Modeling of Panel Zone in Accordance to the Type of Connections in SMRF

www.cafetinnova.org

Indexed in

Scopus Compendex and Geobase Elsevier, Chemical

Abstract Services-USA, Geo-Ref Information Services-USA,

List B of Scientific Journals, Poland,

Directory of Research Journals

ISSN 0974-5904, Volume 06, No. 05 October 2013, P.P. 923-939

#02050607 Copyright ©2013 CAFET-INNOVA TECHNICAL SOCIETY. All rights reserved.

Modeling of PZ in accordance to the type of Connections in SMRF R. AHMADY JAZANY

1 AND A. GOLARA

1

1International Institute of Earthquake Engineering and Seismology (IIEES), Tehran, Iran.

Email: [email protected] , [email protected]

Abstract: This study aims at evaluating and modeling the cyclic behavior of panel zone in Special Moment

Resisting Frame (SMRF), according to their connection type, such as welded unreinforced flange-bolted web

(WUF-B) connections, cover plate connection, one sided haunch and double sided haunches. In this research, some

test specimens of experimental works conducted by SAC joint venture (report) were used to investigate the behavior

of PZ. In the analytical models of this study, all component of connection including bolts, and weld are modeled

using a surface contact for shear tab and bolt and shank, to evaluate the connection behavior precisely. Then

analytical response of the model including cyclic behavior of beams and PZ obtained from analytical models and

experimental models were used for training of Neural Network (N.N). Then new analytical data generated by Neural

Network (N.N.) used for empirical modeling of PZ according to their corresponding connection types. The results of

this study showed that the panel zone seismic behavior depends on the type of connections, and plastic and elastic

shears strain values of panel zone changes regarding type of connections and the proposed model presented in this

research predicts the panel zone seismic behavior accurately.

Keywords: Panel zone, SMRF, WUF-B, Analytical models, cyclic behavior, Neural Network

1. Introduction:

Moment resisting frames (MRFs) are widely used in

steel structures as lateral force resisting system due to

superior ductile behavior and energy absorption. MRFs

behave in ductile manner through flexural yielding of

beam and shear yielding of the panel zone. During of

sever ground motion, a huge amount of plastic

deformation is expected at each member in MRFs [1-5].

In Northridge earthquake, sever damage in welded

connection of steel moment frames occurred.

Nevertheless, the occurrence of the connection fractures

has resulted in changes in design and construction of

steel moment frames. The fracture of moment

connection in the Northridge earthquake exhibits a

variety of origins and path; In general, the fracture of

beam to column joint initiates at the root of beam flange

to column flange CJP groove weld and propagate

through the beam flange, the column flange, or weld

itself. In some instances fracture extended through

column flange and web. The backing plate, which was

generally left in place, produced a mechanical notch at

the weld root and this may be a reason for connection

fracture. Also based on last studies [1-5], the excessive

shear strain of panel zone may result in CJP groove

weld fracture; thus, the PZ seismic behavior can have a

important role in controlling of fracture and yielding

modes of the connection.

Studies on Panel zone (PZ) behavior were initiated in

the late 1960s and early 1970s to comprehend the

inelastic behavior of joints in moment-resisting frames

[1-5]. These studies indicated that the contribution of

top displacement of the sub-assemblage was highly

influenced by the panel zone distortion. Becker [5] also

pointed out the importance of considering the influence

of joint deformations of the frames in terms of stiffness

and energy absorption. In addition, the effect of axial

loads on the performance of connections subjected to

shear was investigated. Several years later, Slutter [6]

and Popov et al. [7-8] studied the extreme loading

condition effects on the beam to column joint cyclic

behavior. They showed that the column flange

contributes the nonlinear behavior of the PZ effectively.

They also emphasized that the PZ could have a high

reserve of strength after yielding, high ductility and

significant strain hardening properties. With regard to

the high nonlinear capacity of PZ, these observations

resulted in a decrease in the PZ demand that was

specified in ICBO [9] and AISC [10]. These codes

therefore capped at 80% the transferred shear

corresponding to the beam plastic moment, but large

distortions of the PZ could lead to local kinking of the

column flange at the corner of the joints, where the

beam flanges are connected to the column flanges.

Results of a study by Tsai and Popov [11] indicated that

panel zones designed according to the above-mentioned

mailto:[email protected]

mailto:[email protected]

924 Modeling of PZ in accordance to the type of Connections in SMRF

International Journal of Earth Sciences and Engineering

ISSN 0974-5904, Vol. 06, No. 05, October 2013, pp. 923-939

provisions could undergo large inelastic shear distortion

before reaching their rated shear capacity. Weak PZ

behavior may have played a role in the failures that

occurred during the Northridge earthquake [12-13]. The

Northridge California earthquake in 1994 triggered a

large amount of research activity in the United States

and additional experiments was conducted to further

understand the balance of energy dissipation between

the panel zone and the beam by varying the panel zone

capacity in terms of the moment capacity of the beam.

Participation of the panel zone to the inelastic response

contributed to the reduction of the demands on the

beams in terms of deformation. Analogous research was

carried out in Europe by Dubina et al. [14] and Ciutina

and Dubina [15] to understand the cyclic performance

of beam-to-column joints. In FEMA-355D [16], which

originated from the SAC joint venture, the proposed

method of design was substantially altered so that it

became completely different to that specified in

previous codes [10], being based on the idea that the

framing beams and panel zone should yield at the same

time to achieve balanced behavior. It defined the yield

point of the beam and of the PZ as the base line for this

balanced condition. Jin and El-Tawil [17] were,

however, of the opinion that this method was

unsatisfactory since such balanced beam and PZ

capacity could not guarantee controlled distortion of

these elements, and that it would not be possible to

establish simultaneous yield mechanisms in the beam

and PZ.

