See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/255483341

Development and analytical verification of an inelastic reinforced concrete

joint model

Article  in  Engineering Structures · July 2013

DOI: 10.1016/j.engstruct.2013.02.032

CITATIONS READS

15 186

2 authors:

Mehmet Unal Burcu Burak

CoreLogic Middle East Technical University
15 PUBLICATIONS   49 CITATIONS    9 PUBLICATIONS   74 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Virtual Earthquake Reconnaissance Team (VERT): Earthquake Engineering Research Institute (EERI) View project

All content following this page was uploaded by Burcu Burak on 06 October 2017.

The user has requested enhancement of the downloaded file.

 Engineering Structures 52 (2013) 284–294

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Development and analytical verification of an inelastic reinforced

concrete joint model
Mehmet Unal a, Burcu Burak b,⇑
a
Civil Engineering Department, Pennsylvania State University, University Park, PA, United States
b
Civil Engineering Department, Middle East Technical University, K2-311 Ankara, Turkey

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

Article history: Previous experimental research indicated that beam-to-column connections of reinforced concrete (RC)
Received 21 February 2012 moment resisting frame structures experience considerable deformations under earthquake loading and
Revised 20 February 2013 these deformations have a major contribution to the story drift of the building. In current analysis and
Accepted 23 February 2013
design applications, however, the connection regions are generally modeled as rigid zones and the inelas-
tic behavior of the joint is not considered. This assumption gives rise to an underestimation of the story
drifts and hence to an improper assessment of the seismic performance of the structure. In order to
Keywords:
implement the effect of shear distortions observed in these regions into the seismic design and analysis
Beam-to-column connections
Seismic loading
of buildings, a model that represents their inelastic behavior needs to be developed. In this study, a para-
Joint model metric model that predicts the joint shear strength versus strain relationship is developed by investigat-
Joint shear ing several previous experimental studies on RC beam-to-column connection subassemblies subjected to
Joint distortion cyclic loading and establishing an extensive database. Based on this experimental database, parameters
that significantly influence the joint behavior are determined by employing statistical correlation method
and these key parameters are assembled to form a joint model. This model is then verified by comparing
the results obtained from the dynamic earthquake analysis carried out by using Perform 3D with the
experimental ones. Based on the verification results, it is observed that the analysis including the devel-
oped joint model improves the prediction of the seismic behavior.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction to cyclic loading significantly affects the overall behavior of the

For more than 40 years, many experimental studies have been
carried out focusing on the behavior of beam-to-column connec-
tions under cyclic loading. Several of these studies indicated that
joint deformations have a significant contribution on the story drift
[1–6]. Therefore, the seismic behavior of beam-to-column connec-
tions significantly influences the earthquake response of reinforced
concrete frame structures. If these regions are not properly de-
signed and detailed, they may fail prematurely under strong earth-
quake ground motions.
In the design of reinforced concrete moment resisting frame
structures, the ‘‘strong column–weak beam’’ philosophy is rec-
ommended to ensure the formation of beam plastic hinging
rather than column hinging at large displacement levels. Beam-
to-column connections, therefore, are expected to be in the elas-
tic range. However, experimental studies indicated that they un-
dergo large inelastic shear deformations even when the strong
column–weak beam design philosophy is followed. It was also
observed that the inelastic behavior of connections subjected
⇑ Corresponding author. Tel.: +90 312 210 5477.
E-mail addresses: mehmet@psu.edu (M. Unal), burcubc@metu.edu.tr (B. Burak).
specimen and should not be neglected. However, in current anal-
ysis and design applications, the connection regions are gener-
ally modeled as rigid zones or regions with reduced stiffness
based on the stiffness of adjoining members and the inelastic
behavior of this region is not accurately taken into account. This
assumption leads to underestimation of the story drifts and
overestimation of beam end rotations and hence to improper
assessment of the seismic performance of the whole structure
and structural members. Therefore, a model needs to be devel-
oped to properly represent the inelastic seismic behavior of
beam-to-column connections.
In order to define the inelastic behavior of these regions, the
factors influencing the seismic behavior of beam-to-column con-
nections should be determined. Many researchers agree that the
key parameters affecting the connection performance are col-
umn-to-beam moment strength ratio, confinement provided by
the lateral reinforcement and beams framing into the connection
region, anchorage of the beam longitudinal reinforcement and
shear stress level in the joint. In addition, material properties, sec-
tion dimensions, eccentricity between the centerlines of beam and
column, axial load acting on the column and presence of wide
beams or slab also affect the connection behavior.

