Sie sind auf Seite 1von 177

6)

.<

ANALYSIS OF FRAMES WITH PARTIALLY


RESTRAINED CONNECTIONS

BY

. RAISAHMAD

DEPARTMENT OF CIVIL ENGINEERING


BANGLADESH UNIVERSITY OF ENGINEERING & TECHNOLOGY (BUET)

March, 1999.

lilliimill 11111111111
1111III
#93299#
ANALYSIS OF FRAMES WITH PARTIALLY
RESTRAINED CONNECTIONS

BY

RAISAHMAD

A thesis submitted to the Department of Civil Engineering of


Baugladesh University of Engineering & Technology, Dhaka
in partial fulfillment of the degree of

MASTER OF SCIENCE IN CIVIL ENGINEERING

March, 1999.
ANALYSIS OF FRAMES WITH PARTIALLY
RESTRAINED CONNECTIONS
. BY

RAISAHMAD

Thesis approved as to the style and content for the degree ofM.Sc. Engineering (Civil).

~. ~
Dr. Md. Zakana Ahmed Chairman
Associate Professor (Supervisor)
Department of Civil Engineering,
BUET, Dhaka-lOOO.

Dr. Md. Hossain Ali : Member


Professor and Head
Department of Civil Engineering,
BUET, Dhaka-lOOO.

Dr. Ahsanul Kabir Member


Professor
Department of Civil Engineering,
BUET, Dhaka-lOOO.

Dr. Muhammad Mahbubul Alam Member


Associate Professor (External)
Department of Mechanical Engineering,
BUET, Dhaka-lOOO.
DECLARATION
/

I hereby certifY that the research work, embodied in this thesis, has been performed by the
author under the supervision of Dr. Md. Zakaria Ahmed, Associate Professor of the
Department of Civil Engineering, BUET. Neither this thesis nor any part of it has been
submitted or is being concurrently submitted else where for any other purposes (except
for publication).

January, 1999
RaisAhmad

.!
ACKNOWLEDGEMENT

The author wishes to express his profound gratitude and sincere appreciation to Dr. Md.
Zakaria Ahmed for his continuous guidance, invaluable suggestions and affectionate

encouragement at every stage ofthis research work.

The author is indebted to the Civil Engineering Department, BUET for its facilities that
were enjoyed by the author. The author also wishes to thank Mr. AsifIslam Khan and Mr.
Sirajul Islam for his kind help in preparing the computer aided drawings. The author
wishes to thank all his lecturer colleagues for their kind help and moral support.

The author wishes to dedicate this work to his mother and all the members of his family

who have been a constant source of encouragement.

ii
ABSTRACT

Steel structures have long been widely used in different countries because of their relative
economy and ease of construction. The connections of the constituent members of these
types of constructions play an important role in affecting the behavior of the structure. In
the conventional methods of steel frame analysis, the behavior of joints is represented by
idealized models, either as a rigid joint or a pinned joint. The available design techniques
are mostly based on these idealized conditions. In reality, most connections are semi-rigid
and possess certain amount of rotational stiffuess. They are termed as partially restrained
connections or semi-rigid connections. Experimental studies have shown that the
connection moment-rotation characteristics are non-linear and the analysis procedure
must follow non-linear approach.

A computet program capable of analyzing frames with semi-rigid connections has been
developed. The program is a non-linear type, iterative in nature and based on incremental
or stepwise loading. The program is a generalized one and can be used to analyze frames
with any type of known semi-rigid connections. The program can handle general types of
loading implementing through a series of load increments. The connection flexibility
effect has been incorporated by the matrix formulation given by Monforton and Wu. The
non-linear behavior of the connections has also been incorporated by the polynomial
moment-rotation relationship given by Frye and Morris polynomial model for different
types of connections.

With the help of the developed program several investigations have been performed to
study the behavior of steel frames with semi-rigid connections. The first study describes
the effect of flexible connections on the behavior of frames, its internal distribution 'of
forces and lateral drift. The effect of bracing, as a sway resisting system on high rise
building frames have also been studied. The effect of bracing on the internal force
distribution and in controlling the sway of frame has been studied. The behavior of frame
with mixed use of rigid and semi-rigid connections has also been encountered. Efforts
have been directed to find the best-suited orientation of mix-mode c~a steel
frame in this investigation. A comparison of the results ~s obtained by the non-linear
analysis of frame with those of linear analysis using single stiffuess of connection has

iii
also been included in this work. Finally, the effect of connection size parameters of semi-
rigid connections on the behavior of frame has also been studied ..

IV
CONTENTS

Page

DECLARATION
ACKNOWLEDGEMENT II

ABSTRACT 1II

CONTENTS v
LIST OF FIGURES IX

LIST OF TABLES Xlii

LIST OF NOTATIONS XV

CHAPTER 1 : INTRODUCTION 1

1.1 GENERAL I
1.2 OBJECTIVES 5
1.3 SCOPE OF THE STUDY 6
1.4 OUTLINE OF THE THESIS 7

CHAPTER 2 : LITERATURE REVIEW

2.1 GENERAL 8
2.2 BEHAVIOR OF CONNECTIONS 8
2.2.1 Causes for the Non-Linear Behavior of Connections 12
2.3 TYPES OF SEMI-RIGID CONNECTIONS 13
2.4 MODELING OF PARTIALLY RESTRAINED CONNECTIONS 17
2.4.1 Linear Model 17
2.4.2 Polynomial Model 18
2.4.3 B-Spline Model 19
2.4.4 Power Model 20
2.4.5 Exponential Model 22
2.4.6 Data Base Prediction of Connection Behavior 23

v
Page

2.4.7 COlmection Behavior Based on Finite Element Model 25


2.5 METHODS OF INCORPORATING CONNECTION 26
FLEXIBILITY
2.5.1 Beam-line Method 27
2.5.2 Modified Methods 28
2.5.2.1 Modified Slope-Deflection Method 28
2.5.2.2 Modified Moment-Distribution Method 29
2.5.3 Matrix Methods 30
2.6 NON-LINEAR ANALYSIS OF SEMI-RIGIDLY
CONNECTED PLANE FRAME 32
2.7 CONCLUDING REMARKS 32

CHAPTER 3 : ANALYTICAL MODELING 34

3.1 GENERAL 34
3.2 CONNECTION MODELING 34
3.2.1 Assumptions and Limitations of Frye and Morris 35
Polynomial Model
3.2.2 Standardization of Moment -Rota tion Curves with 35
Polynomial Model
3.3 STIFFNESS OF BEAM-COLUMN ELELMENT WITH 40
FLEXIBLE CONNECTIONS
3.3.1 Beam-Column Element 40
3.3.2 Connection Element 41
3.3.3 Modification of Beam-Column Element for Flexible 43
Connections
3.3.4 Generation of Correction Matrix 44
3.3.5 Modified Stiffuess Matrices 47
3.4 ITERATIVE ANALYSIS PROCEDURE 50
3.4.1 Incremental Loading Procedure 50

vi

;

Page

3.4.2 Convergence Criteria 52


3.4.3 Analysis Module 53
3.5 CONCLUDINGREMARKS 53

CHAPTER 4 : COMPUTER PROGRAMMING 54

4.1 GENERAL
54
4.2 THE SOLUTION SCHEME
54
4.3 COMPONENTS OF THE COMPUTER PROGRAM
55
4.3.1 The Main Program
55
4.3.2 The Subroutines
59
4.3.2.1 Subroutine INPUT
59
4.3.2.2 Generation of the Modified Element stiffuess
60
Matrix
4.3.2.3 Subroutine PLOAD
61
4.3.2.4 Subroutine STRESS
64
4.3.2.5 Subroutine ICONY
64
4.4 CONCLUDING REMARKS
65

CHAPTER 5 : BEHAVIORAL STUDY ON PARTIALLY RESTRAINED 66


FRAMES

5.1 GENERAL 66
5.2 VERIFICATION OF THE COMPUTER PROGRAM 66
5.3 EFFECT OF SEMI-RIGID CONNECTIONS ON THE 69
DISTRIBUTION OF FORCES AND LATERAL SWAY
5.3.1 Effect under Gravity Loading 72
5.3.2 Combined Action of Gravity and Lateral Loading 74
5.3.3 Summary 89
5.4 EFFECT OF BRACING IN SEMI-RIGID FRAME
91
5.4.1 Effect of Bracing
93

vii

\.
Page

5.4.2 Summary 100


5.5 COMBINED USE OF SEMI-RIGID AND RIGID CONNECTIONS 107
5.5.1 Behavior of Frames Using Mix-Mode Connections 109
5.5.2 Summary 114
5.6 SEMI-RIGID FRAME USING SINGLE STIFFNESS 115
LINEAR MODEL
5.6.1 Comparison under Gravity Loading 117
5.6.2 Comparison under Combined Action of Gravity 121
and Lateral Loading
5.6.3 Summary 128

~
CHAPTER 6: PARAMETRIC STUDY OF CONNECTIONS 129

6.1 GENERAL 129


6.2 SCOPE OF THE PARAMETRIC STUDY 129
6.3 PARAMETRIC STUDY ON DOUBLE WEB ANGLE 131
CONNECTION
6.4 PARAMETRIC STUDY ON TOP & SEAT ANGLE 135
CONNECTION
6.5 PARAMETRIC STUDY ON T-STUB CONNECTION 138
6.6 CONCLUDING REMARKS 143

CHAPTER 7 : CONCLUSIONS AND RECOMMENDATIONS 144

7.1 GENERAL 144


7.2 SUMMARY OF CONCLUSION 144
7.3 RECOMMENDATIONS FOR FUTURE WORKS . 146

REFERENCES 147

APPENDIX A: Sample data file used for the analysis of the frames. 153

viii
LIST OF FIGURES

Page

Fig.2.l Effect of end restraint on moment 9


Fig.2.2 Rotational deformation of a connection 9
Fig.2.3 Moment-rotation curves of different types of semi-rigid connections 10
Fig.2A Moment-rotation behavior of semi-rigid connection II
Fig.2.5 loading/unloading characteristics of connections 12
Fig.2.6 (a) Different types of soft semi-rigid connections 14
(b) Different types of intermediate stiff and stiff semi-rigid connections IS
Fig.2.7 (a) Beam with equivalent rotational spring 17
(b) Linear M -S, models 17
Fig.2.8 Behavior ofthe Ang-Morris power model (Adapted from Ref.29) 21
Fig2.9 Beam-line method 27

Fig.3.1 Family of moment-rotation curves for connections with different 37


values ofpj
Fig.3.2 Comparison of moment-rotation curves for Double web angle connection 38
Fig.3.3 Beam-column element in member and structure coordinates 41
Fig.3A Connection element and connection loading I unloading behavior 42
Fig.3.5 (a) Conjugate beam for semi-rigid connection 44
(b) Sign convention 45
(c) Forces acting at the ends of a member (Monforton and Wu [18]) 45
Fig.3.6 Modification of connection flexibility (Adapted from Ref.25) 52

FigA.I Flow diagram of the main program


56
FigA.2 Flow chart ofthe main program
58

Fig.5.1 Verification of the computer program


67
Fig.5.2 (a) Gravity loading on a 2 bay frame
70
(b) Gravity and lateral loading on 2 bay and 3 bay frames 70
Fig.5.3 Critical moment ratio of beams for gravity load 73
(a) with story level, (b) with connection types
Fig.5A Critical moment ratio of beams with story level for 75
(a) for 2 bay frame, (b) for 3 bay frame.

ix
Page

Fig.5.5 Critical moment ratio of beams with connection types for 76


gravity and lateral loads (a) for 2 bay frame, (b) for 3 bay frame.
Fig.5.6 Critical moment ratio of exterior column with story level 79
for gravity and lateral loads (a) for 2 bay frame, (b) for 3 bay frame.
Fig.5.7 Critical moment ratio of exterior column with connection types 80
for gravity and lateral loads (a) for 2 bay frame, (b) for 3 bay frame.
Fig.5.8 Design moment ratio of exterior column for gravity and 81
lateral loads (a) for 2 bay frame, (b) for 3 bay frame.
Fig.5.9 Critical moment ratio of interior column with story level for 85
gravity and lateral loads (a) for 2 bay frame, (b) for 3 bay frame.
Fig.5.l0 Critical moment ratio of interior column with connection types 86
for gravity and lateral loads (a) for 2 bay frame, (b) for 3 bay fClune.
Fig.5.ll Design moment ratio of interior column for gravity and 87
lateral loads (a) for 2 bay frame, (b) for 3 bay frame.
Fig.5.12 !\ / H with connection types 88
Fig.5.l3 Effect of bracing 92
Fig.5.l4 Critical beam moments (a) for braced frame, (b) for unbraced frame 94
Fig.5.15 Moment reduction factor (M.R.F.) for beams. 95
Fig.5.16 Critical moments for column 1 (a) for braced frame 96
(b) for unbraced frame.
Fig.5. I 7 Critical moments for column 2 (a) for braced frame 97
(b) for unbraced frame.
Fig.5.18 Critical moments for column 3 (a) for braced frame 98
(b) for unbraced frame.
Fig.5. I 9 Critical moments for column 4 (a) for braced frame 99
(b) for unbraced frame.
Fig.5.20 Design moment for exterior columns (a) for braced frame, (b) for 101
unbraced frame.
Fig.5.21 Moment reduction factor (M.R.F.). 102
Fig.5.22 Design moment for interior columns (a) for braced frame, (b) for 103
braced frame.
Fig.5.23 Moment reduction factor (M.R.F.). 104
Fig.5.24 !\ / H for different types of connections (for both 105
braced and unbraced frames).
Fig.5.25 Different cases of mixed use of semi-rigid and rigid connections. 108

x
Page

(a) Critical beam moments for different cases, 110


Fig.5.26
(b) M.A.F. for different cases.
(a) Exterior column design moments for different cases, 111
Fig.5.27
(b) M.A.F. of exterior column for different cases.
(a) Interior column design moments for different cases, 112
Fig.5.28
(b) M.A.F. of interior column for different cases.
113
Fig.5.29 '" / H for different cases.
Building frame used for linear analysis 116
Fig.5.30
Beam moments for gravity loads at story level 1 118
Fig.5.31
(a) for positive moments, (b) for critical moments.
Percentage of under or over design of beam positive 119
Fig.5.32
moments for gravity loads (a) with connection types, (b) with
story levels.
Percentage of under or over design of beam positive 120
Fig.5.33
Moments for gravity loads (a) with connection types, (b) with
story levels.
(a) Critical beam moments for gravity and lateral loads, 122
Fig.5.34
(b) Exterior column critical moments at story levell,
(c) Iinterior column critical moments at story level 1.
Percentage of under or over design of beam critical 123
Fig.5.35
moments for gravity and lateral loads (a) with connection types,
(b) with story levels
Percentage of under or over design of exterior column 124
Fig.5.36
critical moments for gravity and lateral loads (a) with connection
types, (b) with story levels.
Percentage of under or over design of interior column 125
Fig.5.37
critical moments for gravity and lateral loads (a) with connection
types, (b) with story levels.
Percentages of under or over design of columns 126
Fig.5.38
(a) for exterior columns, (b) for interior columns.
127
Fig.5.39 '" / H with connection types

Building frame as used for parametric study 130


Fig.6.1
(a) Critical beam moment with angle depth 132
Fig.6.2
(b) Critical column moment with angle depth for
double web angle connection.
'" / H with angle depth for double web angle connection. 134
Fig.6.3

Xl
Page

Fig.6.4 (a) Critical beam moment with angle length


136
(b) Critical column moment with angle length for
top & seat angle with double web angle connection.
Fig.6.S 1'./ H with angle length for top & seat angle
137
with double web angle connection.
Fig.6.7 (a) Critical beam moment with length ofT-stub (b) Critical 139
column moment with length of the T-stub for T-stub connection.
Fig.6.8 1'./ H with length ofthe T-stub for T-stub connection. 140
Fig.6.9 (a) Critical beam moment with fastener diameter 141
(b) critical column moment with fastener diameter for T-stub connection.
Fig.6.9 1'./ H with fastener diameter for t-stub connection 142

xii
LIST OF TABLES

Page

Table 2.1 Slope-deflection equation coefficients. 29


Table 2.2 Comparison of moment-distribution factors. 30
Table 3.1 Standardized connection moment-rotation function [Frye and Morris). 39
Table 5.1 Comparison of support reactions and top deflections using Rigid 68
and T-stub connection.
Table 5.2 Comparison of moments in members using Rigid and T-stub connection. 68
Table 5.3 Connection size parameters. 71
Table 5.4 Loading on the frames. 71
Table 5.5 Maximum moments (kip-in) in beams of2 bay 4 story frame 74
for gravity loading.
Table 5.6 Maximum moments (kip-in) in beams of2 bay 4 story frame 78
for gravity and lateral loading.
Table 5.7 Maximum moments (kip-in) in beams of 3 bay 4 story frame 78
for gravity and lateral loading.
Table 5.8 Maximum moments (kip-in) in columns of2 bay 4 story franle 83
for gravity and lateral loading.
Table 5.9 Maximum moments (kip-in) in columns of3 bay 4 story frame 83
for gravity and lateral loading.
Table 5.10 Percentages of over estimation in the design moment of beams 90
using Type2 analysis.
Table 5.11 Percentages of over estimation in the design moment of columns 90
using Type 2 analysis.
Table 5.12 Loading on the frame. 91
Table 5.13 Connection size parameters. 93
Table 5.14 Percentages of reduction in member forces (moments). 106
Table 5.15 Connection size parameters. 107
Table 5.16 Loading on the frame. 109
Table 5.17 Connection size parameters. 117
Table 5.18 Loading on the frame. 117
Table 5.19 Comparison oflinear analysis with rational analysis for different 128
types of connections for the design of beams.

xiii
Page

Table 5.20 Comparison oflinear analysis with rational analysis for different 128
types of connections for the design of columns.
Table.6.1 Connection size parameters. 131
Table.6.2 Loading on the frame. 131

xiv
LIST OF SOME COMMON NOTATIONS

Ax Area of the cross section


CAB, CBA Carry over factors
CI, C1, C3 Curve fitting constants
d Depth of the beam
E Young's modulus of elasticity of a material
Fy Yield stress of material
g Gage length
I Moments of inertia of a section
K Standardization constant
L Length of the member
I, Length of the top angle
M Developed moment
AI' Fixed end moment
Initial Moment
Ultimate moment
M-Br Moment-rotation relationship
p Pitch
Initial StifJiless
SE Secant Stiffness of connection
Stiffness Factors
Thickness of the top angle
Column flange thickness
End plate thickness
Thickness of the web angle
w Intensity of uniformly distributed load
VI, fj, YI. Jj Shear force at the member ends
XI. X; Axialforce
ZA, ZB Semi-rigid connection factors

to,':'

xv
Greek Symbols

a Inclination of the member


Top lateral deflection of a frame
(J Stress
rotation of connection
Relative rotation of connection

Change in moment during a load increment


Change in relative rotation during a load increment
fixity factors
Semi-rigid connection factor
e Tolerance limit

Matrices and Vectors

[ke] Element stiffness matrix


[ke]", • [kf], [kif] Modified stiffness matrix
[C], Correction matrix
[FIl Force matrix
[DIl Deformation matrix
[kf ]", Modified stiffness matrix
[T] Transformation matrix
[Tf Transpose of the Transformation matrix
[K] The structure stifjiless matrix
[S] Summation of structure stiffness matrices
{LIP! } Incremental load vector
{LIDJ} Incremental displacement vector

Abbreviations

AISC American Institute of Steel Construction


ASD Allowable Stress Design
BNBC Bangladesh National Building Code

xvi
FR Fully restrained
LRFD Load and Resistance Factor Design
MA.F Moment Amplification Factor
MR.F. Moment Reduction Factor
PR Partially restrained

xvii
CHAPTER-l
INTRODUCTION

........ _. __ ..

1.1 GENERAL

For many years civil engineering design was mainly an art based on intuition and
experience. The development of applied mechanics toward the end of the nineteenth
century gradualIy introduced the use of mathematical operations and more and more
sophisticated mathematical models into analysis and design. Steel structures have long
been used in different countries because of their relative economy and ease of
construction. The steel frame performs as an integrated structure deriving its strength
from individual elements that are properly selected to transmit loads through the structure .
.
"

Since connections are integral parts of a steel frame, their behavior has an effect on the
frame's performance. In customary steel frame analysis, assumption is made that beam-
column connections are either fulIy fixed or idealIy pinned. The rigid joint assumption
implies that fuII slope continuity exists between the adjoining members and that the filII
or a .substantial percentage of gravity moment is transferred from the beam to the column.
On the other hand, the assumption of idealIy pinned connections implies that the beams
wiII behave as simply supported members and that the columns will carry no gravity
moments from the beams. Although the assumption of fulIy rigid or idealIy pinned
connection behavior drasticalIy simplifies the analysis and design procedures, the validity
of these assumptions may be questionable for cases in which the rigidities of the
connections are intermediate between the fulIy rigid and idealIy pinned cases. In reality,
experimental studies [1,2] have shown that the connection moment~rotation
characteristics are non-linear over the entire range of loading for almost alI types of
connections. This means, most connections are semi-rigid, non-linear in nature and faII
under the PartialIy Restrained (PR) type construction category. These PR type
connections possess a certain amount of rotational stiffuess depending upon the amount.of
moment that can be mobilized at the joint. The most obvious of a design using semi-rigid
connections is that beam moments are reduced leading to lighter beams. When a beam
with simple connections is loaded, the span moment is critical, whereas when rigid
connections are assumed, the end moments are critical for design. If semi-rigid
connections are assumed, these two moments may be more nearly balanced. Another
possible source of accuracy lies in the columns where a better understanding of actual
restraint conditions and end moments may well lead to more rationally based conservative
methods of design. Thus, a more economical and conservative design would result if the
effect of semi-rigidity of connections was included in analyzing a frame.

Realizing the potential influence of connection flexibility on frame design, The American
Institute of Steel Construction (AISC) [3,4] has introduced provisions to allow designers
to consider explicitly the behavior of connections in the design of structural steel frames.
The Allowable Stress Design (ASD) Specifications (AISC, 1989) [3], for example, list
three types of constructions.

I. Type 1 or Rigid Framing: This type of construction assumes that the beam-
column connections have sufficient rigidity to maintain the original geometric
angle between intersecting members. Type I connections are assumed for elastic
structural analysis.

2. Type 2 or Simple Framing: This type assumes that the beam and girder
connections transfer only vertical shear reactions without bending moment when
the structure is loaded with gravity loads. In other words, beam-to-column
connections are assumed to be pinned under gravity loads and the connections are
allowed to rotate without restraint. But for lateral loading, Type 2 frame analysis
assumes moment resistant (rigid) connections at the beam-to-column joint.

3. Type 3 or Semi-rigid Framing: This type assumes that the cOimections can
transfer both vertical shear as well as some moment. The connections are
considered to have moment capacity in between complete fixity and the pm
condition.

The Load and Resistance Factor Design (LRFD) Specifications (AISC, 1986) [4],
designate two types of constructions in their provisions.

2
1. Type FR:. Type FR (fully restrained) corresponds to ASD Type 1.

2. Type PR.: Type PR (partially restrained) includes Type 2 and Type 3


constructions of ASD Specifications. If Type PR construction is used, the effect of
connection flexibility must be taken into account in the analysis and design
procedures.

The Bangladesh National Building Code (BNBC, 1993) [5] recognizes FR and PR types
of construction. The BNBC requires that the use of PR construction depend on the
evidence of predictable proportion of full end restraint. If the connection restrained is
ignored, i.e. for simple framing, it is assumed that under gravity loads the ends of the
beams and girders are connected for shear only and are free to rotate. When .the rotational
restraint of the connections is used in the design of the connected members or for the
stability of the structure as a whole, the capacity ofthe connections for such restraint must
be established by analytical or empirical means.

The problem with using the Type 2 frame analysis (AISC-ASD Specification) is that the
connections, irrespective of types, are considered to have adequate stiffuess to provide
resistance against lateral load, while end restraints are ignored for gravity load. This
method of analysis under estimates the sway and results in an under design for the column
and over design for the beams of a structure. The use of rigid or pinned connections
simplifies the analysis but by no means accounts for the actual behavior of the structure.
Since the actual stiffuess of the connections, other than rigid or pinned, is ignored, the
method would yield the same force distribution for all types of semi-rigid connections.
On the other hand, although design procedure for FR construction is available in the
Specification (AISC - LRFD), no specific design procedure for PR type of construction
has been recommended yet. In order to overcome these difficulties, a realistic analysis
method, which accounts for the connection flexibility effect, needs to be employed for the
design of semi-rigid frames.