Hosseini hashemi and Ahmady jazany [18-19] have

conducted six full-scale experiments to understand the

behavioral differences of SMRF with unequal beam

depths with respect to different connection detailing

arrangements. The results showed that haunch

connection system with flange plate connection have

more total plastic rotation with less strength degradation

compared to other connection detailing arrangement.

They [19-21] also analytically and experimentally

showed that the test specimens with flange plate

connection have superior seismic behavior rather than

those with cover plate connection.

This study was based on 22 test specimens reported by

SAC [22] joint venture and other some experimental

works has been done by popov et. al [23] .Fragile

behavior of WUF-B observed in SAC experiments,

resulted in change of formation and connection types,

therefore some kind of modification were made for

connections, such as cover plate, one sided haunch and

double sided, to improve the performance of

connection. The weld joining of the beam flanges to the

face of relatively thick column flange is highly

restrained. This restraint would cause somewhat more

brittle behavior. In the context of earthquake damage to

SMF buildings, the term of repair is used to address the

restoration of strength, stiffness, and inelastic

deformation capacity of structural elements to their

original level. For example, the plastic strain of simple

WUFs specimen rarely reaches 0.02 radians. Chi et.al

[24] reported that larger panel zone caused a larger

ductility demand. EL-Tawil [25] studied the effect of

panel zone distortion on plastic rotation capacity. Most

of pre-Northridge connections behaved similarly. It was

shown that capacity and stiffness of such connections

respectively have small value. Enhancement of the

strength, stiffness, or deformation capacity of either

damaged or undamaged structural elements, would lead

to improvements in their seismic resistance and that of

structure as a whole. Modification may also involve

stiffening by cover plate or haunches and removal of

existing weld. Considering of PZ seismic behavior of

each category exhibit that plastic rotation of each

specimen depends on ratio of capacity of PZ to capacity

of beam, type of connection, and number of continuity

plates provided. Popov et al. [26] have studied WUF-B

connections and triangular haunches at the bottom of

beam flange. Regarding last studies [27] haunch

connection system can improve beam plastic rotation;

on the contrary, it decreases plastic rotation of PZ.

Moreover, in double haunch connection system, PZ

shear strain values do not enter nonlinear phases

significantly. In other word, PZ of this type of

connections behaves, linear. This result was observed

regarding SAC reports [22], another phenomenal which

is observed from double haunches connection, was

degradation of capacity in the experimental result, this

characteristic was known as a defect point in these

categories, while huge amount of beam plastic rotation

was a power point. Another category of this research

was cover plate on top and bottom flange. The plastic

behavior and energy absorption of PZ and beam was

stable and significant; and no significant degradation of

the capacity was observed according to SAC report

[22].

2. Modeling procedure:

2.1 General concept:

Some experimental studies reported by SAC [22] were

used for verification of Finite element models of this

study. Table 1 represents the test specimens in this

research originated from SAC reports [22]. The

objective of this research was to investigate on the

improved post Northridge connection such as WUF-B

connection, cover plate connection and haunch

connection system. The test set up include one beam

and a column like an exterior joint, in which beam web

connects to column through shear tab and bolt. The

beam-ends are simply supported and applied load on

beam end generate moment on connection. In addition,

the end of column was simply supported in two ends,

925 R. AHMADY JAZANY AND A. GOLARA



which reflect inflection point of column in actual

structure. All mechanical properties and geometries of

section has been presented in table 1.also Yield strength

(Fy) and ultimate strength (Fu) was included in this

table.

2.2 Finite element modeling:

ANSYS [28] multi-purpose finite element program was

used to perform the numerical modeling of connections.

FE models were implemented using the ANSYS

parametric design language. The geometrical and

mechanical properties of the connection model were

defined as parameters in ANSYS such as root opening

geometries and Yield strength (Fy) and ultimate strength

(FU). Numerical modeling of connection was performed

including the following considerations: Using eight-

node-first order SOLID45 elements; also, bolt shanks

are modeled using SOLID64 element. ANSYS can

model contact problem using contact pair element:

CONTA174 and TARGE 170 [28], which work together

in a way that there is no penetration occurrence during

the loading process. The interaction in adjacent surface

between shear tab and web are modeled using

mentioned contact element. Bolt heads and nuts are

modeled as hexagonal and similar to real shape to

simulate the frictional forces. Coulomb coefficient was

assumed as 0.3, which produce the best result. Nearly in

most of literature for class a type steel surfaces, the

coulomb coefficient is one third of the usual value of

.33, which is proposed. It means 0.1 is proper amount

for converging and yielding result, however, because of

some differences in type of steel grade, the best

agreement between analytical and experimental seismic

response was occurred when it is considered about 0.3.

Table1: Geometric properties of the reference experiment [22]

