You are on page 1of 65

A CONNECTION ELEMENT FOR MODELLING

END-PLATE CONNECTIONS IN FIRE

Dr Zhaohui Huang
Department of Civil & Structural Engineering, University of Sheffield

22 September 2009

Presented at STiFF Meeting on 22 September 2009

1. INTRODUCTION

Presented at STiFF Meeting on 22 September 2009

Three approaches for modelling the behaviour


of connections in fire
To represent the momentrotation characteristics of
a connection by mathematical expression in the form
of curve-fitting equations.
To use component-based (also known as springstiffness) models for predicting the connections
behaviour.
To model the connection as assembly of 3D finite
shell, brick and contact elements using general
commercial software, such as ABAQUS or ANSYS.

Presented at STiFF Meeting on 22 September 2009

2. DEVELOPMENT OF THE BOLTED


END-PLATE CONNECTION ELEMENT

Presented at STiFF Meeting on 22 September 2009

Two-noded connection element configuration


Central line of

Node of connection

column

element (beam side)

Reference plane

Slab element

Offset = l

Node of connection
element (column side)

Connection element

z
z
w

y
u

x
x

Local coordinates

Presented at STiFF Meeting on 22 September 2009

Beam element

Stiffness matrix of connection element

F = K u
Fx ,1
k
0
0 0
0
k11l
11

Fy ,1
0 k 22 0 0
0
0

F
0
0 k33 0
0
z ,1 0

M x ,1
0
0 0 k 44
0
0

M y ,1 k11l 0
0 0 (k55 + k11l 2 ) 0

k 66
0
0 0
0
M z ,1 0
=

k11l
0 0
0
Fx , 2 k11 0
F 0 k 22 0 0
0
0
y,2
0 k33 0
0
0
Fz , 2 0
0

0 0 k 44
0
0
M x , 2
M k11l 0 0 0 (k55 + k11l 2 ) 0
y,2

0 0 0
0
k 66
M z , 2 0

Presented at STiFF Meeting on 22 September 2009

(1)
k11

0 k 22 0 0
0
0
0
0 k33 0
0
0

0
0 0 k 44
0
0
k11l 0
0 0 (k55 + k11l 2 ) 0
0
0
0 0
0
k 66

k11 0 0 0
0
k11l
k 22 0 0
0
0
0

0
0 k33 0
0
0
0
0 0 k 44
0
0
k11l 0 0 0 (k55 + k11l 2 ) 0

k 66
0
0 0 0
0
0

k11l

u1

v1
w1

x ,1

y ,1
z ,1

u 2
v2

w2

x,2
y , 2

z , 2

( 2)

Stiffness matrix of connection element

In this model only the in plane (x-z plane) behaviour


of the connection is considered.

It is reasonable to assume that the stiffness


coefficients of k 22 , k 44 , k 66 in Eq. (2) have
infinite magnitude (assumed to be 10 9 kN / mm )

Presented at STiFF Meeting on 22 September 2009

Modelled two-dimensional steel frame in fire


The columns were fire protected and beams were
uniformly heated with load of 25 kN/m

305x305x97UC
4m

9m
Connection

9m

Symmetry line
9m

533x210x92UB
B

4m

Fire

The connection was represented as an axial pinned or


rigid spring with different stiffnesses for modelling
pinned or rigid connection
Presented at STiFF Meeting on 22 September 2009

Predicted deflections at Position A for the connections


using axial pinned spring with different stiffnesses
0

Deflection (mm)

-350
9

10 kN/mm
120 kN/mm

-700

40 kN/mm
14 kN/mm

-1050

7 kN/mm

-1400
0

200

400

600

800
o

Beam temperature ( C)

Presented at STiFF Meeting on 22 September 2009

1000

Predicted deflections at Position B for the connections


using axial pinned spring with different stiffnesses
0

Deflection (mm)