The effect of connection flexibility can be included if the practical moment-rotation


relationship or M-El, curve of a connection is defined. A quite good number of

3
experimental studies have been carried out to predict the connection moment-rotation
relationship [1, 2, 6, 7, 8, 9,10,11, 12, 13, 14, 15, 16]. Using these experimental test data,
the researchers developed analytical or mathematical moment-rotation models of partially
restrained connections.

In the early investigations, the single stiffness linear models were proposed by the
researchers [17, 18, 19] for flexibly jointed frame. The initial stiffness of the moment-
rotation curve is utilized to represent the connection behavior for the entire range of
loading. This model may be acceptable up to serviceability state of loading, but beyond
this limit, the model becomes unacceptable. Bi-linear [20, 21], tri-linear [22], and piece
wise linear models [23] were used to obtain a better representation of connection
behavior. All these linear models are easy to use, but the inaccuracies and the sudden
jump in stiffness at the transition points make their practical use difficult. Analytical or
mathematical models using the curve fitting techniques to experimental data were
introduced to describe the M-8, relationship. Sommer [24], for example, fitted
experimental moment-rotation data'to determine standardized moment-rotation curves'in
the form of non-dimensional polynomial series. This research work was carried out for
header plate connection. Frye and Morris [25], expanded this to predict the behavior of
other types of connections. Analytical expressions developed by Frye and Morris give
good agreement with the original experimental curves. The main advantage of this model
is that, a wide variety of connection behaviors can be represented by this model.

Jones et al. [26] used the B-Spline representation of the experimental moment rotation
data of connections. A large number of data are required in this curve fitting process.
Three parameter and four parameter power models were developed by the researchers [27,
28, and 29] to fit the connection moment-rotation curve. Vee and Melchers [30] and Wu
and Chen [31] proposed four parameter and three parameter exponential models
respectively to fit the connection M-8, curve. Several researchers developed analytical
representations of connection behavior using finite element techniques [32, 33].

The connection database, which is a collection of experimental tests for several types of
beam column connections, was also employed in assessing the performance of steel frame

4
[34]. In order to analyze connection using data base all the details and dimensions of the
beam, the column and the connection have to match exactly with the connection for
which the behavior is compiled in the data base. But, all connections with all possible
details and dimensions have not been tested and documented and the inclusion of
enormous test data is a troublesome task. In this aspect, a simple analytical model is more
advantageous over the connection database to predict non-linear behavior of connections.

Several research works have been carried out to study the behavior of partially restrained
frames [35, 36, 37, 38, 39, 40). But these works were performed for the low-rise multi-
storied (one or two story) frames or flexibly connected sub assemblage. Dhillon and
Majid [41) developed structural analysis program considering connection flexibility of
steel frames. But this study was focussed on the performance of the frame with End-plate
with column stiffeners and T-stub connection i.e. stiff semi-rigid connections only.

Commercial software or packages as available in the design office for structural analysis
do not account the effects of connection flexibility. In other words, the available packages
can be used only for rigid frame analysis. This difficulty puts forward a major problem
towards the use ofPR construction recommended by the codes [3, 4, 5). Thus a computer
program, having the capacity to analyze steel frames with partially restrained connections,
is desired in modem office practice. Moreover, a detail investigation for wide variety'of
commonly used semi-rigid connections and loading conditions, is required to evaluate the
behavior of individual members and complete frame. This work will be useful to develop
simplified analysis methods for PR frames.

1.2 OBJECTlVES

The aim of this research work is to develop a computer program using an analytical
model for the analysis of steel frames including connection flexibility effects. This would
help engineers to analyze and design steel frames accurately and economically taking
connection flexibility effect into consideration.

Thus the objectives of the research work are:

\
• To incorporate an analytical connection model in a frame analysis program to enable
treatment of non-linear behavior of connections in steel frame structures.
• To investigate the distribution of internal forces in the members and the load-
deflection behavior of the frame for different types of connections and loading
conditions.

• To study the effect of bracing in partially restrained frame.


• To study the behavior of frames with rigid and semi-rigid connections.
• To investigate the effects of connection size parameters such as depth, angle thickness
etc. of different types of connections.

1.3 SCOPE OF THE STUDY

The program as developed in this study is applicable to 2-D frame. This program is a non-
linear, iterative type based on incremental or stepwise loading. The effect of flexible
connections at the ends of a beam element is incorporated in the analysis according to the
matrix method given by Monforton and Wu [18]. The polynomial model of moment-
rotation relationship [10, 25] for different types of connections is implemented in the
analysis. The type of connections included in this research work are Double web angle,
Header plate, Top and seat angle, Top and seat angle with double web angle, End plate,
T-stub and perfectly Rigid connections. The stiffhess of the connection element within a
load increment is derived as the secant to the moment rotation curve. In other words, the
connection element is represented by a rotational spring with stiffuess equal to change in
moment over the change in relative rotation for the load increment under considerati,?n.
The complete solution for a given load is implemented through a sequence of load
increments with a provision to update the stiffness matrix at the start of each iterative
cycle.

This program is based on static loading such as gravity and lateral loading. No provision
is made for cyclic or dynamic loading in the program. While developing the program, it is
assumed that the members are ideal i.e. no residual stresses and initial imperfections exist
in the members.

6
1.4 OUTLINE OF THE THESIS

The thesis consists of seven chapters. Apart from this chapter, the reminder of the thesis
has been divided into another six chapters.

Chapter 2 presents a literature review covering research works in the field of semi-rigid
connections. Chapter 3 consists of brief description of the methodology of this research
work. Analytical modeling of connections, generation of element stiffuess matrix

including connection flexibility effects has been described in this chapter. ..-

Chapter.4 includes the flow chart of the computer program developed in the research
work. Definition of different subroutines and how they work have been pre~ented in this
chapter. A brief description of input and output files has also been presented.

Chapter 5 presents different studies performed in this research work. Studies include,
behavior of frames with semi-rigid connections, effect of bracing on the behavior of
partially restrained frames, behavior of frames with mixed use of semi-rigid and rigid
connections and linear analysis using single stiffuess of connection. Graphs and tables for'
these studies and their related discussions have been presented in this chapter. Chapter 6
includes the parametric study of the connections. Effect of different size parameters of the
connection on the behavior of frames has been presented in this chapter.

Finally, Chapter 7 includes the conclusion part. The findings of the studies and their
recommendations are presented in this chapter. Apart from the regular chapters there is an
appendix. In Appendix A, a sample data file used in this research work is included.

7
CHAPTER-2
LITERATURE REVIEW

2.1 GENERAL

A connection is a medium through which forces and moments are transmitted from one
member to another. For a beam-to-column connection, a general set of forces that is
transmitted includes axial force, shear force, bending moment and torsion. For an in plan
study, the effect of torsion can be neglected. Furthermore, for most connections, the axial
and shearing deformations are very usually small compared to the rotational deformation.
For practical purposes, only the connection's rotational deformation needs to be
considered. In customary steel frame analysis, assumption is made that the beam-column
connections are either pinned or fixed. These assumptions are not entirely consistent with
the conditions often encountered in practice. In reality most connections are semi-rigid
and possess a certain amount of rotational stiffness. In other words, most connections fall
under the partially restrained (PR) type construction category. Fig.2.1 shows the moment-
rotation diagram of a beam subjected to uniform loading with three different support
conditions i.e. pinned, partially restrained and fixed. It can be noted from the figure that
the design moment of a beam with partially restrained end is always less than that of
pinned or fixed end. Since most of the connections are partially restrained it is logical to
design the beam as partially restrained member. Experimental studies [1,2] reveal that the
connection moment-rotation (M-8,) characteristics are non-linear over the entire range of
loading for almost all types of connections. In the following sections the behavior and
modeling of connections are discussed. Several analysis methods to incorporate the effect
of connections are also included.

2.2. BEHAVIOR OF CONNECTIONS

The rotational deformation (8,) is customarily expressed as a function of the connection


moment ( M ). The rotation (8,) represents the change in angle between the beam and the
column from its original configuration. Fig.2.2 shows the rotational deformation of a
connection.

.;..
5wL'
~ .........'3;;e1
~._-~~-~

QJJJIITI , <-- '.l


••••• ~!l~EI

f-------f

p:LIIIIq. ... ...•. ~


I

Fig. 2.1 Effect of end restraint on moment.

8,

Fig.2.2 Rotational defom1ation of a connection.

9
TOF' .3. SEAT ANGLE

HE.o\OER ?LATE

DOUBLE 'NE3 ANGL'"

$lflGL!: WE3 ANGLE

ROTATION H.

Fig.2.3 Moment-rotation curves for different types of semi-rigid connectIOns [Ref. 25]

Over the past decades, a large number of investigations into the behavior of beam-to-
column connections have been reported. The flexural behavior of a connection is best
described by the M-8, relationship. Fig.2.3 shows schematically the M-8, behavior of a
wide variety of commonly used semi-rigid connections. Several observations can be made
from this figure.

I. All types of connections exhibit M-8, behaviors that falls between the extreme
cases of ideally pinned (the horizontal axis) and fully rigid (the vertical axis)
conditions.

11. For the Same moment, the more flexible the connection, the larger the value of 8,.
is. Conversely, for a specific value of 8" a more flexible connection will transmit
less moment between the adjoining members.

111. The maximum moment that a connection can transmit decreases with more
flexible connection.

10
IV. The M-8, relationships for the semi-rigid connection are typically non-linear over
virtually the entire range ofloading.

In general, steel frame connections are very ductile. The connection curves show
pronounced strain hardening. The angular deformation extends to a relatively large
rotation displaying no sign of brittle failure. In fact, some connections show an abrupt rise
of stiffness due to contact between beam end and column flange. This type of situation is
likely to occur after a substantial rotation. If no serious defect exists in the fabrication of
the connection, the ductility characteristic is almost sure to exist. Connection ductility is a
significant quality as far as the ultimate capacity of the structure is concerned, particularly
for those designed to resist seismic forces.

When a connection is subjected to a moment, the connection rotates according to the


curve shown in Fig.2.4. However, if the direction of the moment is reversed, the
connection will unload and follow a different linear path with a slope equal to the initial
slope, Rk;, of the M-8, curve. This characteristic can best be illustrated by an example ofa
simple portal frame shown in Fig.2.5, subjected to gravity load only. Under the action of
the gravity load, the connections at the ends of the beam will experience a moment of Mo.
Now, if a lateral load is applied to the frame, the leeward connection will continue to
load, but the windward connection will unload. As a result, the apparent stiffness of the
connections under the action of this lateral force will be different.

loading

/'

~I
z'
~I
."" I Rkt - Tangent Sliffness

i
!
ROTATION. 8,
----~-~--

Fig.2.4 Moment-rotation behavior of semi-rigid connection.

11
I
I
----- - \
I
I
I
9. I I
\ I

I • ,

Loads

,-------
I Connection Unloads I

,
I I

9. I
I
Connection
/
LOllds I
: 9,
I I

Fig.2.S (a) Loading characteristics of connections.


(b) unloading characteristics of connections.

2.2.1 Causes for the Non-Linear Behavior of Connections

The non-linearity of connections [S due to a number of factors [42,43]. Some of the


important ones are as follows.

I. Material discontinuity of connection assemblage itself. The connection IS

composed of various combinations and arrangements of bolts and structural


shapes. This formation allows for irregular slip and movement of the components
relative to one another at different stages ofloading.

[I. Local yielding of some component parts of a connection assemblage. This is the
primary factor related to the non-linear behavior of a connection.

Ill. Stress and strain concentrations caused by the holes, fasteners and bearing
contacts of elements used in a connection assemblage.

[v. Local buckling of flanges or web of the beam and the column in the vicinity of a
connection.

12
v. Overall geometric changes under the influence of applied loads.

2.3 TYPES OF SEMI-RIGID CONNECTIONS

Fig.2.6 presents the most commonly used semi-rigid connections in steel structures. A
brief description for these connections is given in the following.

A. Single Web Connection

A Single web angle connection consists of an angle either bolted or welded to both the
column and the beam web (Fig. 2.6i). The most popular fonn is when the angle is welded
to the column in the fabricating shop and the beam is bolted to the angle on. the site. The
connection has little moment-rotation rigidity and are very flexible.

B. Double Web Angle Connection

Double web angle connections consist of two angles, either welded or bolted to both the
column and the beam web (Fig. 2.6ii). Experimental tests have indicated that these
com:ections are capable of developing as much as 20% of the full fixed end moment of
the beam at working loads especially for deep connections [8]. However, the AISC-ASD
Specifications [3] consider this type of connection of as a Type 2 construction connection
(simple connection or shear connection).

C. Header Plate Connection

A Header plate connection consists of an end plate whose length is less than the depth of
the beam, welded to the beam web and bolted to the column flange (Fig.2.6iii). The
moment rotation characteristics of these connections are similar to those of Double web
angle connections and belong to Type 2 framing of the AISC-ASD Specifications [3].
Therefore, a Header plate connection is used mainly to transfer the reaction of the beam to
the column.

13
,

0
0
ld
J
I
-Jgl-
0
0
ld
J
~

~
-Jgl-

-+
-It"L -It"L
Single web angle
•... Double web angle(A)
•• (i)
(ii)

-.J
rJ
1 ~
J ~It.
-Jgl-

-It"L -ItL

Header plate(B)
Top ans seat angle(C)
(iii)
(iv)

Fig.2.6a Different types of soft semi-rigid connections


,-ni
",'-

"1
,

f-
I~ ~ o J

to
,,, "on 0 g:J I
d
0
(J
... - .

c:JJte I
0 0
o 0
-Igl....-

-IlL
o

Top and seat angle with double web angle(D)


End plate(E)
(v)
(vi)
...
C)l

l o 0
J
I f J
d d

- o
o 0 0

-ltL -I L t
P

End plate with column stiffeners rr-Stub(F)


(vii) (viii)

Fig.2.6b Different types of intermediate stiff and stiff semi-rigid connections


D. Top and Seat Angle Connection

A typical Top and seat angle connection IS shown in Fig.2.6iv. The AISC-ASD
Specifications [3], describe the top and seat angle connection as follows:

I. The seat angle transfers only vertical reaction and should not give siguificant
restraining moment on the end of the beam.

II. The top angle is merely for lateral stability and is not considered to carry any
given loads.

However, according to the experimental results, these connections are able to resist sOlpe
amount of end restraining moment of the beam.

E. Top and Seat with Double Web angle Connection

This type of connections is a combination of a top and seat angle connection and a double
web angle connection. A typical Top and seat angle connection with double web angle
connection is shown in Fig.2.6v. This type of connection is considered as Type 3 framing
of the AlSC-ASD Specifications [3].

F. End Plate Connection (With or Without Column Stiffeners)

End plate connections are a popular means of connecting the beam to the column when a
moment connection is required. The end plate is welded to the beam end along both the
flanges and the web in the manufacturing plant and bolted to the column in the field. End
plate connections are classified into two types as Extended End-plate connection (Fig.
2.6.t) and Flush End-plate connection is shown in Fig.2.6vi. Both Extended and Flush
end-plate connections are considered as Type I framing connections by the AlSC-ASD
Specifications [3].

G. T-Stub Connections

A T-stub connection consists of two T-stubs bolted to both the beam and the column at
the top and at the bottom flange of the beam (Fig.2.6viii). This type of connection is

16
considered as one of the stiffest semi-rigid connections, especially when used toget~er
with double web angles.

2.4 MODELING OF PARTIALLY RESTRAINED CONNECTIONS

The most commonly used approaches to describe the M-8, relationship involve the curve
fitting of experimental data to simple expressions or the development of simple analytical
procedures to predict the behavior of the connections if no test data are available for the
specific connection details. Numerous experiments on connections have been performed
in the past resulting in a rather large body of M-8, data [1, 2, 13, 14]. Using these
available data, various M-8, models have been developed.

2.4.1 Linear Model

The influences of partially restrained connections on frame behavior are modeled by


attaching rotational springs to the beam-ends as shown in Fig.2.7a. The stiffnesses of the
rotational springs are determined from moment-rotation characteristics of the connections
as obtained from the experimental studies. In linear models, the moment-rotation relations
are assumed to be linear.

nOTA,TION fl,

Fig.2.7 (a)Beam with equivalent rotational spring, (b) Linear M-8, models.

17
These linear models are applicable to the "modified slope-deflection", "modified
moment-distribution" and "matrix stiffness" methods of analysis. Fig. 2.7b shows three
simple linear models. The single stiffness linear model was proposed by Rathbun [17],
Monforton and Wu [18] and Lightfoot and LeMessurier [19]. All of them utilized the
initial stiffness, Rk; to represent the connection behavior for its entire range of loading.
The validity of this model deteriorates as the moment increases beyond the serviceability
limit of the connection. Romstad and Subramaniam [20] and Lui and Chen [21] proposed
bilinear model. A bilinear model shows a better representation of the connection behavior
in which the initial slope of the moment-rotation line is replaced by a shallower line at a
certain transition moment. In the piecewise linear model [23], the nonlinear M-8, curve is
approximated by a series of straight line segments. Although these linear m~dels are easy
to use, the inaccuracies and the sudden jump in stiffness at the transition points make their
practical use difficult.

2.4.2 Polynomial Model

The experimental M-8, curve, although not following any simple mathematical function,
may be approximated to a polynomial type of function using curve-fitting techniques.
Sommer [24] in 1967, first fitted moment-rotation data to standardized moment rotation
curves in the form of non-dimensional polynomial series. Sommer worked with only
Header plate connection. The form of the polynomial series function is

$ = f(CM) [2.1]

Where, C is a factor to allow for the size effects or dimensions of the connection. The
standardized moment-rotation function is applicable to all connections of the same type.
The influence of different sizes and dimensions is accounted for by the size effect factor,
C.

Frye and Morris [25] extended Sommer's work to develop a polynomial model to predict
the behavior of several types of connections. In this model, the M-8, relationship is
represented by an odd power polynomial of the form

18
[2.2]

Where, K is a standardization parameter dependent upon the connection type and


geometry, and C" C, and C3 are curve-fitting constants. The values of these parameters
vary for different types of connections. The values of the curve-fitting constants and the
standardization constants are presented in Table.3.I of Chapter 3. This model represents
the M-8, behavior reasonably well. The main advantage of this model is that, it provides
the tool for prediction ofthe response of a wide variety of connection types.

Altman et ai. [10] developed prediction equations to describe the M-8, behavior for Top
& seat angle with double web angle connection. The equation of Altman et ai. [10] has
the form

[2.3]

Where,

[2.4]

Where t, is the top angle thickness (in), d is the beam depth (in), tw is the web angle
thickness (in), It is the length of the top angle (in) and g is the distance from the heel of
the top angle to the lower edge ofthe fastener holes (in).

2.4.3 B-Spline Model

A more accurate representation of the connection behavior can be achieved by using cubic
B-Spline curve-fitting techniques. Jones et ai. [26] used this method to fit the curve of the
experimental data of connections. The range of experimental M-8, data was divided into a
number of smaller ranges. A cubic B-spline curve is then used to fit each and every range
of data with continuities of first and second order derivatives enforced in their
intersections. This method has been shown to produce close and smooth curve
representations of experimental moment-rotation data. Although, this model represents

19
the non-linear M-e, behavior extremely well, a large number of data are required in this
curve fitting process.

2.4.4 Power Model

Power models were introduced by the researchers to represent the non-linear behavior of
connections. Colson and Louveau (1983) [27), for example, introduced a power function
based on three parameter elasto-plastic stress-strain model of the form

[2.5)

Where, Rk; is the initial connection stiffuess, Mu is the ultimate moment capacity of the
connection and n is the shape parameter of the M-e, curve. Since the model has only three
parameters, it is not as accurate as B-spline model. However, the number of data required
for this model is reduced.

Ang and Morris, [29) presented prediction equation adopting the standardized Ramberg-
Osgpod [44) function for fiVe types of connections. The equation is in the power form

[2.6)

Where, (e,)o ,(KI.."I)o and n are constants that define the shape of the standardized
function. The constants (e,)o, and (KM)o define the position of point I (Fig.2.8), through
which a family of Ramberg-Osgood [44)curves passes. Constant n defines the sharpness
of the curvature for anyone of the curves and eo and M are the rotational deformation of
the connection and the moment resisted by it respectively. The values of standardization
factor K can be obtained from Reference 29.

20
<M

n] •••.•••.

--------,-------- II
I I
I r
I •
I •
I "'>"2>")
I
I
I
I
I
I

2.l H,).,

Fig. 2.8 Behavior ofthe Ang-Morris power model (Adapted from Ref. 29).

This dimensionless factor, K, scales the ordinates on the curve, by accounting for their
dependence upon the connection size parameters. It has the form

[2.1]

Where,
qj = numerical value of the jth size parameter,

aj = dimensionless exponent which indicates the effect of the jth size parameter on the
moment-rotation relationship,
m = number of size parameters.

The evaluation of the exponent ~ follows the procedure as described by Frye and Morris
[25]. A pair of experimentally obtained moment-rotation curves for two connections that
are identical except for parameter % is considered.

The Kishi and Chen's [28] power model is a semi-empirical connection model which is.in
the form

21
[2.8]

Where, Rk; is the initial connection stiffuess, M" is the ultimate moment capacity and n is
the shape parameter. The initial stiffuess and the ultimate connection moment capacity are
evaluated analytically, while the shape parameter is obtained by a curve-fitting technique.
Kishi and Chen applied the model to predict the behavior of three types of connections:

• Single/Double web angle connection.


• Top and seat angle connection.
• Top and seat angle with double web angle connection.

2.4.5 Exponential Model

Lui and Chen [21] proposed the multi parameter exponential model,

M = fC{I-e-10,1/2ja )+Mo +Rkfl8rl [2.9]


. J=I

Where, Mo is the starting value of the connection moment to which the curve is fitted, R
kf

is the strain hardening stiffness of the connection, a is a scaling factor and Cj is a curve
fitting constant obtained from the linear regression analysis. This model gives as good a
curve fitting to test data as that of the cubic B-spline model. However, there is a sharp
change in the slope of the M-8, curve.

Yee and MeJchers [30] proposed a four parameter exponential model to represent the non-
linear M-8, behavior of bolted connections. The model is of the form

[2.10]

Where, Mp is the plastic moment capacity of the connection, Rk; is the initial elastic
connection stiffiless, Rkp is the strain hardening connection stiffuess and C is the constant

22
'.

that controls the slope of the curve. The parameters M. ' Rk; and RkP are determined
analytically and C is obtained empirically by curve-fitting to experimental data.

Wu and Chen [31] proposed a three parameter model to represent the moment-rotation
behavior of top and seat angles with or without double web angle connections: The model
has the form

~ = n[ln(I..ol- ~)] [2.11 ]


M u n80

Where, M" is the idealized elastic-plastic mechanism moment, 80 is a reference rotation


and n is a shape parameter. The parameters M" and Rk; are obtained analytically, while n
is obtained by calibration of experimental data.

2.4.6 Data Base Prediction of Connection Behavior

Database is a collection of experimental tests and their results. From the early stages of
the twentieth century quite a lot of number of experimental tests based on beam-column
connections were performed. These data are compiled with the corresponding details and
dimensions of the beam, the column and the connections. The types of steel used were
also included. These data were compared with several selected moment-rotation
prediction models recommended for use in a steel frame analysis.

Numerous experimental works were performed by varIOUS researchers to provide


moment-rotation data for bolted, riveted and welded connections. Hechtman and Johnston
[6] at Lehigh University in 1947 considered the behavior of 47 riveted connections.

Goverdhan [I] collected extensive connection data from experimental tests conducted
after 1950. The moment-rotation data were compiled in the computer in the form of data
base. Several prediction equations were presented for each type of connection. The
experimental moment-rotation curves were compared against the available moment-
rotation prediction equations for each of the connection types. The validity and drawbacks
of the equations were discussed and recommendations regarding their use by the designer

23
were given. This collection of test results were mainly for following SIX types of
connections:

• Single web angle.


• Double web angle.
• Header plate.
• Top and seat angle.
• Top and seat angle with web angles.
• End plate.

