Journal of Constructional Steel Research 177 (2021) 106417

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Seismic design of non-dissipative embedded column base connections

Hiroyuki Inamasu a, Amit M. Kanvinde b, Dimitrios G. Lignos a,⁎
a
Department of Architecture, Civil, Environmental Engineering, École Polytechnique Fédérale de Lausanne, Station 18, Lausanne 1015, Switzerland
b
Department of Civil and Environmental Engineering, University of California, Davis, CA 95616, United States of America

a r t i c l e i n f o a b s t r a c t

Article history: Embedded Column Base (ECB) connections in seismically resistant steel moment frames are commonly designed
Received 5 June 2020 to be stronger than the connected column, to protect them from inelastic actions. This relies on estimation of de-
Received in revised form 18 September 2020 mands induced by the column and the strength capacity of the connections themselves. However, recent re-
Accepted 13 October 2020
search indicates that prevalent approaches may be unconservative for both demand and capacity estimation,
Available online 24 October 2020
with the consequence of unintended damage/failure of the connection. Motivated by the above, this research
Keywords:
(1) characterizes the seismic demands induced in ECB connections, and (2) evaluates various strength models
Embedded column bases for these connections, in support of improved design approaches. This is done through a virtual test program
Wide flange columns that uses validated continuum finite element simulations of interactive column-ECB subassemblies. The results
Capacity design suggest that the current design approaches for ECB connections are unconservative, i.e., they underestimate de-
Cyclic loading mands while also overestimating the ECB resistance, with the possibility of unintentional nonlinear behavior
Finite element model within the embedded portion of the ECB connection. This is undesirable, because the ECB connections are not
usually detailed to provide plastic deformation capacity. A method is proposed to provide improved estimates
of the anticipated flexural demands of non-dissipative ECB connections, along with recommendations regarding
the strength models. To accomplish this, the method incorporates local cross-sectional slenderness, gravity-
induced axial load demand in conjunction with load combinations imposed by current seismic design standards.
The sensitivity to the steel column material grade, the imposed loading history, and axial load demands is studied,
and limitations of the approach are outlined.
© 2020 Elsevier Ltd. All rights reserved.

1. Introduction In current design practice, it is commonly assumed that the steel col-
umn base portion above the RC footing is the dissipative element during
Column base connections in steel moment-resisting frames (MRFs) an earthquake (through the formation of a column plastic hinge),
withstand combinations of axial load, flexural and shear demands that whereas ECB connections themselves are non-dissipative (i.e., assumed
are transferred to the foundation system during earthquake loading. to be brittle, and designed to be stronger than the attached column).
While exposed column base connections are a standard practice in The primary source of guidance for the strength-based design of these
low-rise steel construction, embedded column bases (ECBs) are com- connections is the AISC Seismic Design Manual [1]. This follows a
monly used to achieve a fixed boundary condition at the lower end of strength model, which was originally proposed for the design of a steel
first story steel columns in seismic resistant mid- to high-rise steel coupling beam bending within a concrete shear wall [2,3]. Despite of
MRFs. Fig. 1 schematically illustrates a typical ECB detail, indicating its some similarities, this situation has significant differences with respect
main components, including the steel column. The column is usually to ECB connections, including: (1) the absence of axial load in the cou-
welded to a base plate that rests on a concrete layer for leveling of the pling beam, which is present in the ECB column, (2) limited confinement
steel column. The embedded portion of the steel column is encased by in the thin shear wall, (3) the absence of the base plate, as well as differ-
a reinforced concrete (RC) foundation. Face bearing plates are often ences in reinforcing bar patterns. As a result, the current design ap-
employed at the top concrete layer to transfer axial compression and fa- proaches are not directly applicable to ECB connections.
cilitate the formation of a shear panel. Referring to Fig. 1, the flexural re- Moreover, in these as well as other approaches (e.g., [4]), it is com-
sistance of non-dissipative ECB connections is provided through monly assumed that the ultimate flexural resistance of ECB connections
horizontal bearing of the column flanges as well as vertical bearing of is controlled by concrete bearing failure ahead of the compression
the base plate and the face bearing plates. flange of the column. Recent experiments and simulations [5] suggest
that other failure modes are active as well – these include pryout of con-
⁎ Corresponding author. crete on the tension side of the connection due to the uplift of the base
E-mail address: dimitrios.lignos@epfl.ch (D.G. Lignos). plate, as well as cracking or failure in the web panel inside the RC

https://doi.org/10.1016/j.jcsr.2020.106417
0143-974X/© 2020 Elsevier Ltd. All rights reserved.
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

Fig. 1. Typical embedded column base connection detail and resisting mechanisms.

footing, further suggesting that the existing design guidance may be Motivated by these issues, the main objectives of this paper are to:
inadequate. These experiments also indicate that the behavior of (1) critically examine the assumptions commonly used to design ECB
ECB connections originates prior to reaching their assumed ultimate connections, focusing on both capacity and demand assessment, and
flexural resistance; this is due to gapping between the tension flange (2) provide recommendations for improved design of non-dissipative
of the column and the adjacent concrete, as well as nonlinearity of ECB connections. To achieve these objectives, the paper first reviews
the concrete on the bearing side adjacent to the compression flange. the current state-of-the-art in terms of the seismic design of non-
This is further corroborated by field observations from recent earth- dissipative ECB connections. This is accomplished by direct comparisons
quakes [6,7] in which appreciable damage or inelastic behavior is with pertinent experimental data available in the literature. Subse-
noted in ECB connections, that are designed as non-dissipative quently, the expected seismic demands of typical ECB connections are
(i.e., to be stronger than the adjoining column). Collectively, these ex- quantified through Continuum Finite Element (CFE) analyses of a virtual
perimental and field observations suggest that current methods for test matrix. A validated CFE model is used for this purpose. Within the
strength characterization of ECB connections may be inadequate, virtual test matrix, several parameters, that may affect the seismic
and quite possibly unconservative. demands of ECB connections, are interrogated. The findings are synthe-
Similar concerns may also be noted on the demand side. Specifically, sized into coherent recommendations for the seismic design of non-
ECB connections are usually capacity-designed to develop the strain dissipative ECB connections in seismic resistant steel MRFs.
hardened capacity of the column – usually, this is determined as 1.1
times the expected full-plastic resistance of the column for A992 Gr. 2. Review of prevalent strength models in the United States and
50 (i.e., nominal yield stress, fy, n = 345MPa). However, research on Japan
the inelastic behavior of wide-flange steel columns under multi-axis cy-
clic loading [8–12] suggests that the steel column force demands may be The current AISC provisions [22] and the Design Manual [1] adapt
significantly higher than the above estimate. Moreover, the column the flexural/shear resistance of a steel coupling beam embedded to an
force demands are strongly influenced by (a) the steel material cyclic RC wall for the seismic design of non-dissipative ECB connections. The
hardening; (b) the loading conditions (i.e., coupled axial load and lateral stress block distributions, which are shown in Fig. 2a, are employed
drift demands); and (c) the cross-sectional slenderness that controls for this purpose. Concrete bearing controls the design shear resistance
the onset of local buckling. However, the influence of the above param- of the ECB connection, which may be computed as follows,
eters has not been considered in the seismic design process of ECB con-
nections. Finally, although assumed to be fixed, ECB connections exhibit qffiffiffiffiffi  0:66 !
significant rotational flexibility [5,13–21], contributing to as much as 0 bw 0:58−0:22∙β1
V ECB
Rd,AISC ¼ 4:04 f c ∙ ∙β1 ∙bf ∙dembed ∙ ð1Þ
0.4–0.8% to the first story lateral drift demands. bf 0:88 þ d L
embed

2
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

Fig. 2. Available design models for non-dissipative embedded column base connections.

