European Commission

technical steel research

Steel products and applications for building, construction and industry
Large web openings for service integration in
composite floors
(
1
) C. Mller, O. Hechler, (
2
) A. Bureau, D. Bitar, D. Joyeux,
(
3
) L. G. Cajot, T. Demarco, (
4
) R. M. Lawson, S. Hicks, P. Devine,
(
5
) O. Lagerqvist, E. Hedman-Ptursson, E. Unosson, (
6
) M. Feldmann
(
1
) RWTH Mies-van-der-Rohe Strae 1, D-52074 Aachen
(
2
) CTICM Domaine de Saint-Paul, 102, route de Limours, F-78471 Saint-Rmy-ls-Chevreuse
(
3
) ProfilArbed 66, rue de Luxembourg, L-4221 Esch-sur-Alzette
(
4
) The Steel Construction Institute Silwood Park, Ascot SL5 7QN, Berkshire, United Kingdom
(
5
) University of Lule S-97 187 Lule
(
6
) Universitt Kaiserslautern Postfach 3049, D-67653 Kaiserslautern
Contract No 7210-PR/315
1 July 2001 to 31 December 2003
Final report
Directorate-General for Research
2006 EUR 21345 EN
LEGAL NOTICE
Neither the European Commission nor any person acting on behalf of the Commission
is responsible for the use which might be made of the following information.
A great deal of additional information on the European Union is available on the Internet.
It can be accessed through the Europa server (http://europa.eu).

Cataloguing data can be found at the end of this publication.

Luxembourg: Office for Official Publications of the European Communities, 2006

ISBN 92-79-01723-3

ISSN 1018-5593

European Communities, 2006
Reproduction is authorised provided the source is acknowledged.

Printed in Luxembourg

PRINTED ON WHITE CHLORINE-FREE PAPER
Europe Direct is a service to help you find answers
to your questions about the European Union
Freephone number (*):
00 800 6 7 8 9 10 11
(*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed.


Foreword
This report is subdivided into the following parts:
Part 1 contains a summary, general information on the project and a brief description of the works
performed.
In Part 2 test reports, analyses, evaluations and design models are presented in detail.
Design recommendations and predesign aids are given in Part 3 of the report.
The CD ROM enclosed to the report contains:
Part 1, Part 2 and Part 3 of the report in pdf-format
a computer software for the design of steel and composite beams with single large web openings
an interactive document including hypertext links for the predesign of cellular steel and
composite beams fabricated from rolled I-sections; design aids such as loading tables are
presented in terms of graphs and tables
3




Summary
The aim of the project "Large web openings for service integration in composite floors" was to
experimentally and theoretically investigate the behaviour of steel and composite beams with web
openings.
Therefore the works performed in the frame of this project have been subdivided into the following
main topics:
design of steel and composite beams with multiple regular or single large web openings under
normal (cold) conditions
design software for steel and composite beams with single large web openings and design aids
for steel and composite cellular beams
design of steel and composite beams with multiple regular web openings under fire conditions
improved manufacture of cellular beams

Concerning the design of steel and composite beams with multiple regular or single large web
openings under normal (cold) conditions the following tests were performed:
3 large scale tests on composite beams with multiple regular web openings with the aim to
study:
the composite action at the supports of simply supported beams
the behaviour of elongated openings
the influence of load introduction via secondary beams
the influence of highly asymmetrical steel sections
1 large scale test on a pure steel beam to study the effect of different types of stiffening of web
openings
8 large scale tests on steel and composite beams with unstiffened and stiffened single large web
openings in order to study:
the various failure modes
the influence of the size and the position of single large web openings
the influence of the concrete slab
the influence of different types of stiffening
All tests could be verified by adequate Finite Element simulations, so that it was possible to
perform further parameter studies in order to obtain information in areas not covered by the test
programme. Further on extensive Finite Element analyses were performed to get information on the
influence of non-centrally placed web openings leading to a simplified theory for the buckling
resistance of web posts and an alternative compression field or strut model combining compression
and bending in the web post.
For composite beams with multiple regular web openings the following main conclusions can be
drawn:
in general the failure mode of local instability, i.e. web post buckling, should not be decisive
and be avoided by stiffening if necessary; ring stiffeners have been proven to be the most
effective way of stiffening
the influence of the load introduction via secondary beams is negligible, if minimum
requirements (e.g. for the reinforcement, the strength of the secondary beams and its connection
to the main girder, etc.) are met
5


the influence of asymmetry is not as critical as first though and the limitation given in the codes
is too conservative
the stiffness of beams is largely influenced by multiple regular web openings leading to larger
deflections; thus the serviceability limit state is often governing the design
For steel and composite beams with unstiffened and stiffened single large web openings the
summary of the conclusions is as follows:
in general the adequate position of single large web openings should be chosen carefully in
order to avoid premature failure due to local effects
for the shear buckling verification a more accurate formula has been developed in terms of a
reduction factor to the shear buckling resistance of webs without openings
for the design model for the Vierendeel mechanism more beneficial limits for the cross sectional
classification of the Tee section are proposed
concerning elongated openings it could be verified that these can be treated in the same way as
rectangular web openings
the lateral torsional buckling check according to EN 1994-1-1 has been extended to cases with
single large web openings; further on conditions are defined where no lateral torsional buckling
check is necessary
for semi-continuous joints recommendations for the position of the web opening were
developed to guarantee that the joint behaviour is not influenced
A complete design guidance in terms of a step by step procedure has been elaborated giving a
generalised method applicable to hot rolled and fabricated steel sections with discrete and regular
openings. New aspects incorporated in this design guidance are
the treatment of bending and tension interaction effects in the bottom Tee section
the consideration of highly asymmetric sections
the treatment of local composite effects at large web openings
Re-calculations of some of the experiments performed have been carried out showing that the
procedure leads to satisfactory and conservative results.

6


A further outcome of the project is the development of a design software. This software designs and
analyses composite and non-composite beams, under distributed and/or point loads and covers:
composite and non-composite construction
use of standard deck profiles, user defined decks or a solid slab
propped and unpropped construction (of both the beam and the decking)
use as an edge or an internal beam
circular, rectangular and elongated openings, with or without stiffeners
additional point loads and additional uniformly distributed loads
use of normal weight and light weight concrete
use of headed shear connectors (including Hilti HVB shear connectors)
groups of shear connectors along the beam
The background of the software is documented in terms of an engineering logic.
Also a predesign tool for steel and composite cellular beams has been elaborated. It is presented in
terms of a specific document including hypertext links to graphs and tables giving some general
guidance on the predesign, the range of depth for cellular beams, section properties and loading
tables. This simple and efficient tool has been developed on the basis of 50000 case studies on
doubly symmetric cellular beams fabricated from European I-profiles for spans up to 24 m.