0141-0296/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2013.02.032
 M. Unal, B. Burak / Engineering Structures 52 (2013) 284–294
Nomenclature
Ao cross-sectional area of the transverse reinforcement
Ag gross area of column section
bb beam width
bc column width
bc,core width of the column core
bj effective joint width
d00 depth of the beam between top and bottom reinforce-
ment
e eccentricity between beam and column centerlines
fc0 concrete compressive strength
fy yield strength of reinforcing bars
G shear modulus
H height of the specimen
hb beam depth
hc column depth
hc,core depth of the column core
l span length of the beam
leff effective length of the lateral reinforcement in the load-
ing direction, which is taken as the summation of the
lengths of stirrup legs placed parallel to the loading
direction
Based on experimental results, numerous analytical studies
have been conducted to investigate the effect of different param-
eters on the seismic behavior of the connection region and pre-
dict the load–deformation relationships. Consequently, analytical
models representing the joint shear stress versus strain response
have been proposed. Parra-Montesinos and Wight [7] proposed
an analytical model for estimating shear strength versus strain re-
sponse of reinforced concrete beam-to-column connections.
Experimental results of cyclic connection tests were utilized to
develop a factor for defining principle tensile and compressive
stress versus shear distortion response. The development of this
model was based on the state of plane strains in the joint. This
analytical model also included a coefficient that takes into ac-
count the effect of eccentricity.
Lowes and Altoontash [8] developed a joint model that provides
a representation of the nonlinear mechanism of the joint behavior
by developing constitutive relationships of material, geometric and
design parameters and implementing a four-node 12 degree-of-
freedom element. Mitra and Lowes [9] improved this model by
changing the element definition. A numerical simulation of the
parameters in conjunction with finite element analysis to predict
the joint shear strength was utilized in this study. A compres-
sion-strut model was used to simulate the joint core and bond–slip
response was also considered in addition to nonlinear joint core re-
sponse to simulate the joint stiffness. An extensive experimental
database of beam-to-column subassemblies was used in order to
evaluate the accuracy of the model.
Shin and LaFave [10], Kim and LaFave [11] investigated the ef-
fects of some key parameters such as concrete compressive
strength, joint reinforcement and axial load effect using the data
from 26 beam-to-column connection tests. Afterwards, an analyti-
cal model was proposed to estimate the hysteretic joint shear
stress versus strain behavior by employing modified compression
field theory. Burak [1] also used experimental results and devel-
oped a joint model that takes into account the joint shear strain
and predicts the joint shear behavior. The model presents an opti-
mized equation including concrete compressive strength, geome-
try and eccentricity parameters. A five story building was
analyzed by using both rigid connections and the developed joint
285
Lp plastic hinge length
Mn nominal moment strength
N column axial load
n number of layers of transverse reinforcement in the
effective confined area
s center-to-center spacing of stirrups
Vj joint shear force
vj joint shear stress
c joint shear distortion
qcore volumetric confinement reinforcement ratio deter-
mined considering the core volume of the connection
region
qgross volumetric confinement reinforcement ratio deter-
mined considering the gross volume of the connection
region
qonelayer volumetric confinement reinforcement ratio deter-
mined considering the effective volume that contains
one layer of joint transverse reinforcement
h beam rotation, rad
Dj the contribution of joint shear distortion to the total
story drift
model. It was concluded that if the joint model was not included
in the analysis the roof drifts could be underestimated up to 25%.
Canbolat [2] developed a parametric joint model that takes into ac-
count the material properties, geometric properties and confine-
ment provided by the joint hoop reinforcement. The model was
used in the dynamic time history analysis of a five-story RC build-
ing. The results showed that the roof drifts increased up to 25%
when the analytical joint model was utilized.
More recently, Kim and LaFave [12] used statistical methods to
evaluate the effect of key parameters such as concrete compressive
strength, panel geometry, confinement due to joint reinforcement,
column axial compression, and bond characteristics of the longitu-
dinal reinforcement on the joint behavior. It was concluded that
joint shear capacity mostly depends on concrete compressive
strength, however, joint panel geometry has only a slight effect
on the seismic performance. After having determined the most
influential parameters on joint shear stress–strain behavior, an
equation representing joint shear strength was developed by using
Bayesian parameter estimation approach.
In this paper, a database of specimens is presented and the
outputs of experimental and analytical studies on beam-to-col-
umn connections are synthesized to develop a joint model. The
shear strength prediction equation developed by the authors,
the details of which are explained in a prior article [13], is used
as the ultimate shear strength of beam-to-column connections
in this model. The shear distortion value corresponding to the
ultimate strength and two other key points are defined and a
backbone curve for the joint model is developed, then the subas-
semblies in the database are analyzed utilizing the developed
model. When the joint model is included in the analysis, not only
the prediction of the lateral load versus story drift response of the
structure, but also the inelastic behavior of members such as the
connection region and the beams are improved significantly. The
main objective of this study is to develop a joint model that can
be incorporated into a commercial software program that can be
of practical use to the engineering practice. Therefore, the enve-
lope curves are obtained and conservative values for the lateral
load, story drift, beam end rotation, joint shear stress and distor-
tion are estimated.
286 M. Unal, B. Burak / Engineering Structures 52 (2013) 284–294

Table 1
Database of beam-to-column connection specimens.

Research team Specimen Joint fc0 fy bc hc bb hb N=ðAg fc0 Þ e qonelayer (%) bj, ACI 352
type, JT (MPa) (MPa) (mm) (mm) (mm) (mm) (mm) Eq. (1) (mm)