-250
9

10 kN/mm
120 kN/mm

-500

40 kN/mm
14 kN/mm

-750

7 kN/mm

-1000
0

200

400

600

800
o

Beam temperature ( C)

Presented at STiFF Meeting on 22 September 2009

1000

Predicted deflections at Position A for the connections


using axial rigid spring with different stiffnesses
0

Deflection (mm)

-350
9

10 kN/mm
120 kN/mm

-700

40 kN/mm
14 kN/mm

-1050

7 kN/mm

-1400
0

200

400

600

800
o

Beam temperature ( C)

Presented at STiFF Meeting on 22 September 2009

1000

Predicted deflections at Position B for the connections


using axial rigid spring with different stiffnesses
350

Deflection (mm)

0
9

10 kN/mm
120 kN/mm

-350

40 kN/mm
14 kN/mm

-700

7 kN/mm

-1050
0

200

400

600

800
o

Beam temperature ( C)

Presented at STiFF Meeting on 22 September 2009

1000

Determination of axial and vertical stiffness


coefficients, k11 , k 33
A very simplified approach was used for the current model
to determine k11 , k 33:
z Before the connection failure, k11 = 10 9 kN / mm
z When the connection fails due to tension, bending or
.
vertical shear, k11 = 0
z When the connection failed by compression,

k11 = 10 9 kN / mm
z

Before the connection failure due to vertical


shear, k33 = 10 9 kN / mm

z After the connection fails by vertical shear, k 33 = 0


Presented at STiFF Meeting on 22 September 2009

The detail of bolted end-plate connection


between steel column and beam

ep

Tension bolts

Fr 3, Rd

i =1

ep

Fr1, Rd

z
Shear bolts

d3

Presented at STiFF Meeting on 22 September 2009

F
i =1

p(3)
p(4)

p2

p3 (1)

(a) Flush end plate

Tension bolts

Fr 5, Rd

p2

ri , Rd

p(1)

Fr 4, Rd

p(3)

ec

p(2)

Fr 3, Rd

p(2)

ex

d1

Fr 2, Rd

p(1)

Fr 2, Rd

ec

ex

Fr1, Rd

Fr 4, Rd

Shear bolts

ri , Rd

(b) Extended end plate

p3 (1)

d3

Rotational stiffness of a beam-to-column


connection, S j

Sj =

Ez

1
ki

(3)

E = Youngs module

ki

= stiffness coefficient for basic connection


component i

z = lever arm

= stiffness ratio,S j ,ini / S j


Presented at STiFF Meeting on 22 September 2009

The initial rotational stiffness S j ,ini of the connection is


given by Eq. (3) with = 1

S j ,ini

E z2
=
1
k
i
i

Presented at STiFF Meeting on 22 September 2009

( 4)

Column web panel in shear ( k1 )


For unstiffened single-side or double-sided connection in
which the beam depths are similar:

0.38 AVC
k1 =
z
= transformation parameter

AVC = shear area of the column

Presented at STiFF Meeting on 22 September 2009

(5)

Column web in compression ( k 2 )

k2 =

0.7 beff ,c , wc t wc
d c ,c

(6)

d c , c = clear depth of the column web

t wc

= thickness of the column web

beff ,c ,wc = effective width of the column web in


compression

Presented at STiFF Meeting on 22 September 2009

Column web in tension (

k3 =

0.7 beff ,t , wc t wc
d c ,c

k3 )

(7 )

beff , t , wc = effective width of the column web in tension


and equal to the effective length, l eff , c
l eff , c is calculated based on bolt-row considered
individually for this bolt-row for an unstiffened
column flange

Presented at STiFF Meeting on 22 September 2009

End-plate in bending (

k5 =

3
0.9 l eff , p t p
3
mp

(8)

tp =

thickness of end-plate

l eff , p

= effective lengths for an end-plate

m p = a parameter related to geometry of the


connection

Presented at STiFF Meeting on 22 September 2009

k5

Bolts in tension (for a single bolt-row,

k10
As

1.6 As
=
Lb

(9)