Nethercot [13] reviewed data over 70 separate experimental studies on steel beam-colunm
connections and compiled the useful data for analysis. The curve fitting ohhe data was
conducted in the study and some preliminary comparative studies of the role of different
joint parameters on moment-rotation curves were incorporated. The following connection
types were included in the database.

I. Single web angle.


2. Double web angle.
3. Flange angle.
4. Header plate.
5. Flush / Extended end plate.
6. Combined web and flange angles.
7. T-stub.
8. Top and seat angle.
9. T-stubs and web angles.

Kishi and Chen [2] conducted a comprehensive search to review on beam-column


connection data, published since 1936 until 1986. This study provided moment-rotation
characteristics and the corresponding parameters of beam-column connections

The connection types included in this study had semi-rigid moment-rotation


characteristics similar to those reported in Goverdhan's data banle Each experimental

24
datum included not only the moment-rotation characteristics, but also all the parameters
used in the beam-column prediction equation.

2.4.7 Connection Behavior Based on Finite Element Model

Tarpy and Cardinal [32] used a finite element model and performed a parametric study on
97 different connections. A multiple linear regression analysis was used to obtain a
prediction equation in the form

[2.12]

And the moment capacity (in kip-ft) is given by

[2.13]

Where, d is the beam depth (in), g is the gage distance (in), p is the pitch (in), t is the
p

end-plate thickness (in), t,cis the column flange thickness (in), b, is the beam flange width
(in) 'and Fy is the yield stress of the column and end-plate material (ksi).

Krishnamurthy et al. [33] used a two-dimensional finite element model for end plate
connections. The column flange was assumed to be rigid or stiffened. The equation is in
the form

e __C c__
MISS
r - tl.38 [2.14]
p

Where,

[2.15]

25
[2.16]

[2.17]

Where, Sx is the section modulus, Pf (in) is the bolt distance, <It, (in) is the tension bolt area
per row, fb (ksi) is the average stress in the beam, Fy (ksi) is the plate and beam material
yield stress, Fby (ksi) is the bolt material yield stress, d (in) is the depth of the beam, bf
(in) is the beam flange width, tf (in) is the beam flange thickness, tw (in) is the web
thickness and tp (in) is the end plate thickness. The factors 13and ~ are functions of the
beam dimensions and material properties. These factors are designated as beam factor and
material factor respectively.

2.5 METHODS OF INCORPORATION OF CONNECTION FLEXIBILITY

With the realization of the importance of the flexibility effects of semi-rigid connections,
researchers have been performed to enhance the possibility of incorporating the end
restraint of connections in the analysis of steel framework. Batho and Rowan [45] for
example, developed a useful graphical method, known as beam line method for the
analysis of frames with partially restrained connections. Other researchers [17,48]
proposed modified methods applicable to the frame analysis by slope deflection method
and moment distribution method. With the availability of digital computers, Monforton
and Wu [18] developed a matrix method which included the initial stiffuess of the
connections. Utilizing this concept and using analytical non-linear connection model,
Frye and Morris [25], Ang and Morris [29], and Dhillon and Majid [41] employed matrix
methods to predict the behavior of partially restrained frames.

The methods of structural analysis employed in the references noted above are described
briefly in the following sections.

26
2.5.1 Beam-Line Method

Batho and Rowan [45] conducted tests to find a relationship between the moment applied
at a riveted connection and the corresponding rotation. They developed a graphical
method to predict the end restraint provided by a connection for which the experimentally
obtained moment-rotation relationship was known. This method was first used by
Sourochnikoff [46J and is known as 'beam-line method'. In this method moment-area
principles were used to derive the beam-line equation [47] which gives the end restraining
moment as a linear function of the angle of connection rotation as

M = MF_ (2EI~)/L
[2.18]

,
! ~lon1enl Q

II
Mr .
fleslroinl line

Beam line

B
o 11010lion

Fig. 2.9 Beam-line method

To detem1ine the end restraint provided by a certain type of connection a beam-line is


drawn on the connection's moment-rotation curve as shown in Fig.2.9. The beam-line and
the restraint line intersect at a point, P, where the values of moment and angle of rotation
represent the end restraint conditions that would exist at the end of a member. The
advantage of this method is that it uses the actual moment-rotation relationship. However,
the method requires experimental moment-rotation data to be available for every
connection analyzed.

27
2.5.2 Modified Methods

Methods of incorporating semi-rigid end restraint into slope deflection and moment
distribution methods were proposed by both Baker [48] and Rathbun [17] independently.
Both used a semi-rigid connection factor Z, defined as the angle change per unit moment.

[2.19]

This method assumes that a linear moment-rotation relationship exists, despite the fact
that experiments showed that moment-rotation curves were non-linear over the whole
range of loading for most of the connections. This assumption makes the method only
strictly applicable for very low values of rotation.

2.5.2.1 Modified Slope-Deflection Method

The modified slope-deflection method was applied to frames with semi-rigid connections
in 1936, both by Baker [48] and Rathbun [17]. The modified slope-deflection equation
incorporating the end restraint effect for a beam oflength L and flexural rigidity E1 loaded
by a uniformly distributed load w and for no sway case is

[2.20]

The values of the coefficients A, B, C and D for both the conventional and modified
methods are given in Table 2.1. The value of the coefficients a and P in the table are
defined as

[2.21]

[2.22]

28
Where, ZA , and ZB are the semi-rigid connection factors for end A and B for the beam
respectively.

Table.2.r Slope-deflection equation coefficients.


Coefficient The value of the coefficient of Value of the coefficient of modified slope-
slope-deflection method deflection method
AAa I I I [ 3aj3 + 2a + 213+ I ]
BAB 2 (313+2)
CAB I I
DAB I (313+1)
ABA I II [ 3aj3 + 2a + 213+ I ]
BBA I I
CBA 2 (3a + 2) -
DBA -I - ( 3a + I )

2.5.2.2 Modified Moment-Distribution Method

The conventional moment distribution method of analysis was modified to allow for
flexible connections. The conventional method is confined to frames with perfectly rigid
connections. The method when applied to frames with semi-rigid connections is the same
as the usual method except for numerical differences in the values of stiffuess factors,
carry-overs and fixed end moments as shown in Table.2.2. Methods of incorporating the
flexibility effect of connections into the moment distribution method were proposed by
Baker [49J, Rathbun [17J, Johnston and Mount [50], and Gere [51]. For any particular
frame with semi-rigid connections the factors in Table.2.2 can be evaluated and once
obtained the analysis proceeds in exactly the same fashion as for rigidly connected
frames. The main drawback of this method is that they are not very suitable for digital
computers.

29
Table 2.2 Comparison of moment-distribution factors
Factors Conventional Modified method including effects of semi-rigid connectians
moment-distribution
method
Fixed End Moment

MA.F -roL'l 12 -(3P + I) roL'/12 [3ap + 2a + 2p + 1 ]


.

M.AF roL'1 12 '.


(3a + I) roL'/12 [3ap + 2a + 2p + 1 ]
Carry Overs

CA. 1 1 1( 2 + 3p )
C.A 1 1 1( 2 + 3a ) .

Stiffness Factor

SA. 1 (3P + 2) 1 2 [ 2ap + 2a + 2p + 1 ]


.
S.A 1 (3a + 2) 12 [ 2ap + 2a + 2p + 1 ]

2.5.3 Matrix Methods

Before the development of computers the methods of analyzing steel frames with semi-
rigid connections was very tidious and cumbersome. When the frames are large the
analysis becomes laborious for the designers. After the development of the electronic
digital computers, the application of computers using matrix stiffues$ methods to the
analysis of large structural frames became very easy .._Computers also made it possible to
incorporate systematic procedures into methods of analysis to give a better representation
of connection behavior. Iterative methods of correcting connection stiffuess and end
restraint values produced better representation of the real connection behavior.

By 1960, the analysis of plane frames by matrix stiffuess methods using electronic
computers had been established. In 1961 Lightfoot and Baker [52] produced a computer
solution to the problem of plane frames with elastic connections, using the generaliz'ed
slope-deflection equations in matrix forms. The semi- rigid connection restraint was
incorporated into the analysis by the use of correction matrices to amend the initial
assumption of fully rigid connections. Monforton and Wu [18] first incorporated the
effects of semi-rigid connections into the matrix stiffuess analysis. The linear semi-rigid

30
connection factor, A, is used to modifY the member stiffness matrices and the fixed-enrl-
forces vector. Relationships between forces and displacements at the ends of a member
with elastically restrained ends were derived. The force components, in terms of the end
displacements, were arranged in matrix form. Stiffness matrices of a member with elastic
restraints at the ends were presented in the form of stiffuess matrices for members with
rigid connections modified by a correction matrix [c], whose elements are functions of
two parameters, Yi and Yj, designated as 'fixity factors' of the member. A formula, derived
from the joint equilibrium equation, for the force components at the ends of all members
of a frame was established.

The formula for the modified stiffness matrix is:

k=[sJ[c]
[2.23]

Where, [s] is the stiffness matrices of a member with rigidly ends, and [c] is the correction
matrix for semi-rigid connections. The force components is designated as:

F=JkJ[D] [2.24]

Where, [ F ] is the force matrix, [ k ] is the modified stiffness matrix and [ D ] is the
displacement matrix.

The two dimensionless fixity factors comprising the correction matrix are:

L
Yi = (L+3ElyAj) [2.25]

&
L
Yj = (L+3ElyA) [2.26]

Where, A=cp/M [2.27]

31
The detail of this method is presented in the following chapter. The advantages of the
matrix stiffness method are that it can be programmed so that relatively large frames c'an
be analyzed with ease and that the use of interactive techniques would permit the
inclusion of improvements in the end reentrant representation.

2.6 Non-Linear Analysis of Semi-Rigidly Connected Plane Frames

Frye and Morris [25] developed a program for analyzing steel frames usmg matrix
method of analysis. The procedure involves repeated cycles of an iterative procedure,
whose purpose is to determine the appropriate flexibility characteristics for the various
connections in a structure. Frye and Morris verified their procedure by two sample
problems using double web angle and t-stub connections.

Ang and Morris [29] developed a method to analyze three-dimensional rectangular frames
incorporating several types of commonly used semi-rigid connections. A method is also
described for expressing the moment-rotation behavior of the flexible connections using
Ramberg-Osgood function. The analysis procedure is an iterative method based on matrix
method of analysis. Repeated approximations are made in this method to assume stiffness
characteristics of all connections in the structure. Ang and Morris verified their procedure
by two examples using double web angle and Header plate connections.

Dhillon and Majid [41] developed an integrated analysis and design computer program
which applies PR type of construction. The program is a non-linear, iterative and
interactive aided analysis and design tool based on matrix method and polynomial
connection model. Dhillon and Majid took the same sample problem as taken by Frye and
Morris [25] earlier and compared the results. Dhillon and Majid took two stiff
connections to present their analysis procedure.

2.7 CONCLUDING REMARKS

In this chapter the behavior and characteristics of different types of beam-column


connections have been discussed. The chapter emphasizes the importance of including

32
non-linear behavior of connections in the analysis of steel constructions. Research works
earlier done by various researchers to model semi-rigid connection and to implement in
the analysis methods are presented here. Incorporation of the non-linear behavior of
flexible connections has been done by taking Frye and Morris polynomial model. Frye
and Morris polynomial model has been so chosen because it incorporates several types of
flexible connections (seven types) varying from soft semi-rigid connections to stiff semi-
rigid connections. The incorporation of non-linear effect of semi-rigid connections taking
Frye and Morris polynomial model to the analysis of frames has been presented in the
next chapter.

33

'.
CHAPTER 3
ANALYTICAL MODELING

3.1 GENERAL

The connections constitute a small percentage of the total weight of a structure, but they
have a relatively high labor content and hence represent a substantial percentage of the
total framing cost. In addition, connection deformation is sometimes responsible for a
substantial portion of the overall deflection of a structure and it has a significant bearing
on the internal force distribution. As mentioned earlier, in the custommr steel frame
analysis assumption is made that beam-to-column connections are either pinned or fixed.
The available design techniques are based on these idealized conditions. In reality m<;Jst
connections are partially restrained or semi-rigid. In partially restrained type construction,
the moment rotation capability oLa given type of connection must be established by
analytical or experimental means. Most experiments have shown that the connection
moment rotation characteristics are nonlinear over the entire range of loading for almost
all types of connections. Most structures behave non-linearly before reaching the ultimate
resistance capacity. This phenomenon results in the development of non-linear analysis
methods and explicit consideration in the specifications. The Allowable Stress Design
(ASD) Specifications [3] and the Load and Resistance factor Design (LRFD)
Specifications [4] of AISC permit to consider explicit behavior of connection in the
analysis. In other words, the use of non-linear analysis of steel frame is encouraged in the
AISC Specification. Several investigations have been conducted to establish relations
between moments and relative rotations at the beam-to-column connections for several
cOlmection types. Attempts have also been made to develop a mathematical model
representing moment-rotation characteristics of the connections.

3.2 CONNECTION MODELING

The flexural connection behavior is primarily represented by the moment-rotation


relationship, which relates the moment transmitted by the connection to the relative

34
rotation of the intersecting members. The moment-rotation curves for all types of
connections are of non-linear type. In this reseffi-ch work, the polynomial model of Frye
and Morris [25] is used as connection model. In Frye and Morris polynomial model,
available experimental data and Sommer's method [24] were used to express the moment-
rotation characteristics of seven commonly used connections in a non-dimensional form.
The connection size parameters were used and the appropriate non-dimensional functions
were developed to~enerate the force-deformation characteristics for various connections.

3.2.1 AssumptioB8.and Limitations of Frye and Morris Polynomial Model

The assumptions and limitations of Frye and Morris Polynomial model [25] ~e:

• The effects of shear and axial force load on connection deformation are ignored.
• All members are prismatic and straight.

• The structure is loaded with statically concentrated or uniformly distributed loads


only.

• Possible buckling of individual members or portion ofthe structure is ignored.


• The effects of strain hardening are neglected.
• The material in the members is linearly elastic.

• Deflections are sufficiently small that they do not significantly affect the geometry of
the structure.

• The structure behaves linearly except for the non-linear force deformation
characteristics of the connections.

3.2.2 Standardization of Moment-Rotation Curves with Polynomial Model

There are two ways that connection moment-rotation relationships can be incorporated
into structural analysis program.

a) The moment-rotation information for every connection of every type can be stored.
Since for any given type of connection, there are a number of 'size parameters' such

35 '."
,i.:
.://
as beam size, angle depth, angle thickness, etc., this reqUires the storing of an
extremely large amount of information.

b) Since the moment-rotation characteristics for all connections of a given type are quite
similar, a 'standardized' moment-rotation relationship can be derived as function for
the size parameters for that type of connection. The moment-rotation characteristics
for a particular type of connection can then be generated by substituting its size
parameters into the 'standardized' relationship.

The latter procedure drastically reduces the amount of connection information that must
be stored. The standardization procedure involves the representation of .the moment-
rotation curves for all connections of a given type by a single function of the form

x .
e= LCj(KM)'
[3.1]
i=I

Where,

e =.rotational deformation ofthe connection (radian).


C; = constant associated with connection type.
M = applied moment (in-kips).

K = Dimensionless standardization coefficient whose values depends on the connection


size parameters.

The standardization coefficient is assumed to have the form

K = I1p/J [3.2]
j=1

Where,

Pj = numerical values ofjth size parameter.

aj = a dimensionless exponent which indicates the effect of jth size parameter on the
moment-rotation relationship.
m = total number of size parameters.

36
The evaluation of the exponents a; in Equation.3.2 can be illustrated by considering a
family of experimentally obtained moment-rotation curves for connections which are
identical except for parameters Pj as shown in Figure. 3. I .

'P
ROTATION

Fig). I. Family of moment-rotation curves for connections with different values of Pj.

A pair of curves is considered and the relationship between moments M, and M, at a


particular rotation, 8, is assumed to have the form

~ =[pi Ja
M2
2

Pil
j

[3.3]

Where, Pj' and Pj' are the numerical values of parameter p for connections I and 2
(corresponding to curves I and 2) respectively. M, and M, are the moment values. at
rotation 8 for curves I and 2 respectively. Equation 3.3 can be written in the form

[3.4]

Equation 3.4 is used to calculate a; values corresponding to several rotations for each
combination of experimental curves. When average values have been calculated for all m

37
exponents ll; In Equation 3.2, they are plotted on standardized moment-rotation (KM
versus 8) diagram. A least square curve fitting procedure is used to derive the
standardized moment-rotation relationship in the form of Equation 3.1.

The accuracy of the standardization procedure is illustrated in Fig.3.2. The figure sho)Vs
the moment-rotation curves generated by the standardized equation and corresponding
experimentally obtained curves for two double web angle connections. With few
exceptions, the procedure was found to produce accurate moment-rotation curves for
connections within the range of test results available. The standardized moment-rotation
functions and standardization constants for each of the seven types of connections are
listed in Table 3.1. The connection size parameters as used in the standardization constant
expression are given in Chapter 2 (Fig.2.6). Altman et al. [10] followed the procedure of
Frye and Morris and developed polynomial connection model for Top and seat angle with
double web angle connection. The standardization moment-rotation function and the
standardization constant ofthis connection is also listed in Table 3.1.

'00

-•
.0 ---::;::::::.
_- 1 ROW!: OF
F' AS"
EN E R 5

i. ~ 600
1
~ /

~
"~ '00 -
/
;;
,;/ -_. - EXP[RIIAErHAL :URVE~

•.
'
- - -" CURVES GEN[Rr..7ED ~Ol",
.. STANDARDIZED FUN':"j"IOl\'''
,
~
200

o
o 2 .c [) 8 10 1£

ROTATION { RAOltHIS J ,1031

Fig.3.2 Comparison of moment rotation curves for double web angle connections.

38
Table. 3.1.
Standardized connection mom;;nt-rotation functions for different types of connections [10, 25] (Fig.2.6).

CONNECTION TYPE STANDARDIZED MOMENT-ROTATION FUNCTION


(Ref. Fig.2.6 i-viii) STANDARDIZATION
(Ref. Eq.3.1) CONSTANT (Ref. Eq.3.2)
Single web angle connection B= 4.28(KM) x 1O'3+1.45(KM)3 x 10-9 + 1.51(KM)' X 10-16
K= d-2A rl.81 gO.I'
Double web angle connection B= 3.66(KM) x 10-4+ 1.15(KM)3 x IO--<i
+ 4.57(KM)' X 10-8
K= d-2A rl.81 gO.I'
Header plate connection B= 5.1O(KM) x 10-' + 6.20(KM)3 x 10-1°+ 2.40(KM)' x 10-13
K= 1-1.6g 1.6d-2.3 wO.s
Top and seat angle connection B- 8.46(KM) x 10-4+ l.OI(KM)3 x 10-4+ 1.24(KM)' x 10-8
K= I-<J.' d -I.' JJ.lI-o.7
Top and seat angle with double B= 0.223(KM) x 10-4+ O.185(KM)3 X 10-7 - O.319(KM)' X 10-11
web angle connection K~ d-l.2S? t a -I.l281 tw -0.4145 r().6941 J.3499
g
End plate connection without B= 1.83(KM) x 10-3 - l.04(KM)3 X 10-4+ 6.38(KM)' X 10-6
column stiffeners K = d -2A I -<JAf J.I
End plate connection with B= l.79(KM) x 10-3 + 1.76(KM)3 X 10-4+ 2.04(KM)' X 10-4
column stiffeners K = d -2.4I -<J.6

T-stub connection B= 2.IO(KM) x 10-4+ 6.20(KM)3 X 10-6_ 7.60(KM)' X 10-9


K= d-J.5 I -0.5 rJ.l 1-fJ.7
3.3 STIFFNESS OF BEAM-COLUMN ELEMENT WITH FLEXIBLE
CONNECTION

The beam-column element with flexible end connections may be viewed as a combination

of two types of elements:

• A beam-column element,
• A connection element.

For the conventional stiffness formulation, the beam-column element stiffness is modified
to represent the flexibility of the connection.

3.3.1 Beam-Column Element

The beam-column element with full fixity at both ends can be represented by stiffness
matrix, relating the element end actions and displacements. The relationship is linear until
large deflections take place and the resulting structure geometry changes. For a member
shown in Fig.3.3. the linear stiffness matrix [k, ] is:

EAx
L
12EI
0
L3
6EI 4El
0 - symmetrical
L2 L
[keJ = [3.5]
_ EAx EAx
0 0
L L
12El 6EI l2EI
0 0
L3 L2 L3
6EI 2EI 6El 4EI
0 - 0 -
L2 L L2 L

40
Fig. 3.3. Beam-column element in member and structure coordinates

Where,
L = length of the member
A, = cross-sectional area
I = moment of inertia
E = modulus of elasticity

3.3.2 Connection Element

Although connection constitute a small portion of the steel frame, their effect is
significant to the overall frame performance. While the restraining effect of the
connection was established, there were no methods for incorporating it into the analysis of
steel frame analysis. In Fig.3.4b the flexural connection behavior is shown by the
moment-rotation curve, which relates the moment transmitted by the connection to the
relative rotation of the intersecting members. The stiffuess of the connection is
represented by the slope of its moment-rotation curve at any particular moment value. The
non-linear nature of the curve indicates that the stiffness decreases as the loading
increases for all semi-rigid connections. From the available moment-rotation relationship,
the stiffuess of the connection can be derived as the secant or tangent to the curve. The
secant stiffuess approach, which is a simple repetition of first order analysis, is easier to
implement in analysis than the tangent stiffuess approach, which is sensitive to local
variations. In this research work the secant stiffuess method is adopted.

41
The connection element as shown in Fig.3.4a is represented by a rotational spring with
stiffuess equal to the change in moment over the change in relative connection rotation for
the load increment under consideration.

M1
SE=-.-
i18 [3.6]

Where,

SE = secant stiffuess of connection


11M = change in moment during a load increment

i1B= change in relative rotation during the load increment.

The connection behavior under loading and unloading is also shown in the figure. It has
been observed that the connections behave elastically during unloading regardless of the
load level. The stiffuess of the connection during unloading approximately equals the
initial tangent stiffness of the loading curve.

! j-~~'-----=@f

(a)

l£lAOIIlG

ROTl\flO'{ 0

(b)
Fig.3.4 Connection element and connection loading and unloading behavior.

42
3.3.3 Modification of Beam-Column Element for Flexible Connection

The effect of flexible connections at the ends of a beam-column element can be


incorporated in the analysis by modifYing the standard elastic stiffuess matrix [k c] of the
beam-column element. As the secant stiffuess method is applied to the connection curve
in an iterative solution procedure, the secant stiffuess of the connection can be modeled as
the constant of linear spring. This makes the technique developed by Monforton and Wu
[18] applicable to the solution procedure. Monforton and Wu presented the member
stiffness with flexible connections at the ends, in the form of stiffuess matrices for
members with rigid connections modified by a "correction matrix". The elements of this
correction matrix are functions of two dimensionless parameters, Yi and Yj , called 'fixity
factors'. These fixity factors are given in the followings:

L
Yi = (L + 3ElzAJ [3.7a]

L
Yj = (L + 3EIzAJ [3.7b]

Where, the subscripts i and j refers to the ends of a beam element. The value of Y depends
on the known semi-rigid connection factor A, and the geometrical and elastic properties of
the members. It varies from zero for a frictionless pin connection to unity for a perfectly
rigid connection.

A=~ = !lei = _I_


I M.I !lM.I (SE).
\, 1
[3.8a]

[3.8b]

The modified stiffness matrix [k c]m is the product of the beam-column elastic matrix [k,]m
multiplied by the correction matrix [ C ], i.e.

[3.9]

43

/
3.3.4 Generation of Correction Matrix

The generation of the correction matrix is based on the derivation of the force-
displacement relationship at the ends of a member with semi-rigid connections. The
derivation is based on the 'conjugate beam method' of analysis [18] as shown in Fig.3.5a.
The adopted sign convention is shown in Fig. 3.5b. The forces acting at the ends of the
beam are also shown in Fig.3.5c.

'010'"
r--,
~
---
l
I .
I,13- 1

141
rj J _ I

1 ,.
. M R

. /,/['~- -~- ~f:'?!_


. y. $.

'"I .

I. L

(0) Member Under Loading

M'
-'
E "

Fig.3.5a Conjugate beam for semi-rigid connection.