Beam Column

Specimen Web Flange Web Flange

Number Specimen name Section Fy Fy Section Fy Fy Connection type

1 RFSPN1 (Whittaker etal) W30x99 55.7 50.3 W14x176 69.5 69 WUF-B

2 RFSPN2 (Whittaker etal) W30x00 57.4 48.6 W14x176 70.8 68.9 WUF-B

3 RFSPN3 (Whittaker etal) W30x99 53.4 47.2 W14x176 72.5 68.4 WUF-B

4 RFSRN1 (Whittaker etal) W30x99 55.7 50.3 W14x176 69.5 69 WUF-B

5 UCSDPN1 (Uang etal) W30x99 57.1 46.6 W14x176 67.2 68.2 WUF-B



8 UCSDR2 (Uang etal) W30x99 57.1 46.6 W14x176 67.2 68.2 WUF-B

9 SAC-N06 (SAC joint) W24x76 50.2 44.2 W14x132 66.1 66.4 WUF-B

10 SAC-N07 (SAC joint) W24x76 50.2 44.2 W14x132 66.1 66.4 WUF-B

11 RFSAN1 (Whittaker etal) W30x99 55.7 50.3 W14x176 69.5 69 Cover plate

12 UCBAN1 (Popove etal) W36x150 40.3 40.3 W14x257 67.8 67.8 Cover plate

13 UTA-4 (Engelhardt) W36x150 45.5 39.5 W14x257 69 69 Cover plate

14 SAC NO09 (SAC joint) W24x76 39.1 38.3 W14x132 66.1 66.4 Cover plate



17 UCSDR1 (Uang etal) W30x99 57.1 46.6 W14x176 67.2 68.2 one side haunch

18 UCSDR3 (Uang etal) W30x99 57.1 46.6 W14x176 67.2 68.2 one side haunch

19 UCBR2 (Popve etal) W36x150 60.6 60.6 W14x257 67.8 67.8 one side haunch

20 UCB RN2 (Popove etal) W36x150 60.6 60.6 W14x257 67.8 67.8 one side haunch

21 UTAR1 (Engelhardt) W36x150 45.5 39.5 W14x257 69 69 one side haunch

22 UTAR1B (Engelhardt etal) W36x150 45.5 39.5 W14x257 69 69 one side haunch

23 UCBRN2 (Popove etal) W36x151 60.6 60.6 W14x258 67.8 67.8 double side haunches




24 UCBRN3 (Popove etal) W36x152 60.6 60.6 W14x258 67.8 67.8 double side haunches

25 UTA-3 (Engelhardt etal) W35x150 45.5 39.5 W14x257 69 69 double side haunches

2.3 Boundary condition and applied load:

To satisfy boundary condition of analytical model, the

end of beam was fixed against to the lateral

displacement. Also because of existing lateral bracing

system in real model on the flange of beam of test

specimen, some points at middle length of the beam

flanges of analytical models were fixed. Since there was

no information about the situation of bolt regarding pre-

tension or ordinary twisting of bolt, it was considered as

ordinary bolt which would not permit shear tab to slip

outward the plan of web. Displacement control was

used for loading of beams end, and loading pattern was

based on ATC 24 loading protocol [29], which was used

for reference experiments reported by SAC [22-23]

2.4 Material properties:

The material properties of these models had kinematic

behavior with strain hardening in nonlinear phase to

predict the material specification precisely. The stress-

strain curve for all connection components except for

the bolts was represented using a three-linear

constitutive model. An isotropic hardening rule with a

von Mises yielding criterion was used to simulate

plastic deformations of the analytical model. ASTM 36

steel was used for the beam and ASTM 52 steel was

utilized for the column and connection details. In the

current study, the mechanical properties of beam

column and connections are taken from table 1 [22].

The yield stress and ultimate strength of bolts are

assumed to be based on nominal properties of A325.

The yield stress and ultimate stress of weld are assumed

to be based on nominal properties of E71T-8(AWS

A5.20) [30]. Modulus of elasticity and Poisson’s ratio

was taken 29000 kips/in2 and 0.3.

3. Verification of finite element model:

To evaluate the accuracy of finite element modeling

approach, 25 finite element models were used according

to actual test data mentioned in Table 1. Figs 1 to 3

shows Analytical and experimental cyclic responses for

the test specimens PN3 with WUF-B connection, test

specimens AN1 and UTA4 with cover plate connection

and test specimen UTAR1 and UTA-3 with haunch

connection system. Based on these Figures, it can be

seen the results obtained from finite element models

have good agreement with test data. Differences

between the numerical simulation and test result may be

the result of several causes like numerical modeling

simplification, test specimen defect or residual stress. It

can be mentioned that the material properties, which are

used in FE, are from average, but in reality steel is not a

homogenous material and amount of every coupon test

result could affect the actual result. The differences

between the test data and the numerical models grow in

nonlinear phase of curve.

Fig1: (a) experimental and (b) analytical cyclic response for the test specimen PN3 with WUF-B connection




Fig2: (a), (c) experimental and (b), (d) analytical cyclic responses for the test specimens AN1 and UTA4 with cover

plate connection

Fig3: (a), (c) experimental and (b), (d) analytical cyclic responses for the test specimens UTAR1 and UTA-3 with

haunch connection




Based on above-mentioned verification in this section;

the panel zone cyclic behavior of analytical models of

the test specimens were calculated and the backbone

curves of these analytical models were obtained in

parallel with panel zone experimental cyclic behavior.

These curves are used for training neural network

program, and new curve as Neural Network (N.N)

output were produced from Neural Network (N.N)

program. Neural Network (N.N) consisted of three-

layered PERCEPTRON [31]. It is worth mentioning that

data of four groups were the basis of training of Neural

Network program [31].

The first step for training of neural network was use of

the backbone curve of analytical models and

experimental models responses i.e. total moment versus

PZ shear strain .Figs 4, 5, 6 and 7 shows backbone

curve for analytical model and experiments and

corresponding neural network curve. Comparing this

curves for each test specimens showed that training of

neural network was acceptable and corresponding

Neural Network outputs are in a good agreement with

analytical and experimental responses. It is mentioned

that 25 models were used for training of neural network

and then five models were used to test the Neural

Network results. Finally, all curves were reproduced

from neural network as shown in the Figs 4, 5, 6 and 7

regarding each connection type.

RFSPN1

0

5

10

15

20

0 0.005 0.01 0.015

Shear strain of panel zone(radian)

Mo

men

t(kip

s-i

n)

F.E

N.N

Experiment

a

RFSPN2

0

5

10

15

20

0 0.005 0.01 0.015


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

b

RFSPN3

0

5

10

15

20

0 0.005 0.01 0.015


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

c

RFSRN1

0

5

10

15

20

0 0.005 0.01 0.015


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

d

USCD1

0

5

10

15

20

0 0.005 0.01 0.015


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

e

USCD2

0

5

10

15

20

0 0.005 0.01 0.015


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

f




USCD3

0

5

10

15

20

0 0.005 0.01 0.015


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

g

USCD-2R

0

5

10

15

20

0 0.005 0.01 0.015 0.02 0.025


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

h

SAC-No7

0

5

10

0 0.005 0.01 0.015


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

i

SAC-No6

0

5

10

0 0.005 0.01 0.015 0.02 0.025

Shear strain of panel zone(radian)M

om

en

t(k

ips

-in

)

