Connection Modelling in Fire

Ian Burgess

Definition of joint and connection

Joint

Connections

EC3-1.8 Stiffness classification of joints

M

φ

Semi-rigid

Pinned S >0,5EIb/Lb

Rigid

S >kbEIb/LbKb=8 (braced frames)Kb=25 (other frames)

Other classifications

Strength:

• “Full-strength”: Bending strength > Strength of member.

• “Partial-strength”

• “Pinned”: Bending strength < 0.25 Strength of member.

Ductility:

• “Ductile/Class 1”: Sufficient rotation capacity to develop plastic mechanism.

• “Semi-ductile”

• “Brittle/Class 3”: Can only be used in elastic design.

Component Modelling at Ambient Temperature

Beam web

Principal component zones of end-plate

M

Column web

Beam flange, web

Column web

Compression

Beam web

Column web Column

flange End plate

Tension bolts

Tension

EC3-1.8 extended end-plate joint model

M

Beam web tension

Column web tension

Column flange bending End plate

bending

Bolttension

Beam flangeColumn web

compressionColumn

web shear

Equivalent spring model

ce

M

z2

ke,2

ke,1

z1

,

, , , , ,

11 1 1 1 1et i

cwt i cfb i bt i epb i bwt i

k =

k k k k k+ + + +

Simplified spring model

ce

keq

zeq

M

Moment-rotation behaviour in fire

Semi-rigid behaviour of connections in fire: original work

Late 80s – early 90s

• Design opportunity? Interest particularly based on using connection residual stiffness and strength to enhance the fire resistance of “simple” steel beams.

• Analytical studies to assess the likely advantages/problems.

Semi-rigid behaviour of connections in fire: cruciform tests

Mid – 1990s

• Cruciform tests to create small database of M-φcurves. Semi-empirical models to rationalise results.

• 2 successive experimental projects at BRE Garston (Lennon) in “portable” furnace.

1. First series (Leston-Jones) tested a limited range of small non-composite flush-endplate connections.

2. Second series (Al-Jabri ) tested flush and extended endplate connections in composite and non-composite arrangements, including some Cardington connections.

Garston cruciform tests: test arrangement

Column head restrained in position

Load Load

Beam restrained horizontally

Specimen located within furnace

Test Furnace

Specimen

Garston cruciform tests: “portable” furnace

Garston cruciform tests: End-plate Type 2

50

50

150

1010 90

60

375

254x254x89UC(50)

356x171x51UB(50)

6mm FW all round

All holes 22mm dia. forM20 Grade 8.8 bolts

190

High-temperature M-φ-θ characteristics of endplate joints

Rotation (mrad)

Flush End-plate

20 40 60 80 100

20°C 200°C 400°C

500°C

600°C

700°C

35 40 Moment (kNm)

25 20

15 10 5

0 0

20

40

60

80

100

20 40 60 80 100 Rotation (mrad)

Curves at 100°C intervals

Moment (kNm)

20°C/100°C

Flexible End-plate

30

Cardington beam-column joint after fire test

Buckling due to high axial force on heating

Component zones in end-plate joint

Column web in tension

Column web in compression

Beam flange incompression

Beam web in tension and compression

Column flange andend plate in bending

Bolts in tension

Tension Compression Shear

M

Column web in shear

Beam web in shear

Component zones in end-plate joint with axial thrust

Reduced tension

Highercompression in web

Beam flange in highercompression

Beam web mainly incompression

Reduced force on column flange

Lower tension

Tension Compression Shear

Column web in shear

M

F

Beam web in shear

Component Behaviour at High Temperatures

Spyrou Component testing 1998-2001: Objectives

Compression Zone

• Examine experimentally the effect of elevated temperatures on column web buckling.

• Develop simplified/semi-empirical model of column web compression component behaviour for end-plate joints.

• Check both against finite element modelling.

Generally

• Check flush endplate moment-rotation predictions against previous cruciform furnace tests.

Tension Zone

• Do experiments on T-stubs at high temperatures.