The corresponding flexural resistance of an ECB connection may deffective ¼ dref ≤dembed ð5Þ
then be estimated as follows, 2 8  93
1 4 < 2 2
dL þ dU =
MECB
Rd,AISC ¼ V ECB
Rd,AISC ∙Lc ð2Þ M ECB ¼
0
β∙β1 ∙f c ∙bj ∙ dL ∙deffective − 5 ð6Þ
Rd,Grilli
1−α : 2 ;
In Eq. (1), fc’ is the specified concrete compressive strength (in MPa);
bw (in mm) is the width of the concrete foundation perpendicular to the In which, bj is equal to (bf + B)/2 and accounts for the concrete com-
plane of bending; bf (in mm) is the column flange width; Lc is the dis- pression field outside the panel zone (in which, B is the base plate width
tance from the top surface of the RC foundation to the inflection point as shown in Fig. 1); Es and Ec are the Young's modulus of the steel and
of the column (in mm); β1 is a factor relating the depth of the equivalent concrete materials, respectively; I, is the second moment of inertia of
rectangular compressive stress block to the neutral axis depth as de- the steel column with respect to the direction of lateral loading. In
fined in [23]; and dembed is the embedment depth of the steel column. Eq. (4), C is an empirical factor, assumed equal to 1.77. Referring to
In Eq. (1), the term (bw/bf)0.66 accounts for the confinement effect of Fig. 1, the parameters, dL and dU are the depths of the lower and upper
the concrete. Referring to Fig. 2a, fb is the assumed maximum bearing horizontal concrete bearing equivalent rectangular stress blocks, re-
stress of the concrete; c is the distance from the concrete top surface spectively. These may be determined by solving the force (i.e., shear
to the neutral axis in the assumed stress distribution; εf and εb are the and bending) equilibrium equations of the steel column inside the RC
assumed concrete strain demands at the top and bottom of the embed- footing. Grilli and Kanvinde [24] proposed that the maximum horizontal
ded portion of the column. bearing resistance is determined when dL + dU reaches 60% of the effec-
Grilli and Kanvinde [24] indicated that Eq. (1) does not consider the tive embedment depth.
base plate contribution to the overall shear resistance of the ECB con- The AIJ [4] design provisions rely on the concrete bearing to estimate
nection (see Fig. 1). Moreover, the term reflecting the effect of confine- the flexural resistance of ECB connections. Referring to Fig. 2b, depend-
ment on the corresponding bearing stress is unbounded. Accordingly, ing on the assumed stress distribution, a designer may either estimate
they introduced two factors, namely a and β to account for the base the flexural resistance at yield or at ultimate (peak) of the respective
plate contribution to the shear resistance of ECB connections and the ECB. The latter may be employed in the seismic design process of non-
concrete confinement, respectively. While the former (see Eq. (3)) dissipative ECB connections in steel MRFs. This may be estimated as fol-
varies with respect to the embedment depth, dembed, and the reference lows,
depth, dref (see Eq. (4)), the latter is assumed to be constant and equal qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
to 2.0 after following the work of [25,26]. As such, the flexural resistance M ECB
2 2
Rd,AIJ ¼ f cu ∙Bc ∙Lc ∙ ð2Lc þ dembed Þ þ dembed −ð2Lc þ dembed Þ ð7Þ
of a non-dissipative ECB connection should be computed as follows,

dembed In which, fcu is the maximum bearing strength, assumed equal to the
α ¼ 1− ð3Þ
dref specified concrete compressive strength fc’; Bc is the column width
(i.e., equal to bf for wide flange cross-sections). In Fig. 2b, xu is the dis-
C tance from the concrete top surface to the neutral axis in the assumed
dref ¼  14 ð4Þ
E stress distribution. While hollow structural sections are more common
4∙Es ∙I
in the Japanese steel construction, Eq. (7) was originally developed

3
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

based on experimental work on ECBs featuring wide flange steel col- was associated with pryout at large lateral drift demands. Table 1 sum-
umns [13,14,18,19,21,27]. These represented interior steel MRF col- marizes these tests including their key characteristics. Three specimens
umns. In exterior column ECB connections, the AIJ [4] design tested by Grilli et al. (#1, #2 and #4) were subjected to axial load in ad-
provisions recommend specification of reinforcing bars where horizon- dition to lateral cyclic loading. However, the influence of axial load on
tal bearing concrete compressive stresses are expected to be large. the ECB response was not found to be important [5].
Fig. 3 depicts the typical hysteretic behavior of ECB connections from
3. Available experimental data and assessment of current design a prior test (i.e., Test #3) of the experimental program by [5]. The figure
models illustrates the base moment Mbase versus the column drift ratio, θ. Sev-
eral quantities of interest are extracted from this figure – namely, the
To assess the efficacy of various capacity design approaches, forty- maximum attained moment, Mmax, p and Mmax, n (positive and nega-
nine physical experiments on ECB connections featuring interior wide tive, respectively), and column drift ratio corresponding to Mmax [=
flange steel columns available in the literature were reviewed max (|Mmax, p|, |Mmax, n|)], θmax; the effective elastic stiffness, Ke, which
[5,13,15,18,19,21,27,28]. In 11 out of these 49 tests, the steel column is defined as the secant stiffness using the point when Mbase firstly
remained elastic. The peak flexural/shear resistance of these ECBs reaches My = 0.7Mmax; the yield rotation, θy, defined as the column
were determined by concrete bearing, which is consistent with the as- drift ratio at My, the maximum attained moment of the ECB: MECB max =
sumed failure mode by the two aforementioned design models. How- Mmax; and the yield moment of the ECB: MECB y = My. The corresponding
ever, the final failure mode in some of the available experimental data rotations of the ECB, θECB ECB
max, and, θy ,are deduced by subtracting the

Table 1
Experiments on ECB connections

Reference ID Column cross section (mm) dembed/dc Foundation size (mm) Concrete compressive Axial loadc (kN)
(dc × bf × tw × tf)a (bw × hf × lf)b strength fc′(MPa)