..-,.,-., ,

44 )
0'
\
j
.,~ \
z .

y y

Force Components Displacement Component 5


-
Fig.3.5b Sign convention.

z,
/:,~
~V M.
J
';:;f
z
j

x,
. M'
C- tf -------- -_. __ c
7l
ylYe '-, ' X

.J.,MY I
M Y JjJ,
M.
;
r ' J ,
Y, Y
J

Fig.3.5c Forces acting at the ends of a member (Monforton and Wu [18])

The shears at the ends of the conjugate beam are equal to the net angle changes at the
ends of the actual beam.

Vi = ef - <Pi- R (MZLJ [MZL]+ (Ab)


= 3~Iz - 6E~z ElzL [3.IOa]

V-e
J -
Z

J
-
<PJ
-R - (MJLJ-(MJLJ+(~)
- 3El
z
6El
z
ElzL [3.lOb]

45

,
\
•I
Substituting, ljl i = Mi' A.;and, ljl j = M;' Aj into Equation [3.10a] and [3.10b] and
rearranging the terms yields

(L+3ElzA;)~f = [3Elz (Sf~R)J+[ ~j J-e~b) [3.lla)'

S
(L+3ElzAJ)~f = [3EIJ f ~R)J+[ ~f J-e~a) [3.lIb]

From the conjugate beam analysis the internal forces at the member ends are:

A. The moments at the end are

[3.12a]

M' = (4EI,)[ 3Yj J()'_(6EI,)[Yj(2+Yi)Jy.


) L (4-YiyJ ) L (4-YiyJ }
2

[3.12b]

B. The shear forces are

[3.13a]

46
[3.13b]

C. The axial forces are

[3.14a]

[3.14b]

3.3.5 Modified Stiffness Matrices

The.modified stiffness matrix can be written in the form

[3.15a:]

[3.15b]

Where,
[F; ] = force matrix
[D; ] = deformation matrix
[k;; ] = [k;j] = modified stiffuess matrix.

The elements of the modified stiffuess matrix [I<:;;


] or [k;j] are

47
EA
-- 0 0
L
12EIz Yi + Yj + YiYj _ 6Elz Yj(2 + yJ
kii = 0 , (3.16a]
L 4-YiYj L2 4-YiYj
6Elz Yi(2+Yj) 2Elz 3YiYj
0
L2 4-YiYj L 4-YiYj

EA
-- 0 0
L
12EIz Yi + Yj + YiYj 6Elz Yj(2+yJ
kij = 0 , (3.16b)
L 4 - YiYj L2 4- YiYj
6Elz Y;(2 + yJ 2EIz 3YiYj
0
e 4-Wj L 4-YiYj

EA
L
o o
k JI.. = 0 12EIz Yi +Yj +YiYj _ 6Elz Yj(2 + yJ.
(3.16c]
L' 4-YiYj e 4-YiYj
6Elz r;(2+Yj) 2Elz 3YiYj
o 2
L 4-YiYj L 4-YiYj

EA
- 0 0
L
12Elz Yi + Yj + YiYj
k .. = 0
JJ , 6Elz y/2 + yJ
-7 4 -YiYj (3.16d)
L 4-YiYj
6Elz Yj(2+yJ 4Elz 3Yj
0
2
L 4 - YiYj L 4 -YiYj

48
As mentioned earlier the modified stiffuess matrix [k.]m is the product of element stiffuess
matrix [k,] multiplied by correction matrix [C] (Equation 3.19). The elements of the
correction matrix are

[3.17]

Where,

I 0 0
4Yj - 2Yi + YiYj
[Cid = [cj;) = 0 -2L Yi(I-Yj)
[3.18a]
4-YiYj 4-YiYj
6 Yi -Yj 3 Y;{2-Yj)
0
L 4-YiYj 4-YiYj

I 0 0
4Yi - 2Yj + YiYj y(I- y)
[CijJ= [Cjj] = 0 -2L J I
[3.18b]
4-YiYj 4-YiYj
6 Yi - Yj 3 Yj(2 - yJ
0
L 4-YiYj 4 - YiYj

The resulting stiffuess matrix [k, ]m for the member is then transferred to the structure
coordinates by multiplying it with transformation matrix [T] and transpose of the
transformation matrix. The transferred stiffuess matrix in global form is referred as [k]. .

[3.19]

49
Where,

c s 0 0 0 0
-s c 0 0 0 0
0 0 I 0 0 0
[T]=
0 0 0 c s 0 [3.22]
0 0 0 -s c 0
0 0 0 0 0 I

Where,

C = cosa
s=sma

a = inclination of the member.

3.4 ITERATIVE ANALYSIS PROCEDURE


The non-linear response of a partially restrained frame to increasing loads is a result of
non' linear nature of the beam-colunm connections. The beam-column stiffuess matrix
includes non-linear coefficients which depend on the connection stiffness. As a result, the
entire structure stiffuess matrix formed by superimposing non-linear beam-column
stiffuess matrices has a non-linear nature and an iterative solution procedure is required.

3.4.1 Incremental Loading Procedure

As mentioned earlier that the analysis is non-linear and needs iterative solution procedure,
the applied loads should be in increments, i.e. the loading procedure should be on
incremental basis. The applied loads are divided into a series of small increments. The
structure stiffuess matrix changes with each increment of load. The stiffuess equation
solved in the incremental loading form is

[3.21]

Where,

50
[Sf] = I[k ] = summation of Structure stiffuess matrices,
{t>.P f } = incremental load vector,
{t>.Df} = incremental displacement vector.

The incremental solution is implemented through a sequence of linear steps with a


provision to update the stiffuess matrix at the start of each iterative load cycle. It requires
calculation of the connection secant stiffuess at the beginning of each cycle information
from the previous cycle. Convergence is obtained when the difference between the
displacements of two consecutive cycles at all joints come to within a specified tolerance
limit.

Modification of the connection secant stiffuess at each cycle for the first load increment is
shown in Fig.3.6. For the first cycle the connections are assumed to be fixed. The member
end moment, M, is obtained from this analysis and the corresponding connection rotation
is obtained using the polynomial model (Equation 2.2). The connection secant stiffuess is
calculated as

(SE)J = M1 -O(zero)
. 81 -O(zero) [3.22]

The initial secant connection stiffuess, (SE)" is used to develop the coefficients of the
correction matrix for beam-column element as given by Monforton and Wu [18] and also
to modifY the beam-column stiffness coefficients. The solution of the stiffuess equation
is then performed using the updated stiffuess matrices. A new connection moment M" is
obtained and the same procedure is repeated until convergence is obtained.

A convergent solution of load increment forms an initial estimate for the first iteration of
the next load increment and the iterative solution procedure continues until all load
increments are considered. The convergent solutions of all load increments are
accumulated to obtain the total non-linear response.

51
." ..

~'Q\"'\

I
,.,v~1
~!

e, I
i,,
"

,, . ;

