MODELING OF SEMI-RIGID STEEL CONNECTIONS

BY MULTI-LINEAR EQUATION
Prakit Premthamkorn and Prakit Chomchuen

Abstract: The mathematical model for semi-rigid connection purposed in Eurocode 3

renders advantages of clear analytical connections between mechanical properties and
mathematical parameters. However, its M-θ relationship based on the tri-linear
equation is still rather crude approximation. This study proposes an improvement by
employing multi-linear technique. A parameter called “shape factor” is introduced.
The shape factor, processed from database of test results, defines the shape of the
relationship curves.

1. Introduction

Conventional analysis of steel frames assumes either fully rigid or ideally

pinned connections. In reality, most connections behave somewhat between the two
extremes. The behavior of a frame connection subjected to moment is defined by
relationship between moment and connection relative rotation (M-θ curve) as shown
in Fig. 1. The rigidity of the connections effects the distribution of negative and
positive bending moments along the members, and rotational deformation of the
structural elements. The later greatly influences the accuracy of calculation of P-Δ
effect which is very important to the stability of the structures.

Figure1. Connection Moment-rotation Curves

Analysis and design of semi-rigid frames requires the M-θ relationship of the
connections. Several mathematical models of the relationship have been developed.
Among the most popular models are: Frye-Morris Polynomial Model, Kishi and Chen
Power models, and EuroCode3’s Tri-linear model. Each model has its own
advantages and drawbacks.
The EuroCode3’s Tri-linear model estimates the M-θ relationship by a tri-
linear relationship. It employs the so-called “component method” to derive the
connection initial stiffness, and plastic moment. This study proposes an improved
model based on the multi-linear equation. The three basic parameters for the models
are initial stiffness (K), plastic moment (Mp), and shape factor (n). K and Mp are the
basic mechanical properties of the connections. Finally, the shape factor defines
degree of reduction in stiffness of the connection as the rotation increases. This
parameter is dependent of the performance of all components of the connection as
moment and rotation increases. Although not impossible, this is rather complex to
analyze by an equivalent mechanical system such as finite element analysis(FEA).
However it is not practical to perform FEA to all connections during the analysis and
design of the whole structures. Therefore, it is more convenient to predetermine this
characteristic by empirical method. The following outlines the formulation of the
modeling.

2. Multi-linear equation

Multi-linear equation is composed of a series of linear equations. The curve is

derived from the assumption that its slope decreases as the moment increases, and
slope is zero when moment reaches plastic moment of the connection as show in
Figure 2 and Equation 1.

θj = ∑
j
[M i − M i −1 ] (1)
i =1 ⎡ ⎛M ⎞
n
⎤
K ⎢1 − ⎜ i ⎟ ⎥
⎢ ⎜⎝ M p ⎟
⎠ ⎥
⎣ ⎦

where K is initial stiffness of connection, M p is plastic moment of connection, and

n is shape factor.

K
MP

Figure 2. Basic concept of Multi-linear equation

In Equation 1, the parameter that influences the shape of moment-rotation

curve is “shape factor” ( n ) as shown in Figure 3. The higher the shape factor is, the
sharper the curve is.

3. Determination of K and Mp by component method

Component method provides a convenient tool to calculate the initial stiffness

and plastic moment of steel connections. The component method is delineated in
Eurocode 3 [1]. It can be used to calculate initial stiffness and plastic moment for
three types of connections: welded connections, bolted end-plate connections, and
bolted connections with angles. Calculation consists of the following steps [4]:
− listing of the components of the beam-to-column connection to be
examined;
− evaluation of the force versus displacement relationship of each
component;
− assembly of the components for evaluating the initial stiffness and
plastic moment of the whole connection.

45
M
40
M
35 p n2
K n1
30

25
Mi
20
M i −1
15

10 n1 < n2
5

0
0 0.005 0.01 0.015 0.02 θ
0.025

Figure 3. Influence of shape factor

As an example, components for a fully welded connection (Figure 4) are

shown in Figure 5. The first three components (cws, cwc, and cwt) govern both the
rotation stiffness and flexural strength of the connection. Conversely, the last two
components (cfb and bfc) provide only a limitation to the connection flexural
resistance as show in Figure 6.

Figure 4. Fully welded connection

 Figure 5. Component of welded connection

Figure 6. Mechanical model of welded connection

It is obvious that the plastic moment ( M p ) is governed by the weakest

component whose design resistance is:

FRd = min{Fcws. Rd , Fcwc. Rd , Fcwt . Rd , Fcfb. Rd , Fbfc.Rd } (5)

therefore:

M p = FRd ht (6)

In addition, the initial stiffness of the connection can be computed from:

ht2
K = (7)
1 1 1
+ +
K cws K cwc K cwt

The bolted connections can be similarly analyzed for initial stiffness and
plastic moment. The components for bolted end-plate connections (Figure 7) are
shown in Figure 8. The equivalent spring elements are shown in Figure 9.
 Figure 7. Bolted end-plate connection

Figure 8. Components of bolted end-plate connection

Figure 9. Mechanical model of bolted end-plate connection

4. Shape factor

As described above, the shape factor defines the sharpness of a multi-linear

equation to math the actual M-θ curve. It is assumed that the shape factor is
Mp
proportional to the ratio of plastic moment and initial stiffness ( ).
K
A regression analysis has been conducted in order to find a form of equation
Mp
that best fits the n - relationship. It was found that the inverse equation
K
(Equation 5) allows the closest approximation of the relationship.

b bK
n=a+ =a+ (5)
θo Mp

For bolted connections with angles, the shape factor is shown in Equation 6
and Figure 10.
 0.0014 K
n = 1.2575 + (6)
Mp

3.5
3
2.5
2 test
n

0.0014 K
n = 1.2575 +
1.5 Mp

1
0.5
0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014
Mp/K

Figure 10. Shape factor for bolted connection with angles

For bolted end-plate connections, the shape factor is shown in Equation 7 and
Figure 11.