Burak and Wight [1] 1-S D 1.25 29.0 441 356 356 203 381 0.053 76 1.251 256
2-S D 1.25 39.0 441 534 356 254 457 0.039 140 0.815 307
3-S D 1.25 29.0 441 534 356 254 457 0.042 140 0.815 307
2-N A 1 39.0 441 356 534 305 457 0.031 0 1.314 439
3-N A 1 29.0 441 356 534 762 305 0.031 0 1.314 458
Chen and Chen [16] JC A 1 19.9 399 500 500 300 500 0 0 1.141 425
JE A 1 19.9 399 500 500 300 500 0 100 1.141 375
Durrani and Wight [17] X1 C 1.25 34.3 352 362 362 279 419 0.054 0 0.847 321
X2 C 1.25 33.6 352 362 362 279 419 0.056 0 1.270 321
X3 C 1.25 31.0 352 362 362 279 419 0.053 0 0.847 321
Ehsani and Alameddine LL8 A 1 55.1 446 356 356 311 508 0.042 0 1.217 333
[18] LH8 A 1 55.1 446 356 356 311 508 0.042 0 2.130 333
HL8 A 1 55.1 446 356 356 311 508 0.073 0 1.278 333
HH8 A 1 55.1 446 356 356 311 508 0.073 0 2.130 333
LL11 A 1 75.8 446 356 356 311 508 0.030 0 1.278 333
LH11 A 1 75.8 446 356 356 311 508 0.029 0 2.130 333
HL11 A 1 75.8 446 356 356 311 508 0.061 0 1.278 333
HH11 A 1 75.8 446 356 356 311 508 0.063 0 2.130 333
LL14 A 1 96.5 446 356 356 311 508 0.019 0 1.278 333
LH14 A 1 96.5 446 356 356 311 508 0.018 0 2.130 333
HH14 A 1 96.5 446 356 356 311 508 0.040 0 2.130 333
Ehsani and Wight [19] 1B A 1 33.6 437 300 300 259 480 0.059 0 1.320 279
2B A 1 34.9 437 300 300 259 439 0.071 0 1.489 279
3B A 1 40.9 437 300 300 259 480 0.060 0 1.759 279
4B A 1 44.6 437 300 300 259 439 0.055 0 1.935 279
5B A 1 24.3 437 340 340 300 480 0.126 0 1.167 320
6B A 1 39.8 437 340 340 300 480 0.066 0 1.090 320
Fujii and Morita [20] A1 C 1.25 40.2 297 220 220 160 250 0.076 0 0.592 190
A2 C 1.25 40.2 297 220 220 160 250 0.076 0 0.592 190
A3 C 1.25 40.2 297 220 220 160 250 0.227 0 0.592 190
A4 C 1.25 40.2 297 220 220 160 250 0.227 0 1.690 190
B1 A 1 30.0 297 220 220 160 250 0.068 0 0.592 190
B2 A 1 30.0 297 220 220 160 250 0.068 0 0.592 190
B3 A 1 30.0 297 220 220 160 250 0.236 0 0.592 190
B4 A 1 30.0 297 220 220 160 250 0.236 0 1.690 190
Gentry and Wight [21] 1 B 1.25 27.6 441 356 356 864 305 0.026 0 0.676 483
2 B 1.25 27.6 441 356 356 762 305 0.026 0 0.676 457
3 B 1.25 27.6 441 356 356 864 305 0.026 0 0.676 483
4 B 1.25 27.6 441 356 356 864 305 0.026 0 0.676 483
Guimaraes et al. [22] J2 E 1.67 27.6 549 508 508 406 508 0 0 0.841 457
J4 E 1.67 31.6 549 508 508 406 508 0 0 0.841 457
J5 E 1.67 77.9 511 508 508 406 508 0 0 2.484 457
J6 E 1.67 92.1 511 508 508 406 508 0 0 2.484 457
Kaku and Asakusa [23] Specimen 1 A 1 31.1 250 220 220 160 220 0.171 0 0.503 190
Specimen 2 A 1 41.7 250 220 220 160 220 0.099 0 0.503 190
Specimen 3 A 1 41.7 250 220 220 160 220 0 0 0.503 190
Specimen 4 A 1 44.7 281 220 220 160 220 0.166 0 0.131 190
Specimen 5 A 1 36.7 281 220 220 160 220 0.090 0 0.131 190
Specimen 6 A 1 40.4 281 220 220 160 220 0 0 0.131 190
Specimen 7 A 1 32.2 250 220 220 160 220 0.124 0 0.503 190
Specimen 8 A 1 41.2 250 220 220 160 220 0.080 0 0.503 190
Specimen 9 A 1 40.6 250 220 220 160 220 0 0 0.503 190
Specimen 10 A 1 44.4 281 220 220 160 220 0.168 0 0.131 190
Specimen 11 A 1 41.9 281 220 220 160 220 0.079 0 0.131 190
Specimen 12 A 1 35.1 281 220 220 160 220 0 0 0.131 190
Specimen 13 A 1 46.4 250 220 220 160 220 0.045 0 0.503 190
Specimen 14 A 1 41.0 281 220 220 160 220 0.081 0 0.129 190
Specimen 15 A 1 39.7 281 220 220 160 220 0.083 0 0.129 190
Specimen 16 A 1 37.4 250 220 220 160 220 0 0 0.496 190
Specimen 17 A 1 39.7 250 220 220 160 220 0 0 0.503 190
Specimen 18 A 1 40.7 250 220 220 160 220 0 0 0.498 190
Kitayama et al. [24] A1 C 1.25 30.6 326 300 300 200 300 0.064 0 0.708 250
A2 E 1.67 30.6 326 300 300 200 300 0.064 0 0.708 250
A3 E 1.67 30.6 326 300 300 200 300 0.064 0 0.708 250
A4 C 1.25 30.6 326 300 300 200 300 0.064 0 0.708 250
LaFave and Wight [25] EWB 1 B 1.25 28.9 482 356 356 864 305 0 0 0.772 483
EWB 2 B 1.25 30.3 482 356 356 864 305 0 0 0.772 483
EWB 3 B 1.25 34.5 482 305 508 940 305 0 0 1.081 464
ENB 1 B 1.25 24.8 482 305 508 305 559 0 0 1.081 432
 M. Unal, B. Burak / Engineering Structures 52 (2013) 284–294 287

Table 1 (continued)

Research team Specimen Joint fc0 fy bc hc bb hb N=ðAg fc0 Þ e qonelayer (%) bj, ACI 352
type, JT (MPa) (MPa) (mm) (mm) (mm) (mm) (mm) Eq. (1) (mm)