= tensile stress area of the bolt

Lb = the bolt elongation length

Presented at STiFF Meeting on 22 September 2009

k10 )

Equivalent stiffness coefficient,

k eq =
hr

keff ,r hr
r

(10)

zeq

= distance between bolt-row r and the centre


of compression

k eff , r = effective stiffness coefficient of bolt-row r


k eff ,r =

1
ki ,r

Presented at STiFF Meeting on 22 September 2009

1
1
1
1
1
+
+
+
k3,r k 4,r k5,r k10,r

(11)

k eq

zeq =

equivalent lever arm

zeq =

2
k
h
eff ,r r
r

keff ,r hr

(12)

Rotational stiffness for one bolt-row in tension,

Sj =

E z2

1
ki

E z2
1 1 1 1 1
1
+ + + + +
k1 k 2 k3 k 4 k5 k10

Presented at STiFF Meeting on 22 September 2009

Sj
(13)

Rotational stiffness for two or more bolt-rows in tension, S

Sj =

E z2

1
ki

E z2
1 1

+ +
k1 k 2 k eq

(14)

E = average Youngs module for the connection


and changes with temperature

E=

Ecw + Ecf + Ebw + Ebf + E p


5

Presented at STiFF Meeting on 22 September 2009

(15)

Tri-linear moment-rotation characteristic


used for the connection element
Mj
C

M j , Rd

2
M j , Rd
3

Id =

Xd =

A
Unloading

S j ,ini

S j ,ini
O

Id

Xd

Presented at STiFF Meeting on 22 September 2009

Cd

2 M j , Rd
3 S j ,ini
2 M j , Rd
S j ,ini

Cd = 5 Xd

For line OA ( Id ):

M j = k55 = S j ,ini

(16)

For line AB ( Id < Xd ):

2
M j = k55 ( Id ) + M j , Rd
3
k 55 =

(17)

M j , Rd

3( Xd Id )

For line BC ( Xd < Cd ):

M j = k55 ( Xd ) + M j , Rd

(18)

Presented at STiFF Meeting on 22 September 2009

k 55 = 0.065 S j ,ini

Resistance of bolt rows in the tension zone, Ftr , Rd

Ftr , Rd = min ( Ft , fc , Rd ; Ft , wc , Rd ; Ft ,ep , Rd ; Ft , wb , Rd ) (19)


Ft , fc , Rd the column flange in bending

Ft ,wc , Rd the column web in tension

Ft ,ep , Rd the end-plate in bending

Ft ,wb , Rd the beam web in tension

Presented at STiFF Meeting on 22 September 2009

Using equivalent T-stub in tension model


There are three failure modes:

Mode 1: complete
flange yielding

Mode 2: bolt failure


with flange yielding

Presented at STiFF Meeting on 22 September 2009

Mode 3: bolt failure

Column flange in bending, Ft , fc , Rd


For failure Mode 1 (without backing plates):
2

M pl ,1, Rd ,r ,c =

FT ,1, Rd , fc =
t fc =

0.25 leff ,c t fc f y ,c

M0

4 M pl ,1, Rd ,r ,c
mc

(20)

(21)

thickness of the column flange

f y , c = yield strength of column


Presented at STiFF Meeting on 22 September 2009

For failure Mode 2:


2

M pl , 2, Rd ,r ,c =
FT , 2, Rd , fc =

mc

0.25 leff ,c t fc f y ,c

M0

2 M pl , 2, Rd ,r ,c + n p ,c Ft , Rd
mc + n p ,c

(22)

(23)

and n p ,c are parameters related to


geometry of the connection

Presented at STiFF Meeting on 22 September 2009

For failure Mode 3:

FT ,3, Rd , fc = Ft , Rd
Ft , Rd

Ft , Rd =

(24)