Washio et al. [21] POSALI 125 × 125 × 7×9 1.0 300 × 500 × 1300 25.3 0
Washio et al. [21] POSAHI 125 × 125 × 7×9 1.0 300 × 500 × 1300 35.0 0
Washio et al. [21] POTALI 125 × 125 × 7×9 1.0 300 × 500 × 1300 25.3 0
Minami et al. [14] LA20 200 × 100 × 5.5 × 8 1.0 400 × 600 × 1600 20.6 0
Minami et al. [14] LA10 200 × 100 × 5.5 × 8 0.5 400 × 600 × 1600 20.6 0
Takeda and Takahashi [19] S2NO 250 × 100 × 9 × 19 1.0 1000 × 600 × 2800 16.2 0
Akiyama et al. [13] ES-C 200 × 200 × 8 × 12 1.1 410 × 800 × 2750 20.0 0
Grilli et al. [5] #1 455 × 418 × 42 × 68 1.1 1830 × 1092 × 3650 29.2 445
Grilli et al. [5] #2 566 × 305 × 39 × 70 0.9 1830 × 1092 × 3650 29.2 445
Grilli et al. [5] #3 455 × 418 × 42 × 68 1.7 1830 × 840 × 3650 29.2 0
Grilli et al. [5] #4 455 × 418 × 42 × 68 1.7 1830 × 840 × 3650 29.2 445
a
dc, bf, tw, and tf denote the depth, width, web thickness, and flange thickness of the column, respectively.
b
bw, hf and lf denote the width, height and length of concrete foundation, respectively.
c
Positive value indicates compressive load.

Fig. 3. Typical hysteretic behavior of embedded column bases along with the response parameters of interest.

4
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

elastic deformations due to flexure and shear of the column from θmax 4. Assessment of embedded column base design procedures through
and θy, respectively. The pre-capping plastic rotation of the ECB, θECB p is virtual testing
defined as the difference between θECB ECB
max and θy .
Fig. 4a shows a comparison of the measured maximum attained ECB 4.1. Development of finite element model
moment, MECB ECB
max, versus the model prediction, MRd , from Eqs. (2), (6) and
(7) for the 11 test specimens, which are summarized in Table 1. In all Referring to the preceding discussion, it is evident that both the de-
cases, the cover concrete is considered part of the embedment depth mand as well as the capacity assessment methods implicit in the current
as suggested in prior studies [2,3,29]. The established comparisons sug- design methods are inadequate from the standpoint of meeting design
gest that, in most cases, the AIJ [4] model underestimates the test re- objectives (i.e., ensuring elastic response of the ECB connection). Be-
sults by at least a factor of two (average test-predicted ratio 3.20, cause the overall performance depends on the interactions between
with a Coefficient of Variation 0.40), whereas the AISC design model the column as well as the ECB connection, it is informative to directly
[1,22] yields more reasonable predictions (average test-predicted simulate these interactions. For this purpose, a CFE model is developed
ratio 1.27, Coefficient of Variation 0.17). The AIJ model seems conser- to simulate the cyclic behavior of wide flange steel columns embedded
vative because both the effect of concrete confinement on the bearing into RC footings. The CFE model is developed in the commercial soft-
stress capacity as well as the contribution of the base plate are con- ware ABAQUS (version 6.14–1) [30]. A schematic illustration of the
servatively neglected. For five of the test specimens, the AISC model CFE model is shown in Fig. 5a. In brief, the steel column is modelled
underestimates the attained ECB flexural resistance by 40 to 50%. by shell elements with reduced order integration (S4R), while the ECB
This may be attributed to the fact that the base plate contribution to hysteretic behavior is condensed in a rotational spring at the bottom
the ECB's lateral resistance is neglected. Referring to Fig. 4a, the accu- of the column that is tied to the bottom surface of the column. The hys-
racy of the Grilli and Kanvinde model [24] is noteworthy excluding teretic behavior of this spring is discussed hereinafter. The column top
one case (i.e., Specimen S2NO), in which the associated error is rotational spring represents the top end in-plane boundary condition
about 65%. This test specimen featured foundation beams perpendic- of a first story column in capacity-designed steel MRFs. Details on how
ular to the loading direction, thereby increasing the effective width, to derive the flexibility of the top end spring are discussed in [31]. The
as well as the confinement of the concrete foundation compared to CFE model is parametrized and may easily be shifted to a cantilever col-
the remainder of the test specimens. umn, by setting the top end spring's in-plane rotational stiffness to zero
Although the data shown in Fig. 4a corresponds to the peak flexural and by considering the proper member length. The CFE model is vali-
resistance of the ECB connection, Grilli and Kanvinde [24] suggested dated with available full-scale physical data on wide flange steel col-
that the yield flexural resistance, MECB y , of an ECB connection may be as- umns [8] as well as embedded column base connections [5]. The
sumed equal to 0.70MECB max as a suitable estimate of the design resistance – boundary conditions mimic the ones from the experimental setup of
this is because the connection exhibits significant nonlinearity after this Suzuki and Lignos (2015) [8] and Grilli et al. (2017) [5]. The former em-
point is reached. The gathered data, including the ones by [24], suggest ploys cantilever columns, which are loaded with respect to the strong
that the corresponding MECB ECB ECB
y, deduced/Mmax (My, deduced denotes the ECB mo- axis cross-sectional orientation. The column bottom is fixed in all six
ment at a tangent stiffness corresponding to 30% of the elastic stiffness of DOFs, whereas the column top end is only fixed in the out-plane direc-
the footing) ratio is, on average, about 0.65 with a standard deviation of tion in the displacement and rotational DOFs due to the lateral bracing
0.08. This is depicted in Fig. 4b in which, the MECB ECB
y, deduced/Mmax ratio is ex- system. In Grilli et al. (2017) [5], each specimen consists of a cantilever
tracted directly from each one of the gathered test specimens. This is column with an ECB connection. Similarly, only in-plane movement was
even more concerning especially when the prediction of the maximum allowed in the test.
flexural strength of ECBs is accurate, because it amplifies the uncon- With regard to the wide flange steel columns, the CFE modelling
servativeness of current methods for design, which rely on the ultimate strategy follows the general modelling recommendations by [11,12,32].
resistance of the ECB connection. The CFE model features the Voce-Chaboche nonlinear multi-axial

Fig. 4. Comparison of flexural strength of ECBs with gathered experimental data: (a) available peak flexural resistance, MECB ECB.
max, (b) yield flexural resistance, My, deduced.