!
J
I
I
I

~~~ " ~'


I

I
I
ROTATION. 1>

Fig. 3.6 Modification of connection flexibility (Adapted from Ref.25).

3.4.2 Convergence Criteria

The stiffness equation of the beam-column element is non-linear and iterative procedure
is necessary to obtain an accurate solution. Each iteration considers one load increment
and after a convergent solution is obtained, the iteration for the next load increment is
applied. The convergence criterion compares the joint rotation obtained in the current
cycle with those of previous cycle. If, for all the joints, the difference in relative rotations
between current and previous cycles is less than a specified tolerance limit, convergence
is assumed to have occurred. The convergence criterion can be written in the form

Ie -e II <
I 1-

led - e [3.23)

Where,
e, = joint rotation

e = tolerance limit (5% to 10%).

52
3.4.3 Analysis Module

The iterative solution procedure can be summarized separately stepwise as following.

I. The applied loads are divided into a series of small increments.


2. The load increment vector {LiP/} is calculated.

3. The modified beam-column stiffuess matrix [k ] IS calculated in the structure


coordinate system.

4. The modified beam-column stiffuess matrices are assembled to form the overall
structure stiffuess matrix [SI]'

5. The incremental stiffuess equation is solved to obtain the incremental displacement


vector {LiD/}.

6. The member end actions are determined.


7. The convergence criterion is checked.
8. The connection secant stiffness is calculated

9. The non-linear terms in the stiffuess matrices are updated using the latest connection
stiffness ..

10. The steps 3 to 9 are repeated until convergence is obtained.

II. At convergence, the incremental deformation and incremental member actions are
added to the corresponding previous accumulated values.

12. If the total load increments applied equal the total load imposed, the analysis part is
completed. If not, step 3 is repeated and the convergent solution for the initial
estimation of the non-linear terms in the next load increment is taken.

3.5 CONCLUDING REMARKS

In this chapter, the total analytical procedure for a partially restrained frame has been
described. How the effect of flexible connections at the end of a beam-column element
can be incorporated to the analysis by modifYing the standard elastic stiffness matrix of
beam-column element, is presented in this chapter. The following chapter (Chapter 4)
high lights about the computer program developed in this research work. The analytical
and mathematical background for developing the computer program has been discussed in
this chapter.

53
CHAPTER 4
COMPUTER PROGRAMMING

4.1 GENERAL

In the previous chapter the analytical modeling of partially restrained connections has
been presented. The purpose of this chapter is to utilize this connection model and to
develop a computer program to carry out the necessary investigations. The program is a
generalized one. It can be used to analyze any 2-D frame, with any type of known semi-
rigid connection that is subjected to general type of loading and can encounter any
boundary conditions. The program is a non-linear, iterative type based on incremental or
stepwise loading. The complete solution for a given load is impl\<lJ.lented through a
sequence of load increments. The program described here has been developed in Fortran
language and can be run on any IBM compatible computer.

4.2 THE SOLUTION SCHEME

Before the individual parts of the program are discussed, the sequence of the solution
procedure maintained during the computer analysis is described. The proposed solution
scheme includes the following steps:

I. The coordinates of joints, the support conditions, member connectivity including type
of connections at the ends, sectional and material properties designations are defined
first. The area of the cross section, the moments of inertia and the material properties
including the modulus of elasticity, modulus of rigidity and the Poisson's ratio are
defined corresponding to the sectional and material properties designator. At the same
time, by means of a series of simple commands, the loading on the nodal points and
members are defined. In this part the connection parameters and the type of the
problem or the analytical procedure (i.e. rigid analysis or semi-rigid analysis) and the
load cases are also defined.

54
2. For the first iteration of the first increment of loading, the member stiffuess matrix is
calculated in the global coordinate system assuming the connections are rigid. The
boundary conditions are then imposed and accordingly the structure stiffuess matrix is
modified.

3. The stiffuess equations are solved to find the incremental displacements and the
member end actions. The secant stiffuess and the corresponding fixity factors are
calculated to incorporate the non-linear effect of connections.

4. The correction matrix is generated using fixity factors and this correction matrix. is
assembled to the member stiffuess matrix to incorporate the flexibility of connections.
This procedure starts from the second iteration ofthe first increment ofloading.

5. After every iteration, the member end actions and the displacements are updated.
After converging at the end of each increment of loading, the incremental deformation
and the incremental member end actions are added to the corresponding previous
accumulated values to obtain the total response ofthe frame.

4.3 COMPONENTS OF THE COMPUTER PROGRAM

4.3.1 The Main Program

The computer program consists of a number of subroutines combined together by a main


program. Each of the subroutines carries out a definite operation and these operations are
linked together by a series of simple statements contained in the main program. In fact,
the main program controls the overall analysis and maintains the sequence of the solution
scheme described in the previous article. Flow charts presented in FigA.l and FigA.2
show how the analysis is carried out and when the subroutines are called in the main
program. The names of the subroutines that appear in the flow chart are actually used in
the program and are described in the next articles. It may be mentioned that there are also
some subroutines that are called by other subroutines.

55

,.
Call subroutine INPUT to read the
input data

Initialize the matrices and starts the iterations of the load


increment.

Call subroutine BEAM to calculate the structure stiffuess


matrix for rigid and semi-rigid frame element.

Call subroutine PLOAD to generate the load matrix.

Call subroutine BOUND to apply the boundary conditions.

Call subroutine EQSOLV to get the displacements of the


structure in global coordinates.

56
Call subroutine STRESS to calculate the member end forces,
due the current increment ofloading.

Call subroutine CONVR to apply the convergence criterion.

NO
Convergence
occurs

YES

Update all the cumulative displacements and


member end forces and go for the next load
increment.

Load cases
completed

YES

Fig.4.l Flow diagram of the main program

57

!
STOP

Fig.4.2 Flow table of the main prolP"am

58 !\
In the main program the input file and the .output files are opened. There is one input file
titled 'INPUT.DAT'. There are also three output files named 'INPUT.OUT',
'FORCE.OUT' and 'DISPL.OUT'. The INPUT.DAT contains the input data for the
structure and loading. The INPUT. OUT contains the output of the data which are given in
the INPUT.DAT file. The FORCE.OUT contains the member end actions after analyzing
the frame. The DISPL.OUT contains the displacements of each joint after analysis.

4.3.2 The Subroutines

In this article, an attempt has been made to give some insight into the subroutines. The
function of the subroutines, how they are fonned, how they work and other features are
described briefly and when needed flow chart of the subroutines has also been presented.
The subroutines are presented in a order how they are called in the main program.

4.3.2.1 Subroutine INPUT

In this subroutine the input data for the structure to be analyzed are given. The input can
be divided into 5 groups.

a) Structure data. In the structure data nodal positions, boundary conditions, element
connectivity, element types, connection types at the end of element, sectional and
material properties of each element of the structure have to be given.

b) Counection data. In the connection data the type of the connections at the end of
each element and their corresponding size parameters are read.

c) Material and Sectional properties. Material properties include the modulus of


elasticity, modulus of rigidity and Poisson's ratio. Sectional properties include cross
sectional area and moments of inertia of each element of the structure.

59
d) Load data. Gravity loads in the form of both joint load and element load can be
given. Element load must be given as member end actions. The lateral loads are given
at the joints only.

e) Problem type data. Problem type data signifies the type of the analysis procedure.
There are two types of analysis procedure:

I. Rigid frame analysis.


2. Semi-rigid frame analysis.

For semi-rigid analysis, the total number of load increments and the type of loading
conditions are defined.

1) Load case data. Load case-I means, the frame is subjected to gravity loads only and
case-2 refers that both gravity and lateral loads are included.

4.3.2.2 Generation of the Modified Element Stiffness Matrix

The generation of the modified element stiffuess matrix is done by the operation of some
subroutines. The subroutines involved in the operation are described below.

A. Subroutine BEAMSLE

In the subroutine BEAMSLE, the conventional element stiffuess matrix III the local
coordinate system is generated. After generating the element stiffuess matrix in the local
coordinate system, it is stored in matrices titled SLE .

B. Subroutine BEAMSLSR

In the subroutine BEAMSLSR the flexibility effect of the connections are incorporated
and assembled to the local element stiffuess matrix SLSR. For the first iteration of a load
increment the analysis is done taking the connections to have the fixity factors same as

60
obtained at the end of previous load increment. The moments and the rotations, derived
from the first iteration, are used to develop the secant stiffuess. The secant stiffuess is
developed and is employed to calculate the fixity factors of the connections at the member
ends. These fixity factors are used to generate the correction matrix [C]' The correction
matrices[C] and element stiffuess matrices SLE, as obtained from subroutine BEAMSLE
are multiplied by calling the subroutine MATMUL to form the matrix SLSR This SLSR
matrix is the modified stiffuess matrices for each element in local coordinate system
incorporating the effect of semi-rigid connections at the member ends.

C. Subroutine BEAMSL

The program developed in this research work has the provision to analyze a frame in two
different methods, namely:

• Rigid linear analysis.


• Semi-rigid non-linear analysis.

When IPROB (type of the problem read from the input data) equals to I, the rigid analysis
of the frame is done taking the member end connections to be rigid. In that case, only
subroutine BEAMSLE is called to the subroutine BEAMSL to generate the element local
stiffuess matrix SLE. The members ofthe matrix SLE is then stored in a matrix titled SL.

If IPROB equals to 2, the semi-rigid non-linear analysis of the frame is done. When
NETYPE (element type as defined in the element connectivity matrix) equals to 1,
(problem type remaining 2) the analysis procedure assumes the member end connections
to be rigid (usually for columns in steel structures). Subroutine BEAMSLE is called to
generate the element local stiffuess matrix SLE and stored in matrix SL. If the element
type (NETYPE) equals to 2 and problem type (IPROB) equals to 2, increment (mCR)
equals to I, iteration (ITER) equals to 1 and load case (ICASE) equals to 1, subroutine
BEAMSLE is called to generate the matrix SLE and stored in matrix SL. Other wise, the
semi-rigid non-linear analysis is performed. In other words, subroutine BEAMSLSR is

6t
called to incorporate the semi-rigid connection effect by multiplying the correction matrix
C with local stiffuess matrix SLE to generate matrix SLSR. The members of the matrix
SLSR is then stored in the matrix SL. However, for the first iteration of first load
increment, the analysis procedure assumes rigid frame analysis and subroutine
BEAMSLE is called to generate the element local stiffness matrix SLE, which is then
stored in matrix SL.

In subroutine BEAMSL, matrix SL is the final element stiffness matrix III local
coordinates for the members of a structure incorporating the flexibility effect of the
connections. In this subroutine a transformation matrix titled T is generated to transfer
local matrix to global matrix in subroutine BEAM.

D. Subroutine BEAM

In subroutine BEAM the stiffuess matrices for all the elements of a structure in the global
coordinates are generated. First, subroutine BEAMSL is called. Subroutine BEAMSL
provides element stiffuess matrix SL and transformation matrix T. These two matrices are
multiplied by calling subroutine MATMUL to generate matrix SEl. Transpose of the
transformation matrix T, is generated by calling the subroutine TRANS. The transpose of
the transformati6n matrix T is stored in a new matrix TT. Matrix TT and Matrix SEI are
then multiplied by calling subroutine MATMUL to generate the element stiffuess matrix
in global coordinates SE. This SE is the final element stiffuess matrix, which incorporates
the semi-rigid non-linear effect of the connections. Matrix SE is stored in the structure
stiffuess matrix, SS by calling the subroutine STORE ..

4.3.2.3 Subroutine PLOAD

Subroutine PLOAD deals with the modification of applied load vector matrix due to
partially restrained connections. A frame may be subjected to different types or patterns
of loading. In this research work we usually deals with two patterns of loading namely:

62
nodal loading and element loading. In the nodal loading pattern loads directly act on the
nodes and in the element loading pattern loads act on the elements. In this program
element loads are applied as member end actions, e.g., uniformly distributed vertical loads
acting on the beams are converted to a vertical shear and a moment acting at both the ends
of a beam element. As it is known that moment at the member end causes relative rotation
in the semi-rigid connections, the applied moments generated due element loading must
be modified to incorporate the flexibility effect of connections. This incorporation of non-
linear effect of partially restrained connection on member end actions is done in
subroutine PLOAD.

4.3.2.4 Subroutine STRESS

After analyzing the frame, the forces and displacements at the member ends are calculated
in the subroutine STRESS. Subroutine BEAMSTR is called to get the element global
displacement matrix UGE. Subroutine BEAMSL is called to get the element stiffuess
matrix SL and transformation matrix T. The global displacements are then converted into
local displacement matrix UL by multiplying matrix T with matrix UGE. To get the
member end force matrix FORCE, matrices SL and UL are multiplied. The member end
forces and displacements are printed in proper format.

4.3.2.5 Subroutine ICONV

Subroutine ICONY deals with the convergence criterion for the analytical procedure.
After every increment of loading the difference in relative rotations of the joints must
converge to a certain degree of tolerance. In fact the convergence is tested after each
iteration. If the results converge, next increment of loading is introduced, if not the
iterative procedure continues until convergence occurs for a given increment of loading.
In the program the convergence is checked according to the procedure given by Frye and
Morris [25]. The tolerance limit for convergence is taken to be 5%.

64
4.4 CONCLUDING REMARKS

The main program and the principal subroutines developed for the computer analysis have
been described. However, there are some more subroutine which are not described above
because of their simple nature. Among these are the subroutines BOUND, EQSOLV,
MATMUL, TRANS, STORE. Subroutine BOUND applies the boundary conditions for
the structure to be solved. Subroutine EQSOL V solves the equation developed in the
stiffness analysis. Subroutine MATMUL is for multiplying two matrices, subroutine
TRANS is for generating the transformation matrix and subroutine STORE is for
assembling the coefficients of element stiffness matrix, SE into the Structure Stiffness
Matrix SS. A file titled FRAME.CMM is also included at the beginning of each
subroutine and the main program. FRAME.CMM contains common statements,
dimensional statements etc. In this chapter, some flow charts for the main program as well
as for some important subroutines are presented to show the sequence of the analytical
procedure clearly.

65

..
'"" ..
CHAPTERS.
BEHAVIORAL STUDY ON PARTIALLY RESTRAINED FRAMES

5.1 GENERAL .

The computer program as developed based on the analytical model described in chapter 3
enable the solution of a wide variety of problems. In this chapter several framed structures
are treated as sample problem to show the application of the program in solving semi-
rigid frames accurately and also to arrive at some important conclusions regarding the
effects of semi-rigid connections on multi-storied steel frames. A wide variety of
connections, flexible to fairly rigid and perfectly rigid (fully welded) connections are used
in these analyses. In this chapter, four investigations are presented. The first investigation
reveals the effect of semi-rigid connections on the behavior of frames. Frames with
different types of semi-rigid connections at the beam-ends are analyzed. The distributi~m
of forces in the members and lateral drift are studied and presented. The second
investigation is related to the effect of bracing on the behavior of frames. Both braced and
unbraced frames are analyzed to study the effect of bracing on the distribution of member
forces and sway of the frame. Mixed use of rigid and semi-rigid connections is also
investigated to study the behavior of frames and to provide some design economy.
Finally, comparisons between rational method of analysis and a single stiffness (initial
stiffness of connection) linear analysis are presented. The verification of the program is
also carried out by taking the example problems used earlier by the researchers [25,41].

5.2 VERIFICATION OF THE COMPUTER PROGRAM

To verify the validity of the computer program as developed in this study, a 2 bay II
story frame as shown in Fig.5.1 is considered. This frame is analyzed earlier by Frye and
Morris [25], and also by Dhillon and Majid [41]. The results are compared with those
obtained by Frye and Morris [25], and also by Dhillon and Majid [41]. Figure 5.1 shows
the frame configuration, dimensions, loading and numbering of joints and the top
deflection of the frame due to the loading. The analysis is performed considering the
following different connections as used by the researchers:

66
/'

\ "'\
AlL 8£AIo4S•• W21 X82
Pl_61 kipl5
~.4B"--+
t---s "-+
P2-71 kips

Ip2
po3-45

I P2
kips

I P2 P,
f-52S"-f
2B
+--037
f---s.36"-f
"-+
I~
5~
::-1
11.7~
P' IP2 I P2 I P2 P' ~I
~I I
It!.
'" ~
<til] ,'';
~ PI ~ -I ~ I '"
-
- .
P2 P'
10.8911.
• ;; :c I e i1
• • •
• ,.; , f!
'" 1 ~
P' P2 P'

I
I ::-1
• H
1 ti ' :
-• "'u
~ P1 P2 P1 ~
10.8911.
;; I

10.0411.
• P, Ip2 P2
W21X83

IP2 P,

• 0,
0
~
I
I
I ,I

,I
I

0•,
1
••• I
I
0'0 10.0411.
-
N
P' P2 Pl g 0 I
I
" •
N

•• -
0

9.60311.
• P'
• P2 P,
• I
I

I I

W21X803
I I
~ Pl Ip2 ;! P2 P2 PI ~ I
9.20311. N
~ N
I
• •• I

8.3711.
• •
• P2 P,
• I
1 (10)
, I

(') (8)
P' P2
10.9511. P'
, ~
• •
~ m
~ (5) • ~ -'-
• •• N
0,


~ (1) (2) (J) ::::

• • ;;,
,
~' 2 (

(a) Unbraced frame (b) Top deflection using rigid connection (c) Top deflection using T-stub connection

Fig.5.! Verification of the computer program

~--..)
• Rigid connections.
• T-stub connections.

The moments and deflections of the analyses are summarized in Table 5.1 and Table 5.2
and the sway along the story levels are shown in Fig.5.!. The tables show that the results
obtained by the developed computer program agree well with the results of the analyses
by Frye and Morris [25] and those of Dhillon and Majid [41].

Table5.!' Comparison of support reactions and top deflections using Rigid and T-stub
connections.
Analysis Connection Moment at joint Vertical Force at joint Top %of
Method Type (kip-in) (kip) Deflection diff.
(in)
1 2 3 I 2 3
Present Study Rigid 6033 8520 5904 888 1531 1154 5.28 0.02%
Ref. [41] 6103 8602
.
5953 889 1531 1171 5.25 -
Ref. [25] - - - - - - 5.25 1.4%

Present Study T-stub 6583 9091 6242 893 1525 1155 6.37 0.02%
Ref. [41] 6666 9194
Ref. [25]
6290 885 1527 1162 6.48 -
- - - - - - 6.36 1.6%

Table5.2. Comparison of moments in beams and columns using Rigid and T-stub
connections.
Analysis Connection Moments in beams (kip-in) Moment in col. (kip-in) %of
Method Type 4 5 9 10 7 6,8 diff.
Present Study Rigid 6721 6377 6609 6428 3888 3619 0.18%-
Ref. [41] 6684 6348 6595 6416 3905 3641 0.6%

Present Study T-stub 6125 5914 6155 6053 4016 3605 0.1%-
Ref. [41] 6167 5809 6231 5971 4102 3598 2.0%

From the above tables it can be observed that the percentage of difference between the
results are only 0.1% to 2%. It can be observed from Fig.5.1b and Fig.5.1c that the top

68
deflection as obtained by this study are 5.28 and 6.37 inch for Rigid and T-stub
connection respectively. The top deflection by Frye and Morris [25] and Dhillon and
Majid [41] are 5.25 inch for Rigid connection and 6.36 inch and 6.48 inch for T-stub
connection respectively. The variation is only 0.6% for Rigid connection and 0.2% to
1.7% for T-stub connection. Thus the comparison shows that the results of the computer
program are in good agreement with those obtained by the researchers [25,41]. This
confirms that the computer program developed in this research work to analyze a steel
frame with rigid or semi-rigid connections is quite representative and accurate.

5.3 EFFECT OF SEMI-RIGID CONNECTIONS ON THE DISTRIBUTION OF


FORCES AND LATERAL SWAY.

To determine the extent to which column and girder moments are affected by connection
flexibility, two frames i.e., a 2 bay 4 story frame and a 3 bay 4 story frame as shown in
Fig.5.2, are analyzed. A 22ft bay width and 12 ft story height is maintained for all the
frames. The frames are spaced at 24ft intervals. Beams and columns sections are so
chosen that are used in practice for medium rise buildings. Beam and column sections
used in this study are W14x34 and W14x38.

Six types of semi-rigid connections are used, ranging from soft semi-rigid connection, i.e.
Double web angle connection to stiff semi-rigid i.e. T-stub connection. In addition a
perfectly Rigid connection (fully welded) is also included in the analyses. Thus the
connection types used in this study are as follows:

1. Double web angle connection (referred as connection type A [Fig.2.6] in the graphs).
2. Header plate connection (referred as connection type B [Fig.2.6] in the graphs).
3. Top and seat angle connection (referred as connection type C [Fig.2.6] in the graphs).
4. Top and seat angle with double web angle (referred as connection type D [Fig.2.6] in
the graphs).

5. End plate connection (referred as connection type E [Fig.2.6] in the graphs).


6. T-stub connection (referred as connection type F [Fig.2.6] in the graphs).
7. Rigid connection (referred as connection type G [Fig.2.6] in the graphs).

69
~
,
N
o

I~ __ I
T 2022' ••44' ,

Fig.5.2a Gravity loading on 2 bay frame

O.lk/fl

O.175k/ft

W14X.:54

'"•,
N

:ll :ll
x ~

;; •
;;
x
•;;

I~ __ ! I>--- ~I
T 2022'-44' , T .:5022'=66' ,

Fig.5.2b Gravity and lateral loading on 2 bay and 3 bay frame

70
The size parameters used for these types of connections are given in Table 5.3.

Table 5.3. Connection size parameters.


Conn. Depth of Total Thickness Thickness Gage Thickness Fastener Length
Type anglelbeam thickness of top of web length of dia of top
/plate (d) of angle/ angle angle / (g) web/plate (f) angle/
flanges of (t,) header (t" / 1,) T-stub
T-stub & plate (I)
column (t) (tJ
A 10.0 - - 0.375 - -
5.5 -
B 9.5 - - 0.375 5.5 0.285 - -
C 14.0 0.89 - - - - 0.875 8.0

D 14.0 - 0.375 0.375 1.75 - - 8.0

E 17.0 - - - - 0.75 1.0 -


F 14.0 1.265 - - - - 1.0 8.Q

* all the dimensions are in inches.

Steel is taken as the material for the members in the analyses. The modulus of elasticity,
the modulus of rigidity, and the Poisson's ratio of steel are taken to be 29000 ksi, 11200
ksi and 0.25 respectively.

The present study has been intended to examine the effect of semi-rigid connections on
the behavior of multistoried building frames under working load. Frames have been
analyzed for both gravity and the combined actions of gravity and lateral loading. The
gravity load includes both the dead loads of the structure and the live loads for residential
purposes. The dead load and the live load applied on the frame are shown in Table 5.4.
The load on the roof is not same to the loads on the others floors.

Table 5.4. Loading on the. frames.


Floor levels Dead Load Live Load Wind load
All floors 47.5 psf 40 psf 25 psf
Roof level 30 psf 20 psf 25 psf

71
The effect of semi-rigid connections on steel frames is mainly characterized by the
member force actions and the lateral drift of the whole structure. In this study the beam
moments, column moments and the top deflection of the frames are determined for semi-
rigid framing (Type 3) and simple framing (Type 2) analyses as recognized by AISC. The
results of the rational analysis (Type 3) are compared with those obtained by Type 2, i.e.
simple framing analysis. As mentioned earlier, for Type 2 analysis, the frames are
analyzed by considering all the connections at the beam ends to be perfectly pinned under
gravity load and perfectly rigid connections against lateral loads.

5.3.1 Effect under Gravity Loading

The effect of gravity loads on beams with semi-rigid connections is presented in Table
5.5. and plotted in Fig.5.3. Table 5.5. presents the critical moment (positive or negative)
developed in the beams for both Type 3 and Type 2 analysis for different types of
connections for the 2 bay frame. Critical moment ratios of the beams are plotted against
connection type as well as story levels for 2 bay and 3 bay frames. Critical moment ratio
is the ratio of the maximum moment (positive or negative) developed due to simple
framing or Type 2 analysis and that of the rational analysis or Type 3 analysis. From
Fig.5.3 it can be found that the critical moment ratios for beams at all floor levels are
greater than 1.0, i.e. simple framing method provides over designing of beams under
gravity loads. The percentages of over design for different types of connections vary from
25% to 95 % under gravity loading. It can also be found that for stiff semi-rigid