Lee and Ko [26] S0 A 1 32.6 471 400 600 300 450 0.089 0 0.423 350
S50 A 1 34.2 471 400 600 300 450 0.085 50 0.423 350
W0 A 1 28.9 471 600 400 300 450 0.101 0 1.134 450
W75 A 1 30.4 471 600 400 300 450 0.096 75 1.134 450
W150 A 1 29.1 471 600 400 300 450 0.100 150 1.134 450
Shiohara [27] J-7 C 1.25 79.2 857 300 300 240 300 0.117 0 0.689 270
J-10 C 1.25 39.2 598 300 300 240 300 0.236 0 0.689 270
Quintero-Febres and IWB1 E 1.67 27.6 503 356 356 889 305 0 0 1.097 489
Wight [28] IWB2 E 1.67 27.6 503 356 356 660 305 0 0 1.097 432
IWB3 E 1.67 27.6 503 330 508 838 305 0 0 0.908 457
Raffaelle and Wight [29] 1 C 1.25 28.6 441 356 356 254 381 0.025 51 0.772 307
2 C 1.25 26.8 441 356 356 178 381 0.026 89 0.772 231
3 C 1.25 37.7 441 356 356 191 381 0.019 83 0.772 244
4 C 1.25 19.3 441 356 356 191 559 0.036 83 0.772 244
Shin and LaFave [30,31] SL 1 D 1.25 29.9 468 457 330 279 406 0 89 0.615 329
SL 2 D 1.25 36.1 468 457 330 178 406 0 140 0.615 227
SL 3 D 1.25 47.4 551 457 330 279 406 0 0 0.615 362
SL 4 D 1.25 31.1 579 279 368 279 406 0 0 1.099 279
Teng and Zhou [32] S1 C 1.25 33.0 440 400 300 200 400 0.111 0 0.885 275
S2 C 1.25 34.0 440 400 300 200 400 0.108 50 0.885 275
S3 C 1.25 35.0 440 400 300 200 400 0.105 100 0.885 245
S5 C 1.25 39.0 440 400 200 200 400 0.110 50 1.287 250
S6 C 1.25 38.0 440 400 200 200 400 0.113 100 1.287 230

ters to accurately represent their effect on the capacity and obtain

the minimum average error and the highest correlation with the
experimental values. While carrying out these steps, the guidelines
given in ACI Committee 318 Building Code Requirements for Struc-
tural Concrete [14], and ACI-ASCE Committee 352 Recommenda-
tions for Design of Beam–Column Connections in Monolithic
Reinforced Concrete Structures [15] are followed. After developing
equations for the maximum joint shear strength and the corre-
sponding strain value (vj,u, cu), two more critical points in the joint
shear strength versus strain curve are determined by using statis-
tical data. These points are selected as the onset of cracking (vj,cr, -
ccr) and the end of cracked-elastic stage (vj,i, ci) (Fig. 1).
Fig. 1. Critical points of the proposed model for the joint shear stress–strain The critical points are defined as parametric equations aiming
diagram of specimen SL2 – Shin and LaFave [31]. to give the best correlation with the experimental values. The
resulting equations to define the strength and strain values of
the critical points are given below, which resulted in an average er-
2. Analytical joint model definition ror of 4.2% and a correlation of 88% with the experimental data:

In order to predict the joint shear strength versus strain behav-

ior, past experimental data on the cyclic testing of beam-to-col- v j;u ðMPaÞ ¼ JT  ðfc  fy Þ1=6  qjoint  EE  CI  NE  WB  SI
umn connection subassemblies, which provide both joint shear
strength and strain measurements are utilized. The parameters
v j;cr ðMPaÞ ¼ 0:4  v j;u
which are believed to be influential on the seismic behavior of v j;i ðMPaÞ ¼ 0:9  v j;u
joints are collected in a database given in Table 1. By using statis- v j;u 0:7 1 hc ð1Þ
ci ¼ ð Þ  
tical correlation methods, the most effective parameters are deter- G JT bj
mined whereas the ones that have a slight effect on the shear ccr ¼ 0:15  ci
behavior are neglected (Table 2). Consequently, an equation to pre-
cu ¼ 2:5  ci
dict the maximum joint shear strength of reinforced concrete
beam-to-column connections subjected to earthquake loading is The parameters used in this analytical study are explained in an
generated [13]. Final parameters in the equation are defined in earlier article by the authors [13], but are also briefly explained
terms of ratios and powers of some of the key individual parame- here.

Table 2
Correlation of key parameters with maximum experimental joint shear strength.

V j;max fc0 fy qcore qgross qonelayer bc hc bb hb N e

Exterior joints 0.6693 0.0790 0.4512 0.3101 0.5356 0.1393 0.3189 0.2319 0.2963 0.3292 0.2471
Interior joints 0.6929 0.0850 0.3608 0.3546 0.5604 0.0996 0.1996 0.1032 0.0353 0.1564 0.3349
P
ðx xÞðyy Þ
where, CorrelationðX; YÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P 2
P 2
.
ðx
xÞ  Þ
ðyy
288 M. Unal, B. Burak / Engineering Structures 52 (2013) 284–294