5
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

Fig. 5. CFE model development and its validation (test data retrieved from [8]).

plasticity law [33,34]. The material model relies on consistent parame- hot-rolled profiles. Indicative values for these imperfections are
ters calibrated through a gradient-based optimization based on work discussed in [11,12].
by [35,36]. Residual stresses due to hot-rolling are considered based on The model validations constitute physical testing conducted by the
the residual stress distribution proposed by [37]. This is confirmed by re- last author and his former students [8,12] in which fixed-end cantilever
cent residual stress measurements on hot-rolled wide flange profiles specimens were subjected to a range of multi-axial cyclic loading histo-
[38]. An idealized fixed-base column may be considered if the in-plane ries. Fig. 5 shows indicative comparisons with such experiments with
rotational spring of the column base is made infinite. In the CFE model, emphasis on tests subjected to collapse-consistent loading histories
local buckling is triggered by considering local imperfections with a mag- coupled with either compressive or time-varying axial load demands.
nitude, which is smaller than the allowable manufacturing tolerance of From the same figure, the agreement between the simulated and

6
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

Fig. 6. CFE simulation model and validation with experimental data (test data from [5]).

experimental results is noteworthy in terms of global and local defor- For this purpose, we developed and implemented in the ABAQUS fi-
mation patterns, deduced moment-column drift ratios as well as axial nite element software (version 6.14–1) a user-defined element (UEL).
shortening versus column drift ratios regardless of the imposed loading Referring to Fig. 6a, the model follows a trilinear backbone curve in-
histories. cluding the basic loading, unloading and re-loading paths. These are
As noted in Fig. 6, the hysteretic behavior of ECB connections is dom- controlled by two pinching parameters px and py [40]. The model is
inated by pinching and unloading stiffness deterioration due to the con- able to simulate basic strength and unloading stiffness deterioration
crete bearing failure. Experiments on ECBs by the second author and his based on the energy-based deterioration rule proposed by Rahnama
former students [5] suggest that the above failure mode within the con- and Krawinkler [40,41] and subsequently modified by Lignos and
crete footing is not strongly influenced by the compressive axial load de- Krawinkler [42]. This rule assumes that the ECB connection has a refer-
mand. Consequently, its hysteretic response may be represented ence hysteretic energy dissipation capacity, EECB
c , which may be esti-
through a concentrated hinge model with a pinched behavior as illus- mated as follows,
trated in Fig. 6a. A similar approach was also adopted in prior studies
[39]. Note that local responses (e.g., base plate and concrete) cannot be
disaggregated and discussed separately with this modelling approach. EECB
c ¼ λECB ∙θECB ECB
p ∙M y ð8Þ

7
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

Fig. 7. Embedded column base design example as adopted from the Seismic Design Manual.

in which, λECB, is the reference cumulative rotation capacity, which in story 3-bay steel MRF building. Fig. 7a and b illustrates the plan view
turn may be estimated through calibrations with ECB physical tests and elevation of the steel frame building, respectively. Fig. 7c illustrates
exhibiting concrete bearing failure. the final design of a typical first story ECB connection. This is comprised
Fig. 6b and c illustrate comparisons of the simulated versus mea- of a 4270 mm long W14x176 steel column. The cross-section profile is
sured hysteretic response with noteworthy accuracy. Referring to made of ASTM A992 Gr. 50 (i.e., nominal yield stress, fy, n=345 MPa).
Fig. 6b and c, while flexural strength deterioration may not be evident The steel column is embedded into an RC footing, which is made of
in the ECB hysteretic response, the primary deteriorating mechanism 28 MPa concrete. The foundation is reinforced with longitudinal and
is associated with unloading stiffness deterioration, which is induced transverse reinforcing bars such that the concrete bearing becomes
by concrete bearing failure. The cumulative rotation capacity, λECB was the critical failure mode. Furthermore, deformed anchor bars are at-
found to be 2.6 and 25.8 for specimens 1 (dembed = 508 mm) and 3 tached to the embedded column flanges as transfer reinforcement. A
(dembed = 762 mm), respectively. The corresponding pinching parame- base plate is not present in this example. Following the Seismic Design
ters are determined as px = 0.8, py = 0.05. These parameters are Manual, the column base may be designed for a flexural demand that
adopted hereinafter. is the lesser of (a) 1.1ZRyfy, n or (1.1/1.5)ZRyfy, n, (i.e., Ry, fy, n, and Z are
the ratio of the expected yield stress to the specified minimum yield
stress, specified minimum yield stress of steel, and the plastic modulus
4.2. Simulation of a characteristic design case study
of the steel column cross section, respectively); and (b) the flexural de-
mand, which is calculated by the load combinations according to [43].
The CFE model is used to simulate the hysteretic response of a char-
The maximum ECB flexural and shear resistances for the specific design
acteristic design case, which is documented in the AISC Seismic Design
example are found to be, MECB ECB
max = 1964kNm, Vmax = 921kN, respectively.
Manual [1]. The example features an ECB connection as part of a 4-

8
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

Fig. 8. Base moment – column drift ratio along with the decomposition of the total and accumulated rotations versus the loading excursion.

Notably, according to the current design provisions [22], there is no column–RC footing subassembly follows the modelling principles pre-
distinction with regard to the corresponding steel material hardening sented earlier. At the column top end, the in-plane rotational flexibility
depending on the steel grade of the column (e.g., A992 Gr. 50 versus of the first-floor beam-to-column connections is explicitly considered.
A913 Gr. 65). Moreover, the ECB connection may be designed for a mo- This indicates a more realistic boundary condition of first story steel
ment demand that is smaller than the corresponding plastic flexural re- MRF columns [9,31]. Based on prior work by the authors [5,31], the elas-
sistance of the embedded steel column. The CFE model of the steel tic stiffness of the ECB is estimated to be 5EsI/L of the column (L is the

9
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

length of the column). The remaining degrees of freedom in the CFE Table 2
model are idealized as fixed. To reflect the associated variability of con- Virtual test matrix.

crete in Eq. (1) with regard to the estimation of the maximum flexural Cross Web local Steel Lateral loading Axial load protocol
resistance of the ECB connection, three cases are considered, including section slenderness material protocol
an “as-designed” case (Case I) with nominal properties and two other (h/tw)