= resistance of individual bolt

k 2 f ub As

M2

(25)

k 2 = 0.9
f ub = ultimate tensile strength of the bolt,

As

= tensile area of the bolt,

Ft , fc , Rd = min FT ,1, Rd , fc ; FT , 2, Rd , fc ; FT ,3, Rd , fc


Presented at STiFF Meeting on 22 September 2009

( 26)

Column web in transverse tension, Ft ,wc , Rd

Ft , wc , Rd

beff ,t , wc t wc f y ,c
=
M0

(27)

= reduction factor to allow for the interaction


with shear in the column web panel

1
beff ,c , wc t wc
1 +1.3
Avc

Presented at STiFF Meeting on 22 September 2009

(28)

End plate in bending, Ft ,ep , Rd


For failure Mode 1 (without backing plates):
2

M pl ,1, Rd ,r ,b =

FT ,1, Rd ,ep =

0.25 leff , p t p f y , p

M0

4 M pl ,1, Rd , r ,b
m p1

(29)

(30)

f y , p = yield strength of end-plate


Presented at STiFF Meeting on 22 September 2009

For failure Mode 2:


2

M pl , 2, Rd ,r ,b =
FT , 2, Rd ,ep =

0.25 leff , p t p f y , p

M0

2 M pl , 2, Rd ,r ,b + n p ,ep Ft , Rd
m p1 + n p ,ep

(31)

(32)

m p1 and n p ,ep are parameters related to


geometry of the connection

Presented at STiFF Meeting on 22 September 2009

For failure Mode 3:

FT ,3, Rd ,ep = Ft , Rd

(33)

The resistance of the end plate in bending is

Ft ,ep , Rd = min FT ,1, Rd ,ep ; FT , 2, Rd ,ep ; FT ,3, Rd ,ep

Presented at STiFF Meeting on 22 September 2009

(34)

Beam web in tension, Ft , wb , Rd

Ft , wb , Rd =

beff ,t , wb t wb f y ,b

(35)

M0

beff ,t , wb = effective width of the beam web in


tension and equal to

twb =

leff , p

thickness of the beam web

f y , b = yield strength of beam

Presented at STiFF Meeting on 22 September 2009

Compression Resistance,

Fc , Rd = min Fc , wc , Rd ; Fc , fb , Rd

Fc , Rd

(36)

Fc , wc , Rd column web in transverse compression


Fc , fb , Rd beam flange and web in compression

Presented at STiFF Meeting on 22 September 2009

Resistance of column web in transverse


compression, Fc,wc,Rd
For an unstiffened column web:
k wc beff ,c , wc t wc f y ,c k wc beff ,c , wc t wc f y ,c

;
Fc , wc , Rd = min
M0
M1

k wc

= reduction factor

= reduction factor for plate buckling

Presented at STiFF Meeting on 22 September 2009

(37)

Resistance of Beam flange and web in


compression, Fc , fb , Rd

Fc , fb , Rd =

M c , Rd

(h

hb

t fb

(38)

= depth of the connected beam

M c , Rd = moment resistance of the beam


cross-section

t fb

= flange thickness of the connected beam

Presented at STiFF Meeting on 22 September 2009

If the height of the beam including the haunch exceeds


600 mm the contribution of the beam web to the
compression resistance should be limited to 20%:

Fc , fb ,max
bb

bb t fb f y ,b
0.8 M 0

(39)

= width of the beam section

Fc , fb , Rd Fc , fb , max
Presented at STiFF Meeting on 22 September 2009

( 40)

Force distribution in bolt rows


The first condition that the effective tension
resistance has to satisfy is:

Fc , Ed Fc , Rd

( 41)

Fc , Ed = Ftr , Rd

( 42)

r =1

N = total number of bolt rows in tension

Presented at STiFF Meeting on 22 September 2009

If Fc , Ed > Fc , Rd the force distribution in bolt rows


should be adopted to make sure that:
N

Fc , Ed = Ftr , Rd = Fc , Rd
r =1

Presented at STiFF Meeting on 22 September 2009

( 43)