JT is a parameter that takes into account the effect of the con-
finement provided by the surrounding beams as defined in Fig. 2.
This parameter is based on the ACI-ASCE Committee 352 [15] joint
classification for Type 2 connections and the arrangement of beams
in the test setup.
The joint shear p
resistance
ffiffiffiffi is based on (fcfy)1/6, which gives sim-
ilar results to the fc approach of ACI-ASCE Committee 352 Rec-
0
ommendations [15] for regular strength concrete and joint
transverse reinforcement, but more improved results for connec-
tion regions with lower yield strength transverse reinforcement.
G is the elastic shear modulus defined by the following
equation:
E
G¼
2ð1 þ mÞ
qjoint is a parameter that depends on the volumetric joint transverse
reinforcement ratio for one layer of confinement reinforcement:
account the effective beam width defined in ACI Committee 318
Building Code Requirements [14] and the reinforcement placed
in the flange are considered. The nominal moment strength for
the flanged section is calculated and divided to that of the rectan-
gular beam section having same depth and web width:
M n ðFlanged SectionÞ
Mn ðRectangular SectionÞ
; when slab is present
Slab Index ðSIÞ ¼
1; when slab is not present
ð8Þ
3. Analytical verification of the joint model
ð2Þ
After the parametric joint model that predicts the experimental
behavior is generated and calibrated, the specimens are analyzed
using Perform 3D v4.0 [33] which is a commercially available non-
linear dynamic analysis program. Beams, columns and connection
regions are modeled individually and cyclic loading that was ap-

b
1  hbb  b j ; when wide beams are present in the loading direction
Wide Beam Effect ðWBÞ ¼ b b ð7Þ
1; when wide beams are not present in the loading direction

qjoint ð%Þ ¼ 1:0 if qonelayer < 1:0 Ao  leff plied in the experiment is imposed to the specimen models by cre-
and qonelayer ¼
qjoint ð%Þ ¼ ðqonelayer Þ0:5 if qonelayer P 1:0 hc;core  bc;cores ating a ground acceleration record. Finally, the analytical results
ð3Þ and the experimental results are compared for the verification of
the joint model.
EE shows the reduction in joint strength due to the presence of The model for interior connection regions consists of two col-
eccentricity: umns, two beams and a joint panel zone element; whereas for
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the exterior connections, two columns, one beam and a joint panel
1 zone element are used (Fig. 3). Majority of the selected experi-
Eccentricity Effect ðEEÞ ¼ ð4Þ
1 þ e=bc ments had a test set-up similar to the one shown in Fig. 3a, where
the lateral load was applied to the top of the column. However, for
CI is the column index based on column aspect ratio that is used
some experiments on exterior connections, the beams were
to account for the reduction of effective joint area in rectangular
aligned vertically as in Fig. 3b and the load was applied at the beam
columns:
end. The details of each individual subassembly model are given
qffiffiffiffi
bc
when bc
< 1:0 below:
hc hc
Column Index ðCIÞ ¼ ð5Þ
bc
1:0 when hc
P 1:0
3.1. Beam element
NE defines the confinement effect due to axial load:
N The beam element is defined as a frame compound component
Axial Load Effect ðNEÞ ¼ 1 þ ð6Þ including three basic components which are rigid end zone at the
Ag  fc
beam end, semi-rigid moment connection hinge next to the rigid
WB gives the reduction in strength when wide beams are pres- end zone and a uniform elastic cross-section. For the elastic sec-
ent in the loading direction: tion, cracked stiffness, that is taken as 35% of the gross stiffness
SI defines the confinement of the connection region due to the [14], is considered (Fig. 4a) and the rigid end zone length is se-
presence of a floor system. In order to take the presence of slab into lected as half of the column width.

A B C D E

Fig. 2. Joint types and joint type index (JT) values for computations in SI units.
 M. Unal, B. Burak / Engineering Structures 52 (2013) 284–294 289

Fig. 3. Specimen models.

(a) (b)
Fig. 4. Details of the subassembly components.

Fig. 5. Beam moment versus plastic hinge rotation relationship.

Fig. 6. Moment through connection versus shear strain relationship.
Semi-rigid moment connection hinges are the main compo-
nents that represent the inelastic activity in the beams. In order
to get the inelastic moment–rotation relationship, first the mo-
zone at the end of the column, an inelastic P-M-M hinge and
ment–curvature behavior of the beams is obtained by Response
an elastic column cross-section at the free end. The cracked iner-
2000 [34]. Afterwards, the rotation values are obtained by multi-
tia of the column elastic segment is taken as the 70% of the
plying the curvature values by the plastic hinge length that is taken
gross inertia [14]. The column compound element is also illus-
as the half of the effective beam depth (Lp = 0.5d). As a result, the
trated in Fig. 4a. The rigid end zone length is taken as half of
moment–rotation relationship shown in Fig. 5 is used in defining
the beam depth.
semi-rigid moment connection hinges. A 20% strength decrease
Zero-length P-M-M hinges are employed, which are rigid-
was assumed to occur at higher rotation values based on the exper-
plastic hinges that remain elastic up to a moderate load level.
imental results.
Because, the inelastic activity in the columns is not as signifi-
cant as the one in the beams and they remain mostly in the
3.2. Column element elastic range due to strong column–weak beam design approach.
For defining these hinges, moment versus axial load yield (inter-
Similar to the beam element, the column element was mod- action) surfaces are plotted and utilized in defining the
eled as a frame compound component consisting of a rigid end component.
290 M. Unal, B. Burak / Engineering Structures 52 (2013) 284–294

Fig. 7. Lateral load versus story drift responses – specimens of Burak and Wight [1].

(a) (b)
Fig. 8. Joint shear stress versus joint shear distortion responses – specimens of Burak and Wight [1].

3.3. Joint element (Fig. 4b). This element has only one node that is taken at the inter-
section of beam and column axes. The rigid end zones of beam and
In order to represent the inelastic activity in the joint region, column elements are connected to rigid links of the panel zone
connection panel zone element in Perform 3D is utilized (Fig. 4a). A rotational spring attached to the panel zone has a non-
 M. Unal, B. Burak / Engineering Structures 52 (2013) 284–294 291

Fig. 9. Beam moment versus beam plastic hinge rotation responses – specimens of Burak and Wight [1].