cases, where the MECBmax, reflect a 10% and 20% increase compared to the W14x370 6.9 A992 Gr. SAC symmetric, Constant: Pg/Py = 5%,
“as designed” case (Case II and Case III, respectively). The corresponding W24x370 14.2 50, Collapse 20%, 35%, 50%,
W24x229 22.5 A913 Gr. consistent Variable: Pg/Py = 5%,
embedment depth, for the ECB to remain elastic, are 559 mm, 615 mm
W24x146 33.2 65 15%, 25%
and 671 mm, respectively, for the three cases discussed above. W24x103 39.2
The steel column-ECB connection subassembly is subjected to the W24x84 45.9
SAC symmetric cyclic loading protocol [44] coupled with a constant
compressive axial load that is 9% of the cross-sectional yield load calcu-
lated with the expected yield stress as indicated by the design example
[1]. Fig. 8 illustrates the simulated results for the ‘as-designed’ case 4.3. Parametric study
(Case I) in terms of deduced column base moment, Mbase, versus column
drift ratio as defined in Fig. 8a. The analyzed case is compared with an Using the methodology outlined in the previous sections, a virtual
ideally fixed-base steel column without the presence of the RC founda- test matrix (see Table 2) is assembled to examine the design space in
tion (denoted “fixed-base”). Referring to Fig. 8b, the simulation results which ECB connections concentrate their inelastic deformations solely
suggest that the “as designed” case exhibits inelastic behavior in the in the steel column right above the RC footing, as implied in capacity de-
RC footing rather than the steel column itself. This is clearly seen in signed steel MRFs. While the ECB is assumed to behave elastically in the
Fig. 8c that depicts the decomposition of the total into its subcompo- examined cases, its flexibility is explicitly considered in the CFE model
nents from the steel column portion and the ECB connection. For lateral as discussed in [31]. Referring to Table 2, six wide flange steel profiles
drift demands of about 1%, the accumulated rotation between the two commonly used in seismic resistant steel MRFs [45,46], are selected.
components is roughly equal, acknowledging the elastic contribution Their web local slenderness ratio, h/tw, ranges between 7 ≤ h/tw ≤ 46.
of the ECB to the total lateral drift. However, above 2% lateral drift de- Cross sections near the lower bound are unlikely to experience local
mands, the accumulated rotation from the ECB is larger than that of buckling at high lateral drift demands [47], thereby inducing high flex-
the steel column above the RC footing. Evidently, the current design ap- ural demands due to the associated material cyclic hardening. On the
proach, does not prevent the inelastic behavior within the RC founda- other hand, slender cross sections near the current slenderness limits
tion, which is the non-dissipative (and possibly brittle) element in this for highly ductile members tend to limit the flexural demands due to
case. A 10% and 20% increase in the embedment depth of the ECB con- the onset of nonlinear geometric instabilities at modest lateral drifts
nection decreases the inelastic contribution of the RC footing as [9,10]. The steel material is either A992 Gr. 50 (fy, n=345 MPa) or
shown in Fig. 8d and e, respectively. A913 Gr 65 (fy, n=450 MPa). While the former is a standard choice in
While the mischaracterization of connection capacity is one issue, seismic resistant steel MRFs, the latter is considering its prospective
another key reason for the insufficient design of the ECB connection ac- use in future seismic design. The employed model parameters of the
cording to the Design Manual [1] pertains to the fact that the flexural de- Voce and Chaboche multiaxial plasticity model are based on the model-
mands on the ECB may exceed the estimated flexural resistance of the ling by [35,36].
steel column itself. The mechanistic reasons of this increase are associ- Referring to Table 2, two loading protocols are used: (1) the sym-
ated with (a) cyclic hardening which is a function of the material metric SAC symmetric protocol [44,48], which is likely to produce the
type; (b) cross sectional compactness that controls the onset and pro- highest flexural demands, and (2) a collapse-consistent loading proto-
gression of local and/or lateral torsional buckling of the steel column; col [49] that is more representative to be used for the quantification of
(c) the influence of the loading history that may manifest the material the seismic demands of ECB connections prior to incipient structural
hardening prior to the onset of local buckling; and (d) the axial load collapse. The lateral loading history is applied in the presence of either
demands imposed to the steel column. To determine the relative and in- a constant compressive axial load, ranging from 5% to 50% of Py
teractive contributions of these effects, a parametric study is conducted (i.e., axial yield strength of the column based on expected material
as outlined in the next section. properties, Py = Ryfy, nA), or a time-varying axial load in which the

Fig. 8 (continued).

10
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

steel columns may experience tensile axial loading during a lateral drift in Fig. 11 to establish general trends that inform the prospective devel-
excursion. Albeit the former is common in interior columns, the latter is opment of refined design recommendations for non-dissipative ECB
representative in exterior columns due to dynamic overturning effects. connections. The following observations are made:
The axial load variation is synchronized with the respective lateral load-
ing history as depicted in Fig. 9. This variation is established based on • Referring to Fig. 10a and Fig. 11a, steel columns with stocky cross sec-
the methodology discussed in [49]. tions (i.e., h/tw in the range of 7–14) with compressive axial load ratios
The reduction in the variation of axial load demand is attributed to Pg/Py ≤ 0.20, attain a Mmax/fyZ ratio of about 1.6. On the other hand, the
force redistribution within the overall frame after the onset of geometric same ratio, for cross sections near the AISC 341–16 compactness limits
instabilities. The steel columns summarized in the virtual test matrix of (i.e., h/tw in the range of 33–46) for highly ductile members, does not
Table 2, are subjected to progressively increasing loading excursions till exceed 1.2. In the latter, the onset of local buckling at modest lateral
their peak flexural strength deteriorates cyclically by at least 20%. drift demands tends to inhibit the column's flexural demands as
shown in Fig. 10b for two analyzed cases featuring a stocky and slen-
der cross section profiles.
4.4. Results and discussion • Referring to Fig. 10c and d, in column cross sections near the AISC
341–16 [22] compactness limits for highly ductile members, when
Fig. 10 summarizes selected results regarding the influence of each subjected to variable axial load demands, the Mmax/fyZ ratio does not
variable on the hysteretic response of steel columns in terms of the col- exceed 1.2.
umn base moment versus the column drift ratio as defined in Fig. 8a. • With regard to the steel material (see Fig. 10e), while A913 Gr. 65
The moment is normalized with respect to the plastic flexural resistance steel columns exhibit modest isotropic hardening compared to their
of the respective column cross section. For each case, the maximum A992 Gr. 50 counterparts, the larger kinematic hardening of A913
attained flexural demand, Mmax, is extracted and collectively presented Gr. 65 further delays the onset of local buckling. However, the

Fig. 9. Lateral and axial loading histories on a W24x146 column with gravity offset, Pg/Py = 0.15 (A992 Gr. 50).

11
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

Fig. 10. Influence of investigated parameters on the hysteretic response of steel columns

12
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

differences with wide-flange profiles made of A992 Gr. 50, in terms of the imposed lateral and/or axial loading history has a notable influence
the Mmax/fyZ ratio, appear insensitive to the steel material (see Fig. 11a on the plastic deformation capacity of steel columns, particularly near
and b). the incipient collapse limit state, it does not appear to influence the col-
umn flexural demands design basis (i.e., probability of exceedance of 10%
in 50 years) or maximum considered earthquake (i.e., probability of ex-
• Fig. 10f, 11b and d suggest that flexural demands in embedded steel
ceedance of 2% in 50 years) intensity. This is consistent with earlier find-
columns are sensitive to the imposed lateral loading history only
ings on engineering demand parameters of steel MRFs under seismic
when (a) they feature stocky cross-sections (i.e., h/tw ≤ 15); and
loading [50]. Based on these observations, a general regression-based
(b) they are subjected to time-varying axial load demands. Local
model is suggested to estimate flexural demands in ECB connections in
buckling-induced softening caps the flexural demands near the ECB
steel MRFs. Particularly, the flexural demand ratio may be estimated as
connection in all other cases.
follows,

4.5. Implications for prospective design    

M max h Pg
¼ a0 þ a1 ∙ þ a2 ∙ þε ð9Þ
f yZ tw Py
Based on the above discussion, it is noted that the maximum attained
flexural demands in capacity-designed steel MRF columns as part of non-
dissipative ECB connections are strongly influenced by the steel column In which a0 is the intercept; a1 and a2 are the estimated coefficients
position (i.e., interior versus exterior that result in constant versus time- of the variables h/tw and Pg/Py; ε is the standard error of the predictor.
varying axial load), and the cross section geometry (i.e., stocky versus The flange local slenderness ratio, bf/2tf, is strongly correlated with
slender profile). The distinction between A913 Gr. 65 and A992 Gr. 50 h/tw for hot-rolled cross sections manufactured in the United States
is statistically insignificant with respect to the corresponding Mmax/fyZ and Europe [51,52]; hence, bf/2tf is excluded from the multivariate re-
ratio based on a standard t-test at the 95% confidence interval. While gression. However, this assumption may not hold true if built-up cross

Fig. 11. Representative trends of normalized maximum attained column moment.

