections: Experiments and analytical models. J. Struct. Eng. 138, 549–560.
726 https://doi.org/10.1061/(ASCE)ST.1943-541X.0000495
727 Krawinkler, H., Mohasseb, S., 1987. Effects of panel zone deformations on seismic response. J.
729 Lignos, D., Krawinkler, H., 2011. Deterioration Modeling of Steel Components in Support of
730 Collapse Prediction of Steel Moment Frames under Earthquake Loading. J. Struct. Eng.
32
732 Medina, R.A., Krawinkler, H., 2005. Evaluation of drift demands for the seismic performance
734 https://doi.org/10.1061/(ASCE)0733-9445(2005)131:7(1003)
735 National Institute of Standards and Technology (NIST), 2010. Evaluation of the FEMA P695
738 Newell, J.D., Uang, C.-M., 2008. Cyclic behavior of steel wide-flange columns subjected to
740 9445(2008)134:8(1334)
742 Torres-Rodas, P., Zareian, F., Kanvinde, A., 2018. A Hysteretic Model for The Rotational
743 Response of Embedded Column Base Connections. Soil Dyn. Earthq. Eng. Under
744 Review.
745 Torres-Rodas, P., Zareian, F., Kanvinde, A., 2017. Rotational stiffness of deeply embedded
747 https://doi.org/10.1061/(ASCE)ST.1943-541X.0001789
748 Torres-Rodas, P., Zareian, F., Kanvinde, A., 2016. Hysteretic Model for Exposed Column-Base
750 541X.0001602
751 Trautner, C.A., Hutchinson, T., Grosser, P.R., Silva, J.F., 2017. Investigation of steel column-
752 baseplate connection details incorporating ductile anchors. J. Struct. Eng. U. S. 143.
753 https://doi.org/10.1061/(ASCE)ST.1943-541X.0001759
33
View publication stats
754 Trautner, C.A., Hutchinson, T., Grosser, P.R., Silva, J.F., 2016. Effects of detailing on the cyclic
755 behavior of steel baseplate connections designed to promote anchor yielding. J. Struct.
757 Tryon, J., 2016. Simple Models for Estimating the Rotational Stiffness of Steel Column-to-
759 Vian, D., Bruneau, M., 2003. Tests to Structural Collapse of Single Degree of Freedom Frames
760 Subjected to Earthquake Excitations. ASCE Journal of Structural Engineering. 129 (12),
761 1676-1685.
762 Zareian, F., Kanvinde, A., 2013. Effect of column-base flexibility on the seismic response and
764 https://doi.org/10.1193/030512EQS062M
765 Zareian, F., Krawinkler, H., 2012. Conceptual performance-based seismic design using building-
766 level and story-level decision support system. Earthq. Eng. Struct. Dyn. 41, 1439–1453.
767
34