connections the over design factors are higher (55% to 65%) than the soft semi-rigid
connections (25% to 55%). It is noticeable that for connection type D, i.e. Top and seat
with double web angle, the critical moment ratios for the beams are the largest (1.75 to
1.95) and thus the over design factor varies from 75% to 95%. So by using connection
type (D) and employing rational analysis procedure, maximum economy in the design of
beams can be achieved.

It can be observed that for soft semi-rigid connections such as; Double web angle (A),
Header plate (B) and Top and seat angle (C) the critical moment ratios are almost
identical at all the floors (except the top floor). For stiff semi-rigid connections such as

72
2.0
-.-Ior conn. type A
-. - lor conn. type B
-&- for conn. type C
-T- rOt conn. type 0
-+- for conn. type E
1.8 -+- for amn. type F

o
:;::
-X-for conn. type G

-
~
c
Q)

E
1.6

o
E
rou x--------------
:;::
.C 1.4 • •
u

• •
1.2
1 2 3 4
story levels

(a)

-.- for story level 1

2.0 -e- for story level 2


-.A.- for story levelS
-T- for story level"

o 1.8
:;::

-~
c
Q)

E
1.6
o
E
ro
:;::
u 1.4
.C
u

1.2

A B C D E F G

connection types

(b)

Fig.5.3 Critical moment ratio of beams for gravity load


(a) with story level, (b) with connection types.

73
End plate (E) and T-stub (F) connections the variation of critical moment ratios at all
floor levels are 10% or less both for 2 bay and 3 bay frames (Fig.5.3). From Fig 5.3 it can
also be observed that for story level I, 2 and 3, the critical moment ratios for a particular
type of connection does not vary appreciably. The variation occurs only at the top floor
where the loading is different from the other floors. But the variation is within 10% to
20% depending upon the type of connection.

Table.5.5. Maximum moments (kip-in) in beams of2 bay 4 story frame for gravity
loading.
Connection type Beam Moments (kip-in)

Story 1 Story 2 Story 3 Story 4

Type 3
A 1201 1201 1200 629
B 1090 1089 1088 587
C 959 956 956 509
D 873 870 869 443
E -901 . -876 -868 -491
F -964 -937 -929 -542
G -1070 -1031 -1021 -588
Type 2
simple connection 1524 1524 1524 871
- indicates that the critical moment is the maximum negative moment.

5.3.2 Combined Action of Gravity and Lateral Loading

The effect of the combined action of gravity and lateral loading on beams is presented in
Table 5.6 and 5.7 and plotted in Fig.5.4 and Fig.5.5. Table 5.6 and 5.7 presents the
maximum moment generated by both Type 3 and Type 2 analysis. Graphs are plotted
taking critical moment ratio against connection type as well as story level for both 2 bay
and 3 bay frames. From Fig.5.4 and Fig.5.5 two distinct pattern of curves can be
observed. Connection types A, Band C (soft semi-rigid connections) follow one pattern
and connection types E and F (stiff semi-rigid connections) and the Rigid connection G
follow the other pattern. As for connection type D (intermediate stiff connection), it
sometimes follows the pattern of soft semi-rigid connections (for 2 bay frame) and

74
1.6

1.5

-
0
:,::;
~
c:
1.4

1.3
.•. ~...
,.-~~
~

Q)
E
0
E
1.2 . . ~
co
tl
:,::;
'C
tl
1.1
:
+~~
// /~~:: ..•.•""'m. ..".c
1.0 -y-f01conn.IypeO
....•.... for conn. type E
. -+- for conn. type F
X
~
-x- lor conn. typeG
0.9
1 2 3 4
story level
(a)

/
1.8

o
:,::;
1.6

,.-~,.~~
-~
c:
Q)
E 1.4
.•.~ .•.~
o
E
""iii
tl
:,::;
'C
tl
1.2 :-~~.~
~
------
.~
~~
---------
'=i=fOrOOflrl.
.-
type A

for conn. type B


________ .4- lor conn. type C

1.0 X •••
+-
=:;:~~:::~
for corm. type F
X-for conn type G

1 2 3 4
story level
(b)

Fig.5.4 Critical moment ratio of beams with story level for gravity
and lateral loads (a) for 2 bay frame, (b) for 3 bay frame.

75
1.6
Y?,. .
1.4

-C
Q)
E
/.~ '--.~Y
1.2
o
Y~' •
E

1.0
:~
• -.-forsloryleveI1
-e- forstorylevel 2
:~.~
~.~ """.

-A- for story level 3


-,.-- for slory level 4 •

A B c D E F G

connection type

(a)

1.8

o
:;::
1.6 /~\
-~
C
Q)
E
1.4
Y// ~I~
o
E
ro 1.2
.~:/ .~.~:
C,)
:;::
'C
C,)
i~ -.-forstoryleveI1 .,--.~.

-e- for story level 2 •


1.0
-A- for slory level 3
-T- for slory level 4

A B c D E F G

connection type

(b)
Fig.5.5 Critical moment ratio of beams with connection types
for gravity & lateral loads. (a) for 2 bay frame,
(b) for 3 bay frame.

76
sometimes follows the pattern of stiff semi-rigid connections (for 3 bay frame). In other
words, the behavior of the frame with connection type D depends on the slenderness ratio
of the frame. It is a transition type of connection between flexible and stiffer connections.
Again from Fig.5A and Fig.5.5 it can be observed that critical moment ratios of beams are
greater than 1.0 for all types of connections. This indicates that the beams are consistently
over designed following Type 2 or simple framing analysis method. Depending upon the
slenderness ratio of the frames, the percentage of over design varies from 10% to 85% for
different types of connections. For soft semi-rigid connections the over design factor
varies from 10% to 60%. For stiff semi-rigid connections it is 10% to 50%. But for
intermediate stiff semi-rigid connection (connection type D) the over design factor varies
from 50% up to 85% (critical moment ratio 1.5 to 1.85). It is also observed that for the
case of combined action of gravity and lateral load the degree of over design in beams is
lower than that for the case of gravity load only.

Soft connections, such as Double web angle (A), Header plate (B) and Top and seat angle
connection (C), show quite similar critical moment ratios for all stories. The critical
moment ratios for these 3 types of connections vary up to 10% which is negligible. The
critical moment ratios for intermediate and stiff semi-rigid connections like Top and seat
angle with double web angle (D), End plate (E) and T-stub (F) are more sensitive. With
respect to story level, the critical moment ratios for stiffer connections vary considerably.
In addition, it can also be noted that difference in over design for stiff semi-rigid
connection depends on the slenderness ratio of the frame. Over design is more for frames
with higher slenderness ratio. In other words, End plate and T-stub connections the
variation of critical moment ratios are very wide for higher slenderness ratio (2 bay
frame). The variation ranges up to 40% with the story level. For soft semi-rigid
connections such as A, Band C the ratio does not vary considerably with the story levels.
The critical moment ratio for beams is lowest at the ground floor and increases in upper
floors and reaches the maximum at the top floor.

The effects of gravity and lateral loading on columns are divided into two parts, such as:

77
Table.5.6 Maximum moments (kip-in) in beams of2 bay 4 story frame for gravity and
lateral loading.
Connection type Beam Moments (kip-in)

Story I Story 2 Story 3 Story 4

Type 3
A 1410 1439 1384 745
B 1255 1272 1206 639
C 1116 1110 1054 552
D 1079 1067 989 570
E -1432 -1340 -1141 -607
F -1462 -1373 -1185 -653
G -1634 -1502 -1285 -690
Type 2
simple framing 1524 1524 1524 871
- indicates that the critical moment is the maximum negative moment.

Table.5.7 Maximum moments (kip-in) in beams of3 bay 4 story frame for gravity and
lateral loading.
Connection type Beam Moments (kip-in)

Story I Story 2 Story 3 Story 4

Type 3
A 1349 1360 1313 694
B 1194 1198 1163 616
C 1066 1058 1019 535
D 1014 999 946 471
E -1274 -1197 -1054 -570
F -1318 -1242 -1104 -617
G -1465 -1358 -1202 -656
Type 2
simple framing 1524 1524 1524 871
- mdIcates that the en/ieal moment is the maximum negatIve moment.

(a) Exterior Columll

The effects of gravity and lateral loading on the exterior columns of frames are presented
in Table 5.8 and Table 5.9 and plotted in Fig5.6 and Fig.5.7. From Fig.5.6 and Fig.5.7, it
can be observed that connections A, B, C and D follows one pattern and E, F and G

78
0.8

-
""C1l

E
L.

C
Q)
0.6

0 ~
E 0.4
C1l -.- for conn. type A
U -e- for conn. type B
~
'C
""U -4- for conn. lypeC
-y- for conn. type 0
-+- for conn. type E
0.2 -+- for conn. type F
-X- for conn. type G

1 2 3 4
story level
(a)

0.8

-
""~
C
Q)
E
0.6

o
E 0.4 .- for conn. type A
C1l -e- for conn, type B
U
-4.- for conn. type C
'C
""U -..,- for conn. type 0

- for conn, type E


+- for conn. type F
0.2 -X- for conn. type G

1 2 3 4
story level
(b)
Fig.5.6. Critical moment ratio of exterior column with story levels
for gravity and lateral loads. (a) for 2 bay frame, (b) for
3 bay frame.

79
1.0

-.- for story level 1


.....• ~ for story level 2
0.8 -.A.- for story level 3 •
-T-for slorylevel4 ~.---'-

------------- -
-~-
o ./'
--------
:;::: A -7 -----.
-~
c
Q)
0.6
.•.-------...
--.It..~
""
"" e_e .
E
o
E
0.4 .•.. _- -- .•.--- .•.
rou y~ '
:;::: y_y
'C
u 0.2 ----y--- '---'y---y

0.0
A 8 c o E F G

connection type
(a)
1.0

-.- for story level 1

0.8 -e- forstorylevel 2


-.&- for story level 3
-T- for story level 4
.---------
,_--1_---
-
o
:;:::
~ 0.6 ..~:---------- ----
------
c
Q)
E
o 0.4
A

~
---.

.•.-------...
--
_

.~ ~
.•.________.
e e__ •
E
rou .•.- .•.,--- .•.
:;:::
'C
u 0.2 y -------------
y_y
_yo Y' y. y

0.0
A 8 c o E F G

connection type
(b)

Fig.5.7 Critical moment ratio of exterior column with connection types for
gravity and lateral loads. (a) for 2 bay frame, (b) for 3 bay frame.

80
0.85

0.80 "
..--"~
0.75

"/"
. /"----
o 0.70
""~
-c
Q)
E
0.65

0.60-
o "~.
E
c
OJ
.Vi
Q)
0.55-

0.50 .~ .-------."'" -
"0 0.45
• & 21 "'"
r
for slory 1
0.40

0.35
-e- forslory3&41 e ._

,
--.
A B c D E F G
connection types

(a)

0.8

0.7 ~" -"--"


o .-----"
""~ /"
-
c
Q)
0.6
"
------"
E
o
E 0.5 .~
c
OJ
.Vi .
.~.~.--.-
'----------
Q)
"0
0.4 .--------..
-.- for story 1& 2
-e- forstory 3 & 4
0.3

A B c D E F G

connection types
(b)
Fig.5.B Design moment ratio of exterior column for gravity and
lateral loads. (a) for 2 bay frame, (b) for 3 bay frame.

81
follows the other pattern. It is quite evident from the graphs that the critical moment of
column moments at the first story level is 0.58 to 0.82 i.e. below 1.0. This means that
Type 2 method provides under design for the columns in the first story or base level. For
soft semi-rigid connections the percentage of under design is quite noticeable at the base
level. It reaches up to 40% (critical moment ratio 0.6) for connections like Double web
angle (A), Header plate (B) and Top and seat angle (C) connection. For stiff semi-rigid
connections such as End plate (E) and T-stub connections (F), the critical moment ratios
are 0.75-0.82, i.e. close to I and the percentages of under design, following Type 2
method, range from 20% t025% for exterior columns. For intermediate stiff semi-rigid
connection (connection type D) the percentage of under design varies from 25% to 35%
(0.67 to 0.72). For all other floors, the critical moment ratio for exterior column in ihe
case of stiff semi-rigid connections (connection type E and F) decreases with the increase
in story level reaching the minimum at the top floor, i.e. for connections E and F the
percentages of under design is very high (max. 80%) at the top floor.

Generally for low rise building frames usually one section is provided for each type of
columns. But for high rise building frames column sections are varied throughout the
height of the frame. So, for low-rise buildings the design moment for a particular type of
column is generally limited to one. For high rise buildings there may be several design
moments for different story levels. It is known that when gravity and lateral loads act on
frame, the maximum column moments develop at the base level for exterior columns. So,
for the 4-story frames as used in this study, the moment developed at the base level of a
column may be considered as the design moment for that column for all story levels . .In
that case the design moment ratio is, therefore, the critical moment ratio at the base level.
The design moment ratios of the exterior column for different types of connections are
presented in Fig.5.8. It can be observed from this figure that the design moments for all
types of connections are below 1.0, which clearly states that Type 2 method of analysis
for columns provides severe under design of columns, reaching up to 40%. However, if
two design moments are considered (Fig.5.8) for the design of the exterior columns such
as one for first two stories and the second for the top two stories, it can be observed that
for both the conditions the design moment ratios for the exterior column are well below

82
1.0 (70% under design in column for the top two stories). This means that Type 2 method
provides under design for exterior column which is very high at the upper levels.

Table.5.8 .Maximum moments (kip-in) in columns of 2 bay 4 story frame for gravity and
lateral loading. .
Connection ExteriorColumn Moments (kip-in) InteriorColumn moments (kip-in)
type Story 1 Story2 Story 3 Story 4 Story 1 Story 2 Story 3 Story 4
Type 3
A 1354 538 456 352 1386 631 514 405
B 1320 593 545 456 1323 607 525 405
C 1168 596 593 509 1159 574 515 340
D 1122 610 609 562 1112 592 510 325
E 1016 809 720 567 952 630 456 231
F 1004 823 733 594 936 632 452 226
G 965 857 751 604 891 637 433 204
Type 2
simple 791 385 288 115 891 637 433 204
framing

Table.5.9 Maximum moments (kip-in) in columns of 3 bay 4 story frame for gravity and
lateral loading.
Connection ExteriorColumn Moments (kip-in) InteriorColumn moments (kip-in)
type Story 1 Story 2 Story 3 Story 4 Story 1 Story 2 Story 3 Story 4
Type 3
A 989 424 377 272 996 434 371 272
B 968 481 455 403 947 416 377 272
C 867 504 501 456 828 405 363 226
D 839 535 522 510 791 417 358 214
E 776 689 625 520 685 440 317 152
F 770 707 643 547 672 440 313 149
G 747 738 664 56! 644 449 295 14f
Type 2
simple 576 269 200 77 638 434 295 134
framing

83

\,
\
(b) Interior Columns

The effect of the combined action of gravity and lateral loads on the interior columns of
the frames are given in Table 5.8 and 5.9 and also plotted in Fig.5.9 and Fig5.10. Again it
can be observed that for most of the story levels including the first story, the critical
moment ratios for the interior columns are less than 1.0, i.e. simple framing or Type 2
method provides under design for interior columns too. For soft semi-rigid connections
(A and B) the critical moment ratio reaches up to 0.64 at the first story or base level. For
End plate (E) and T-stub connections (F) the critical moment ratio is close to 1.0 but less
than 1.0. This demonstrates that under design is more for soft semi-rigid connections than
the stiff semi-rigid connections following Type 2 analysis. It is interesting to note that the
critical moment ratio for all the connections at story level 2 is 1.0 or close to 1.0 (1.0-
1.07) for interior columns. This means that for all connections the moment~ generated at
story level 2 using rational analysis (Type 3) are very close to the moment generated by
Type 2 analysis. It is also found from Fig.5.1 0 that the critical moment ratio increases for
the stiff semi-rigid connections, which means the risk of unconservative in the design of
interior columns can be reduced significantly by providing stiff semi-rigid connections
such as End plate (E) and T-stub (F) connections.

The' design moment ratios for the interior columns for different types of connections are
presented in Fig.5.II. As mentioned earlier, when two design moments for column design
are considered, it can be found that both the design moment ratios are less than 1.0 for all
types of connections and thus Type 2 analysis provides under design for the interior
columns in all floors. The percentages of under design for End plate and T-stub
connections are around 5% to 10%. For more flexible connections it reaches up to 35%.

Lateral Drift (top deflectioll)

For lateral loading the top deflection of building frame is an important criterion for the
design. ASD Specification of AISC [3]specifies that the ratio of top deflection to the total
height of the building (LVH)should not be greater than 1/400. It is natural that the stiffer
the connections are the less will be the top deflection. Increase in the number of bays also
decreases the lateral drift of the frame. Fig.5.12 shows the ~/H for different types of

84
1.2

1.0 x
0 +
"
."~ •
-
<1l

c
OJ
E
0.8 ,.
0 •• ~.-
---e~ for
for conn. type A
conn. type B
E -.0\- for conn. type C
<1l
<.) •• -"f'- for conn. type 0
-+- for conn. type E

."
'C
<.)
0.6 -+- for conn. type F
-X-for conn. type G

0.4
1 2 3 4
story level
(a)
1.2

1.0 x
0 + ~
." •
-
~
C
OJ
E 0.8 ,.
~

0 •• .~
.-
for conn. type A
for conn. type B
E A.- for conn. type C
<1l
<.) •• -"f'- for conn. type 0
-+-
."
for conn. type E
'C 0.6 +- for conn. type F
<.) x- for conn. type G

0.4
1 2 3 4
story level
(b)

Fig.5.9 Critical moment ratio of interior column with story levels for
gravity and lateral loads. (a) for 2 bay frame. (b) for 3 bay
frame.

85
1.2

1.0 .-------
.---------- .-----------... .~--.'---'.
o
~~:=='=/
:;::;

-~
c:
Q) 0.8
.. .. .. _:j;?/r-- T

E
o
E ./ •.------- =:=:::::~:::::;
0.6
.----

/T____' T -T-f""0", '",',


-.&.- for slory level 3

T T

0.4
A B c D E F G

connection type
(a)
1.2

1.0
o
:;::;

-~
c:
Q)
E
0.8
o
E
rou -.- for story level 1

:;::; 0.6 -e- for slory level 2


'C -4- for story level 3
u -T- for story level4

0.4
A B c D E F G

connection type
(b)

Fig.5.10 Critical moment ratio of interior column with connection


types for gravity & lateral loads. (a) for 2 bay frame,
(b) for 3 bay frame.

86
1.1.,------- ---,

1.0

o
""~
-
c
OJ
E
0.9

o
E 0.8
c
0>
.Ui _ forS10l'y1&21
OJ
"0 0.7
I -e- for story 3 & 4 I

1.0

-
c
OJ
E
0.9

o
E 0.8
c
0>
.Ui
OJ
"0 0.7 I .-fors!ory1
-e-
&21
for story 3 & 4 I

0.6 -'---.,--,---,---,---,-----,,---,r---,-----J
ABC D E F G

connection types
(b)

Fig.5.11 Design moment ratios for interior column.


(a) for 2 bay frame, (b) for 3 bay frame.

87
0.010

0.008
.~ -.- for 2 bay 4 story frame
....• - for 3 bay 4 story frame

I
<1
0.006
.~'\.
.~~.
0.004

!>IH~11400 _
.~.~ _~.~.~.
0.002 e _


A 8 c o E F G
connection types

Fig.5.12 . L'./ H with connection types

88
connections. Only End plate (E) and T-stub (F) connections for the 3 bay frame satisfy the
criterion (iVIl ~ 1/400). For the 2 bay frame the limit is satisfied only by rigid connection
and not by any other semi-rigid connections. Thus by providing stiff semi-rigid
connection and increasing the number of bays the limit can be achieved. However, the top
deflection also depends on the size, shape and material properties ofthe beam and column
members.

5.3.3 Summary

The results indicate that the design of beams and columns are highly sensitive to
connection flexibility. Simple framing or Type 2 analysis consistently overestimates beam
moments and under estimates column moments especially exterior columns. The over
design of beams, following Type 2 analysis reaches up to 40% for soft connections, 60%
to 80% for intermediate stiff semi-rigid connections and 40% to 50% for stiff semi-rigid
connections. The under design of interior columns following Type 2 analysis ranges up to
30% to 40% for soft connectio'ns, 20% to 25% for intermediate stiff semi-rigid
connections and 10% to 15% for stiff semi-rigid connections. The percentages of over and
und~r estimation in moments of beams and columns using different types of semi-rigid
connections are presented in Table 5.10 and Table 5.11. The values in the Tables 5.11 are
the single design moment for columns for all the floors. It should be noted that the degree
of risk in the under design of exterior column is very high compared to interior columns.
Over designing of beams may be structurally safe, but under designing of columns may
equally be fatal. So, for the design of columns, both exterior and interior, the effect of
connection flexibility must be considered properly and accurately.

89
Table 5.10 Percentages of over estimation in the design moment of beams using Type-2
analysis.
Connection type Gravity Gravity + Lateral
+(%) - (%) +(%) - (%)
Double web angle 25 - 40 - 5 - 25 -
.

Header plate 40 - 50 - 20 - 40 -
Top & seat angle 55 - 70 - 35 - 60 -
Top & seat with D.W. 75 - 95 - 40 - 85 -
End plate 65 - 80 - 5 c 50 -
T-stub 55-70 - 5 - 40 -
+ indicates over estimation
- indicates under estimation

Table 5.11 Percentages of under estimation in the design moment of columns using
Type-2 analysis.
Connection type Exterior Columns Interior Columns
+(%) - (%) + (%) - (%)
Double web angle - 40 - 45 - 35 - 40
Header plate - 35 - 45 - .30 - 35
Top & seat angle - 30 - 35 - 20 - 25
Top & seat with D.W. - 25 - 30 - 15 - 25
End plate - 20 - 25 - 5 - 10
T-stub - 20 - 25 - 5 - 10
+ indicates over estimation
- indicates under estimation

90
5.4 EFFECT OF BRACING IN SEMI-RIGID FRAME

From the previous investigations it is observed that even low rise frames with soft to
intermediate stiff semi-rigid connections do not satisfy the criterion of lateral drift (MH :$

1/400). So, controlling the lateral drift is an important criterion in the design of steel
frames. The most commonly used sway resisting system is the use of bracing. In this
section, a tall building frame under working load has been investigated with or with
bracing for different types of connections.

A 3 bay 10 story frame as shown in Fig.5.13 is being analyzed for both gravity and lateral
loading to investigate the effect of bracing on lateral drift of the frame in this section.
Once again the bay width and the story height of the frame are taken as 22 ft and 12 ft
respectively. The frames are spaced at 24 ft intervals. The bracing are provided diagona!ly
in the mid panel of all the floors shown in Fig. 5.13.

Beams, columns and bracing sections are so chosen that are used in practice for high rise
buildings. The beam section used in this study is W18x60 for all floors. The column
sections are changed with the height of the frame. The section of all the columns up to 3'd
floor is taken as W 14x132, from 4th floor to 6th floor W 14xl09 and form 7th floor to 9th
floor W 14x90. The section of the bracing are taken as W 14x22. Material properties of
the members are taken as same as that of earlier investigations. The dead load, the live
load and the wind load applied on the frame are shown in Table 5.12.

In this investigation Double web angle (A), Top and seat angle (C), End plate (E) and
Rigid connection are used. The parameters used for these types of connections are given
in Table 5.13.

Table5.12 Loading pattern on the frame.


Floor levels . Dead Load Live Load Wind load
Ground - 3" floor 47.5 psf 40 psf 25 psf
4th - 6th floor 47.5 psf 40 psf 30 psf
7" - 9th floor 30 psf 20 psf 35 psf

9t
Coil Col2 Col3 Col4
I I I I
I I I I

o o o
'"
x
..,. '"
x
..,. '"x..,.
;:

'0
N
~
II
N

@
0

'"
0 '"
0
x
..,. x
..,.
;: ;:

"'x..,. "'x..,.
;: ;:

t 3@22'=66' t

Fig.5.13 Effect of bracing

92
~.;, ..

\
Table 5. I 3 Connection size parameters.
Conn. Depth of Total Thickness Thickness Gage Thickness Fastener Length
Type anglelbeam thickness oftop of web length of dia of top
/plate (d) of angle/ angle angle / (g) web/plate (I) angle/
flanges of (tJ header (" / lp) T-stlib
T-stub & plate (I)
column (t) (tJ
A 12.0 - - 0.5 5.5 - - -
C 18.0 1.21 - 0.5 - - 0.875 8.0

E 21.0 - - - - 0.75 1.0 -

* all the dimensions are in inches.

5.4.1 Effect of Bracing

The variation of critical beam moments for different types of connections using both
braced and unbraced frames are presented in Fig.5.14. It is noted from the graphs that the
critical moments of the beams for unbraced frames are much higher than those for brac.ed
frames. A new term called 'Moment Reduction Factor' (M.R.F.) is introduced in this
study. The M.R.F. is the ratio of critical moments found in the braced frame and that of
the unbraced frame analysis. Variation of M.R.F. with story levels for different types of
connections is also presented in Fig.5.I5. From Fig5. 15 it can be observed that the M.R.F.
for beams are much lower than 1.0 at the lower level stories than the upper level stories.
At story level 2 and 3 the M.R.F. reduces up to 0.57 for Double web angle connection and
0.6 for Top and seat angle connection. This indicates that the effect of bracing on beam
moment is much prominent on the lower level stories than the upper level stories and the
reduction can reach up to 40%. For stiff semi-rigid connections the M.R.F. is somewhat
lower than the soft semi-rigid connections. The patterns of graphs are almost similar for
all types of connections used.

Variation of critical moments of columns for both braced and unbraced frames usihg
different types of connections are presented in Fig.5.16, Fig.5.17, Fig.5.18 and Fig.5.19. It
can be observed that the patterns of the graphs are same for all the columns. It is also
observed that the critical moments generated due the use of bracing are much less than

93
2000
.------- •
-------.-.~.-.
1600
.-.-----.-.~ ~.----.---."
c
1200
.
.-.~.-------.~ ~
.•......••
.•...•
I=L=:=I~~:=:=~~:~
c 800
Q)
E
o -.- for conn type G "

E 400 --e- for conn. type E


-.&- for conn. type C
- .•.~ for conn. type A

o
o 1 2 3 4 5 6 7 8 9 10 11
story level

2800

,2400

2000
c
Q.
1600
:.52
.•......••
.•...• 1200
c
Q)

E 800
o
E 400

o
o 2 4 6 8 10

story level
Fig,5.14 Critical beam moments. (a) for braced frame.
(b) for unbraced frame.

94
1.2

-c
c.
~
••....... 0.8
.•....
c
Q)
E -.- conn. type G
0 -e- conn. type E
E -A- conn. type C
-'Y- conn. type A

o 2 4 6 8 10

story level

Fig.5.l5 Moment reduction factor (M.R.F.) for beams.

95
1000
jY

~
C
800
:7)\
. . /:
-.-
-e-
for connection type
for connection type E
-.0\- for connection
-T- for connection
type
type A
G

600
c.
:Q
-
~
c
Q)
400 /:
/ •.
E

/
0
E
200 •
+

0
0 2 4 6 8 10

story level
(a)

3000

-.- for connection type G


c -e- forconnection type E
- .• ~ fOf connection type C
c.
:Q
2000 • -T- for connection type A

-
c
Q)
E
o 1000
E

o
o 2 4 6 8 10
story level
(b)

Fig.5.16 Critical moments for column 1. (a) for braced frame.


(b) for unbraced frame.

96
2000

1600

-.- for connection type G


C

0.
~
1200
~A' -e- for connection type E
-.&- for connection type C
-y- for connection type A

-
~
c 800
•.•.•
Ql
E
0
E ::f;~.
400

0
0 2


4 6
-1 .~~ 8 10
story level
(a)

••
, 4000
.•.

-.- (or connection type G


c 3000 -e- for connection type E
-A- for connection type C
0. -...,~ for connection type A
~

-
~
c
Ql
E
20'00

0
E
1000

o
o 2 4 6 8 10
story level
(b)

Fig.5, 17 Critical moments for column 2, (a) for braced frame,


(b) for un braced frame,

97
2000

1600

~A'
.~ for connecllon type G
C -e- forconnection type E
1200 ~ •. - for connection type C
C. -"'- for connection type A
:52 •• ••
-
~
c 800 ~

:!/
1:'.~
Q)
E •
0
E
400 ..

~l[

0 •
0 2 4 6 8 10
story level
(a)

••
'4000
••
~
,!: 3000
. -.- for connection type G

.~
..\
~ -e- forconnection type E
C. -.0\- for conneclion type C
:52

-
-"'- for connection type A
~
c 2000
.•
~:~
~.~
Q)
E .~t~I---------I
----- -----
0
E
1000
~
.~.-----I~
...• ~
~
•~I~
.~I
0
0 2 4 6 8 10
story level
(b)

Fig,5,18 Critical moment for column 3, (a) for braced frame,


(b) for unbraced frame,

98
1600
-.- for connection type G
-e- for connection type E

• -1..- for connection type C


-T- for connection type A
~
c
1200 •
I ••••
a.
:;;::
~
-c
OJ
E
800


0
E 400

• ••
0 ••
0 2 4 6 8 10
story level