13
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

Fig. 12. Measured versus predicted peak flexural strength ratios.

sections (or sections that do not follow this correlation) are employed. with the equations. This is assumed to be the case for the variability due
The proposed equations are as follows: to the actual geometric properties of the steel column as well as the
Interior steel columns: gravitational load demand. The shear demand force may be estimated
    using the flexural demand derived herein (i.e., (Mmax + Ryfy, nZ )/L).
Mmax h Pg The axial force demand may be estimated according to the current pro-
¼ 1:89−0:016∙ −0:996 , R2 ¼ 0:93, COV ¼ 0:07 ð10Þ
f yZ tw Py cedure described in [1]. Similarly, the design procedure according to [1]
may be retained to establish the required embedment depth of the ECB
Exterior steel columns: connection.
 
Mmax h
¼ 1:60−0:009∙ , R2 ¼ 0:749, COV ¼ 0:05 ð11Þ 5. Conclusions
f yZ tw
This paper presents findings and implications for the seismic design
The range of applicability of Eqs. (10) and (11) is as follows, of non-dissipative (or brittle) embedded column base (ECBs) connec-
7 ≤ h/tw ≤ 46, and 0 ≤ Pg/Py ≤ 0.5. For larger web local slenderness ratios tions as part of capacity designed moment-resisting frames (MRFs).
(i.e., for the moderately ductile cross-sections), the value corresponding These are based on a rigorous evaluation of the current seismic design
to h/tw = 46 is likely to be conservative. Fig. 12 illustrates the perfor- models of ECB connections [1,4,22] with past experimental data along
mance of the equations relative to simulation data. Referring to with CFE simulations of wide flange steel columns embedded into rein-
Fig. 12, the accuracy of the proposed formulas for both exterior and in- forced concrete (RC) footings. The parametrized CFE model is validated
terior columns is nearly the same as indicated by the associated COV against full-scale experiments of steel columns and ECB connections un-
values of the ratio between the measured and predicted strength quan- dergoing highly inelastic deformations due to cyclic loading. The main
tities. It is evident that the flexural demands in non-dissipative ECB con- findings of the study are as follows:
nections may only be smaller than the corresponding plastic flexural
resistance of the steel column in cases that the gravity-induced axial • The AIJ strength design model [4] underestimates the peak flexural re-
load ratio is larger than 20% and the corresponding local web slender- sistance of ECB connections, MECB
max, by a factor of two. The AISC design
ness ratio is h/tw ≥ 35. Steel MRFs comprising stocky profiles model [1,22] yields more reasonable predictions, but only when the
(e.g., heavy W14) are likely to experience considerable inelastic de- base plate contribution to the overall lateral resistance of the ECB is
formations in their ECB connections with the current design procedure modest.
of non-dissipative ECB connections as described in the AISC design • The Grilli and Kanvinde model [24] is fairly accurate in terms of
manual [1]. predicting peak flexural strength MECBmax, of the assembled experimen-
Eqs. (10) and (11) may be used to establish lower bound design tal data, excluding cases in which foundation beams are present in
limits for ensuring elastic response of non-dissipative ECB connections. the perpendicular loading direction. The additional concrete confine-
This may be achieved by designing the ECB connections such that their ment produced to these beams is unaccounted for in this model.
reliable capacity exceeds the demands implied by Eqs. (10) and (11). • The yield ECB flexural resistance, MECB
y may be estimated as 65% of the
From a mechanistic standpoint, this reliable capacity may be deter- maximum attained peak flexural resistance, MECB max, of an ECB connec-
mined as, ϕ ∙ MECB y , in which, ϕ = 0.90, and, My
ECB
= 0.65MECB
max based tion. The relatively modest coefficient of variation (i.e., COV = 0.08)
on experimental observations by [24]. Particularly, the associated vari- suggests that the above value is fairly invariant across various config-
ability mostly comes from the yield stress of the employed steel mate- urations.
rial. In that respect, this evolves around the expected yield stress of • The detailed CFE simulations suggest that the AISC [1,22] design ap-
typical A992 Gr. 50 and A913 Gr. 65 steels (i.e., Ry = 1.1); thus, ϕ = proach, unintentionally does not prevent the inelastic behavior of
0.90. The associated variability of Eqs. (10) and (11) are fairly minor, ECB connections within the RC foundation, which is the non-
as depicted by the relatively small COV values that are reported together dissipative portion of the ECB connection. This is because flexural