Fig. 10. Story shear versus story drift responses – specimens of Raffaelle and Wight [29].

linear moment–rotation relationship that enables the definition of
the inelastic behavior.
As described before, the inelastic behavior of the joint is defined
by parametric equations. Using these equations, inelastic joint
shear strength versus shear strain relationships are generated for
connection regions. However, since the connection element re-
quires the joint moment versus shear strain relationship of the
connection region in Perform 3D, shear stress values are converted
to moment values by multiplying them by the effective joint width
(bj) and the column depth (hc) to obtain the joint shear force and
292 M. Unal, B. Burak / Engineering Structures 52 (2013) 284–294
Fig. 11. Joint shear versus joint deformation responses – specimen 2 of Raffaelle and Wight [29].
Fig. 12. Beam moment versus rotation response – specimen 2-S of Burak and Wight
[1].
the depth of the beam between the top and the bottom reinforce-
ment layers (d0 0 ) to obtain the moment through connection (Eq.
(9)).
M j ¼ v j  bj  hc  d00 ð9Þ
The above-mentioned procedure was carried out for all critical
shear strain points discussed earlier and eventually joint moment
versus shear strain relationship was obtained. The test results indi-
cated a connection strength decrease of about 20% at high shear
strain levels, therefore, 20% reduction of the ultimate moment is
considered. The resulting moment through connection versus
shear strain relationship is illustrated in Fig. 6.
4. Results of analytical verification
The proposed joint model is verified analytically by using Per-
form 3D. As it is stated earlier, the subassemblies that were tested
under quasi-static loading are modeled and ground acceleration
records that exactly simulates the displacement histories applied
to the subassemblies during the tests are imposed to the speci-
mens, because of the limitation of Perform 3D in defining cyclic
loading. The analytical results are presented below and compared
with the experimental results in order to test the validity and accu-
racy of the model.
4.1. Specimens of Burak and Wight [1]
Specimen 2-S, 3-S, 2-N and 3-N of this experimental study,
which have the same displacement history applied during the
experiment, are analyzed under dynamic earthquake loading. The
material strengths and geometric properties of these specimens
are presented in Table 1. The main variables considered in this
experimental study are eccentricity, beam width to column width
ratio, column aspect ratio and the presence of slab. Both experi-
mentally and analytically obtained lateral load versus story drift
responses are shown in Fig. 7. As it can be observed from this fig-
ure, analytical lateral load response represents the experimental
one realistically, except for pinching, which cannot be modeled
using Perform 3D. As a result of this limitation of the software,
wider loops are observed in the hysteresis curves. Nevertheless,
this does not affect the main objective of this study which is to
estimate the envelope behavior of the connection region and the
maximum values for lateral load, story drift, joint shear stress
and distortion. Specimens 2-S and 3-S have good predictions of
maximum story shear, therefore, even the cyclic degradation in
higher drift levels closely matched with the experimental ones.
Specimen 2-N and 3-N also have an acceptable accuracy, however,
the prediction in the negative loading direction is not adequate for
Specimen 3-N. This may be due to the wide beam of this specimen,
which was intentionally designed not to comply with the ACI-ASCE
Committee 352 Recommendations [15] in terms of the member
depth required to provide adequate confinement to the connection
region.
The comparison of the analytical and experimental response of
joint shear strength versus distortion is presented in Fig. 8. The
maximum joint shear strengths of the analytical models are signif-
icantly close to the experimental values. After the maximum shear
strength has been reached, the analytical behavior diverged from
the experimental behavior, especially for specimens such as 3-S,
which does not have adequate confinement provided by the
adjoining members [3].
Finally, the beam end moment versus plastic hinge rotation
curves are compared in Fig. 9. As for the load–deformation re-
sponse, the analytical model has good correlation with the exper-
imental results in terms of beam plastic hinge response.
Therefore, a realistic assessment of the beam behavior in terms
of moment capacity, maximum inelastic rotations and their contri-
bution to the total story drift can be performed by using these ana-
lytical results.
 M. Unal, B. Burak / Engineering Structures 52 (2013) 284–294 293

Table 3
Displacements due to joint shear distortions and their contribution to total story drift at 2% drift level.

Specimen Displacement due to joint shear distortion (mm) Total displacement (mm) Contribution of joint shear distortion (%)
Burak and Wight [1] 2-S 20.84 51.80 40.24
3-S 14.07 51.80 27.17
2-N 6.12 51.80 11.82
3-N 21.83 51.80 42.13
Kaku and Asakusa [23] Specimen 1 3.94 30.80 12.80
Specimen 2 2.93 30.80 9.52
Specimen 3 5.70 30.80 18.52
Specimen 6 6.29 30.80 20.42
Specimen 9 6.10 30.80 19.81
Specimen 12 6.43 30.80 20.88
Raffaelle and Wight [29] 1 7.49 44.72 16.75
2 8.62 44.72 19.28
3 6.46 44.72 14.44
4 6.56 44.72 14.66
Shin and LaFave [30,31] SL1 22.95 58.94 38.94
SL3 6.58 58.94 11.16
LaFave and Wight [25] EWB1 2.65 44.72 5.92
EWB2 2.48 44.72 5.55
EWB3 5.45 44.72 12.18
ENB1 1.48 44.72 3.31
Average 11.40 44.80 23.24