F.E

N.N

Experiment

g

Fig4: Comparison between backbone curve of response of experiment, FE and N.N. for the test specimens with

WUF-B connection

UBCAN1

0

5

10

15

20

0 0.005 0.01 0.015 0.02 0.025 0.03


Mo

men

t(kip

s-i

n)

F.E

N.N

Experiment

a

UTA4

0

5

10

15

20

25

30

35

0 0.01 0.02 0.03 0.04


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

b

RFSAN1

0

5

10

15

20

0 0.005 0.01 0.015 0.02 0.025 0.03 0.035


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

c

SAC No.9

0

5

10

0 0.01 0.02 0.03


Mo

men

t(kip

s-i

n)

F.E

N.N

Experiment

d




SAC No. 12

0

5

10

15

0


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

e

SAC No 13

0

5

10

15

20

0 0.005 0.01 0.015 0.02 0.025


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

f

Fig5: Comparison between backbone curve of response of experiments, FE and N.N. for the test specimens with

cover plate connection

UCBR2

0

5

10

15

20

25

30

35

40

0 0.005 0.01 0.015 0.02


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

a

USCD-1R

0

5

10

15

20

0 0.005 0.01


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

b

USCD-1R

0

5

10

15

20

0 0.005 0.01


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

c

UTA-1R

0

5

10

15

20

0 0.005 0.01 0.015


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

d

UTA-2R

0

5

10

15

20

25

30

35

40

0 0.005 0.01


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

e

Fig6: Comparison between backbone curve of response of experiments, FE and N.N. for the test specimens with

one sided haunch connection




UCBRN1

0

5

10

15

20

25

0 0.005


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

a

UCBRN2

0

5

10

15

20

25

0 0.005


Mo

me

nt(

kip

s-i

n)

F.E

N.N

Experiment

b

Fig7: Comparison between backbone curve of panel zone response of experiments, FE and N.N. for the test

specimens with double sided haunch connection

Fig8: Typical moment-rotation curve of connection and its reference moment

4. Test specimen with WUF-B connection:

First category of this group was test specimens with WUF-B connection. To evaluate the first group of this

ensemble, hysteretic behavior of PZ was obtained from analysis performed in this study then the idealized curve was

produced from the backbone curve of analytical and experimental cyclic responses based on FEMA 273 [32] as

shown in Fig.8 .These idealized curves consists of two idealized lines as shown Fig 8. The yield moment values and

plastic rotations of panel zone were obtained from this idealized curve. Figs.9 (a) and (b) show the FE model and

Von Mises stress distribution for test specimens test specimen PN1 with WUF-B connection

Fig9: (a) FE modeling of test specimenPN1 with WUF-B connection (b) Von Misses stress distribution of test

specimen PN1 with WUF-B connection




Fig 10 shows and ratio of Mm/Mp versus ratio of

Vy/VPZMy (Mm is the maximum moment at the column

face and Mp is the plastic moment capacity of the

connected beam). It is mentioned that initial stiffness is

the tangent of hysteretic curve at the start point .this

amount are normalized by GK , which is obtain by

FEMA 355-D [16]:

wcG tdGK ...95.0 (1)

G is shear of steel modulus, tw is thickness of column

web, and dc is column depth. Also VpzMy is presented as

follows:

))(2

(h

dh

ldL

L

d

MV b

pcb

yPZMy

(2)

In this equation; VPZMY is the shear which is projected

from beam to PZ, L is beam length, h is column height

and db is beam depth, also My and Mp are ultimate

elastic and plastic moment of beam. Fig.10 shows

moment ratio of (Mm/MP) with respect to a panel zone

strength ratio (Vy/VPZMy). The moment ratio is an

indicator that beam can reach its moment capacity

before the connection fails; if ratio of (Mm/Mp) does not

go exceeds the value of 1 ,it means that plastic moment

capacity does not reach full plastic moment capacity.

Regarding Fig. 10, ratio of (Mm/Mp) exceeds 0.9 where

ratio of Vy/VPZMy reaches 1. The study by Whan Han

[33] confirms this finding. also the ratio of (Mm/Mp) for

all test specimens of this study are in small range

between 0.9 and 1.0 .The difference between results of

this study and other study[33,16] may be originated

from using deeper beam in this study which results in

smaller plastic rotation capacity.

This is due to use of different beam section in this study.

In the study of Whan Han [33], the comparison has been

made for the ratio of (Mf/MP) and Mf is moment at

fracture point, and Mm is maximum moment. In this

study, since there is no significant strength degradation

in reference test specimens with WUF connection and

Mf approximately reaches to Mm .also based on Fig 10,

panel zone strength ratio (Vy/VPZMy) should be equal to

1.0 if moment capacity of beam reaches 0.9 value of

plastic moment capacity. Based on Fig 11(a)

dependency of PZ elastic rotation of test specimens with

WUF-B connection to panel zone strength ratio

(Vy/VPZMy) is insignificant, also regarding Fig 12(b), the

PZ plastic rotation does not reach 0.02 radian for all

corresponding analytical and experimental value of the

reference experiments with WUF-B connection

presented in Table 1.