To study the behaviour of cellular composite beams under fire conditions two fire tests on
specimens similar to those tested under normal (cold) conditions were carried out:
an unloaded test in order to study the influence of the web openings and the way of application
of the fire protection on the heat distribution
a loaded test to investigate the influence of the web openings on the fire resistance
For both tests a sprayed fire protection has been used.
As conclusion from the test and the accompanying simulations on the heat distribution it can be
stated that in general the inner edge of openings should be protected in order to avoid a specific
assessment of the temperature field. The simple approach using massivity factors leads to safe-sided
results.
Concerning the fire resistance under loading the comparison of the test result with numerical
simulations leads to a good correlation as the failure mode, the time to failure and the vertical
displacement could be reproduced. First thoughts on a very simple approach for the design of the
fire protection thickness have been presented. Nevertheless there is only one test available and
further investigations will be necessary.

7


Concerning the improvement of the manufacture of cellular beams the main focus was on the
optimisation of the welding procedure of the web posts. The investigations have been performed in
two steps.
In the first step 13 small scale tests on plates representing the web posts were carried out with
different welding procedures, configurations and parameters with the aim to minimise the
deformations from welding. It can be concluded that the two sided welding, without chamfering for
web thicknesses smaller than 12 mm and with chamfering for web thicknesses greater than 12 mm,
are the most efficient and economic solutions.
In the second step 8 real scale tests on short beams with symmetric and asymmetric cross sections
loaded mainly in shear were made. If, from the overall design, full penetration butt weld were not
necessary the welds were carried out as partial penetration butt welds in order to optimise the
welding procedure and the consumption of the weld material. The test results could be verified by
Finite Element analyses and further parameter studies were carried out.
As conclusion it can be stated that in many cases full penetration butt weld are not necessary, so
that even chamfering can be avoided.



8


Contents Page
1 Introduction and general 11
1.1 Project objectives 11
1.2 Ways and means 11
2 Contribution from RWTH 13
2.1 Introduction 13
2.2 Description of test programme 13
2.2.1 General arrangements 13
2.2.2 Test 1 14
2.2.3 Test 2A 15
2.2.4 Test 2B 17
2.2.5 Test 3 18
2.2.6 Test 4 19
2.3 Fabrication and execution 20
2.4 Testing 20
2.4.1 General 20
2.4.2 Experimental results 22
2.5 Numerical Simulations 29
2.5.1 General 29
2.5.2 Comparison between simulations and tests on composite beams 31
2.5.3 Comparison of test results and FE analysis 31
2.5.4 Parametric Studies 34
2.6 Summary and conclusion 39
2.7 Design recommendations 41
2.8 Literature 42
3 Contribution from CTICM 43
3.1 General 43
3.2 Predesign tools for cellular beams 44
3.2.1 Introduction 44
3.2.2 Steel cellular beams 44
3.2.3 Composite cellular beams 49
3.3 Fire behaviour of cellular beams 51
3.3.1 Experimental results on the unloaded beam 53
3.3.2 Experimental results on the loaded beam 55
3.3.3 Numerical mechanical behaviour of the loaded beam 57
3.3.4 Analysis 60
9


3.3.5 Conclusion 62
3.4 References 63
4 Contribution from PROFILARBED-RESEARCH 65
4.1 Summary 65
4.2 ACB ARCELOR Cellular Beams fabrication 65
4.3 Fabrication of the test specimen 65
4.4 Improvement of the welding procedures for cellular beams 66
4.4.1 Actual assembling procedure for cellular beams 66
4.4.2 Small scale tests 67
4.4.3 Large scale tests 72
5 Contribution from SCI 87
5.1 Finite Element analysis of web post buckling 87
5.2 Design guidance for composite beams with large openings for services 88
5.2.1 Existing publications 88
5.2.2 Scope of publication 89
5.2.3 Design codes 89
5.2.4 Material strengths 89
5.3 ARCELOR web post buckling tests 90
5.3.1 Symmetric Beams 90
5.3.2 Asymmetric Beams 91
5.3.3 Conclusions 91
5.4 Engineering logic for the analysis and design software 92
5.5 Design Software 102
5.6 References 105
6 Contribution from UNIV LULEA 107
6.1 General 107
6.2 Main objectives 107
6.3 Slender web beams 108
6.3.1 Test results 108
6.3.2 Shear buckling 109
6.3.3 Vierendeel bending 112
6.3.4 Web-post buckling 113
6.4 Elongated circular openings 113
6.5 Beams in compression including lateral restraints - Lateral torsional buckling
of composite beams with web openings 114
6.6 Semi-continuous joints 119
6.7 References 120
10



1 Introduction and general
1.1 Project objectives
Currently, no design recommendations exist at a European level to permit design of large web
openings in composite beams without particular tests for the individual purpose.
Within this project the following issues concerning steel and composite beams and composite floors
with web openings have been studied by tests and theoretical investigations:
behaviour under normal conditions (effects from web buckling, bending resistance, shear
resistance, additional deflection from the web openings, effects of local composite action)
behaviour under fire conditions including the effects from fire protection
improvement of the manufacturing process of beams with single and regular web openings
On the basis of the results from the experimental and theoretical investigations design
recommendations as well as simplified design aids have been elaborated. Further on a computer
software for the design of steel and composite beams with single web openings has been developed.

1.2 Ways and means
The works performed in the frame of the project have been subdivided into 5 working packages
with several sub-tasks; for each working package one partner acted as the working package
coordinator.

WP 1: Tests on composite beams with web openings (WP coordinator: RWTH)
1.1 Review existing tests and their performance. Identify unexpected performance and gaps in
current tests.
1.2 Identify new test series to complement existing tests and to provide test information in areas
of current practice.
1.3 Set up tests on composite beams, based on the objectives identified in 1.2.
1.4 Analyse the performance of these tests.
1.5 Prepare test reports.

WP 2: Improved manufacture of beams with web openings (WP coordinator:
PROFILARBED)
2.1 Review current manufacturing processes.
2.2 Address improved manufacturing techniques.
2.3 Address requirements for connections.
2.4 Tests on modified or improved weld details and on distortion due to welding.
2.5 Address the use of high strength steel (S460) for perforated beams and the associated
manufacturing requirements.

11


WP 3: Fire resistance of perforated beams (WP coordinator: PROFILARBED)
3.1 Review the methods in Eurocode 3 and 4 Parts 1.2 for establishing the fire resistance and fire
protection required for steel and composite beams.
3.2 Identify fire tests on perforated beams.
3.3 Plan and execute fire tests.
3.4 Develop design rules for steel and composite beams with perforated webs taking account of
the temperature distribution in the web and bottom flange in comparison to solid web beams.
Prepare these rules in accordance with EC3 and EC4 Parts 1.2.
3.5 Prepare simplified design tables for perforated beams.

WP 4: Preparation of design guidance and software (WP coordinator: SCI)
4.1 Back-analyse all tests in accordance with the proposed design method. Present model factors
to demonstrate the level of safety of the proposed method.
4.2 Prepare draft recommendations.
4.3 Prepare simplified design tables.
4.4 Prepare resp. extent existing computer software for analysis of steel and composite beams
with large web openings.
4.5 Investigate practical issues of service integration for the cases identified above.
4.6 Prepare design recommendations.