(a)

-.- for connection type G


3000 -e- forconneclion type E
-.&- for connection type C

-"'- for connection type A


~
.S
a. 2000
:;;::
-
~
c
OJ
E
0
E 1000

o
o 2 4 6 8 10
story level
(b)

Fig.5.19 Critical moment for column 4. (a) for braced frame.


(b) for unbraced frame.

99
those generated without the use of bracing. The design moments for both exterior and
interior columns for braced and unbraced frames are presented in Fig.5.20, Fig.5.21,
Fig.5.22 and Fig.5.23 respectively. The columns are divided into 3 groups such as group I
for story 1-4; group 2 for story 5-7 and group 3 for story 8-10 and consequently 3 design
moments are obtained for each column. The M.R.F. for the design moments of the
exterior and the interior columns for both braced and unbraced frames are presented in
Fig.5.21 and Fig.5.23. From the figures, it can be seen that the M.R.F. for exterior and
interior columns can reduce up to 0.37 for Double web angle and Top and seat an~le
connections i.e. bracing provides reduction up to 60% in column moment for group 1
columns (story 1-4) when soft semi-rigid connections are used. For connections like End
plate connection the moment ratio (M.R.F.) is higher than that of soft connections like
Double web angle connection. For other column groups, the M.R.F. is higher compared to
group 1.

The variation of LVII with different types of connections are presented in Fig.5.24. Only
the rigid connection satisfies the lateral drift criterion (LVII ::; 1/400) for unbraced frame.
But for the braced frame the criterion for top deflection is satisfied by all the connections
(both soft and stiff semi-rigid).

5.4.2 Summary

This investigation is done to show the effect of bracing on high rise steel structural frames
with partially restrained connections. It is observed from the investigation that bracing is
not only needed to control the lateral drift of the frame but also to lower the design
moments for the columns as well as for beams and thus allow to design economically.
The additional cost due to the inclusion of bracing can be minimized by the reduction in
cost for reduced size of both beams and columns. Table 5.14 shows the reduction of
beam, exterior column and interior column moments due to the use of bracing.

100
-.- for story 1-4
1600 -e- for story 5-7
~ -A- for story 8-10 •
C

C.
I


_.
32

-
~ 1200
__________ e
C
Q)
E
0 ~e
E - e __________ A
c e
Ol 800
'w
Q) A
A------------ A

A c E G

connection type

3600

3200 .~ -.-
-e-
for story 1-4
for story 5-7

C.
2800
.~ -A- for story 8-10

32

-
c
Q)
2400

2000
.~ •
E
0
E 1600
c
Ol
'w
Q) 1200 ______ e e
0 e- e--------
______ A
A- A -A
800
A C E G

connection type

Fig.5.20 Design moments for exterior columns. (a) for braced frame,
(b) for unbraced frame.

101
1.2

-.- for story 1-4


-.- for story 5-7
1.0 -.t.- for story 8-10 .t.

.t.~
t,-- .===-=-----=---- .
0.8

LL
0:::
2
0.6

0.4

A C E G

connection type

Fig.5.21 Moment reduction factor (M.R.F.) for exterior columns.

102
2000

~
c
1600
•------.------.'------ .
Q. -a- forstory 1-4
:i2 -e- forstory 5.7
~
-c
OJ
1200
-A.- for story 8-10

E ______ e e. e
0 e
E
c
Cl
'u; 800
OJ A A.______ A
0

A c E G

connection type
(a)
5000,----- --,

-_________ -.~forstory 1-4


4000 ~ -e- forstory 5-7
.~ -A-forstoryB-10
c

Q. 3000
:i2

-
c
.--------- •
OJ
E
o
E
c ------e
Cl
'u;
OJ ------A
o

Fig.5.22 Design moments for interior columns. (a) for braced frame.
(b) for unbraced frame.

103
1.2

-.- for story 1-4


1.0
-.- for story 5-7
- ••.- for story 8-1 a

------- .•.------- .-
0.8
.•. •
U-
n::
• _.-------- -------.
.
~ 0.6
~

0.4 .------- •
0.2
A c E G

connection type

Fig.5.23 Moment reduction factor (M.R.F.) for interior columns.

104
0.010

0.008

~
0.006 • -.- for unbraced frame
-e- forbraced frame

0.004

.------.~.:~~
~/ H = 1 /400
0.002
.
'------.
A c E G

connection types

Fig.5.24 !l. / H for diffrerent types of connections


( for both braced and unbraced frames)

105
Table.5.14 Percentage of reduction in member forces (moments)
Connection type Beam Ex. Column Int. Column
(%) (%) (%)
Double web angle 0-40 10 - 60 25 - 60
Top & seat angle 0-40 10 - 55 20 - 55
End plate 0-40 5 - 40 10 - 40

106
5.5 COMBINED USE OF SEMI-RIGID AND RIGID CONNECTIONS

The behavior of frames with combined use of semi-rigid and rigid connections is
examined in this section. For low rise building frames semi-rigid connection is generally
preferred in order to reduce cost of construction. However, depending upon the
slenderness ratio of the building frame and due to the use of semi-rigid connection, the
drift may be large and exceed the allowable limit i.e. MH :5 1/400 as recommended by
AISC Specification [3]. In this study an attempt has been made to keep drift within the
allowable limit by considering a provision to use both semi-rigid and rigid connections in
a frame. The objective of this study is to find out the best suited arrangement of semi-
rigid connections in a steel frame when mixed use of semi-rigid and rigid connection is
desired. A total number of 3 cases for placing semi-rigid connections at the beam ends are
shown in Fig.S.2S. A 3 bay 6 story frame is taken for this investigation. The most widely
used connection, Top and seat angle (C) is employed in this study. In each case 50% of
the connections i.e. 18 of the 36 connections are made semi-rigid. The positions of the
semi-rigid connections for different cases are shown in Fig.S.2S.

Like earlier, the bay width and story height of the frame are taken to be 22 ft and 12 ft
respectively. The frames are spaced at 24 ft intervals. Beam and column sections used in
this study are W 14x48 and W l4x61. The connection parameters for the Top and seat
angle (C) are presented in Table 5.15. The material properties of the members are taken as
same as used in earlier investigation. The dead load, the live load and the wind load
applied on the frame are shown in Table 5.16.

Table 5.15 Connection size parameters (all the dimensions are in inches).
Conn. Depth of Total Thickness Thickness Gage Thickness Fastener Length
Type anglelbeam thickness of top of web length of dia of top
/plate (d) of angle/ angle angle / (g) web/plate (I) angle/
flanges of (tJ header (I,. / 1,) T-stub
T-stub & plate (I)
column (t) (tJ
C 14.0 1.02 - - - - 0.875 8.0

107
W14X48 W14X4!1

;;; ;;; ;;; ;;; ;;; ;;; ;;; ;;;


~ •
• ~ •• ~ •• •• •

;; ;; ;; ;; ;;
~I,.- ~ ~l,.- ~I,.- ~l".- ~~ ~
-- -- ~l".-

3022'-66' 1 1------------<1
, 3022'_66'

Cale-l Cale-2

W14X4!1

;;; ;;;
~ •• ~ ~
;; ;; ;;
~I,.- ~I,.- .~ ~I,.-
1 1
Caee-3

Fig,5,25 Different cases of mixed used of semi-rigid and rigid connection

lOB
Table.5.16 Loading on the frame.
Floor levels Dead Load Live Load Wind load
All floors 47.5 psf 40 psf 25 psf
Roof level 30 psf 20 psf 25 psf

5.5.1 Behavior of Frames Using Mix Mode Connections

Behavior of frames using combination of semi-rigid and rigid connections is presented in


Fig.5.26 to 5.29. The critical moments for beams are calculated and plotted in Fig.5.26. A
new term called 'Moment Amplification Factor' (M.A.F.) is introduced in this section.
The M.A.F. is the ratio of the moment as obtained due to analysis of the frame with semi-
rigid and rigid connections and that of the analysis of frame with rigid connections only.
The M.A.F. for different cases (different arrangements of semi-rigid connection) are
plotted in Fig.5.26.

The story levels at which rigid connections are considered, it can be observed from
Fig.5.26b that the values ofM.A.F. for case 2 (1.0-1.07) are close to 1.0 than for case I
(1.08-1.16) and case 3 (1.05-1.14). In other words, for case 1 and case 3 the M.A.F.
values are higher than that for case 2. The story level at which semi-rigid connections are
included, case 1 (0.65-0.78) and case 3 (0.69-0.73) gives lower values of M.A.F. than
case 2 (0.79-0.91). The extent of variation in the M.A.F. values for the whole frame
(considering all the stories) is smaller for case 2 than for case 1 and case 3. so, for the
design of beams of the frame case 2 gives a more economical arrangement for the
combined use of semi-rigid and rigid connections.

The design moments of exterior and interior columns are presented in Fig.5.2? and 5.28.
The M.A.F. for design moments for exterior and interior columns show that case 2 offers
the best result. The values ofM.A.F. for exterior columns varies between 0.95 to 1.0 and
for interior columns 1.0 to I. I I for case 2 condition.

It is obvious that the stiffer the connections are the less will be the top deflection. For the
cases investigated in this study (Fig.5.29), case 2 gives the minimum top deflection

109
110

'-
1300

1200
.~.~.
~
"
c
c.
:l2 1100
~
~
c
Q)
E 1000
o
E
8 900
;;!
Q)

800 .~.
case1 case2 case3
cases
(a)
1.30 .

1.25

1.20

0 1.15
(,)

;;!
Q) 1.10
~
.E
LL 1.05
~
~ 1.00

0.95
:~:
0.90
case1 case2 case3
cases
(b)

Fig.5.27 (a) Exterior column design moment for different cases.


(b) MAF. of exterior column for diffrent cases.

111
1300

~
1200

1100
.~.~.
"
c
a.
:.>2 1000
-
~

E
c
Q) 900 -.- for story 1.3
0 -e- forstory 4-5
E 800
0
II
700
...•c
600

500
case 1 case 2 case 3
cases
(a)
1.7

1.6

1.5
0
II 1.4
.•.•
C -.- for story 1-3
•... 1.3 -e- forstory 4-5

LL
1.2
~

:~:~.
:2:
1.1

1.0

case 1 case 2 case 3


cases
(b)
Fig.5.28 (a) Interior column design moment for different cases.
(b) M.A.F. of interior column for different cases.

112
0.0034

0.0032

.~.
0.0030

0.0028
--
J:
<1
0.0026

0.0024 /1/ H = 1/400

0.0022

0.0020
case1 case2 case3
cases

Fig.5.29 LlI H for different cases.

113
(LVR=0.00277). Though it does not satisfy the criterion for top deflection (LVR :> 1/400),
but the LVRratio for case 2 is not more than 10% higher than the allowable limit, which
can be neglected.

5.5.2 Summary

From the study presented in this section, it can be observed that placement or arrangement
of semi-rigid connections in combination with rigid connections has important influence
on the design of the structure. Case 2 provides maximum possible design economy for
both columns and beams and minimum top deflection. Further to add that when
combination of semi-rigid and rigid connections are used it is better to place the semi-
rigid connections at the top story levels to obtain minimum top deflection and maximum
possible design economy.

114
5.6 SEMI-RIGID FRAME USING SINGLE STIFFNESS LINEAR MODEL

Researchers employed different approaches to quantify and represent the connection


flexibility. The most commonly adopted method was to take a linear approximation of the
M-8, curves. This approach includes representation of the connection stiffuess by the
initial tangent ofthe ofthe M-8, relationship as mentioned earlier.

It is very tedious to include the non-linear effect of the connection flexibility into usual
office practice. For routine office practice it may be convenient to take the connection
stiffuess as a linear response and thus considering the total response of the frame as linear.
In this section a comparison between the linear analysis and the non-linear analysis in
calculating beam moments, column moments and lateral drift of a semi-rigid frame is
presented. The purpose of this study is to provide the designers, an in sight idea of the
variation between the rational analysis and linear analysis using single stiffuess when
incorporating the effect of the connections.

A 3 bay 4 story frame is being analyzed for two different loading cases (Fig.5.30). Once
again the bay width and the story height of the frame are taken as 22 ft and 12. ft
respectively. The frames are spaced at 24 ft intervals. Beam and column sections used in
this study are W 14x34 and W l4x38 respectively. Member sections and properties have
not been varied along the width or height of the frame. Material properties of the
members are taken as same as that taken in earlier investigations.

In this investigation the three types of semi-rigid connections are used: two soft and one
stiff semi-rigid connection. The connection types used in this study are:

1. Double web angle connection (A),


2. Top and seat angle connection (C).
3. T-stub connection (F).

The parameters used for these types of connections are given in Table 5.17. The dead
load, the live load and the wind load applied on the frame are shown in Table 5.18.

115
W14X34
'.,
".
N
II

".

., ., ., .,
'"x '"x '"x '"x
". ". ". ".

10--- -----<1
, 3@22'=66' ,

a) Gravity loading

W14X34
.,
".
II
N
@

".

., ., ., .,
'"
x '"x '"x '"x
,.". ,.". ,.". ,.".
~ 3@22'=66' ~

b) Gravity and lateral loading

Fig.5.30 Building frame used for linear analysis

116
Table 5.17. Connection size parameters
Conn. Depth of Total Thickness Thickness Gage Thickness Fastener Length
Type angle/beam thickness of top of web length of dia of top
/plate (d) of angle/ angle angle / (g) web/plate (I) angle/
flanges of (,) header (" / t,.) T-stub
T-stub & plate (I)
column (t) (tJ
A 10.0 - - 0.375 5.5 - - -
C 14.0 0.89 - - - - 0.875 8.0

F 14.0 1.265 - - - - 1.0 8.0

* all the dimensions are in inches.

Table 5.18 Loading pattern.


Floor levels Dead Load Live Load Wind load
All floors 47.5 psf 40 psf 25 psf
Roof level 30 psf . 20 psf 25 psf

5.6.1 Comparison Under Gravity Loads

Variation of positive and critical beam moments for different types of connections for
story level I are presented in Fig.5.31. The comparison between the linear analysis and
the rational analysis is performed for both positive and critical moments of beams. The
percentages of over estimation or under egimation are shown in Fig.5.32 and Fig.5.33. It
can be observed that for soft connections like Double web angle (A) and Top and seat
angle (C), linear stiffuess analysis gives under estimation for beam moment up to 20%.
For stiff connections such as T-stub (F), the percentage of under or over estimation is
negligible (1%-3%).

117
1300

~
1200 •
.,
c
n.
-.- linear analysis
1100 -e- exact analysis
:i2
-
~
c
Q)
E
1000

0 •
E 900
E
CO
Q) 800
..c
Q)
> 700
.iii
""
0
n.
600

A C F
connection types

(a)

1200 •
~ 1150
.,n.
C

1100
-.-
-e-
linear analysis
exact analysis

:i2
-
~
c
Q)
E
1050

0 1000
E •
E
CO
950 •
Q)
..c 900
CO
()

.C
850
""()
800

A C F
connection types
(b)

Fig.5.31 Beam moments for gravity loads at story level 1.


(a) for positive moments, (b) forcritical moments.

118
0
c:
'T
C.
-2
;g.
<:
Ql -4 -.- for story level 1

E -e- for slorylevel 2


0 -4- for story level 3
E -6 -y- for slory level 4
.5
c:
-8

,,-----"
0
~
E -10
~ Ql

-!.- -12
Q;
"~
c:
::J
-14
0
-16
:£~ •
Ql -18

-
0
>
0

-20
.-=::::~.

* A c F
connection types

(a)
10
'2
"6.
g
-.- for conn. type A
OJ
E -e- forconn. lype C
0 -4- for conn. type F
E 0
.5
c:
:g0 •• ------, •.------ •. -------- ..
.S
<;;
,
Ql
-10 •

--._--_./"
~ Ql

"
c:

-._--_.
::J
~
0 •
:£~ -20
• ------.
Ql

-
>
0
0
~
0

2 3 4
story level

(b)
Fig.5.32 Percentage of over or under estimation of beam positive moment
for gravity load. (a) with connection types, (b) with story level

119
-C
Ql
E
10

-
0
E
0 -.~ fo'r story level 1
c 0
-.~ for story level 2
.9
co - .•. - for story level 3
E -5 -y- for story level 4
~
Ql
~
..!-
~
Ql
-10 y-----
----y
"0
C
::J
~

.~--~.
0
-15

:£ •
~
Ql
>
-20 ---
- 0
0
<f!. -25
A C F
connection types

(a)

- C
Ql
E
20

15

-
0
E -.- for conn. type A
10
0 -e- forconn.typeC
c -A- for conn. type F
:.=ell0
E
:u;
Ql
5

0
.• ----- .• ----- ..• ----.•
2 -5
~
Ql
"0
c -10
::J
~
0


~
-15
Ql

->
0
0
<f!.
-20

-25
2 3 4
story level
(b)
Fig.5.33 Percentage of over or under estimation of beam critical
moment for gravity loads. (a) with connection types,
(b) with story level.

120
5.6.2 Comparison under Combined Action of Gravity and Lateral Loading

The critical moments developed both by the linear and non-linear solutions under gravity
and lateral loading are presented in Fig.5034a. The percentages of over or under
estimation of beam critical moments are presented in Fig.5.35. For soft connections linear
analysis gives under estimation of beams moments but for stiffer connections it gives
almost same as the rational solution (1.01-1.04). The maximum variation is up to 4%
which is negligible.

Comparison of critical moments for the columns developed in linear and non-linear
analyses are presented in Fig.5034b and 5034c. The percentages of over estimation or
under estimation for exterior and interior columns are presented in Fig.5036and Fig.5.3?
Linear analysis over estimates the exterior column moments for soft connections except at
story level 1. The percentage of over estimation for Double web angle connection reaches
up to 40%. For stiff connections such as T-stub, the linear analysis gives almost identical
results with exact analysis. The design moment ratios as shown in Fig.5.38a demonstrates
that exterior columns are under estimated at lower story levels and over estimated at
upper story levels. For interior columns the linear analysis under estimates critical
moments for almost all the connection, Fig.5038b shows that interior columns are
thoroughly under estimated at all floor levels by linear analysis method.

The variation of ~ for different types of connections are presented in Fig.5.39 for both
linear analysis and rational non-linear analysis. Linear analysis gives smaller lateral drift
and consequently lower ~. Exact analysis gives a higher value of ~ as expected. It
can also be observed that both linear and non-linear analysis give almost same ~ for
stiff connections (T -stub). For this connections the lateral drift satisfies the criterion for
top deflection required by the ASD Specification of AISC (~ ~ 11400). Though linear
analysis gives lower values of ~ for soft connections like Double web angle and Top
and seat angle connections, but it does not satisfY the deflection criterion (~ ~ 11400).

121
1400 -e- forlinearalllllysis
~ .••- for exact analysis •
C
'7
C.

g 1300
CQ)

E 1200
0
E
E
C1l
Q)
.a 1100
Cii
g
'C 1000 o
u

A c D F
connection types
(a)

I•
c 1000 for Imear analYSIS I
"6- ••. -A- for exact analYSIs
g
cQ)

E
o 900
E
Cii
u
"""5
.~.~-.
800
ou o~~
~. 0____ •
o ---0

2
iij 700 -'-_-,- --,- -, .,----"
A c D F
connection types
(b)
1000

I•
-A_
for linear analYSIs
for exact analYSIs
I
I
900

.~.
800
~ ~.
700 o~
--------- - e ~ °

A c 0 F
connection types
(c)

Fig.5.34 (a) Critical beam moments for gravity and lateral loads,
(a) Exterior column critical moments, (b) variation of
interior column critical moments, at story level 1.
122
20

c 15
0 -.- for slory rever 1
-e- for slory level 2
""Cll 10
E -A- for story level 3
-~- for story level 4
""a>en 5
~,
~ 0
L-
a> -5
""0
c
:J
-10
~
L-
0 ••
+ -15
~
a>
L-

-20
••
>
0
.•... -25
0 0
?f?-
-30

A C F
,connection types
(a)
20

c
o -.- for connection type A
""
Cll 10 . -e- forconnection type C
E -.A.- for connection type F
~
a>
~,
~ o :~-----.. ..----- .•
.~.~.
L-
a>
""0
C
:J
L- -10
o
;:-
~.
~
L-
a> -20
>
o
.•...
o
?f?-
.--.---------. .

-30
1 2 3 4
story level
(b)

Fig.5.35 Percentage of over or under estimation of beam critical


moment for gravity and lateral load (a) with connection
types, (b) with story level.

123
,

40 4
c
0
~
ctl
30
E
~
en • •
Q)
~0 20
~
~ 4
Q)
""0
c 10
::l -.- for story level 1
~ -e- forstorylevel 2 'Y_________ ~ "'"
0
~ 0 -.4- for story level 3 ~
+
~ -y- forstory level 4 _

----------- .
~
Q)

-
>

~
0
0
0
-10

-20

A c F
connection types
(a)

40
t:
0
~ -.- for conn. type A
ctl
E -e- forconn.typeC
30
~ - .•.- for conn. type F
en
Q)
~0
~ 20
~
Q)
""0
C
::l 10
~
0
~
+
~
0 4 ,4 ,4
~
Q)

-
>
0
0
~
0
-10

2 3 4
story level

(b)

Fig.5.36 Percentage of over or under estimation of exterior column


critical moment for gravity and lateral load. (a) with
connection types, (b) with story level.

124
10

0
+::
c
Cll
E
5
.~
+::
en
<])
0 •
~,
~ -5
•...
U
C
<])

::::J -10
•...
~
+
~
0

•...
<])
-15 .~. -.- for slory level 1
>
0 -20 -e- forstorylevel 2
'0 -4- for story level 3
~
0
-y- for slory level 4
-25
A c F
Connection types
(a)

10

c
0 5 -.- for conn. type A
+::
Cll -e- for conn. type C
E -.4- for conn. type F
0
~
<])
~,
~ -5
•...
<])
U
C
::::J -10
•...
0
~
+
~ -15
•...
<])
>
0 -20
'0 ••
~
0
-25
2 3 4
story level
(b)

Fig.5.37 Percentage of over or under estimation of interior column


critical moment for gravity and lateral load. (a) with
connection types, (b) with story levels.

125
40
0
c •
:;::;
co
-.- % for story 1 & 2
E 30
:;::; -e- % for story 3 & 4
(/j
Q)
~,
~ 20
•...
""0
C
Q)

::l 10
•...
0
~
+
~
....
Q)
0
>
0
0+-
0 -10
~
0

A c F
connection types
(a)
10

c
0
:;::; 5
CO -.- % for story 1 & 2
E
:;::; -e- % for story 3 & 4
(/j
Q) 0
~,

.~.~.
~
....
Q)
""0 -5
c
::l
•...
0
~ -10
+

.~ .
~
•...
Q)
>
0 -15
0+-
0
~
0
-20
A c F
connection types
(b)

Fig.5.38 Percentage of over or under estimation of column design


moments (a) for exterior columns, (b) for interior columns.

126
0.006

0.005
-.- for linear analysis
-e- forexact analysis

0.004
--
I

<l
• •
0.003

0.002
6/ H = 1/400 ... ~~:

A .c F
connection types

Fig.5.39 ~ / H with connection types.

127
5.6.3 Summary

In this scope of the research work it can be concluded that for stiff connections like T-
stub, linear approximation of connection stiffuess can be used for the analysis of a frame.
The percentage of variations between the linear analysis and the exact non-linear analysis
are very negligible (not more than 5%). But for soft semi-rigid connections linear
approximation should not be applied. The linear approximation can lead from 5% to 25%
of under estimation of member forces. The variation is much higher for soft connections
than that found for stiff connections. It is suggested that for soft to intermediate stiff semi-
rigid cOimections exact non-linear analysis of the should be performed. Table.5.19 and
Table.5.20 show the comparison of linear analysis with rational analysis for member
forces for different types of connections.

Table 5.19 Comparison oflinear analysis with rational analysis for different types of
connections for the design of beams.
Connection type Gravity Gravity + Lateral
Linear Exact Variation Linear Exact Variation
Moment Moment (%) Moment Moment (%)
(k-in) (k-in) (k-in) (k-in)
Double web angle 992 1205 -18 1001 1360 -26
Top and seat angle 802 959 -16 1088 1066 +2
T-stub 989 960 +3 1377 1318 +4
+ indicates percentage of over estimation
- indicates percentage of under estimation

Table 5.20 Comparison oflinear analysis with rational analysis for the design moments
of columns.
Connection type Gravity Gravity + Lateral
Linear Exact Variation Linear Exact Variation
Moment Moment (%) Moment Moment (%)
(k-in) (k-in) (k-in) (k-in)
Double web angle 885 989 -11 832 996 -16
Top and seat angle 813 867 -6 734 828
. -12
T-stub 760 770 -1 661 672 -2
+ indicates percentage of over estimation
- indicates percentage of under estimation

128
CHAPTER 6.
PARAMETRIC STUDY OF CONNECTIONS

6.1 GENERAL

The mam purpose of this chapter is to study the influence of the connection size
parameters on the behavior of frames. To this objective a parametric study has been
. conducted using the computer program developed in this research work. In this study
attempt has been made to determine the parameters affecting the serviceability conditions
of a semi-rigid frame, particularly the lateral drift and the member end forces. The results
of this study may help identifying the parameters that influence the controlling factors in
the design of the frame most. This in turn will enable to select the suitable connection
parameters and lead us to a cost-effective design.

6.2 SCOPE OF THE PARAMETRIC STUDY

In this investigation we considered three types of connections namely: Double web angle,
Top and seat angle with double web angle and T-stub connections for parametric study.
The reason for taking these three types is that, double web angle is very flexible, Top and
seat with double web angle is fairly semi-rigid and T-stub is the stiffest of the semi-rigid
connections. The results of the present study are subjected to the limitations inherent.to
the scope ofthe range of the parameters.

In the scope of this research work, a 2 bay 5 story frame as shown in Fig.6.1 has been
taken for the parametric study. The bay width and the story height of the frame are 22 ft
and 12 ft respectively. The frame is spaced at 24 ft intervals. The frame is being analyzed
for combined action of gravity and lateral loading. The beam section used in this study is
W 18x60 for all floors. The column section are changed with the height of the frame. The
section of all the columns in ground floor and 1" floor are taken as WI4x74 and form 2"d
floor to 4th floor W14x53.

129
W18X60

"0 '" '" '"


to
II
'".,.
X
'".,.
X
'".,.
x
N ;;: ;;: ;;:
@

'"

.,. .,. .,.


.,."
X
".,.
X
".,.
X

;;: ;;: ;;:

+ 2@22';44' +
Fig.5.! Building frame as used for parametric study

130
The size parameters involved with these connections are listed in Table 6.1, presenting
their range of variation. The dead load, the live load and the wind load applied on the
frame are shown in Table 6.2.

Table.6.1 Connection size parameters and their range of variation.


Conn. Depth of Total Thickness Thickness Gage Thickness Fastener Length
Type anglelbeam thickness of top of web length of dia. of top
/plate (d) of angle/ angle angle / (g) web/plate (I) angle/
flanges of (~) header (lw / 1,) T-stub
T-stub & plate (1)
column (t) (tJ
A 5.5-11.5 - - 0.3125- 4.5 - - -

0.625
D 14.0 - 0.3125- 0.375 1.75 - - 5.5-11'.5
0.625
F 14.0 1.16- - - - - 0.625- 5.5-9.5
1.535
1.25
* all the dimensions are in inches.

Table 6.2 Loading on frame.


Floor levels Dead Load Live Load Wind load
All floors 47.5 psf 40 psf 25 psf
Rooflevel 30 psf 20psf 25 psf

6.3 PARAMETRIC STUDY ON DOUBLE WEB ANGLE CONNECTION

Double web angle has only three parameters; angle depth (d), thickness of the angle (t,)
and the gauge length (g). In this study 4 different thickness of the web angle has been
taken. The thickness are 5/16", 3/8", liz" and 5/8". The gauge length is varied from 5.5
inch to 11.5 inch with an increment of 1.0 inch.

The variation of critical moments of beam and column with gauge length for different
angle thickness is presented in Figure 6.2. The variation of beam end moment with gauge
length for different angle thickness has been presented in Fig.6.2a. For thinner angles

131
1700

-.- t = 5/16"
1600 •• -e- 1=318" 1_
~ .•. ~. - • ----=::::::::::. ------- -&- t = 1/2"

~.~.~.
c -T- t = 5/B"
~~ e •
1500
:'2
c. :~.
-
~
c
Q)
E
1400 -
••~... e
0
-
.0
E
E
ro
Q)
1300

1200 .
.~. .
••
1100
5.5 6.5 7.5
,
8.5
-.
9.5 11.5
,

angle depth (in)


(a)
4000

-.- t = 5/16"
. 3500 -e- t = 3/8"
- ••.- 1= 1/2"
c - .•.~ 1=5/8"

3000

. -c
Q)
2500
E
o
E
c 2000
E
:::J
8
1500

5.5 6.5 7.5 8.5 9.5 11.5


angle depth (in)
(b)
Fig.5.2 (a) Critical beam moment with angle depth,
(b) Critical column moment with angle depth.
for Double web angle connection.

132
(angle thickness 5/16" and 3/8") the critical beam moment initially increases and then
starts decreasing with the increase in angle depth. Critical beam moment also decreases
with the increase in angle thickness. Column moment decreases with increase in both
angle depth and angle thickness (Fig.6.2b).

Variation of MH with angle length for various angle thickness has been presented in
Fig.6.3. Deflection decr.eases with the increase in angle depth. The higher the angle depth