0.75

0.8

0.85

0.9

0.95

1

1.05

1.1

0.8 0.9 1 1.1 1.2 1.3 1.4

Vy/VpzMy

Mf/M

p

Test result

FE

Neural Network

Fig10: moment ratio (Mm/MP) versus Vy/VPZMy for

(WUF) connection

y=-.0062(Vy/VpzMy)+0.011

0

0.005

0.01

0.015

0.02

0.8 1 1.2 1.4

Vy/VpzMy

PZ

,ela

sti

c r

ota

tio

n,

y

Test Result

FE

Neural Network

a

p=-.0081(V/VpzMy)+.0167

0

0.01

0.02

0.03

0.04

0.05

0.06

0.8 0.9 1 1.1 1.2 1.3 1.4

Vy/VpzMyz

PZ

pla

sti

c r

ota

tio

n,

pz

Test Result

FE

Neural Network

b

Fig11: (a) PZ elastic rotation versus Vy/VPZMy for (WUF)

connection (b) PZ elastic rotation versus Vy/VPZMy for

(WUF) connection

5. Test specimens with cover plate connection:

Regarding FEMA 355D [16] ,This type of connection

has acceptable seismic behavior compared to WUF-B

connection and it is qualified as SMRF (special moment

resisting frame[10]. According to SAC report [22], this

type of connection provides more plastic rotation up to

0.03 radians. The FE model of RFSAN1 is shown in

Fig.12 (a) also Fig 12(b) shows moment ratio (Mm/MP)

with respect to a panel zone strength ratio (Vy/VPZMy).




Strength ratio (Mf/MP) is indicator of connection

capacity and it shows if the connected beam reaches its

moment capacity before the connection is failed.

Because of completion of data in this section, other FE

models were used to cover the gap between data and

then all statistical have been made based on additional

information. Fig 12(a) and (b), show elastic and plastic

panel zone shear behavior respectively. As it was

obvious, the cover plate connection provides the panel

zone plastic shear strain value more than 0.02 radians.

0.8

0.9

1

1.1

1.2

1.3

0.8 1 1.2 1.4 1.6 1.8

Vy/Vpz

Mf/M

p

Test result

FE

Neural Network

b

Fig12: (a) FE model of Specimen (RFSAN1) (b)

moment ratio (Mm/Mp) versus Vy/VPZMy for cover plate

connection

y=-.0150(Vy/VpzMy)+.0234

0

0.005

0.01

0.015

0.02

0.025

0.03

0.8 1 1.2 1.4 1.6 1.8

Vy/vpzMy

PZ

ela

sti

c r

ota

tio

n,

y

Test Result

FE

Neural Network

a

pz=-0.0086(V/VpzMy)+.0362

0

0.01

0.02

0.03

0.04

0.05

0.06

0.8 1 1.2 1.4 1.6 1.8

Vy/VpzMyz

PZ

pla

sti

c r

ota

tio

n,

pz Test result

FE

Neural Network

b

Fig13: (a) PZ elastic rotation versus Vy/VPZMy for the

test specimens with cover plate connection (b) PZ

plastic rotation versus Vy/VPZMy for the test specimens

with cover plate connection

And for Vy/VPZM <1 ,it reach 0.03 radians and the slope

of regression line is reducing, however the value of PZ

elastic rotation for the cover plate connection is more

than corresponding value of the models with WUF-B

connection. Based on Fig. 12(b), the minimum values of

(Mm/Mp) are more than 1 and the maximum

corresponding values reaches 1.3 and it increases when

ratio of (Vy/VPZM) increases. In conclusion, presence of

cover plate could enhance nonlinear properties of PZ.

6. Test specimens with one sided haunch

connection:

Welding a tapered haunch to the beam bottom has been

used for many years in steel structures; this method will

be very effective for improvement of cyclic

performance of damaged moment connections or new

construction [34]. Reinforcing the beam with a welded

haunch can be considered as a means of increasing the

section modulus of the beam at the face of the column

.many studies [35-38] suggested that a more appropriate

approach is to consider the flange of the weld haunch as

a diagonal strut. The strut action drastically changes the

force transfer mechanism of this type of connection

[39].

In this study, some of experimental works, which

included one sided haunch connections, have been

modeled. Fig. 14(a) and (b) shows FE models of test

specimens USCD-2R and UTAR1B. In most of the test

specimens that are used in this study, a vertical plate

was provided in front of haunch flange on beam web to

prevent beams local buckling of beam flange. Fig. 14(c)

and (d) shows Von Mises stress distribution for

analytical model of test specimens USCD-2R and

UTAR1B. Because of shortage of information, it was

necessary to produce enough data, therefore six FE

models were included which was similar to

corresponding six experimental works as presented in

table 1, but the yield stress value in the web column of




these new models were ten present lower than reference

FE models.

Fig14: (a), (b) FE modeling of test specimen USCD-2R

and UTAR1B (c), (d) Von Misses stress distribution of the

test specimen USCD-2R and UTAR1B

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

0.8 1 1.2 1.4 1.6 1.8

Vy/Vpz

Mf/M

p

SAC

FE

Neural Network

Fig15: (Mm/MP) versus Vy/VPZMy for one sided haunch

connection

y=-.0057(Vy/VpzMy)+.0106

0

0.01

0.02

0.03

0.04

0.05

0.9 1.1 1.3 1.5 1.7

Vy/VpzMy

PZ

ela

sti

c r

ota

tio

n,

y

SAC

FE

Neural Network

a

pz=-.0065(V/VpzMy)+0.0126

0

0.01

0.02

0.03

0.04

0.05

0.06

0.8 1 1.2 1.4 1.6 1.8

Vy/VpzMyz

PZ

pla

sti

cro

tati

on

, p

z

SAC

FE

Neural Network

b

Fig15: (a) PZ elastic rotation versus Vy/VPZMy for test

specimens with one sided haunch connection (b) PZ

plastic rotation versus Vy/VPZMy for test specimens with

one sided haunch connection

The strength ratio (Mm/MP) for this group was presented

in the Fig.15. As it is obvious for all values of Vy/VPZMy,

the strength ratio (Mm/MP) is larger than 1.1. It means

that use of this type of connection results in

improvement of the ultimate moment to full plastic

moment before fracture or significant strength

degradation. The average of strength ratio (Mm/MP) is

1.2 ; in other word , haunch connection system provide

significant plastic rotation for the beam and ,the test

specimens with haunch connection system enter strain

hardening before collapse happens, FEMA-355D[16]

account for strain hardening for reinforced connections

as follows:

2.max

tbyb

imum

FFZM

Where Z is plastic modulus of beam, tbF is expected

tensile stress of steel, ybF is yield strength of steel and

maxM is maximum moment before fracture. Further

studies [35-37] show that this over strength could reach

to 40% to 50% and results of this study confirm this

finding. PZ elastic and plastic rotation versus Vy/VPZMy

for one sided haunch connection was presented in Fig

16(a) and (b) respectively. As it is mentioned, plastic

rotation does not reach 0.02 radians. In conclusions, PZ

of this type of connection does not enter nonlinear phase

significantly. This is due to presence of three continuity

plates and strut action of a haunch flange that is fixed to

the boundary of PZ.