Moment resistance, M

M j , Rd = hr Ftr , Rd

j , Rd

(44)

Ftr , Rd = effective design tension resistance of bolt-row r

hr

= distance from bolt-row r to the centre


of compression

r = the bolt-row number

Presented at STiFF Meeting on 22 September 2009

Resistance for individual bolt subjected


to vertical shear forces, Fb , Rd

Vb , Rd = min Fv , Rd ; Fb ,cf , Rd ; Fb ,ep , Rd

( 45)

Fv , Rd shear resistance of one bolt


Fb ,cf , Rd bolts in bearing on column flange

Fb ,ep , Rd bolts in bearing on end plate

Presented at STiFF Meeting on 22 September 2009

V f u ,bolt As
Fv , Rd =
M2

(46)

f u ,bolt = ultimate tensile strength of bolt


For classes 4.6, 5.6 and 8.8 bolts: V = 0.6
For classes 4.8, 5.8, 6.8 and 10.9 bolts: V = 0.5

Presented at STiFF Meeting on 22 September 2009

Fb,cf , Rd =

Fb,ep , Rd =

d=

k1,c b,c f u ,c d t fc

M2
k1, p b , p f u , p d t p

M2

(47)

(48)

nominal bolt diameter

f u , b = ultimate tensile strength of column


f u , p = ultimate tensile strength of end-plate
Presented at STiFF Meeting on 22 September 2009

Resistance of the connection with the axial


force N j , Ed in the connected beam

M j , Ed
M j , Rd
Mj,Rd

N j , Ed
N j , Rd

1.0

(49)

= moment resistance of the connection,


no axial force

N j , Rd = axial resistance of the connection,


no applied moment

Presented at STiFF Meeting on 22 September 2009

The moment resistance of the connection which


consider the influence of axial force:

'
j , Rd

N j , Ed
= 1.0

N j , Rd

M j , Rd

(50)

The axial resistance of the connection which consider


the influence of applied moment:

N 'j , Rd

M j , Ed
= 1.0

M j , Rd

Presented at STiFF Meeting on 22 September 2009

N j , Rd

(51)

For tension resistance:


N

N j , Rd = Ftr , Rd

(52)

r =1

For compression resistance:

N j , Rd = Fc , Rd
Presented at STiFF Meeting on 22 September 2009

(53)

Connection behaviours at elevated temperatures


The model presented above can be extended into
elevated temperatures by relating all material
properties, such as yield strength; ultimate tensile
strength and Youngs module to the temperature.

It is assumed that the material degradation of bolt at


elevated temperatures is the same for the structural steel.

The model specified in Eurocode 3 Part 1.2 is adopted in


this research.

Presented at STiFF Meeting on 22 September 2009

3.

VALIDATIONS

Presented at STiFF Meeting on 22 September 2009

Bolted end-plate connection tested at


ambient temperature

The detail of ambient temperature test (Leston-Jones 1997)

Presented at STiFF Meeting on 22 September 2009

Comparison of predicted and measured momentrotation curves (Leston-Jones 1997)


40

Moment (kNm)

30

Test (Leston-Jones)

20

Prediction (current model)


Prediction, bolt rows
10

individually (Block et al)


Prediction, bolt rows as a
group (Block et al)

0
0

20

40

60

Rotation (Millirads)

Presented at STiFF Meeting on 22 September 2009

80

100

The detail of tests at elevated temperatures


(Leston-Jones 1997)

The load levels applied to Test 1, Test 2,


Test 3 and Test 4 were 5 kNm, 10 kNm,
15 kNm and 20 kNm, respectively
Presented at STiFF Meeting on 22 September 2009

800

Beam bottom flange temperature ( C)

Comparison of predicted and measured connection


rotations for Test 1 and Test 3 (Leston-Jones 1997)

600

Test 1 (tested, Leston-Jones)

400

Test 1 (predicted, Block et al)


Test 1 (predicted, current model)
Test 3 (tested, Leston-Jones)