WP 5: Special cases (WP coordinator: UNIV LULEA)
5.1 Develop design rules for slender web beams, including:
tension field action
vertical stiffeners
5.2 Develop design rules for elongated circular openings.
5.3 Develop design rules for beams subject to compression, including lateral restraints.
5.4 Develop design rules for semi-continuous construction through end-plate connections.

12


2 Contribution from RWTH
2.1 Introduction
The objectives of work package 1 Tests on Composite Beams with Web Openings has been to
analyse the behaviour of composite beams with multiple regular web openings under normal
conditions. Preliminary existing solutions have been reviewed to identify experimental
investigations providing test information in areas of current practice. Consequently the following
test programme has been fixed.
Test 1, Test 2 and Test 3 are tests on composite beams using metal sheets as a common composite
slab system. Also a common concrete grade is used. Test 4 is a test on a pure steel beam. The
following tests have been agreed:
Test 1 is the basic test with the main focus on the composite action in the end cell; Test 1 could
be extended to 2 tests 1A and 1B. After the expected failure at an opening at one end the
specimen could be fixed and the test could be continued until failure at the opening at the other
end.
Test 2A investigates elongated openings; Test 2B tests the load introduction via secondary
beams. The distance of the shear studs is kept similar to Test 1 or 3.
Test 3 has an highly asymmetrical section (A
f,b
: A
f,t
= 1 : 4).
Test 4 investigates the behaviour of stiffened openings.
This report presents a short summary of each single test. Furthermore this document includes a
rough overview of numerical analyses of the tests performed. In addition an extract of the
parametrical studies with the focus on the influence of material properties, geometry and load
introduction on the failure load and failure mechanism are enclosed. Finally the conclusions derived
from the test programme are given.

2.2 Description of test programme
2.2.1 General arrangements
The following arrangements for all tests have been agreed:
All steel parts are in grade S355.
One row of shear studs 19 x 100 with a distance of 150 mm has been applied.
The composite slab consists of steel sheet and concrete topping.
The overall slab thickness is 130 mm.
The width of the slab has been chosen to 1,8 m, i.e. larger than the effective width L/4.
Concrete grade C25/30 is used.
HOLORIB

steel sheets HR 51 / 150 with a thickness of 1,25 mm are applied.

For Test 1 and 3 the minimum reinforcement of 0,4 % in longitudinal and transversal direction
is applied.
For Test 2 the reinforcement for a good design is used.
Stiffeners are provided at support and under each point of load introduction.
The test specimens were propped during erection.
For Test 1 and Test 3 the sheets have been orientated perpendicular to the beam, for Test 2 parallel
to the beam. In both cases the sheets have primarily been punched and afterwards put over the shop-
welded studs.
13


2.2.2 Test 1
The scope of Test 1 is to study a basic test with multiple regular web openings to which the further
tests can be related to. Additionally the composite slab has been left away at one of the supports to
investigate the composite action in the end cell.
Thus a simply supported composite beam with regular web openings has been chosen, see
Figure 2-1. It has been loaded by four point loads P as close as possible to 1/8, 3/8, 5/8 and
7/8 L to simulate a uniformly distributed load.

Figure 2-1: Longitudinal section, cross section and loading - Test 1
The symmetrical cross section of the beam has been based on IPE 400 profiles; twelve regular
openings have uniformly been distributed along the beam span.
Further more Test 1 has been divided into 2 tests to achieve more information about the composite
action in the end cell:
Test 1A
Test 1B (reloading after failed opening has been closed on non-concreted side)
For test 1B the test specimen from Test 1A has been fixed by a transversal plate, see Figure 2-2,
closing opening 11 and 12 to avoid repeated failure on this side of the specimen.
14



Figure 2-2: Sealed openings 11 and 12 - Test 1B

2.2.3 Test 2A
The scope of Test 2A is to study the behaviour of an elongated opening. The opening is located in
the area of moment and shear interaction.


Figure 2-3: Test set-up - Test 2A
Therefore a simply supported composite beam with one elongated opening and several circular web
openings has been selected for Test 2A, see Figure 2-4. It has been loaded by a single point load P
centric over the cross section at midspan.
The beam has got a symmetrical cross section identical to test 1 and ten regular openings have
uniformly been distributed along the beam span interrupted by one elongated opening, see
Figure 2-4 and Figure 2-5. The elongated opening is gained by the fusion of two regular openings
and is situated in the middle of the overall beam.
15



Figure 2-4: Longitudinal section, cross section and loading - Test 2A
16


2.2.4 Test 2B
The scope of Test 2B is to study the influence of the load introduction via secondary beams on
composite beams with regular web openings whereas the secondary beams are also in composite
action. The beams are connected by end plates.
The test specimen of Test 2B is the same as for Test 2A under varying load and support conditions.
Hence Test 2B has been a simply supported composite beam with one elongated opening and
several circular web openings. It has been loaded by 4 point loads P introduced via secondary
beams, Figure 2-5.

Figure 2-5: Longitudinal section, cross section and loading - Test 2B

17


2.2.5 Test 3
The scope of Test 3 is to study the behaviour of an highly asymmetrical steel section with web
openings as a composite beam.
Additionally the composite slab has been left away at one of the supports to check the composite
action in the end cell.
Consequently Test 3 has been a simply supported composite beam with 12 regular web openings. It
has been loaded by four point loads P as close as possible to 1/8, 3/8, 5/8 and 7/8 L to simulate a
uniformly distributed load, see Figure 2-6.

Figure 2-6: Longitudinal section, cross section and loading - Test 3
The beam consisted of an asymmetrical steel section with A
f,b
/ A
f,t
> 3 (here 4:1), greater than
allowed according to ENV 1994-1-1: 1992, see Figure 2-7. The upper chord has been made of an
IPE 300, the lower chord of an HEB 340. Twelve regular openings have uniformly been distributed
along the beam span.
18



Figure 2-7: Ratio upper flange to lower flange - Test 3
2.2.6 Test 4
The scope of Test 4 is to check the behaviour of ring-stiffened, half-closed or closed circular web
openings.
The test is performed as a 3-point bending test of a pure steel beam, see Figure 2-8, Figure 2-9 and
Figure 2-10. 6 circular openings have uniformly been distributed along the beam span.
Ring stiffeners have been provided at opening 1 to 3, openings 4 and 5 are half-closed by infills and
opening 6 is fully closed by an infill.


Figure 2-8: Longitudinal section and
loading - Test 4
Figure 2-9: Test set-up - Test 4

19





Figure 2-10: Cross section and loading - Test 4

2.3 Fabrication and execution
All girders have been fabricated by PROFILARBED, Luxembourg. The test specimen have been
reinforced and concreted in the laboratories of the University of Kaiserslautern.
All tests have been carried out at the Labor fr konstruktiven Ingenieurbau, University of
Kaiserslautern, Germany.