4.2. Specimens of Raffaelle and Wight [29]
The main variable of the experimental study by Raffaelle and
Wight [29] is the eccentricity between beam and column center-
lines. Besides eccentricity, member dimensions and the reinforce-
ment detailing of beams varies in different specimens. Although
the column dimensions are the same in all specimens, the effective
joint width differs due to changes in eccentricity and the beam
depth. The story shear versus story drift curves for the four eccen-
tric specimens of Rafaelle and Wight are given in Fig. 10. The enve-
lopes of the analytical results match the experimental ones with
significant accuracy, therefore the analytical joint model is verified
one more time for the eccentric interior connections. It is observed
that as the eccentricity increases, the ultimate story shear force de-
creases. Moreover, Specimen 4 attains the highest strength since it
has the deepest beam and thus the highest flexural strength. Up to
3% story drift, the ductility characteristics of all four specimens are
similar. After that point, the beam width and the eccentricity influ-
enced the energy dissipation capacity. For instance the beam width
of Specimen 2 is smaller; therefore it has smoother stiffness degra-
dation as captured by the model.
The joint shear versus joint deformation behavior comparison
for Specimen 2 is presented in Fig. 11. In this figure, the joint shear
stress versus strain relationship of the connection panel element
was compared with the experimental joint response measured
both at the flush (outer) and offset (inner) joint faces. In order to
make a comparison, the average of both figures should be consid-
ered since the panel zone element in the analytical model gives an
average response. It is observed that the maximum joint shear
capacity is predicted successfully by the analytical model. Besides,
the maximum joint strain values also closely match with the aver-
age shear deformation response obtained in the experiments. It
should also be noted that with the presence of a slab the difference
in the joint shear distortions will not be this significant on two
opposite faces of the specimen as in the case of Canbolat [3] and
Shin and LaFave [30].
4.3. Comparison of the joint model with rigid connections
In order to examine the effect of including the joint model in the
nonlinear analysis of RC beam-to-column connection subassem-
blies, all specimens are also modeled by assuming the connection
regions as rigid zones. In other words, the specimens are analyzed
once again by removing the joint model. Fig. 12 shows the moment
versus rotation response for the beams of Specimen 2-S by Burak
and Wight [1]. When the story drift values are kept the same, the
beam rotations are significantly overestimated with the use of rigid
connections, which could lead to improper seismic assessment of
these members.
The influence of joint shear distortion on total story drift is com-
puted at 2% drift level, using analytical shear strains in Eq. (10) [1]
and is presented in Table 3. It can be observed from this table that
the average contribution of the joint shear distortion to the total
story drift is 18%. This clearly indicates the need for considering
a joint model in nonlinear analysis.
 