200

Test 3 (predicted, Block et al)


Test 3 (predicted, current model)
0
0

20

40

60

80

Connection rotation (Millirads)

Presented at STiFF Meeting on 22 September 2009

100

800

Beam bottom flange temperature ( C)

Comparison of predicted and measured connection


rotations for Test 2 and Test 4 (Leston-Jones 1997)

600

400

Test 2 (tested, Leston-Jones)


Test 2 (predicted, Block et al)
Test 2 (predicted, current model)

200

Test 4 (tested, Leston-Jones)


Test 4 (predicted, Block et al)
Test 4 (predicted, current model)

0
0

20

40

60

80

Connection rotation (Millirads)

Presented at STiFF Meeting on 22 September 2009

100

800

Beam bottom flange temperature ( C)

Influence of beam axial forces on the connection


behaviour for Test 1 (Leston-Jones 1997)

600

400
Tested results (Leston-Jones)
Predictions (axial force = 0)
200

Predictions (axial force = 20.9 kN)


Predictions (axial force = 41.9 kN)

0
0

20

40

60

80

Connection rotation (Millirads)

Presented at STiFF Meeting on 22 September 2009

100

The detail of Group 1 (FB1) connection fire tests


(Al-Jabri et al 2005)

Four tests FB11, FB12, FB13, FB14 were conducted at


load levels corresponding to connection moments of
4.4 kNm, 8.2 kNm, 13.12 kNm and 17.1 kNm, respectively
Presented at STiFF Meeting on 22 September 2009

Comparison of predicted and measured connection


rotations for Group 1: FB11, FB13 (Al-Jabri et al 2005)

Beam bottom flange temperature ( C)

800

600

FB11 (tested)

400

FB11 (predicted)
FB13 (tested)
200

FB13 (predicted)

0
0

20

40

60

80

Connection rotation (Millirads)

Presented at STiFF Meeting on 22 September 2009

100

Comparison of predicted and measured connection


rotations for Group 1: FB12, FB14 (Al-Jabri et al 2005)

Beam bottom flange temperature ( C)

800

600

400
FB12 (tested)
FB12 (predicted)
200

FB14 (tested)
FB14 (predicted)

0
0

20

40

60

80

Connection rotation (Millirads)

Presented at STiFF Meeting on 22 September 2009

100

The detail of Group 2 (FB2) connection fire tests


(Al-Jabri et al 2005)

Four tests denoted as FB21, FB22, FB23, FB24 were


carried out using the load levels of 27.4 kNm, 54.8 kNm,
82.1 kNm and 110 kNm, respectively
Presented at STiFF Meeting on 22 September 2009

Comparison of predicted and measured connection


rotations for Group 2: FB21, FB23 (Al-Jabri et al 2005)

Beam bottom flange temperature ( C)

800

600

400

FB21 (tested)
FB21 (predicted)
FB23 (tested)

200

FB23 (predicted)

0
0

20

40

60

Connection rotation

Presented at STiFF Meeting on 22 September 2009

80

100

Comparison of predicted and measured connection


rotations for Group 2: FB22, FB24 (Al-Jabri et al 2005)

Beam bottom flange temperature ( C)

800

600

400
FB22 (tested)
FB22 (predicted)

200
FB24 (tested)

FB24 (predicted)
0
0

20

40

60

Connection rotation

Presented at STiFF Meeting on 22 September 2009

80

100

4. CONCLUSIONS

The current model has the advantages of both the


previous simple and component-based models.

The current model is robust and has a capability to


predict the behaviour of bolted end-plate connection
under fire attack with reasonable accuracy.

Compared to the tested results the predictions of the


current model were mainly on conservative side.

The model can be used for structural fire engineering


design on steel-framed composite buildings.

The idea described in this paper can also easily be


applied to develop other kind of connections.

Presented at STiFF Meeting on 22 September 2009

Presented at STiFF Meeting on 22 September 2009

You might also like