2.4 Testing
2.4.1 General
Each test has been an ultimate limit state test. The test procedure comprises the execution of load
cycles at certain load levels to exclude slippage of the load introduction and supports as well as
friction in the shear joint and structure if existing. Additionally the vertical deflections, inclination
and strains have been measured to guaranty a detailed evaluation of each test. As examples the
following test set-ups of Test 1 and Test 4 are shown in Figure 2-11 and Figure 2-12.


Figure 2-11: Experimental set-up - Test 1 Figure 2-12: Experimental set-up - Test 4

20


Table 2-1 summarises the experimental results:

Table 2-1: Experimental results
Test Failure Load Failure criterium Remarks
Test 1A 805,6 kN Flexural web post buckling
between 11
th
and 12
th
opening
Uplifting of concrete chord
Vierendeel mechanism at opening 2
and 11 could be identified before
failure
The Vierendeel mechanism has slightly
been more developed at opening 11
than at opening 2
Unequal slippage at the different
supports
Test 1B 843,7 kN Failure due to web post
buckling between 1
st
and 2
nd

opening
Vierendeel mechanism at opening 2
could be identified before failure
Test 2A 481,2 kN no failure stopped at a load level of 481,2 kN to
avoid any influence on Test 2B by
excessive cracking of the concrete
chord
Vierendeel mechanism could be
identified at a load level of 337 kN by
uplifting of the concrete chord
Test 2B 780,4 kN Stopped due to excessive
yielding of the bottom
T-section at the elongated
openings at midspan
Slight lifting up of the sheets at 530 kN
at midspan over the elongated opening
Longitudinal cracks on top of the
concrete chord due to transversal
bending at a load level of 600 kN
Test 3 658,0 kN Web post buckling between
11
th
and 12
th
opening
The web post buckling shape has been
influenced by the different web
thicknesses
Vierendeel mechanism at opening 2
and 11 before failure
Vierendeel mechanism at opening 2 has
slightly been less deformed than at
opening 11
Unequal slippage at the different
supports
Test 4 793,0 kN Vierendeel mechanism at
opening 4
Weld cracked at upper part of the half-
moon stiffener along the perimeter
Half-moon stiffener buckled due to
deformations of the opening
Upper flange is extensively deformed

21


2.4.2 Experimental results
Diagrams of the measurements as well as photographs of the failure mechanism are given in the
following figures.
2.4.2.1 Test 1A

0
5
10
15
20
25
30
35
40
0 570 1140 1710 2280 2850 3420 3990 4560 5130 5700 6270 6840
Span [mm]
D
e
f
l
e
c
t
i
o
n

[
m
m
]
P = 208 [kN]
P = 306 [kN]
P = 406 [kN]
P = 509 [kN]
P = 602 [kN]
P = 705 [kN]
P = 806 [kN]

Figure 2-13: Deflection to length - Test 1A


0
100
200
300
400
500
600
700
800
900
0 0.5 1 1.5 2 2.5 3
Deflection [mm]
L
o
a
d

[
k
N
]
MP 70
MP 83
MS 70 MS 83
MP 70 MP 83
MP 70 MP 83

Figure 2-14: Slippage at end span - Test 1A

22



0
100
200
300
400
500
600
700
800
900
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
Strains [mm/m]
L
o
a
d

[
k
N
]
MP 54
MP 55
MP 56
MP 57
MP 58
MP 59
MP 60
MP 61
54
55
56
59
57
61
60
58
55
59
55
57
61 61
55 58

Figure 2-15: Strains around 11
th
opening in tangential direction - Test 1A


Figure 2-16: Flexural web post
buckling between opening 11 and 12 -
Test 1A
Figure 2-17: Vierendeel mechanism at
11
th
opening - Test 1A

23


2.4.2.2 Test 2A
0
5
10
15
20
25
0 570 1140 1710 2280 2850 3420 3990 4560
Span [mm]
D
e
f
l
e
c
t
i
o
n

[
m
m
]
P = 104 [kN]
P = 199 [kN]
P = 299 [kN]
P = 404 [kN]
P = 481 [kN]

Figure 2-18: Deflection to length - Test 2A


Figure 2-19: Vierendeel mechanism at elongated opening - Test 2A

24


2.4.2.3 Test 2B
0
10
20
30
40
50
60
70
0 570 1140 1710 2280 2850 3420 3990 4560 5130 5700 6270 6840
Span [mm]
D
e
f
l
e
c
t
i
o
n

[
m
m
]
P = 215 [kN]
P = 304 [kN]
P = 501 [kN]
P = 700 [kN]
P = 780 [kN]

Figure 2-20: Deflection to length - Test 2B

0
100
200
300
400
500
600
700
800
900
0 10 20 30 40 50 60 70
Deflection [mm]
L
o
a
d

[
k
N
]
MP 13
MP 30
MP 39

Figure 2-21: Deflection at load introduction - Test 2B
25


2.4.2.4 Test 3
0
5
10
15
20
25
30
35
40
45
50
0 570 1140 1710 2280 2850 3420 3990 4560 5130 5700 6270 6840
Span [mm]
D
e
f
l
e
c
t
i
o
n

[
m
m
]
P = 106 [kN]
P = 201 [kN]
P = 300 [kN]
P = 403 [kN]
P = 510 [kN]
P = 599 [kN]
P = 658 [kN]

Figure 2-22: Deflection to length - Test 3

0
100
200
300
400
500
600
700
-20 -15 -10 -5 0 5 10 15 20
Strains [mm/m]
L
o
a
d

[
k
N
]
MS 54
MS 55
MS 56
MS 57
MS 58
MS 59
MS 60
MS 61
54
56
57
61
59 58
57
56 55
54 61 60
55
60
58
59

Figure 2-23: Strains around 11
th
opening in tangential direction - Test 3

26



0
100
200
300
400
500
600
700
0 1 2 3 4 5 6
Deflection [mm]
L
o
a
d

[
k
N
]
MP 70
MP 83
MS 70 MS 83
70
83
MP 70 MP 83

Figure 2-24: Slippage at end span - Test 3


Figure 2-25: Web post buckling -
Test 3
Figure 2-26: Vierendeel mechanism -
Test 3

27


2.4.2.5 Test 4
0
5
10
15
20
25
0 570 1140 1710 2280 2850 3420
Span [mm]
D
e
f
l
e
c
t
i
o
n

[
m
m
]
P = 100 [kN]
P = 200 [kN]
P = 301 [kN]
P = 401 [kN]
P = 498 [kN]
P = 599 [kN]
P = 702 [kN]
P = 793 [kN]