Another type of haunch connection system described in

this chapter is double-sided haunches. FE model of this

type of connection and Von Mises stress distribution

were shown in Figs 16(a) and (b). Based on studies by

Chia-Ming et al. [35], this type of connection provides

significant total plastic rotation for the beams reinforced

by this type of haunch. However, PZ seismic behavior




of this type of connection is essentially linear, and PZ of

this type of connection does not experience nonlinear

phase. Fig. 17 shows PZ elastic rotation versus

Vy/VPZMy, based on this Figure; PZ elastic rotation

experience very small values and dependency of PZ

elastic rotation to ratio of Vy/VPZMy for double haunch

connection is insignificant.

Fig16: (a) FE model of (UCB RN3) with double sided haunch connection (b) Von Misses distribution of FE model

of test specimen (UCB RN3) with double sided haunch connection

To achieve more reliable results, the models with

different of yield strength value according to of three

main category of the connection were built, and results

of analyses were the included for the final conclusions,

as it was pointed out PZ of this kind of connection stay

in linear phase. Also based on Fig 17, elastic rotation of

this type of connection is very small compared to the

test specimens with one-sided haunches. It may be due

to presence of four continuity plates and strut action of

two flange haunches for ensemble, which the PZ is

stiffened considerably,

y=-.0027(Vy/VpzMy)+.0052

0

0.005

0.01

0.9 1.1 1.3 1.5 1.7

Vy/vpzMy

PZ

ela

sti

c r

ota

tio

n,

y

SAC

FE

Neural Network

Fig17: PZ elastic rotation versus Vy/VPZMy for double

sided haunch

The average value of strength ratio(Mm/MP) of PZ of

this type of connection reach 1.25 and minimum value

is 1.15 ,in other word ultimate moment of this type of

connection reaches up to 25 % full plastic capacity. As

a result, it is pointed out, specially in double-sided

haunches, there is no absorption of energy for double-

sided haunches.

7. Validity for proposed model of PZ:

7.1 Previous studies:

A few models of PZ for all type of connection have

proposed by several researches in the past. Gupta and

Krawinkler.h, [39] developed a centerline model having

a connection fracture spring element which is installed

at the end of the beams and emulates the fracture at the

beam flange connection. Whan Han et al. [31] used this

concept for modeling WUF-B connection and the result

is valid with 5% to 7% error in total of beam

displacement. Gupta and Krawinkler et al.[39]

developed a connection model (“M2 model”)

accounting for panel zone deformation explicitly. The

M2 model consists of column, beam and panel zone

elements. This model uses the exact dimension of PZ

with rigid body and two bilinear hysteretic springs,

which simulate a trilinear behavior for the panel zone.

7.2 Analytical models proposed for each category:

In order to simulate the hysteretic of the post Northridge

behavior for each category, the regression line for all

groups are used to create PZ model. This study attempts

to predict the PZ behavior with respect to connection

type therefore the main aim of this research is to present

a model for PZ according to connection. Totally,

emphasis will be on PZ behavior and the beam seismic

behavior, which is obtained from backbone curve of

beam in reference experimental work [22]. Thus, the

proposed connection can account for inelastic panel

zone as well as connection fractures.

In the panel zone model, the boundary elements for the

panel zone are rigid elements with very high axial

rigidity. The panel zone shear strain and stiffness can be

http://www.researchgate.net/researcher/23985086_Sang_Whan_Han/




modeled by two bilinear springs in one of the four

corners as shown in the Fig.18. The two superimposed

springs in the PZ were used in this study to simulate

trilinear behavior of PZ. Detail of the two bilinear was

obtained from study of Gupta and Krawinkler [39]. The

other three corners are modeled as simple pin

connections. PZ with double-sided haunches is

nonlinear and does not follow this model. Because of

focusing on the PZ seismic behavior in this study under

different types of connection (WUF-B, cover plate, one-

sided haunch, and two sided haunches) and number of

continuity plate which is implemented in PZ, modeling

of beam was ignored, and the backbone curve of beam

and column which were obtained from test results are

considered. The strain hardening for PZ reach to 6%

which is proposed by FEMA 355-D [16].

Fig18: Analytical models of PZ and beam

Based on Whan Han study [16], for test specimens with

WUF-B connection, where shear ratio (Vy/VPZMy) is

less than 0.9, the beam moment could not reach its

plastic moment capacity (Mp) while the test specimens

with shear ratio (Vy/VPZMy) more than value of 0.9,

beam reach to its plastic moment (Mp). this shows that

type of connection is effective on beams moment to

reach their own plastic capacity for example haunch

system have more potential to reach their own full

plastic capacity and also it has more independency to

the shear ratio Vy/VPZMy.

7.3 Verification of the proposed PZ model:

The proposed PZ models of this study were validated by

cyclic behavior of reference test model. In order to

perform analytical modeling of PZ, Drain-2DX

Analytical modeling software [40] was used. The spring

specifications used in the Drain-2DX were provided by

the average values for each connection type (regression

line) as shown in Figs 11, 13, 16 and 17. Fig.19 shows

cyclic responses generated by analytical model

including applied moment versus beam rotation and

panel zone rotation with respect to connection type.

Based on this Figure panel zone of the test specimen

with cover plate connection (UCB-AN1) has more PZ

rotation compared to the corresponding value of

analytical model with test specimens WUF-B

connection (RFS-PN3) and PZ rotation of analytical

model of test specimen with one sided haunch

connection (USCD-R3) provides minimum plastic

rotation (See Fig 19).