14
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

demands commonly estimated in ECB connections do not incorporate [5] D.A. Grilli, R. Jones, A.M. Kanvinde, Seismic performance of embedded column base
connections subjected to axial and lateral loads, J. Struct. Eng. 143 (2017),
the effects of steel material cyclic hardening, cross-sectional slender- 04017010. https://doi.org/10.1061/(ASCE)ST.1943-541X.0001741.
ness, loading history as well as the imposed column axial load. [6] G.C. Clifton, M. Bruneau, G.A. MacRae, R. Leon, A. Fussell, Steel structures damage
• The peak ECB flexural demands normalized with respect to the plastic from the Christchurch earthquake series of 2010 and 2011, Bull. N. Z. Soc. Earthq.
Eng. 44 (2011) 297–318, https://doi.org/10.5459/bnzsee.44.4.297-318.
flexural resistance of the steel column, Mmax/fyZ, attain a value of 1.6 [7] G.A. MacRae, G.C. Clifton, M. Bruneau, A.M. Kanvinde, S. Gardiner, Lessons from steel
when relatively stocky profiles are employed (i.e., h/tw < 15) for structures in Christchurch earthquakes, 8th International Conference on Behavior of
gravity-induced axial load ratios of 0.20 or lower (i.e., typical in steel Steel Structures in Seismic Areas (STESSA), Shanghai, China 2015, pp. 1474–1481.
[8] Y. Suzuki, D.G. Lignos, Large scale collapse experiments of wide flange steel beam-
MRFs). The above ratio is capped at 1.2 when slender but still seismi-
columns, Proceedings of the 8th International Conference on Behavior of Steel
cally compact profiles are employed in the seismic design process of Structures in Seismic Areas (STESSA), Shanghai, China, 2015.
the ECB connection. Although these observations are detrimental in [9] A. Elkady, D.G. Lignos, Full-scale testing of deep wide-flange steel columns under
the context of ECB connections, where the column strength imposes multiaxis cyclic loading: loading sequence, boundary effects, and lateral stability
bracing force demands, J. Struct. Eng. 144 (2018), 04017189. https://doi.org/10.
demands on the connection, they may be beneficial in other contexts, 1061/(ASCE)ST.1943-541X.0001937.
e.g., in simulating the proclivity of the frame to sidesway collapse, [10] G. Ozkula, J. Harris, C.-M. Uang, Observations from cyclic tests on deep, wide-flange
wherein an increased column strength may delay the formation of beam-columns, Eng. J. 1 (2017) 45–59.
[11] A. Elkady, D.G. Lignos, Improved seismic design and nonlinear modeling recommen-
soft stories. This is subject to further investigation. dations for wide-flange steel columns, J. Struct. Eng. 144 (2018), 04018162. https://
• The CFE results suggest that the Mmax/fyZ ratio is practically insensitive doi.org/10.1061/(ASCE)ST.1943-541X.0002166.
to the choice of the material steel grade for the two types of materials [12] J. Cravero, A. Elkady, D.G. Lignos, Experimental evaluation and numerical modeling
of wide-flange steel columns subjected to constant and variable axial load coupled
that were investigated herein (i.e., A992 Gr. 50 and A913 Gr. 65).
with lateral drift demands, J. Struct. Eng. 146 (2020), 04019222. https://doi.org/
• The seismic demands in ECB connections appear to be sensitive to the 10.1061/(ASCE)ST.1943-541X.0002499.
loading history only when stocky profiles (i.e., h/tw < 15) are used. [13] H. Akiyama, M. Kurosawa, N. Wakuni, I. Nishimura, Strength and deformation of col-
umn bases embedded in base concrete, J. Struct. Construct. Eng. (Transactions of AIJ)
(1984) 45–53(In Japanese).
[14] K. Minami, Y. Nishimura, K. Kurisawa, Research on Stress Transfer Mechanism of
Considering these various effects, simple design equations are pro- Embedded Column Base Connection, Summary of Technical Papers of Annual Meet-
posed to achieve a seismic performance that is more consistent with ex- ing, Kinki Branch, 1982 293–296.
pectations. The proposed work has several limitations as well, chiefly [15] K. Morita, B. Kato, A. Tanaka, N. Fujita, Experiments investigation on maximum
strength of embedded column bases, J. Struct. Construct. Eng. (Transactions of AIJ)
that experimental data for ECB-column interaction is sparse, necessitat-
(1985) 65–74(In Japanese).
ing the reliance on simulation. Moreover, a rigorous reliability analysis [16] S. Nakashima, S. Igarashi, Behavior of steel square tubular column bases for interior
has not been conducted for the capacity equations, such that the resis- columns embedded in concrete footings under bending moment and shearing force
tance (ϕ-) factors adopted from literature may not be accurate. Further- part 1: test program and load-displacement relationships, J. Struct. Construct. Eng.
(Transactions of AIJ) (1986) 106–118(In Japanese).
more, the ECB strength models (e.g., [24]) are limited by the test details, [17] S. Nakashima, S. Igarashi, Behavior of steel square tubular column bases for interior
and extrapolation to significantly different configurations (e.g., with ad- columns embedded in concrete footings under bending moment and shearing
ditional footing reinforcement) must be conducted with caution. force: part 2 initial stiffness, ultimate strength and mechanism of stress flow, J.
Struct. Construct. Eng. (Transactions of AIJ) (1987) 63–76(In Japanese).
Finally, the study assumes that the ECB connections are entirely non- [18] T. Takeda, Y. Takahashi, Experimental Study on the Column Base of Steel Structure
dissipative and shall be capacity-designed. While this is consistent and Steel Reinforced Concrete Structure Part 1: Comparison of 4 Types of Column
with the intent of current design approaches [22], experimental data Bases, Summary of Technical Papers of Annual Meeting, Kanto Branch (In
Japanese), 1980 265–268.
suggests that ECB connections offer some deformation capacity. Further
[19] T. Takeda, Y. Takahashi, Experimental Study on the Column Base of Steel Structure
studies, as well as detailing guidelines for ECB connections may provide and Steel Reinforced Concrete Structure Part 2: Examination of the Interior Embed-
methods to leverage this deformation capacity for a more economic de- ded Column Bases, Summary of Technical Papers of Annual Meeting, Kanto Branch
sign. Notwithstanding these issues (which may be addressed in the fu- (In Japanese), 1982 229–232.
[20] S. Tsujioka, K. Inoue, K. Imai, M. Hirayama, Strength of improved embedded RHS ex-
ture), the proposed approach is likely to yield significantly improved terior column-footing connections with U-reinforcing bars, J. Struct. Construct. Eng.
designs in terms of earthquake safety of steel MRFs. (Transactions of AIJ) (In Japanese) 401 (1989) 117–127.
[21] K. Washio, T. Suzuki, S. Nakashima, I. Nishimura, Effect of embedment on the steel
column base I part 1: Experimental results, Summary of Technical Papers of Annual
Meeting, AIJ, C-1(II) (In Japanese), 53, 1978, pp. 1289–1290.
Declaration of Competing Interest [22] American Institute of Steel Construction (AISC), Seismic Provisions for Structural
Steel Buildings. ANSI/AISC 341–16, American Institute of Steel Construction, 2016.
[23] American Concrete Institute (ACI), Building Code Requirements for Structural Con-
The authors declare that they have no known competing financial crete and Commentary. ACI 318–14, American Concrete Institute, 2014.
interests or personal relationships that could have appeared to influ- [24] D.A. Grilli, A.M. Kanvinde, Embedded column base connections subjected to seismic
loads: strength model, J. Constr. Steel Res. 129 (2017) 240–249. http://www.
ence the work reported in this paper. sciencedirect.com/science/article/pii/S0143974X1630284X.
[25] G.G. Deierlein, T.M. Sheikh, J.A. Yura, J.O. Jirsa, Beam-column moment connections
Acknowledgements for composite frames: part 2, J. Struct. Eng. 115 (1989) 2877–2896, https://doi.
org/10.1061/(ASCE)0733-9445(1989)115:11(2877).
[26] T.M. Sheikh, G.G. Deierlein, J.A. Yura, J.O. Jirsa, Beam-column moment connections
This study is based on work supported by the Swiss National Science for composite frames: part 1, J. Struct. Eng. 115 (1989) 2858–2876, https://doi.
Foundation (Award No. 200021_169248). The financial support is org/10.1061/(ASCE)0733-9445(1989)115:11(2858.
gratefully acknowledged. Any opinions, findings, and conclusions or [27] K. Washio, T. Suzuki, S. Nakashima, I. Nishimura, Effect of embedment on the steel
column base I part 2: Observations, Summary of Technical Papers of Annual Meet-
recommendations expressed in this paper are those of the authors and ing, AIJ, C-1(II) (In Japanese), 53, 1978, pp. 1291–1292.
do not necessarily reflect the views of sponsors. [28] N. Barnwell, Experimental Testing of Shallow Embedded Connections between Steel
Columns and Concrete Footings, All Theses and Dissertations http://scholarsarchive.
byu.edu/etd/4428 2015.
References [29] B. Gong, B.M. Shahrooz, Steel-concrete composite coupling beams - behaviors and
design, Eng. Struct. 23 (2001) 1480–1490, https://doi.org/10.1016/S0141-0296
[1] American Institute of Steel Construction (AISC), Seismic Design Manual, Second ed., (01)00042-6.
2012. [30] ABAQUS, ABAQUS Analysis User’s Manual Version 6.14–1, Dassault Systems Simulia
[2] K.A. Harries, D. Mitchell, W.D. Cook, R.G. Redwood, Seismic response of steel beams Corp, RI, USA, 2014.
coupling concrete walls, J. Struct. Eng. 119 (1993) 3611–3629, https://doi.org/10. [31] H. Inamasu, A.M. Kanvinde, D.G. Lignos, Seismic stability of wide-flange steel col-
1061/(ASCE)0733-9445(1993)119:12(3611. umns interacting with embedded column base connections, J. Struct. Eng. 145
[3] A.H. Mattock, G.H. Gaafar, Strength of embedded steel sections as brackets, ACI J. 79 (2019), 04019151. https://doi.org/10.1061/(ASCE)ST.1943-541X.0002410.
(1982) 83–93. [32] A. Elkady, D.G. Lignos, Analytical investigation of the cyclic behavior and plastic
[4] Architectural Institute of Japan (AIJ), Recommendations for Design of Connections in hinge formation in deep wide-flange steel beam-columns, Bull. Earthq. Eng. 13
Steel Structures, third ed. Architectural Institute of Japan, 2012. (2015) 1097–1118, https://doi.org/10.1007/s10518-014-9640-y.