Figure 2-27: Deflection over length - Test 4


0
100
200
300
400
500
600
700
800
900
-10 -5 0 5 10 15
Strains [mm/m]
L
o
a
d

[
k
N
]
MP 50
MP 55
55
50
55 50

Figure 2-28: Strains at opening 4 in longitudinal direction - Test 4

28





Figure 2-29: Deformations at opening 4
- Test 4
Figure 2-30: Buckling of half-moon
stiffener at opening 4 - Test 4

2.5 Numerical Simulations
2.5.1 General
To intensify the analysis of the test results and to enlarge the experiences concerning composite
beams with cellular openings numerical simulations have been performed.
The numerical simulation of the non-linear behaviour of composite members has to consider the
particular behaviour of the following elements:
Steel member
Concrete slab
Steel sheets
Reinforcement
Shear connectors
The software MARC/MENTAT [1] used for the numerical simulation works allows to describe and
to combine various material laws.
The steel parts are fully modelled taking the geometry as well as the non-linear material behaviour
into account. As the concrete slab is mainly under compression and the slab itself has hardly any
influence on the shear bearing capacity it has been possible to generate the concrete chord itself
using simplified assumptions for the discretisation. Local failure of the concrete therefore is not
been considered in the analysis whereas the global influence has been regarded.
The load-slip behaviour for the shear connectors of the composite beams has been taken into
account using springs with non-linear behaviour obtained by the Standard Push-Out Test.
29


The following Table 2-2 shows a short description of each model and the used assumptions and
simplifications.

Table 2-2: Discretisation of the models
Test 1A / 1B and Test 3 Test 2A / 2B


Concrete slab Three dimensional 20-
node-brick elements
Linear Mohr-Coulomb
Measured geometry
Concrete slab Three dimensional 8-
node-brick elements
Linear Mohr-Coulomb
Measured geometry
HOLORIB sheets Transversal
Not generated
HOLORIB
sheets
Longitudinal
Quadratic 4-Node Thick
Shell Elements
Von Mises
Shear connectors Springs with non-linear
behaviour
Shear connectors Springs with non-linear
behaviour
Steel beam Quadratic 8-Node Thick
Shell Elements
Von Mises
Measured geometry
Steel beam Quadratic 8-Node Thick
Shell Elements
Von Mises
Measured geometry
Initial
Imperfection
First (positive)
Eigenmode (h
w
/ 250)
Initial
Imperfection
First (positive)
Eigenmode (h
w
/ 250)

Test 4, as a pure steel beam, has completely been modelled by Quadratic 8-Node Thick Shell
Elements taking the measured geometries into account. To obtain the realistic failure behaviour an
initial imperfection according to the first (positive) Eigenmode has been applied for the analysis.
The boundary conditions for the models have been adjusted to the support condition and loading of
each test, also regarding lateral restraint at the load introduction.
As the analyses have been carried out taking geometric non-linearity into account the change in
structural behaviour and loss of structural stability has fully been considered.

30


2.5.2 Comparison between simulations and tests on composite beams
A comparison between the calculated and the measured deflections and strains as well as the failure
mechanisms are given in Table 2-3. These investigations show, that it is possible to model
composite beams with regular web openings with good agreement to the real behaviour.

Table 2-3: Comparison of results
Test FEM Test / FEM [%]
Test 1A Load P [kN] 806 789 98
Test 1B Load P [kN] 852 877 103
Test 2A Load P [kN] 481 546 113
Test 2B Load P [kN] 780 794 102
Test 3 Load P [kN] 659 649 98
Test 4 Load P [kN] 793 808 102

In the case of Test 2A the discrepancy between the FE-analysis and the test behaviour is acceptable
due to the fact, that the main focus of the FE-generation has been on Test 2B, where load
introduction via secondary beams has been investigated. The divergence can be explained by not
properly reflected local effects like uplift of the concrete chord, cracking of the concrete etc. by the
model chosen.

2.5.3 Comparison of test results and FE analysis
The following figures show the comparison of the load deflection curves and the failure modes
observed from the tests and the results from the FE analyses.
0
100
200
300
400
500
600
700
800
900
0 10 20 30 40 50
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM
0
100
200
300
400
500
600
700
800
900
0 10 20 30 40 50
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM

0
100
200
300
400
500
600
700
800
900
1000
0 10 20 30 40 50 60
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM
0
100
200
300
400
500
600
700
800
900
1000
0 10 20 30 40 50 60
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM

Figure 2-31: Load-deflection curve
- Test 1A
Figure 2-32: Load-deflection curve
- Test 1B

31


0
100
200
300
400
500
600
0 5 10 15 20 25
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM
0
100
200
300
400
500
600
0 5 10 15 20 25
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM

0
200
400
600
800
1000
0 20 40 60 80 100
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM
0
200
400
600
800
1000
0 20 40 60 80 100
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM

Figure 2-33: Load-deflection curve
- Test 2A
Figure 2-34: Load-deflection curve
- Test 2B

0
100
200
300
400
500
600
700
0 10 20 30 40 50 60
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM
0
100
200
300
400
500
600
700
0 10 20 30 40 50 60
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM

0
200
400
600
800
1000
0 5 10 15 20 25 30 35 40
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM
0
200
400
600
800
1000
0 5 10 15 20 25 30 35 40
Deflection [mm]
L
o
a
d

[
k
N
]
TEST
FEM

Figure 2-35: Load-deflection curve
- Test 3
Figure 2-36: Load-deflection curve
- Test 4


Scaling Factor 2


Figure 2-37: FEM failure mode - Test 1A Figure 2-38: Test failure mode - Test 1A

32




Scaling Factor 10


Figure 2-39: FEM failure mode - Test 2A Figure 2-40: Test failure mode - Test 2A


Scaling Factor 1


Figure 2-41: FEM failure mode - Test 2B Figure 2-42: Test failure mode - Test 2B


Scaling Factor 2


Figure 2-43: FEM failure mode - Test 3 Figure 2-44: Test failure mode - Test 3

33




Scaling Factor 2


Figure 2-45: FEM failure mode - Test 4 Figure 2-46: Test failure mode - Test 4

2.5.4 Parametric Studies
2.5.4.1 Influence of the material properties
Accompanying the re-calculation of the tests parametrical studies have been carried out with the
aim to enlarge the understanding of the test results. The main focus of the studies has been on the
influence of the material properties on the failure load and mechanism. Therefore the FE-models
calibrated to the tests have been recalculated with varying steel and concrete properties generally
used in construction.
The following tables and figures summarise the outcome of the parametrical study on the material
properties:

Table 2-4: Summary of parametric study on material properties Test 1A
Test Concrete
f
ck
[N/mm]
Steel
f
yk
[N/mm]
Failure load
F [kN]
Failure mechanism
Test 1A measured (42) measured (451/489) 789 Web Post Buckling
Test 1A 20 265 494 Web Post Buckling
Test 1A 20 400 670 Web Post Buckling
Test 1A 20 460 721 Web Post Buckling
Test 1A 40 265 503 Web Post Buckling
Test 1A 40 400 688 Web Post Buckling
Test 1A 40 460 721 Web Post Buckling
Test 1A 80 265 506 Web Post Buckling
Test 1A 80 400 707 Web Post Buckling
Test 1A 80 460 739 Web Post Buckling
Test 1A without concrete measured (451/489) 749 Web Post Buckling
Test 1A without concrete 265 398 Vierendeel
Test 1A without concrete 400 667 Web Post Buckling
34