0.0012 K
n = 1.7472 + (7)
Mp

3.5
3
2.5 n = 1.7472 +
0.0012K
test
Mp
2
n

1.5
1
0.5
0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016
Mp/K

Figure 11. Shape factor for bolted end-plate connections

For welded connections, the shape factor is shown in Equation 8 and Figure 12
 0.0039 K
n = 1.2923 + (9)
Mp

8
7
6
5
4 test
n

3 n = 1.2923 +
0.0039K
Mp
2
1
0
0 0.002 0.004 M/K 0.006 0.008 0.01

Figure 12. Shape factor for welded connections

5. Comparison of Results

The moment-rotation relationship proposed in this study (multi-linear

equation) is compared to testing results and other mathematical models i.e.
polynomial model [5], three parameter power model [2] [3], tri-linear model [1], and
finite element method [6] (only bolted connections with angles)
The M-θ relationships from multi-linear modeling are compared to test results
and other types of modeling for typical samples of connections for bolted connections
with angles, bolted end-plate connections, and welded connections, in Figure 13, 14,
and 15, respectively.

100
90
80 Polynomial Model
Moment (KN-m.)

70
60 Three Parameter Power Model
50 8S1
Finite Element
40
30 Multi-Linear
20
10 Eurocode 3
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
rotation (rad.)
 120

100 Polynomial Model

Moment (KN-m.)
80

60
Three Parameter Power Model
40 8S2
Finite Element Multi-Linear
20 Eurocode3

0
0 0.005 0.01 0.015 0.02 0.025 0.03
rotation (rad.)

120

100
Polynomial Model
Moment (KN-m.)

80
Three Parameter Power Model
60 8S3

40
Multi-Linear
Finite Element
20 Eurocode 3
0
0 0.01 0.02 0.03 0.04
rotation (rad.)

Figure 13. Comparison behavior of bolted connections with angles

450

400
Polynomial Model
350

300
Moment (KN-m.)

t109004
250
Multi-Linear
200
Eurocode 3
150

100

50

0
0 0.01 0.02 0.03 0.04 0.05
rotation (rad.)
 600

500
Polynomial Model

Moment (KN-m)
400 t109005

300 Multi-Linear
Eurocode 3
200

100

0
0 0.01 0.02 0.03 0.04 0.05
rotation (rad.)

140
t101010
120

100
Moment (KN-m.)

Multi-Linear
80 Eurocode 3

60 Polynomial Model

40

20

0
0 0.01 0.02 0.03 0.04 0.05
rotation (rad.)

Figure 14. Comparison behavior of bolted end-plate connections

350
t105018
300
Eurocode 3
250
Moment (KN-m.)

200 Multi-Linear

150

100

50

0
0 0.01 0.02 0.03 0.04 0.05
rotation (rad.)
 450
400 t105019

350 Eurocode 3

Moment (KN-m.)
300 Multi-Linear
250
200
150
100
50
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
rotation (rad.)

700

600 Eurocode 3 t105021

500
Moment (KN-m.)

Multi-Linear
400

300

200

100

0
0 0.005 0.01 0.015 0.02 0.025
rotation (rad.)

Figure 15. Comparison behavior of welded connections

6. Conclusion

A mathematical model for semi-rigid steel connections has been purposed.

The moment-rotation relationship is based on a multi-linear technique. Initial
stiffness, plastic moment, and shape factor are employed as basic parameters. The
former two parameters can be analyzed from a mechanical modeling as suggested by
Eurocode 3’s component method. The shape factor defining the shape of the
relationship curve, is an dependence of the ratio of plastic moment and initial
stiffness. The shape factors for three types of connections were derived.
In comparison to other popular models, the model subjectively emulates the
relationship well for all three types of the connections namely bolted connections with
angles, bolted end-plate connections, and welded connections. The model renders
clear connections with mechanical properties and versatility of analytical models but
allows ease of use of the empirical models.

7. References
[1] CEN, 1993. Eurocode 3 (Annex J). Draft for Development. Comite Europeen de
Normalisation.
[2] Chen, and Kishi, 1989. Semirigid Steel Beam-To-Column Connections: Data Base
and Modeling. Journal of Structural Engineering, 15(1) : 105-109.
[3] Chen et al, 1992. Semi-rigid connections in Steel Frames. United States of
America : McGraw-Hill, Inc.
[4] Fealla et al, 1999. Structural Steel Semirigid Connections: Theory, Design and
Software. United States of America : CRC Press LLC.
[5] Frye and Morris, 1975. Analysis of Flexibly Connected Steel Frames. Canadian
Journal of Civil Engineering, Vol.2 : 280-291.
[6] N. Kishi et al, 2001. Nonlinear Finite Element Analysis of Top- and Seat-Angle
with Double Web-Angle Connections. Structural Engineering and Mechanics,
Vol.12, No.2 : 201-214.

ABOUT THE AUTHORS

Prakit Premthamkorn and Prakit Chomchuen: Mahanakorn University of Technology,

Bangkok, Thailand