However, PZ of Analytical models of test specimen

with double-sided haunches remain in linear phase (Fig.

19).Fig 20 shows comparison between analytical and

experimental response for the test specimens of UCB-

PN3, UCB-AN1, UCB-R3 and UCB-RN3 with different

connection types. Based on Fig 20, cyclic response of

analytical model (Figure 19) agreed well with the cyclic

response of the corresponding response experiments. It

means that proposed model of PZ, which its rotation is a

part of total rotation provide good estimation for PZ

deformation. FEMA 273 [32] suggests a constant

partial ductility factor ( ) for PZ, without

consideration connection type effects on PZ seismic

behavior. Krawinkler et al. [3] showed that the value of

four times of PZ elastic distortion(y4 ) as stable PZ

plastic distortion will be promising value and it can

keep the connection within a safe margin whereas the

results of this study showed that this values strongly

depends on the connection type and shear ratio

(Vy/VPZMy). Considering backbone curve of PZ as

shown in Figs 4 to 7; elastic and plastic rotation of PZ

depend on the connection type ,i.e. cover plate ,flange

plate and haunch connection system, shear ratio

(Vy/VPZMy) and also number of continuity plate on the

panel zone.





Fig19: analytical models of the test specimens (a) RFS-PN1 (b) UCN-AN1 (C) UCSD-R3 (d) UCB-RN3

Fig20: comparison between analytical and experimental response for the test specimens (a) test specimen UCB-

PN3 (b) UCB-AN1 (c) UCB-R3 (d) UCB-RN3




8. Conclusion:

This study investigated on the PZ cyclic behavior

according to connection type. This study emphasized

that the type of connection besides shear ratio

(Vy/VPZMy) affects the PZ nonlinear behavior. The

following conclusions can be made:

1. Analytical results of this study showed that PZ

plastic rotation value was certainly affected by type

of connection; in other word, PZ of analytical

models of the test specimens with cover plate

provide more PZ plastic rotation rather than the

corresponding values of test specimen with WUF-B

connection and haunch connection system. The

above-mentioned values of PZ with WUF-B

connection provide more plastic rotation capacity

than the corresponding values for test specimens

with haunch system.

2. In the haunch connection system, PZ of test

specimen with one-sided haunch provide

insignificant plastic rotation capacity while for the

analytical model with two-sided haunch, PZ hardly

enter nonlinear phase and it behaves linearly. Also

for this connection type, slope of regression line is

less than others. The results of this study showed

that dependency of PZ plastic strain on shear ratio

Vy/VPZMy in this type of connection is less than

others, and presence of three or four continuity

plates respectively for one-sided haunch and two-

sided haunch, in front of haunch flange and beam

flange, reduces plastic rotation of PZ

3. 3- In the analytical models with cover plate

connection, as shown in Figs 19, analytical data

were scattered compared to other categories. Based

on analytical modeling of PZ derived by neural

network and finite element model, the cover plate

connections provide the minimum strength at least

equal to 1 Mp (full plastic moment). In the case of

analytical models with one-sided haunch, the

corresponding value at least reaches to similar

value compared to cover plate connection but the

maximum values reaches to 1.4 full plastic

moments.

4. The results of this study also showed that ductility

ratio of Panel zone, which was equal to 12y based

on FEMA 273[32] ( y is panel zone yielding shear

strain), is not a constant value. This value depends

on type of connection, for example in the case of

cover plate; ductility ratio is greater than the

corresponding value for analytical model of test

specimen with haunch connection system and

WUF-B connection.

5. The results of this research showed that the

proposed analytical modeling of the panel zone

result in total cyclic response of SMRF connections

with acceptable accuracy compared to the main

reference experimental responses and

corresponding analytical data generated by Finite

element modeling of this study.

9. Acknowledgments:

The authors would like to thank Dr. Ghobadi for his

invaluable comments to improve the manuscript and

Prof. D. Venkat Reddy Editor in chief of International

Journal of Earth Sciences and Engineering (IJEE) for

his kind cooperation for publication the article.

10. References:

[1] Naka, T., Kato, B., Watabe, M., and Nakato, M.

1969. “Research on the behaviour of steel beam-to-

column connections”. 4th World Conference of

Earthquake Engineering, Santiago, Chile.

[2] Fielding, D.J., and Huang, J.S. 1971. “Shear in steel

beam-to-column connections”. Welding Journal,

50(7): 313–326.

[3] Krawinkler, H., Bereto, V.V., and Popov, E.P.

1971. “Inelastic behaviour of steel beam to columns

subassemblagement”. EERC Report No. 71-7,

University of California, Berkeley.

[4] Bertero, V.V., Krawinkler, H., and Popov, E.P.

1973. “Further studies on seismic behaviour of steel

beam-to-column subassemblages”. EERC Report

No. 73-27, University of California, Berkeley.

[5] Becker, R. “Panel zone effect on the strength and

stiffness of steel rigid frames”. Engineering

Journal, AISC, 1975; 12(1): 19–29.

[6] Slutter, R. 1981. “Tests of panel Zone behavior in

beam column connections”. Report No.

200.81.403.1, Lehigh University, Bethlehem, Pa.

[7] Popov, E.P., Amin, N.R., Louis, J.J.C., and

Stephen, R.M. 1985. “Cycle behavior of large beam

column assemblies”. Earthquake Spectra, 1(2).

[8] Popov, E.P. 1987. “Panel zone flexibility in seismic

moment joints”. Journal of Constructional Steel

Research, 8: 91–118.

[9] International Conference of Building Officials

(ICBO). “Uniform building code”. 1988 ed. ICBO,

Whittier, Calif.

[10] American Institute of Steel Construction (AISC).

Seismic provisions for structural steel buildings.

1997 American Institute of Steel Construction,

Chicago.