Table 2-5: Summary of parametric study on material properties Test 2A
Test Concrete
f
ck
[N/mm]
Steel
f
yk
[N/mm]
Failure load
F [kN]
Failure mechanism
Test 2A measured (40) measured (383/474) 640 Vierendeel
Test 2A 20 235 408 Vierendeel
Test 2A 20 355 497 Vierendeel
Test 2A 20 460 552 Vierendeel
Test 2A 40 235 439 Vierendeel
Test 2A 40 355 531 Vierendeel
Test 2A 40 460 586 Vierendeel
Test 2A 60 235 462 Vierendeel
Test 2A 60 355 557 Vierendeel
Test 2A 60 460 611 Vierendeel

Table 2-6: Summary of parametric study on material properties Test 2B
Test Concrete
f
ck
[N/mm]
Steel
f
yk
[N/mm]
Failure load
F [kN]
Failure mechanism
Test 2B measured (40) measured (383/474) 794 Web Post Buckling
Test 2B 20 235 498 Web Post Buckling
Test 2B 20 355 657 Positive Moment
Test 2B 20 460 760 Positive Moment
Test 2B 40 235 482 Positive Moment
Test 2B 40 355 617 Web Post Buckling
Test 2B 40 460 729 Web Post Buckling
Test 2B 60 235 494 Positive Moment
Test 2B 60 355 628 Web Post Buckling
Test 2B 60 460 739 Web Post Buckling

Table 2-7: Summary of parametric study on material properties Test 3
Test Concrete
f
ck
[N/mm]
Steel
f
yk
[N/mm]
Failure load
F [kN]
Failure mechanism
Test 3 measured (30,2) measured
(408/467/488/453)
649 Web Post Buckling
Test 3 20 235 414 Web Post Buckling
Test 3 20 355 575 Web Post Buckling
Test 3 20 460 642 Web Post Buckling
Test 3 40 235 419 Vierendeel
Test 3 40 355 584 Vierendeel
Test 3 40 460 650 Web Post Buckling
Test 3 80 235 421 Vierendeel
Test 3 80 355 595 Web Post Buckling
Test 3 80 460 661 Vierendeel
Test 3 without concrete measured
(408/467/488/453)
380 Web Post Buckling
Test 3 without concrete 235 217 Buckling Upper Flange
Test 3 without concrete 355 329 Vierendeel
35


Table 2-8: Summary of parametric study on material properties Test 4
Test Concrete
f
ck
[N/mm]
Steel
f
yk
[N/mm]
Failure load
F [kN]
Failure mechanism
Test 4 pure steel measured (437/449) 808 Vierendeel
Test 4 pure steel 235 433 Vierendeel
Test 4 pure steel 355 650 Vierendeel
Test 4 pure steel 460 838 Vierendeel

0
100
200
300
400
500
600
700
800
900
0 10 20 30 40 50 60
Deflection [mm]
L
o
a
d

[
k
N
]
FEM Test
C20/25 S235
C20/25 S355
C20/25 S460
C40/50 S235
C40/50 S355
C40/50 S460
C80/95 S235
C80/95 S355
C80/95 S460
S235
S355
FEM Test
S460

0
100
200
300
400
500
600
700
0 5 10 15 20
Deflection [mm]
L
o
a
d

[
k
N
]
FEM Test
C20/25 S235
C20/25 S355
C20/25 S460
C40/50 S235
C40/50 S355
C40/50 S460
C60/75 S235
C60/75 S355
C60/75 S460
S460
S355
S235

Figure 2-47: Parametric study - Test 1A Figure 2-48: Parametric study - Test 2A

0
100
200
300
400
500
600
700
800
900
0 10 20 30 40 50 60 70
Deflection [mm]
L
o
a
d

[
k
N
]
FEM Test
C20/25 S235
C20/25 S355
C20/25 S460
C40/50 S235
C40/50 S355
C40/50 S460
C60/75 S235
C60/75 S355
C60/75 S460
S235
S460
S355

0
100
200
300
400
500
600
700
0 10 20 30 40 50
Deflection [mm]
L
o
a
d

[
k
N
]
FEM Test
C20 S235
C20 S355
C20 S460
C40 S235
C40 S355
C40 S460
C80 S235
C80 S355
C80 S460
S 235
S 335
S 460
FEM Test

Figure 2-49: Parametric study - Test 2B Figure 2-50: Parametric study - Test 3

0
100
200
300
400
500
600
700
800
900
0 5 10 15 20 25 30 35
Deflection [mm]
L
o
a
d

[
k
N
]
FEM Test
S235
S355
S460
S235
S460
FEM Test
S355

Figure 2-51: Parametric study - Test 4

36


2.5.4.2 Conclusions of parametric study on material properties
As failure of composite beams with cellular openings occurs due to shear the parametric study
verified that the failure load and failure mechanism mainly depends on the steel grade whereas the
concrete strength has a minor influence on the behaviour.
In the case of Test 1A a large rotational capacity for a steel S235 could be confirmed until the
failure takes place due to web post buckling between the 11
th
and 12
th
opening. For the steel S355
and S460 the failure criterion changes into the occurrence of web post buckling between the 11
th

and 12
th
opening without any rotational capacity. For Test 1A generally for all steel grades
Vierendeel mechanism could be preliminarily identified until failure due to post buckling occurs.
For Test 2A obviously failure due to the Vierendeel mechanism at the elongated opening 6 governs
the design as it has been intended for testing. Higher steel grades increase the ultimate failure load.
The parametric study on material properties for test 2B once more shows, that higher strength steels
tend to fail by local instability. Therefore the failure mode of Test 2B changes from failure by
excessive yielding of the bottom chord to web post buckling for higher grade steels like S460,
however with an increased failure load. This change of the failure mode is accompanied by a
decreasing moment rotational capacity of the specimen with increasing steel strength. Furthermore
it has to be remarked that the rotational capacity is also effected by the concrete strength of the
chord.
The results from the analysis of Test 3 analysis is comparative to the results of the FE-simulation
for Test 1A. As failure due to shear is clearly governing the failure mode plausibly the steel strength
only influences the failure load significantly. Since the failure modes web post buckling and
Vierendeel mechanism affect each other the minor influence of the concrete strength causes a
change of the failure mode of Test 3.
All numerical analyses of Test 4 failed by local failure at the 4
th
opening.

2.5.4.3 Influence of the geometry
2.5.4.3.1 Test 1 - End Cell
The scope of Test 1 has also been to study the composite action in the end cell. Therefore
FE-analyses have been performed considering a composite beam with and without the end cell
concreted, see Table 2-9.