-
is the less will be the deflection. Similarly deflec.tiOlw:e.d.u.c~swith the increase in the
angle thickness. For the angle thickness of 5/16" the deflection is maximum and for the
angle thickness of 5/8" the deflection is minimum. The pattern of the graphs is almost
identical.

133
0.030

0.025

0.020
:~
~.~
-.~
~.-
-A-
lorl=5/16"
fort=3/S"
fort= 1/2"
-y- fort= 5/8"

I
---<1 0.015
A~.~.~
0.010 ",,--- A~ .~:~.
~" A __________. ~
____________ ----- A ------ ~ •

. "____ ----------- A ------.


0.005
1>1 H = 1/400 . "------ " --------------- A

- --------- --------- . --------- -------.- ------_.- "


0.000
5.5 6.5 7.5 8.5 9.5 11.5
angle depth (in)

Fig.6.3 ~ I H with angle depth for Double web connection.

134
6.4 PARAMETRIC STUDY ON TOP AND SEAT ANGLE WITH DOUBLE
WEB ANGLE CONNECTION

Top and seat angle with double web angle cOlmection has five parameters; beam depth
(d), web angle thickness (t,), top angle thickness (tJ, length of top angle (I), and gage
length (g). In this study four different thickness of the top angle has been taken. The
thickness are 5/16", 3/8", Yz" and 5/8". The top angle length is varied from 5.5 inch to
I 1.5 inch with an increment of 1.0 inch.

The variation of beam critical moment with angle length for different angle thickness has
been presented in Fig.6.4a. Critical beam moment decreases with the increase in angle
length as well as with the increase in angle thickness. The variation of column moment
with angle length for different angle thickness is presented in Fig.6.4b. Critical moment
decreases with increase in angle length and angle thickness.

Variation of NH with angle length for different angle thickness has been presented in
Fig.6.5. Deflection decreases with the increase in angle length. The higher the angle
length is the less will be the deflection. Deflection also reduces with the increase in the
angle thickness. For the angle thickness of 5/1 6" the deflection is maximum and for the
angle thickness of 5/8" the deflection is minimum. The pattern of the graphs is identical.

From the parametric study on Top and seat angle with double web angle connection, it
can be observed that both length and thickness of the top angle are predominant.
Economic design of the members depends on the optimum use of the top angle length and
thickness.

135
1700

1600 • -.- t = 5/16"


• -e- t = 3/B"
~ • ------------. -4- t = 1/2"
---------------. ----------. ~
~.-~
-,,- t = 5/8"

~.
c 1500
---------------
~
Q.
1400 4~
.-----------. •

-
~
c
Q)
E 1300 4~4 ~.

0
E "~ ~4

E 1200 "~ ------------4

co
Q) " " . ~ 4
.0
1100
------------"~
1000
5.5 6.5 7.5 8.5 9.5 11.5
angle length (in)

(a)

.~
1700

-------. -.- t = 5/16.


1600 ~ -e-I=3/8"
c ------.. • -.A.- t = 1/2 .•
~._______________
-----------.
~ ."~r-t = 5/8"

- 1500 .-----------.
---------- .~.
~.
c
Q)
E
o
E 1400
c
E
OJ

8 1300

5.5 6.5 7.5 8.5 9.5 11.5


angle length (in)

(b)
Fig.6.4 (a) Critical beam moment with angle length.
(b) Critical column moment with angle length.
(for Top and seat angle with double web angle)

136
0.005

for t = 5/16
•~ -.-
-e- fort = 318 M
6

• -.&- fort= 1/2"


0.004

.
.______ ----------. -y- fort = 5/8 •

------. --------- --------- .


:r:
-- 0.003
.-------- .---- .---------. .----------
---------- .
----------
<I Y
---A __________.

--- y ----------.
- ------_.-
d / H = 1/400
~,,~- y
------_._~
y
..•
--------- y
0.002

5.5 6.5 7.5 8.5 9.5 11.5


angle length (in)

Fig.6.5 '" IH with angle length for Top and seat with double web
angle connection.

137
6.5 PARAMETRIC STUDY ON T-STUB CONNECTION

T-stub connection has four parameters; beam depth (d), total thickness of the flanges of
the T-stub and column (t). fastener diameter (f) and the length of T-stub (I). For T-stub
connection two investigations are performed. In the first investigation length of T-stub is
varied for different values of thickness while for taking the fastener diameter to be
constant. In the second investigation, the fastener diameter is varied keeping the length
and thickness constant. The fastener diameters included in this study are 5/S", v..... 7/S".
I",I-I/S' and 1-1/4". Four different thicknesses are considered in this study. The
thicknesses are 1.16". 1.2S5", 1.41" and 1.535". The length of the T-stub is varied from
5.5 inch to 9.5 inch with an increment of 1.0 inch.

Variation of critical moment of beam with length of T-stub for different thicknesses is
shown in Fig.6.6a. Critical moment increases with the increase in length as well as
thickness. Critical moment of beam with fastener diameter is also presented in Fig. 6.Sa
for which the length and thickness are taken to be fixed (I =7.5". t = 1041"). Critical
moment increases with the increase in fastener diameter.

Variation of critical moment of column with length of T-stub for different thicknesses are
presented in Fig.6.6b. taking the fastener diameter to be fixed (f = 0.S75"). It can be
observed that the column moment decreases with increase in length of the T-stub. The
column moment also decreases with the increase in thickness. Variation of column for
different fastener diameters is presented in Fig.6.Sb. Once again moment decreases with
increase in fastener diameter.

Variation of MH with length ofT-stub for different thicknesses is presented in Figure 6.7.
taking the taking the fastener diameter to be fixed (f= 0.S75"). Deflection decreases with
the increase in length of T-stub. Deflection reduces with the increase in the fastener
diameter as shown in Fig.6.9. Deflection also reduces for higher value of thickness.

From the parametric study on T-stub connection, it can be observed that both length and
the thickness of the T-stub and the fastener diameter are influential parameters on the

138
1500

.!:
1400

-
c
Q)
E
-.-1=1.16"
-e- t = 1.285 M

o - .•.-t=1.41"
E 1300 -~- 1= 1.535"

E
m
Q)
.0

1200
5.5 6.5 7.5 8.5 9.5
length ofT-stub (in)
(a)
1350 .

-.-1=1.16"
c -e- t = 1,285"
-A.- t= 1.41"
1300
a. -y- t = 1.535"
32

-
c
Q)
E
o
E 1250
c
E
::l
8
1200
5.5 6.5 7.5 8.5 9.5
length of T-stub (in)
(b)

Fig.6.6 (a) Critical beam moment with length of T-stub, (b) Critical
column moment with length of T-stub (for T-stub connection).

139
0.0026,- __

0.0024 !> I H = 1/400

0.0022

--
I

<1 0.0020

0.0018

0.0016-'----:.r5'~
---::'6. :-5--~7;:;.5~--88)~.5i--99:.55'

5. length of T-stub (in

Fig.5.? <1/ H with length 0f T e e for t -s tub connection.

140
~. ---------.

/ .
•/. I -.- /=1.41",1,=7.5'"

/

5/8 3/4 7/8 1 1-1/8 1-1/4


fastener diameter (in)
(a)
1400

.~
1350
.!:

1300

-
c .~ I. -.-
.~
Q) ~ /=1.41",1,=7.5'"
E 1250
o
E
c
E
:>
ou 1200
.-------------
. --------------

1150
5/8 3/4 7/8 1 1-1/8 1-1/4

fastener diameter (in)


(b)
Fig.6.8 (a) Critical beam moment with fastener diameter,
(b) Critical column moment with fastener diameter.
(for T-stub connection)

141
0.0030

0.0025

.~
- --------- -------.- --------- ------_.- -------'-

M H = 1/400
J:
--
<1 •
. ~ I -.-
0.0020
.~ [0(1=1.41",1,=7.5" I
.-------- .~•
0.0015
5/8 3/4 7/8 1 1-1/8 1-1/4
fastener diameter (in)

Fig.6.9 L\ / H with fastener diameter for I-stub connection.

142
behavior of connections. When the thickness remains constant, top deflection and column
moment decreases with the increase in length ofT-stub as well as fastener diameters. But
for the beam, design moment increases with the increase in these parameters. So,
economic design of the members depends on the right choice of the.length and thickness
of the cOlmection element and fastener diameter.
,
. .p---
6.6 CONCLUDING REMARKS

From the parametric study investigated in this chapter it can be observed that size of the
connection parameter has considerable influence on the behavior of the. While analyzing
a frame with partially rigid connections, all the parameters influencing the connection,
should be incorporated properly and accurately to optimize the design of the frame.

143
CHAPTER 7
CONCLUSIONS AND RECOMMENDATIONS

7.1 GENERAL

Analysis of steel frames with partially restrained connections has been presented. Based
on the matrix method and polynomial connection model, a computer program has been
developed to analyze steel frames with semi-rigid connections. The program is a
generalized one and can be used to any kind of 2-D steel frames with known flexible
connections. The program is a non-linear, iterative type, based on incremental or stepwise
loading. A number of investigations have been performed to study the effects of semi-
rigid connections.

7.2 SUMMARY OF CONCLUSION

The present study has offered important conclusions regarding the selection of semi-rigid
connections and the consequent behavior of the semi-rigid frame. The important
conclusions derived from the behavioral study and from the parametric study are
summarized below:

• Simple framing (Type 2) analysis over estimates the beam moment. For
intermediate semi-rigid connection like Top and seat angle with double web angle
connections, the percentage of over design of beam moments are higher than the
soft semi-rigid connections (Double web angle, Header plate and Top and seat
angle) or stiff semi-rigid connections (End plate and T-stub) when simple framing
method is employed. Other than intermediate stiff connections, soft connections
gives more over design for beams than stiff connections.

• Type 2 method provides under estimation of column moments both for exterior
and interior columns. For stiff semi-rigid connections the percentage of under
design is very low and can be neglected. But for more flexible connections the
percentage of under design is quite alarming and should be taken care by properly

144
analyzing the frame taking connection flexibility effect. The study also confirms
that exterior columns are more under designed than the interior colunms.

• Lateral drift or top deflection high rise frames with flexible connections are quite
large and does not fulfill the allowable limit recommended by the AISC
Specifications. For high rise building it would not be possible to use semi-rigid
connection without providing any sway resisting system. By providing bracing,
the sway of the frames can be controlled and be reduced to an allowable limit.
With the use of bracing, the lateral drift of the frames even with more flexible
connections can be controlled to allowable limit. From the investigations it is
observed that bracing not only reduces the lateral drift but also reduces beam and
colunm moments which results in the economic design of beams and colunms.

• For low to medium rise building frames mixed use of rigid and semi-rigid
connections can be advantageously used to reduce the cost of connections. The
arrangement of the rigid and semi-rigid connections is very important to control
the top deflection. Making the beam-to-colunm connections at the lower level
stories rigid and at the higher level stories flexible this phenomenon can be
achieved.

• To take the pon-linear


, effect of the flexible connections into usual office practice
is very tedious and time consuming. Sometimes it may be convenient to use a
single stiffness of connection or the initial stiffness and perform a linear analysis.
The linear approach is quite simple and easy to use. Frame with stiffer
connections, the linear analysis is quite representative of the non-linear analysis
and gives reasonable results. But for less stiff connections both the beams and the
columns are under estimated taking linear approach. For flexible connections the
beams are underestimated about 15% - 30% and columns are underestimated
about 5% - 20%. For frames with stiff semi"rigid connections the initial stiffness
method can be applicable but for soft to intermediate semi-rigid connections
proper correction factor should be included while employing linear approach.

145
• Connection size parameters have considerable effect on the frame behavior. For
soft semi-rigid connections, increase in the angle depth reduces the beam and
column moments and the lateral drift of the frame.

• For stiff semi-rigid connections like T-stub, the increase in the angle length
decreases the column moment and lateral drift but increases the beam moment.

• For T-stub connections fastener diameter has influence on the variation of member
end actions. The increase in the size of fasteners reduces the column moment and
top deflection, but increases the beam moment. To optimize the design, all the size
parameters of the connections should be taken properly and accurately.

7.3 RECOMMENDATION FOR FUTURE WORK

The objective of this research work was to develop a computer program to analyze muIti-
storied steel frames taking connection flexibility into account.

• In the present study, only gravity and wind loading have been considered. For
-=-
earthquake prone zones and for high rise frames the earthquake loading must be
considered. In that case gypamic analysis of partially restrained frames is required.
The effect of mass and dampin'g must be included to modifY the program to make
. -".--
it suitable for such analysis.

• The developed program assumes linear elastic material for the, members. Material
non-linearity of the frame members can be incorporated and a detailed study can
be carried out in the behavior of semi-rigid frame.

• The P-t> effect or geometric non-linear effect of the members of the frame can be
0ncluded to~~e analysis program. v

• More detailed study on the effect of connection size parameters for all the
connections may be investigated.

146
References

I. Goverdhan, A.V. (1983). A Col1ection of Experimental Moment-Rotation Curves and


Evaluation of Prediction Equations for Semi-Rigid Connections, Master's Thesis,
Vanderbilt University, Nashville, TN, 490 pp.

2. Kishi, N. and Chen, W.F. (1986). Data Base of Steel Beam-to-Column Connections,
Structural Engineering report No. CE-STR-86-26, School of Civil Engineering,
Purdue University, West Lafayette, IN, USA.

3. American Institute of Steel Construction (1989). Al10wable Stress Design


Specifications for Structural Steel Buildings, Chicago, IL.

4. American Institute of Steel Construction (1986). Load and Resistance Factor Desiiln
Specifications for structural Steel Buildings, Chicago, IL.

5. Housing and Building Research Institute (1993). Bangladesh National Building


Code, I" ed., Dhaka.

6. Hechtman, R.A. and Johnston, B.G. (1947). Riveted Semi-Rigid Beam-to-Column


Building Connections, Progress Report No. I, AISC Research at Lehigh University,
Bethlehem, PA.

7. Bel1, W.G., Chesson, E., Jr. and Munse, W.H. (1958). Static Tests of Standard
Riveted and Bolted Beam-to-Column Connections, University of Illinois Engineering
Experiment Station, Urbana, IL.

8. Lewitt, C.W., Chesson, E. and Munse, W.H. (1966). Restraint Characteristics of


Flexible Riveted and Bolted Beam-to-Column Connections, Bul1etin No. 500,
Engineering Experiment Station, University of Illinois, Urbana, IL.

147
9. Lipson, S.L. (1968). Single-angle and single-plate beam framing connections,
Canadian Structural Engineering Conference, Toronto, Canada, pp. 141-1621.

10. Altman, W.G., Azizinamini, A., Bradburn, J.H. and Radziminski, J.B. (1982).
Moment-Rotation Characteristics of semi-rigid Steel Beam-to-Colunm C9nnections,
The Civil Engineering Department, University of South Carolina, Columbia, SC.

11. Marley, M.J. and GerstIe, K.H. (1982). Analysis and Tests of Flexibly Connected
steel Frames, Report to AlSC under Project 199, University of Colorado, Boulder,
CO.

12. Azizinamini, A., Bradburn, J.H. and Radziminski, I.E. (1985). Static and Cyclic
Behavior of Semi-Rigid Steel Beam-column Connections, Structural Research
Studies, Department of Civil Engineering, University of South Carolina, Columbia,
SC.

13.. Nethercot, D.A. (I 985a). Steel Beam-to-Colunm Connections _ A review of Test


Data and its Applicability to the Evaluation of Joint Behavior in the Performance of
Steel Frames, CIRlA Project Record, RP 338, London.

14. Chen, W.F. and Kishi, N. (1989). Semi-rigid steel beam-to-column connections: data
base and modeling, Journal of the Structural Engineering, ASCE, 115, I; 105-119.

15. Wang, Y.C.(l990). Semi-rigid Action in Steel Frame Structures: Data bank of M-</J
Test Results, ECCS Agreement No. 7210 SA/819, Department of Civil and Structural
Engineering, University of Sheffield.

16. Abdallah, K.M. and Chen, W.F. (1995). EXPanded Data Base of Semi-rigid Steel
Connections, Computers and Structures, Vo1.56, NO.4, pp. 553-564.

17. Rathbun, J.e. (1936). Elastic properties of riveted connections, Transactions of


ASCE, 101; 524-563.

148
18. Monforton, A.R. and Wu, T.S. (1963). Matrix analysis of semi-rigidly connected
frames, Journal of Structural Division, ASCE, Vol. 89(ST6), 13-41.

19. Lightfoot, E. and Le Messurier, A.P. (1974). Elastic analysis of frame works with
elastic connections, Journal of the Structural Division, ASCE, 100, ST6; 1297-1309.

20. Romstad, K.M. and Subramanian (1970). Analysis of frames with partial connection
rigidity, Journal of the Structural Engineering, ASCE 96, 2283-2300.

21. Lui, E.M. and Chen, W.F. (1986). Analysis and behavior of flexibly-j()inted frames,
Engineering Structures, 8; 107-118.

22. Moncarz, P.D. and GerstIe, K.H. (1981). Steel frames with non-linear connections,
Trans, ASCE, J. Struct. Div., 107, ST8.

23. Razzaq, Z. (1983). End restraint effect of steel column strength, J. Struct. Div., ASCE
109,314-334.

24. Sommer, W.H. (1969). Behavior of welded Header Plate Connections; Master's
Thesis, University of Toronto, Ontario, Canada.

25. Frye, M.J. and Morris, G.A. (1976). Analysis of flexibly connected steel frames,
Canadian Journal of Civil Engineers, 2, 3; 280-291.

26. Jones, S.W., Kirby, P.A. and Nethercot, D.A. (1982). Columns with semi-rigid joints,
J. Struc. Div., ASCE, Vol. 108(2),361-372.

27. Colson, A. and Louveau, J.M. (1983). cOru1ections Incidence on the inelastic
behavior of steel structures, Euromech Colloquium 174, October.

149
28. Kishi, N. and Chen, W.F. (1990). Moment-rotation relations of semi-rigid
connections with angles, journal of Structural Engineering, ASCE, VoUI6, 1990,
pp. 1813-1834.

29. Ang, KM. and Morris, G.A. (1984). Analysis of three-dimensional frames with
flexible beam-column connections, Canadian Journal of Civil Engineers, 11; 245-
254.

30. Yee, Y.L. and Melchers, R.E. (1986). Moment-rotation curves for bolted
connections, Journal of the Structural Engineering, ASCE, 112,3,615-635.

31. Wu, F.S. and Chen, W.F. (1990). a design model for semi-rigid connections,
. Engineering Structures, 12,2; 88-97.

32. Tarpy, T.S. and Cardinal, J.W. (1981). Behaviour of semi-rigid beam to column end
plate connections, Proceedings Conference, Joints in Structural Steelworks, Halsted
press, London, pp. 2.3-2.25.

33. Krishnamurthy, N., Huang, H.T., Jefferey, P.K and Avery, L.K (1979). Analytical
M-e curve for end plate connections, Journal of Structural Division, ASCE, 105,
ST1; 133-145.

34. Nethercot, D.A. (1985b). Utilization of experimentally obtained connection data in


assessing the performance of steel frames, in Connection Flexibility and and Steel
Frames (W.F. Chen ed.), Procedings of a Session sponsord by the Structural
Division, ASCE, Detroit, pp. 13-37.

35. Jones, S.W., Kirby, P.A. and Nethercot, D.A. (1980). Effect of Semi-rigid
Connection on Steel Column Strength, Journal of Construction steel Research, VoU,
No. I, September pp. 38-46.

t':' . ~
150 li..
.- ..
It •
36. Sugimoto, H. and Chen, W.F. (1982). Small End restraint Effect on Strength of H-
Columns, journal of Structural Division, ASCE, Vol. 108, pp. 661-681.

37. Simitses, G.J. and Vlahinos, S.A. (1982). Stability analysis of a semi-rigidly
connected simple frame, Journal of Construction Steel Research, VoI.2(3), 29-32.

38. Shen, Z.Y. and Lu, L.W. (1983). Analysis ofInitially Crooked End Restrained Steel
Columns, Journal of Constructional Steel Research, Vol. 3, No. I, pp. 10-18.

39. Ackroyd, M.H. and Gerstle, K.H. (1983). 'Strength of Flexibly-Connected Steel
Frames, Engineering Structures, Vol. 5, pp. 31-37.

40. Gerstle, K.H. (1985). Flexibly Connected Steel Frames, Steel Framed Structures:
Stability and Strength, edited by Narayanan, R., Elsevier Applied Science Publishers,
pp. 205-239.

41. Dhillon, B.S. and Majid, S.A. (1990). Interactive Analysis and Design of Flexibly
Connected Frames, Computers & Structures, Vol. 36, NO.2, pp. 189-202.

42. Chen, W.F. and Lui, E.M. (1986). Recent developments in structural connections, in
Advances in tall Buildings (L.S. Beedle, Editor-In-Chief), Van Nostrand Reinhold,
New York, pp. 353-365.

43. Barakat, M. (1989). Simplified Design Analysis of Frames with Semi-Rigid


Connections, Ph.D. Dissertation, School of Civil Engineering, Purdue University,
West Lafayette, IN, 211 pp.

44. Ramberg, W. and Osgood, W.R. (1943). Description of stress-dtrain curves by three
parameters, National Advisory Committee for Aeronautics, Technical Note No. 902,
Washington, D.C.

151
45. Batho, e. and Rowan, H.e. (1931). Investigations beam and stanchion connections,
2'" Report, Steel Structures Research Committee, Department of Scientific and
Industrial Research, HMSO, London, 92.

46. Sourochnikoff, B. (1950). Wind Stresses in Semi-Rigid Connections of Steel


Framework, Transactions, ASCE, 115,382-402.

47. Salmon, e.G. and Johnson, lE. (1980). Steel Structures - Design and Behavior,
Harper and Row Publishers, New York.

48. Baker, J.F. (1931). Methods of Stress Analysis, First Report, Steel Structures
Research Committee, HMSO, London, England.

49. Baker, J.F. (1934). Methods of Stress Analysis, Second Report, Steel Structures
Research Committee, HMSO, London, England

50. Johnston, B.G. and Mount, E.H. (1942). Analysis of Building Frames with Semi-
Rigid connections, Transactions, ASCE 107,993-1019.

51. Gere, J.M. (1963). Moment Distribution, Van Nostrand Publishers, New York.

52. Lightfoot, E. and Baker, A.R. (1961). The Analysis of Steel Frames with Elastic
Beam-Column Connections, Golden Jubilee Congress Symposium on the Design of
High Buildings, Hong Kong University Press, pp. 205-217, Hong Kong.

152
APPENDIX A

SAMPLE DATA FILE

ANALYSIS OF FRAME WITH PARTIALLY RESTRAINED CONNECTIONS

Total No. of Joints


36

Joint No., X Co-ordinate, Y Co-ordinate, X, Y, Z Restraint Conditions


1,0.,0.,1,1,1
2,342.,0.,1,1,1
3,684.,0.,1,1,1
4,0.,252.,0,0,0
5,342.,252.,0,0,0
6,684.,252.,0,0,0
7,0.,408.,0,0,0
8,342.,408.,0,0,0
9,684.,408.,0,0,0
10,0.,564.,0,0,0
11,342.,564.,0,0,0
12,684.,564.,0,0,0
13,0.,720.,0,0,0
14,342.,720.,0,0,0
15,684.,720.,0,0,0
16,0.,876.,0,0,0
17,342.,876.,0,0,0
18,684.,876.,0,0,0
19,0.,1032.,0,0,0
20,342.,1032.,0,0,0
21,684.,1032.,0,0,0
22,0.,1188.,0,0,0
23,342.,1188.,0,0,0
24,684.,1188.,0,0,0
25,0.,1344.,0,0,0
26,342.,1344.,0,0,0
27,684.,1344.,0,0,0
28,0.,1500.,0,0,0
29,342.,1500.,0,0,0
30,684.,1500.,0,0,0
31,0.,1656.,0,0,0
32,342.,1656.,0,0,0
33,684.,1656.,0,0,0

153
~,
"
I',
C.'
, ;."'. '",
34,0.,1812.,0,0;0 I
35,342.',1812.,0,0,0
36,684.,1812.,0;0,0
)" ',' "i .•.•.•.
, '~A ~.i'• 1 ,

Total No., of Elements, Nodes per Element


"J,. . .'. ',' . .,
552 c(t " '. , • I "
' ' ••• " ;' oj "-

' " , ' '" J, ."


st nd
Element No., Element
I, ~
Type, 1 Node, 2 I ' ••
Node, I" Node Connection Type, 2nd Node: .
Connection Type; sectional Type, Material type
J' .'.~-:j-{~'t(l
1,2,1,4~9,~;2;1'1
"
2,2,2,5,9,9,1;1, 'i 1
3;2,3,6,9,9,4;~';' ,
1,2;4;5,7,7,11,1 ;':
5,2,5,6,7,7,11;1' •
6,2,4,7,9,9,2,1" .,. '.:'\.r :;;".
7,2,5,8,9,9,1,1
8,2,6,9,9,9,4,1
9,2,7,8,7,7,11,1 . {t;;;~,;;( . I, ". f'Z7~ :>~~",~'r_
- ~:' '-1i~
19,2,8,9,7,7,11,1 t, i,
11,2,7,10,9,9;5;1 {;-
0
12,2,8,1),9,?,3:1i;; 11
13,2,?,I2,9,9,6,1. " :' ~ '
14,2;10,1 q,7,1 ~,l G;' "
15,2,ll,12J,7~lI,10 0
1.6,2;IO,13;9,9,~;IC 113 • ,
17,2,11,14,9,9,3;1,,"0 ~ ' ~.
18,~,12;15,9,9,6,1 '.
19,2,13,14J,7;1l;l'''1d :'c.<l;,'r;,I i!
20,2,14,15,7,7,11,1
2"1,2,13,16,9,9,7,1
22,2,I4,17,9,9,5,t ,',;P'-, 'ir~;'.',~n~~t',.::
l,-; )1.',
2~,2,15,18,9,9,8, I
24,2,16,17,7,7, II, I
25,2,17;18,7,7,11,1
26,2,16,19,9,9,7,1
27,2,1'7,20,9,9,5,1
.,- " ~ '.-'

28,2,18,21,9,9,8,1
2'9,2,19,20,7,7,11,1
30,2,20,21,7,7,11,1
31,2,19,22,9,9,9, I
32,2,20,23,9,9,5,1
33,2,21,24,9,9,9,1
34,2,22,23,7,7,11,1
35,2,23,24,7,7,11,1 • ~,-.
36,2,22,25,9,9,9,1
37,2,23,26,9,9,5, I
38,2,24,27,9,9,9, I
39,2,25,26,7,7,11,1
40,2,26,27,7,7,11,1
41,2,25,28,9,9,10,1

154
.
;

G,
42,2,26,29,9,9,9,1
43,2,27,30,9,9,10,1
44,2,28,29,7,7,11,1
45,2,29,30,7,7,11,1
46,2,28,31,9,9,10,1
47,2,29,32,9,9,9,1
48,2,30,33,9,9,10,1
49,2,31,32,7,7,11,1
50,2,32,33,7,7,11,1
51,2,31,34,9,9,10,1
52,2,32,35,9,9,9,1
53,2,33,36,9,9,10,1
54,2,34,35,7,7,11,1
55,2,35,36,7,7,11,1
Total No. of Connection Types
8
Connection Type, Connection size Parameters
1,12.,0.5,0.,3.5,0.,0.,0.
2,12.,0.5,0.,6.,0.,0.,0.
3,12.,0.5,0.,6.,0.5,0.,0.
4,21.4,0.5,0.,0.,0.,0.875,8.
5,24.4,0.75,0.,0.,0.,0.75,0.
6,24.4,0.75,0.,0.,0.,0.,0.
7,21.4,2.065,0.,0.,0.,1.25,12.
8,21.4,0.5,0.51,2.,0.,0.,8.
Total No. of Sectional Property Types
11
Sectional Property Type, Area, Moments oflnertia
1,117.0,6000.
2,101.0,4900.
3,92.4,4390 .
. 4,84.3,3910.
5,76.6,3520.
6,64.5,2790.
7,62.0,2660.
8,59.8,2530.
9,49.25,2020.
10,35.3,1380.
11,24.3,1830.
Total No. of Material Types
1
Material Type, Modulus of Elasticity, Poisson 's Ratio, Modulus of Rigidity
1,29000.,0.25,11200.

155
Total No. of Loaded Joints for Gravity loads
33
Joint No., Load in X direction, Load in Y direction, Moment about Z axis
.4,0.,-61.,0.
5,0.0,-71.,0.
6,0.0,-61.,0.
7,0.,-61.,0.
8,0.0,-71.,0.
9,0.0,-61.,0.
10,0.,-61.,0.
11,0.0,-71.,0.
12,0.0,-61.,0.
13,0.,-61.,0.
14,0.0,-71.,0.
15,0.0,-61.,0.
16,0.,-61.,0.
17,0.0,-71.,0.
18,0.0,~61.,0.
19,0.,-61.,0.
20,0.0,-71.,0.
21,0.0,-61.,0.
22,0.,-61.,0.
23,0.0,-71.,0.
24,0.0,-61.,0.
25,0,,-61.,0.
26,0.0,-71.,0.
27,0.0,-61.,0.
28,0.,-61.,0.
29,0.0,-71.,0.
30,0.0,-61.,0.
31,0.,-61.,0.
32,0.0,-71.,0.
33,0.0,-61.,0.
34,0.,-45.,0.
35,0.0,-45.,0.
36,0.0,-45.,0.
Total No. of Loaded Members
22
Member No., Axial, Shear & Moment at the Left End, Axial, Shear & Moment at
the Right End
4,0.,35.5,3035.3,0.,35.5,-3035.3
5,0.,35.5,3035.3,0.,35.5,-3035.3
9,0.,35.5,3035.3,0.,35.5, -3035.3
10,0.,35.5,3035.3 ,0.,35 .5,-3035.3
14,0.,35.5,3035.3,0.,35.5,-3035.3
15,0. ,35 .5,3035 .3,0.,35.5,-3035.3

156
19,0.,35.5,3035.3,0.,35.5,-3035.3
20, 0.,35.5,3035.3,0.,35.5, -303 5.3
24,0.,35.5,3035.3,0.,35.5,-3035.3
25,0.,35.5,3035.3,0.,35.5,-3035.3
29,0.,35.5,3035.3,0.,35.5,-3035.3
30,0.,35.5,3035.3,0.,35.5,-3035.3
34, 0.,35.5,3035.3,0.,35.5, -3035.3
35,0.,35.5,3035.3,0.,35.5,-3035.3
39,0.,35.5,3035.3,0.,35.5,-3035.3
40,0.,35.5,3035.3,0.,35.5,-3035.3
44,0.,35.5,3035.3,0.,35.5,-3035.3
45,0.,35.5,3035.3, 0.,35.5,-3035.3
49,0.,35.5,3035.3,0.,35.5,-3035.3
50,0.,35.5,3035.3,0.,35.5, -3035.3
54,0.,22.5,1923.8,0.,22.5,-1923.8
55,0.,22.5,1923.8,0.,22.5,-1923.8
Total No. of Loaded Joints for L,ateral Load
11

Joint No., Load in X direction, Load in Y direction, Moment about Z axis


4,10.95,-0.,0.
7,8.37,-0.,0.
10,9.23,-0.,0.
13,9.63,-0.,0.
16,10.<J4,-0.,0.
19,10.04,-0.,0.
22,10.89,-0.,0.
25,10.89,-0.,0.
28,10.89,-0.,0.
31,11.73,-0.,0.
34,5.86,-0.,0.

Problem Type, Total No. of Load Increments, Load Case


2,10,2

•.•• *1t~

157

Das könnte Ihnen auch gefallen