[11] Tsai, K.C., and Popov, E.P. 1988. “Steel beam-

column joints in seismic moment resisting frames”.

UCB/EERC Report No. 89/19, University of

California, Berkeley.

[12] Federal Emergency Management Agency (FEMA).

1997. Interim guideline. Advisory No.1,




supplement to FEMA-267. FEMA-267A/Report

SAC-96-03, SAC Joint Venture, Sacramento, Calif.

[13] Lee, C.H., Jeon, S.W., Kim, J.H., and Uang, C.H.

2005. “Effects of panel zone strength and beam

web connection method on seismic performance of

reduced beam section steel moment connections”.

Journal of Structural Engineering, 131(12): 1854–

1865.

[14] Dubina, D., Ciutina, A., and Stratan, A. 2001.

“Cyclic test of double sided beam-to-column

joints”. Journal of Structural Engineering, 127(2):

129–136.

[15] Ciutina, A.L., and Dubina, D. 2006. “Seismic

behaviour of steel beam to-column joints with

column web stiffening”. Steel and Composite

Structures, 6(6): 493–512.


2000a. State of the art report on connection

performance. FEMA-355D, prepared by the SAC

Joint Venture for FEMA, Washington, D.C.

[17] Jin, J., and El-Tawil, S. 2005. “Evaluation of

FEMA 350 seismic provisions for steel panel

zones”. Journal of Structural Engineering, 131(2):

250–258.

[18] Hosseini Hashemi B, Ahmady jazany R.2012.

“Study of connection detailing on SMRF seismic

behavior for unequal beam depths”. Journal of

constructional steel research (JCSR), 68 (1) 150–

164.

[19] Hosseini Hashemi B, Ahmady jazany R.2010.

“Experimental evaluation of cover plate and flange

plate steel moment resisting connections

considering unequal beam depths”. Journal of

seismology and earthquake engineering (JSEE)

12(3):88-102.

[20] Ahmady jazany R, Hosseini Hashemi B.2012.

“Effects of Detailing on Panel Zone Seismic

Behaviour in Special Moment Resisting Frames

with Unequal Beam Depths”. Canadian journal of

civil engineering (CJCE); 39(4):388-401.

[21] Ahmady Jazany R. “study of panel zone seismic

behavior in SMRF for unequal beam depths”. PhD.

Thesis, International institute of earthquake

engineering and seismology (IIEES) No. 90-3-06;

Nov 2011.


1996. “Experimental investigations of beam-

column sub-assemblages”. Report SAC-96-01, Part

1, SAC Joint Venture, Sacramento, Calif.

[23] Popov, E.P., Blondet, M. M. Stepamov, L. and

Stoja dinovic, B 1996 “Full-scale Steel Beam

Coloumn Connection Tests”, SAC 96-01 Part 2.

SACjoint Venture.

[24] Chi WM, Deierilein AR. “Fractures toughness

demands in welded beam –column moment

connections”. J Struct Engrg, ASCE 2000; 126(1).

[25] El-Tawil S. “Panel zone yielding in steel moment

connection” .Engrg J AISC 2000; 3qrt00:120.31.

[26] Popov, E.P, Amin, N.R.Louiej. j.c, and Stephen,

1985. “Cyclic behavior of Large Beam to Column

assemblies” Earthquake Spectra, 1 (2):203-238.

[27] Qi -Song, Uang Chia-Ming, Gross J. “Seismic

rehabilitation design of steel moment connection

with welded haunch”. Journal of Structural

Engineering 2000; 126: 69–78.

[28] ANSYS. User's Manual, Version 5.4. 201. Johnson

Road, Houston: ANSYS Inc.; 1998.

[29] ATC 24. Guidelines for cyclic seismic testing of

components of steel structures. Applied

Technology Council; 1992.

[30] AWS. Structural Welding Code — Steel. Miami

(FL): American Welding Society; 2002.

[31] Ghaboussi, J, & wu, X.soft “computing with neural

network for engineering application”. Fundamental

issues and adaptive approaches, structural

engineering and mechanics 1998, 6(8), 955-969.

[32] FEMA. “Provion for rehabilitation of exsiting

structutes”. Federal management agency, FEMA

273.2002.

[33] Whan Han S, Kwon Gun Up, Moon K.H. “Cyclic

behavior of post Northridge WUF-B connections”

Journal of constructional steel research 2007; 12

(5): 365-374.

[34] Noel, s Yang, c.-m. “Cyclic Testing of Steel

Moment connection s for the for the San Francisco

civic center” Report NO .TR-96/07 ,Division of

structural engineering ,University of California, San

diego,1996

[35] Chia-Ming Uang, Duane Bondad, Cheol-Ho Lee.

“Cyclic performance of haunch repaired steel

moment connections”. Experimental testing and

analytical modeling Engineering Structures 1998;

20:552–61.

[36] Civjan SA, Engelhardt MD, Gross JA. “Retrofit of

pre-Northridge moment-resisting connections”.

Journal of Structural Engineering 2000; 126:445–

52.

[37] Schinder, S.P. and Amidi A. “Seismic Behavior of

Steel Frames with Deformable Panel Zone”.

Journal of Structural Engineering ASCE 1998:124

(1):35-42

[38] Yang, popov E.”Design of Steel MRF before and

After 1994 Northridge Earthquake”. Proceedings;

International Conference on Advance in Steel

Structures,Hong Kong, 1996.

[39] Gupta, Krawinkler h. “Seismic demands for

performance evaluation of steel moment resisting

frames structures”. Blume Engineering Research

Center,Rep No132 Dept . stanford university;1999.

[40] Parakash v, Powell GH, DRAIN-2DX Base

Program Description and user Guide ver1.10.


http://www.researchgate.net/researcher/72151919_Gun_Up_Kwon/

http://www.researchgate.net/researcher/72862138_Ki_Hoon_Moon/

Modeling of Panel Zone in Accordance to the Type of Connections in SMRF

Documents