Table 2-9: Summary of parametric study - Test 1A
Configuration Concrete
f
ck
[N/mm]
Steel
f
yk
[N/mm]
Failure Load
F [kN]
Failure Mechanism
end cell non-concreted 42 measured 794 Web Post Buckling
end cell concreted 42 measured 640 Web Post Buckling

According to the FE-analysis the non-concreted end cell is beneficial for the ultimate limit state
behaviour of the beam. The beam possesses more ductile behaviour for shear load transformation
and Vierendeel mechanism can largely locally develop until failure by web post buckling. Else the
concreted end cell induces the shear directly into the web post and large deformations are prevented
by the concrete chord on top. Therefore earlier buckling of the web post is occurring.

2.5.4.3.2 Test 4 - Stiffening
Since the purpose of Test 4 has been to investigate the influence of stiffening concepts 3 further
geometries have additionally been modelled. In the following study Geometry 1 represents the
37


original configuration tested. Geometry 2 is compatible to Geometry 1 leaving away all stiffening
configurations. Geometry 3 corresponds to Geometry 1 but with closed openings 3 to 6. It focuses
on the effect of ring stiffening. Finally Geometry 4 is related to geometry 1 having all openings
closed to compare the effect of stiffening to a beam with a compact web without any openings. The
results of the comparative study are shown in Table 2-10.

Table 2-10: Summary of parametric study - Test 4
Steel
f
yk
[N/mm]
Failure Load
F [kN]
Configuration Failure Mechanism
235 433 Geometry 1 Vierendeel
355 650 Geometry 1 Vierendeel
460 838 Geometry 1 Vierendeel
235 395 Geometry 2 Vierendeel
355 573 Geometry 2 Web Post Buckling
460 682 Geometry 2 Web Post Buckling
235 396 Geometry 2 - without imperfection Vierendeel
355 593 Geometry 2 - without imperfection Vierendeel
460 763 Geometry 2 - without imperfection Vierendeel
235 468 Geometry 3 Vierendeel
355 701 Geometry 3 Vierendeel
460 903 Geometry 3 Vierendeel
235 573 Geometry 4 M
pl

355 861 Geometry 4 M
pl

460 1109 Geometry 4 M
pl


Generally the stiffening arrangements should shift the failure mode to achieve an increase of the
failure load level and strengthen the specimen.
Comparing the stiffened beam tested to a beam with the same geometry but non-stiffened openings
clearly the stiffening configuration increases the failure load and stiffness, see Figure 2-52. No
stiffening results into an early failure by web post buckling. The non-stiffened geometry 2 without
considering an initial imperfection excludes web post buckling. Nevertheless the failure load is
lower than for the stiffened specimen. Therefore the use of half moon stiffeners is an efficient way
to increase the failure load significantly.
According to the parametric study comparing the ways of stiffening, see Figure 2-53, the appliance
of ring stiffeners is the most effective way to strengthen the openings.
38



0
100
200
300
400
500
600
700
800
900
0 5 10 15 20 25
Deflection [mm]
L
o
a
d

[
k
N
]
Geometry 1 - S235
Geometry 1 - S355
Geometry 1 - S460
Geometry 2 - S235
Geometry 2 - S355
Geometry 2 - S460
S235
S460
S355
Geometry 1 Geometry 2

0
100
200
300
400
500
600
700
800
900
1000
0 5 10 15 20 25 30 35
Deflection [mm]
L
o
a
d

[
k
N
]
Geometry 1 - S235
Geometry 1 - S355
Geometry 1 - S460
Geometry 3 - S235
Geometry 3 - S355
Geometry 3 - S460
S235
S460
S355
Geometry 3 Geometry 1

Figure 2-52: Parametric study on
geometry (1) - Test 4
Figure 2-53: Parametric study on
geometry (3) - Test 4
Comparing the failure load of beams with ring-stiffened openings to the failure load of beams with
a compact web with no openings a reduction of the failure load of 81% with a corresponding loss of
stiffness of 7% has been identified by the numerical studies. Therefore it has to be stated that it is
not possible to reach the load capacity of a beam with a compact web by the stiffening methods
investigated.

2.5.4.3.3 Influence of load introduction
According to the FE-simulations the appliance of stiffeners at the location of the load introduction
slightly increases the failure loads and does not change the failure mode of Test 1A. The stiffeners
were mainly provided to avoid early buckling under concentrated load introduction for the fire test
(see section 3.3) but they are not necessary for cold design. In addition studies varying the load
introduction show that the chosen test setup using 4 point loads instead of a uniformly distributed
load does not influence the outcome of the tests significantly.
The aim of Test 2B has been to identify the influence of load introduction via secondary beams on
the overall behaviour of the test specimen. In accordance to the parametrical study by numerical
simulations no major influence of the load introduction could be detected and the failure mode is
not concerned.

2.6 Summary and conclusion
The objectives of work package 1 Tests on Composite Beams with Web Openings has been to
analyse the behaviour of composite beams with multiple regular web openings under normal
conditions.
The test programme comprehended 3 tests on composite beams with multiple regular web openings
and 1 test on a pure steel beam with stiffened multiple regular web openings as follows:
Test 1 - Composite action in the end cell
Test 2A - Elongated openings
Test 2B - Load introduction
Test 3 - Asymmetric design
Test 4 - Stiffening of web openings
Therefore experimental investigations accompanied by numerical studies have been performed.

39


ad Test 1
The scope of Test 1 has been to study a basic test with multiple regular web openings. Additionally
the composite slab has been left away at one of the supports to investigate the composite action in
the end cell.
Uplifting of the concrete chord during testing shows that Vierendeel mechanism tends to develop
starting at low load levels due to the cellular geometry of the specimen. Finally the beam fails by
flexural web post buckling at the non-concreted opening. An unequal slippage of both sides has
been identified. After closing the failed opening the test specimen has been reloaded and failed due
to web post buckling at the concreted end cell on a significantly higher load level. The composite
action in the end cell therefore is contributing to the ultimate limit state behaviour.
The examination of Test 1 clearly reveals the dependency of failure modes on the steel grade. The
concrete strength hardly affects web post buckling whereas an influence of the concrete strength is
detected for the Vierendeel mechanism. Additionally it has been identified that the rotational
capacity of composite beams with cellular openings and higher steel grades is limited as instability
of the web post is governing the design. The non-concreted end cell is beneficial for the ultimate
limit state behaviour of the beam as the beam possesses more ductility for shear load
transformation. Vierendeel mechanism can largely develop. Else the concreted end cell is inducing
the shear directly into the web post and earlier buckling of the web post occurs.

ad Test 2A
The scope of Test 2A was to study the behaviour of an elongated opening. The opening is situated
in the area of moment and shear interaction.
The failure mode of Test 2A has clearly been the Vierendeel mechanism as expected. Comparing
the ultimate load of Test 2A to the one of Test 2B the bearing capacity in the area of moment and
shear interaction is reduced to approximately 40% due to the elongated opening.