• Develop simplified/semi-empirical model of tension component behaviour for end-plate joints.

• Check both against finite element modelling.

Spyrou furnace and control apparatus

Control panel

View portsfor cameras

Loading actuator

Failure modes for tension T-stubs

Failure Mode 2 Failure Mode 3Failure Mode 1

Spyrou compression zone test arrangement

ActuatorReaction Frame

Specimen

Furnace

Roberts’ model for ultimate strength of stocky webs

Uniform stress σyw

ββ c

Pu1 2 3 4

Simplified model of compression zone

tfb

γ/10

β

γ

w

γe

β c

300

w1

γ/10

Leff=e+(γ/5)

1.Yield of column web

Principles of simple compression zone model

2 3

β

γ'

w

γ'e'

βc

650tfb

w1

Lt

Yielding

2. Yield of column flange (first plastic hinges)

β Δw′

Le=β+c/4

c/2Mpfc

1

3. Yield of column flange (final plastic hinges)

Qian shear panel tests

Restrained high-temperature joint testing

Qian & Tan restrained tests

Qian & Tan restrained specimen

UB305X102X25kg/m

14521652

700

1500

16521452100 100

800

ENDPL

UC254X254X107kg/m

ENDPL

100 100

D29D29

D29

D29

D29

D29

D29 D29

Qian & Tan restrained tests

Qian & Tan restrained tests

Qian & Tan restrained tests

1. Modified Rotational Models

Component Modelling in Fire

Simoes da Silva (2001) component approach

• Find ambient-temperature force-displacement response at ambient temperature according to EC3-1-8 component method principles.

• Apply high-temperature material reduction factors for stiffness and strength to produce high-temperature equivalents.

Simoes da Silva modelling of Sheffield tests

Modified rotational models: Qian & Tan

?

VNb

Nc

Nc

cwt cfb epb bt

bwscws cwc bfc

F

?

FRd

K

F

?

FRd

Elastic-plastic component Rigid-plastic component

cwt cfb epb bt

cwt cfb epb bt

cwt cfb epb btMV

M

Component Modelling in Fire

2. General Connection Elements

The “Component” method with axial force

• In fire axial compression acts together with moment due to restraint to thermal expansion. Component model would deal with this automatically, though M-φcurves change due to thrust.

K1

K2

Kc

The “Component” method with axial force

FcKc

F2

F1K1

K2

Ft

M

• In fire axial compression acts together with moment due to restraint to thermal expansion. Component model would deal with this automatically, though M-φcurves change due to thrust.

The “Component” method with axial force

• In fire axial compression acts together with moment due to restraint to thermal expansion. Component model would deal with this automatically, though M-φcurves change due to thrust.

FcKc

F2

F1K1

K2

Ft

M

F

Component-Based Connection Element (Block)

kC kTekTc

kS

u

w

φ

lT,1

lT,3

lT,2

lC,1

lC,2

21

kC kTekTc

kS

u

w

φ

lT,1

lT,3

lT,2

lC,1

lC,2

21 21Compression Spring- Column Web

Tension Spring –

T-Stub in End-Plate

Shear Spring- Bolts

Comparison of joint element with tests by Leston-Jones

0

100

200

300

400

500

600

700

800

900

0 10 20 30 40 50 60 70 80Connection rotation [mrad]

Stee

l temp

eratu

re [°

C]

BFEP 5 - 5 kNmBFEP 15 -15 kNmBolt rows individuallyBolt rows as a group

Component-based connection element: beam shear panel

kC kTekTc

kS

u

wφ

lT,1

lT,3

lT,2

lC,1

lC,2

21

kC kTekTc

kS

u

wφ

lT,1

lT,3

lT,2

lC,1

lC,2

21 21

Implementation of joint element in software

I

J

K

L

The end …

… nearly …

Next emphasis

Floor beam Shear tab plate

Column tree

Robustness in tying (tension) of typical connection details in fire.

Example: WTC5 system

Failure of tab plates in WTC 5 column trees

Combined catenary tension and shear

Current work on robustness

… Thank you