hb hc
Dj ¼ H  c  1   ð10Þ
H l
5. Summary and conclusions
The main objective of this study is to develop an analytical joint
model to predict the inelastic behavior of reinforced concrete
beam-to-column connections subjected to earthquake loading that
can be incorporated into commercial software programs. For this
purpose, first, an experimental database of reinforced concrete
connection subassemblies is generated which includes material
properties, geometric characteristics and the test results under
cyclic loading [13]. Then, a set of equations are developed to esti-
mate the critical joint strength and strain values which give mini-
mum average error and the highest correlation with the
experimental values. While carrying out these steps, the guidelines
given in ACI Committee 318, Building Code Requirements for
Structural Concrete [14], and ACI-ASCE Committee 352 Recom-
mendations for Design of Beam–Column Connections in Mono-
lithic Reinforced Concrete Structures [15] are followed. The
connection subassemblies are modeled and the cyclic loading his-
tories followed in the experiments are applied analytically using
Perform 3D [33]. Finally, the analytical results obtained by consid-
ering the joint model and rigid connections are compared with the
experimental ones. Based on the results of the analytical study, the
following conclusions can be drawn:
294 M. Unal, B. Burak / Engineering Structures 52 (2013) 284–294
1. The envelope of the nonlinear hysteretic behavior of the joint
can be predicted by considering three critical points at which
significant slope changes are observed. A simple and conserva-
tive trilinear joint model is developed, which represents this
nonlinear behavior, to be used in inelastic analysis and design
applications. The nonlinear analysis is performed using com-
mercially available software, so that this joint model can also
be utilized in practical engineering applications. By slightly
increasing the computational time, more conservative esti-
mates of drift can be obtained, as well as the inelastic responses
of the connection regions.
2. The proposed joint model is integrated to the subassembly
model as a panel zone element. Beams are modeled by consid-
ering their moment versus rotation relationships, and for col-
umns, in addition to moment versus rotation curves, the
interaction diagrams are obtained. The joint model is verified
by comparing the experimental results with the analytical
response in terms of overall lateral load–story drift relationship,
joint element response, and beam plastic hinge response. It is
observed that, in general, the seismic behavior of beam-to-col-
umn connection subassemblies is reasonably predicted not only
for the overall lateral load–story drift response, but also for the
element responses.
3. The significance of concrete compressive strength, axial load
and volumetric joint transverse reinforcement ratio on the joint
shear response is also verified by this analytical study.
4. When analytical results considering the joint model and rigid
connections are compared, it is concluded that assuming a rigid
connection region in modeling results in underestimation of the
story drifts and overestimation of the beam plastic hinge rota-
tions, which leads to improper assessment of the seismic per-
formance of the overall structure and the structural members.
5. The contribution of joint shear distortions to the total story drift
are computed and it is observed that in average 18% of the story
drift is due to joint shear deformations.
References
[1] Burak B. Seismic behavior of eccentric reinforced concrete beam–column–slab
connections. PhD thesis, University of Michigan, Ann Arbor; 2005.
[2] Canbolat BB. Structural applications of a reinforced concrete beam–column–
slab connection model for earthquake loading. In: The 14th world conference
on earthquake engineering, Beijing, China; October 12–17, 2008.
[3] Canbolat BB, Wight JK. Experimental investigation on seismic behavior of
eccentric reinforced concrete beam–column–slab connections. ACI Struct J
2008:154–62 [Title no. 105-S16].
[4] Ghobarah A, Biddah A. Dynamic analysis of reinforced concrete frames
including joint shear deformation. Eng Struct 1999;21(11):971–87.
[5] Dhakal RP, Pan T, Irawan P, Tsai K, Chen C. Experimental study on the dynamic
response of gravity-designed reinforced concrete connections. Eng Struct
2005;27(1):75–87.
[6] Lee J, Kim J, Oh G. Strength deterioration of reinforced concrete beam–column
joints subjected to cyclic loading. Eng Struct 2009;31(9):2070–85.
[7] Parra-Montesinos GJ, Wight JK. Prediction of shear strength and shear
distortion in R/C beam–column joints SP197-10; 2002.
[8] Lowes NL, Altoontash A. Modeling of reinforced-concrete beam–column joints
subjected to cyclic loading. J Struct Eng 2003;129(12):1686–97.
View publication stats
[9] Mitra N, Lowes NL. Evaluation, calibration, and verification of a
reinforced concrete beam–column joint model. J Struct Eng 2007;133(1):
105–20.
[10] Shin M, LaFave JM. Modeling of cyclic joint shear deformation contributions in
RC beam–column connections to overall frame behavior. Struct Eng Mech
2004;18(5):645–69.
[11] Kim J, LaFave JM. Key influence parameters for the joint shear behaviour of
reinforced concrete (RC) beam–column connections. Eng Struct
2007;29(10):2523–39.
[12] Kim J, LaFave JM. Probabilistic joint shear strength models for design of
RC beam–column connections. ACI Struct J 2008:770–80 [Title no. 105-
S71].
[13] Unal M, Burak B. Joint shear strength prediction for reinforced concrete beam-
to-column connections. Struct Eng Mech 2012;41(3).
[14] ACI Committee 318, 2011. Building code requirements for structural
concrete, ACI 318-11. American Concrete Institute, Farmington Hills,
Michigan; 2011.
[15] ACI-ASCE Committee 352, 2002. Recommendations for design of beam–
column connections in monolithic reinforced concrete structures. ACI 352R-
02. American Concrete Institute, Farmington Hills, Michigan; 2002.
[16] Chen CC, Chen G. Cyclic behavior of reinforced concrete eccentric beam–
column corner joints connecting spread-ended beams. Technical paper.. ACI
Struct J 1999:443–50 [Title no. 96-S50].
[17] Durrani AJ, Wight JK. Behavior of interior beam-to-column connections under
earthquake-type loading. ACI Struct J 1985:343–9 [Title 82-30].
[18] Ehsani MR, Alameddine F. Design recommendations for type 2 high-strength
reinforced concrete connections. Technical paper. ACI Struct J 1991:277–90
[Title no. 88-S30].
[19] Ehsani MR, Wight JK. Exterior reinforced concrete beam-to-column
connections subjected to earthquake-type loading. Technical paper. ACI
Struct J 1985:492–9 [Title no. 82-43].
[20] Fujii S, Morita S. Comparison between interior and exterior RC beam–column
joint behavior. Design of beam–column joints for seismic resistance – ACI
1991. p. 145–65.
[21] Gentry TR, Wight JK. Wide beam–column connections under earthquake-type
loading. Earthq Spectra 1994;10(4):675–702.
[22] Guimaraes GN, Kreger ME, Jirsa JO. Evaluation of joint-shear provisions for
interior beam–column–slab connections using high-strength materials.
Technical paper. ACI Struct J 1992:89–98 [Title no. 89-S10].
[23] Kaku T, Asakusa H. Ductility estimation of exterior beam-column
subassemblages in reinforced concrete frames. Design of beam-column
joints for seismic resistance. ACI SP-123; 1991. p. 167–85.
[24] Kitayama K, Otani S, Aoyama H. Development of design criteria for RC
interior beam–column joints. ACI SP-123 design of beam–column joints
for seismic resistance. American Concrete Institute, Michigan; 1991. p. 97–
123.
[25] LaFave JM, Wight JK. Reinforced concrete exterior wide beam–column–slab
connections subjected to lateral earthquake loading. ACI Struct J 1999:577–86
[Title no. 96-S64].
[26] Lee HJ, Ko J. Eccentric reinforced concrete beam–column connections
subjected to cyclic loading in principal directions. ACI Struct J 2007:459–67
[Title no. 104-S44].
[27] Shiohara H. New model for shear failure of RC interior beam–column
connections. J Struct Eng 2001;127(2):152–60.
[28] Quintero-Febres CG, Wight JK. Experimental study of reinforced concrete
interior wide beam–column connections subjected to lateral loading. ACI
Struct J 2001:572–81 [Title no. 98-S55].
[29] Raffaelle SG, Wight JK. Reinforced concrete eccentric beam–column
connections subjected to earthquake-type loading. ACI Struct J 1995:45–55
[Title no. 92-S6].
[30] Shin M, LaFave JM. Seismic performance of reinforced concrete eccentric
beam–column connections with floor slabs. ACI Struct J 2004:403–12 [Title no.
101-S41].
[31] Shin M. Performance of reinforced concrete edge beam–column–slab
connections subjected to earthquake loading. PhD thesis, University of
Illinois at Urbana – Champaign, Urbana – Illinois; 2004.
[32] Teng S, Zhou H. Eccentric reinforced concrete beam–column joints subjected
to cyclic loading. ACI Struct J 2008:139–48 [Title no. 100-S15].
[33] Computers and Structures Inc. (CSI). User manual for PERFORM-3D v4.0,
Berkeley, California, USA; August 2006.
[34] Bentz EC. Version 1.1, user manual, response 2000; 2001.