15
H. Inamasu, A.M. Kanvinde and D.G. Lignos Journal of Constructional Steel Research 177 (2021) 106417

[33] J. Lemaitre, J.-L. Chaboche, Mechanics of Solid Materials, Cambridge university [44] H. Krawinkler, A. Gupta, R. Medina, N. Luco, Development of Loading Histories for
press, 1994. Testing of Steel Beam-to-Column Assemblies, Stanford University, 2000.
[34] E. Voce, The relationship between stress and strain for homogeneous deformation, J. [45] A. Elkady, D.G. Lignos, Modeling of the composite action in fully restrained beam-to-
Inst. Met. 74 (1948) 537–562. column connections: implications in the seismic design and collapse capacity of
[35] A. de Castro e Sousa, Y. Suzuki, D.G. Lignos, Consistency in solving the inverse prob- steel special moment frames, Earthq. Eng. Struct. Dyn. 43 (2014) 1935–1954,
lem of the Voce-Chaboche constitutive model for plastic straining, J. Eng. Mech. 146 https://doi.org/10.1002/eqe.2430.
(9) (2020) 04020097, https://doi.org/10.1061/(ASCE)EM.1943-7889.0001839. [46] National Institute of Standards and Technology (NIST), Evaluation of the FEMA P695
[36] A. de Castro e Sousa, D.G. Lignos, On the inverse problem of classic nonlinear plasticity Methodology for Quantification of Building Seismic Performance Factors, NEHRP
models-An application to cyclically loaded structural steels, Resilient Steel Structures consultations Joint Venture, 2010.
Laboratory (RESSLab), Ecole Polytechnique Fédérale de Lausanne, Lausanne, 2017. [47] J. Newell, C.-M. Uang, Cyclic behavior of steel wide-flange columns subjected to
[37] B.W. Young, Residual stresses in hot-rolled sections, International Colloquium on large drift, J. Struct. Eng. 134 (2008) 1334–1342, https://doi.org/10.1061/(ASCE)
Column Strength, International Association for Bridge and Structural Engineering, 0733-9445(2008)134:8(1334).
Zurich, Switzerland 1971, pp. 25–38.
[48] P. Clark, K. Frank, H. Krawinkler, R. Shaw, Protocol for Fabrication, Inspection, Test-
[38] A. de Castro e Sousa, D.G. Lignos, Residual Stress Measurements of European Hot-
ing and Documentation of Beam-Column Connection Tests and Other Experimental
Rolled I-Shaped Steel Profiles, Resilient Steel Structures Laboratory, École
Specimens, 1997.
Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland, 2017.
[49] Y. Suzuki, D.G. Lignos, Development of collapse-consistent loading protocols for ex-
[39] P. Torres-Rodas, F. Zareian, A. Kanvinde, A hysteretic model for the rotational re-
perimental testing of steel columns, Earthq. Eng. Struct. Dyn. 49 (2020) 114–131,
sponse of embedded column base connections, Soil Dyn. Earthq. Eng. 115 (2018)
https://doi.org/10.1002/eqe.3225.
55–65, https://doi.org/10.1016/j.soildyn.2018.08.015.
[40] L.F. Ibarra, R.A. Medina, H. Krawinkler, Hysteretic models that incorporate strength [50] D.G. Lignos, C. Putman, H. Krawinkler, Application of simplified analysis procedures
and stiffness deterioration, Earthq. Eng. Struct. Dyn. 34 (2005) 1489–1511, https:// for performance-based earthquake evaluation of steel special moment frames, Earth-
doi.org/10.1002/eqe.495. quake Spectra 31 (2015) 1949–1968, https://doi.org/10.1193/081413EQS230M.
[41] M. Rahnama, H. Krawinkler, Effects of Soft Soil and Hysteresis Model on Seismic De- [51] D.G. Lignos, A.R. Hartloper, A. Elkady, G.G. Deierlein, R. Hamburger, Proposed up-
mands, John A. Blume Earthquake Engineering Center Stanford, 1993. dates to the ASCE 41 nonlinear modeling parameters for wide-flange steel columns
[42] D.G. Lignos, H. Krawinkler, Deterioration modeling of steel components in support of in support of performance-based seismic engineering, J. Struct. Eng. 145 (2019),
collapse prediction of steel moment frames under earthquake loading, J. Struct. Eng. 04019083. https://doi.org/10.1061/(ASCE)ST.1943-541X.0002353.
137 (2011) 1291–1302, https://doi.org/10.1061/(ASCE)ST.1943-541X.0000376. [52] D.G. Lignos, A.R. Hartloper, Steel column stability and implications in the seismic as-
[43] American Society of Civil Engineers (ASCE), Minimum Design Loads and Associated sessment of steel structures according to Eurocode 8 part 3, Stahlbau 89 (2020)
Criteria for Buildings and other Structures. ASCE/SEI 7–16, American Society of Civil 16–27, https://doi.org/10.1002/stab.201900108.
Engineers, 2016.

16