ad Test 2B
The scope of Test 2B was to study the influence of the load introduction via secondary beams on
composite beams with regular web openings.
Test 2B shows that the load introduction via secondary beams is neglectable for the design, if
minimum requirements, e.g. for reinforcement, the strength of secondary beams and its connection,
etc. are met. The failure mode is not affected.

ad Test 3
The scope of Test 3 was to study the behaviour of an highly asymmetrical steel section with web
openings as a composite beam. The same beam has also been tested under fire conditions, see
section 3.3.
The test revealed, that Vierendeel mechanism tends to develop at low load levels due to the cellular
geometry of the specimen. Finally the beam fails by flexural web post buckling at the non-concreted
opening. The buckling mode of the post has obviously been influenced by the stronger and thus
strengthening web part of the lower flange. An unequal slippage of both sides has been identified.
As failure due to web post buckling is clearly governing the design of Test 3 the parametric study
plausibly shows that only the steel strength influences the failure load significantly. Since the
failure modes web post buckling and Vierendeel mechanism affect each other the minor influence
of the concrete strength may causes a change of the failure mode.
As the outcome of the Test 3 analysis is comparative to the results of Test 1 it is possible to state
that the limitation for the design of hybrid sections according to ENV 1994-1-1: 1992 to the ratio
40


A
f,b
/ A
f,t
< 3 is too conservative for composite beams. The concrete chord is sufficiently linked to
the upper flange to marginally strengthen the section and avoid local buckling of the compression
chord. Additionally the safety against web post buckling seems to be increased since the upper web
part is strengthened by the lower part due to different web thicknesses evoking a change in the
buckling mode.

ad Test 4
The scope of Test 4 is to check the behaviour of ring-stiffened, half-closed or closed circular web
openings.
Test 4 fails due to Vierendeel mechanism evoking buckling of the half moon stiffener. The upper
flange is extremely deformed.
Comparing the stiffened beam tested to a beam with the same geometry but non-stiffened openings
clearly the stiffening configuration increases the failure load and stiffness. The parametric study
comparing 2 ways of stiffening shows that stiffening by ring stiffeners is the most effective way to
strengthen the openings. The appliance of half moon stiffeners is also an efficient way to stiffen
openings to increase the failure load. Generally the stiffening arrangements should shift the failure
mode to achieve an increase of the failure load level and strengthen the specimen. In general it has
to be stated that it is not possible to reach the load capacity and stiffness of a beam with a compact
web by the stiffening methods investigated.

2.7 Design recommendations
As general design recommendation it should be stated that web post buckling, i.e. local instability,
should not govern the design. If necessary ring stiffeners or equal measures increase the web post
buckling resistance.
Furthermore the effect of the shifted plastic neutral axis due to the concrete chord or asymmetry of
the cross section of the beam is not notably affecting the buckling resistance as far as examined.
Further on different web thicknesses are not critical in terms of instability of the web post. Buckling
of the upper and thinner web part is stabilised by the stronger lower part of the web.
As web post buckling is very sensitive to imperfections buckling curve c should be applied for
design.
As beams with large web openings tend to fail by Vierendeel mechanism it is essential to chose an
adequate location of the large openings, i.e. in areas of low shear, so that the resistance of the beam
is not limited by local failure of the opening.
Additionally the interaction between Vierendeel mechanism and web post buckling should be
checked.
More over the stiffness of the beams is largely influenced by regular web openings leading to larger
deflections. Therefore serviceability is often governing the design of cellular beams.
The shear connection between the concrete chord and the upper flange of the steel beam is
influenced due to large deflections and local Vierendeel effects. These effects have to be considered
when designing an adequate shear connection for the composite beam.

41


2.8 Literature
[1] MSC.Marc Mentat 2003
[2] H. Bode. Euro-Verbundbau, 2. Auflage. Werner Verlag GmbH & Co. KG, Dsseldorf. 1998.


42


3 Contribution from CTICM
3.1 General
This section summarises the activities of the Centre Technique Industriel de la Construction
Mtallique (CTICM), St Rmy ls Chevreuse, in the ECSC Project "Large Web Openings for
Service Integration in Composite Floors" for the period between 1
st
July 2001 and 31
st
December
2003 as well as the main results obtained.
In the frame of this project, CTICM has focused its work especially on the two following topics :
1) Predesign tools for steel and composite cellular beams. The result consists in a specific
document including hypertext links to edit graphs and tables.
2) Fire behaviour of cellular beams. Two tests have been performed by CTICM in Maizires-ls-
Metz.
Moreover CTICM has participated to all the meetings of the project and has been involved in the
discussions concerning the various works performed by the partners in the frame of this research.
The main actions achieved during the research period have been :
1) Predesign tools
definition of the technical specifications
development of a specific software module for the calculation of loading capacity of steel
and composite cellular beams
elaboration of the loading tables for steel cellular beams
elaboration of the loading tables for composite beams
elaboration of graphs
guidance for the predesign of cellular beams
2) Fire behaviour of cellular beams
definition of the tests
numerical simulations to estimate the behaviour
performing the tests at the testing laboratory of CTICM
treating on-site measurements and issuing test report
3) Tests on the programme developed by the Steel Construction Institute
Some calculation tests have been performed by CTICM and remarks were sent to the SCI.
The progress of the work during the research period has been in satisfactory accordance with the
previsions.
43



3.2 Predesign tools for cellular beams
3.2.1 Introduction
The aim of this topic is to develop simple tools for the predesign of cellular beams fabricated from
European I-rolled profiles.
These tools mainly consist in tables giving the loading capacity for a wide range of span lengths,
but they also consist in tables of geometric properties. Graphs have also been plotted in order to
facilitate the predesign.
The loading capacity of cellular beams have been determined by using the software ARCELOR
Cellular Beams (ACB) developed outside of the project, by CTICM for ARCELOR. This software
allows the designer to check a steel or composite cellular beam. The verifications are based on the
Eurocodes 3 [1] and 4 [3], and more particularly the specific rules for cellular beams are those from
Annex N of Eurocode 3 [2]. Some calculation aspects are not in the field of application of the
Annex N, that is why the ACB software also refers to the SCI Publication n100 [4].
However a specific software module has been developed as an interface between the Excel
spreadsheet and the ACB Software. This module includes an iterative process based on a dichotomy
method for the estimation of the loading capacity. For this purpose, several ultimate limit states are
considered such as :
bending resistance,
shear resistance,
local bending with Vierendeel effect,
M-N-V interactions,
web post buckling,

and a deflection condition for the serviceability limit state.
Finally more than 50000 cases of steel cellular beams have been studied (different geometries,
different steel grades, different span lengths). This leads to a great number of tables and it has
been proposed to make them available on a CD ROM only. Therefore only examples of these tables
are given in this final report.

3.2.2 Steel cellular beams
3.2.2.1 Basic assumptions
The main assumptions concerning the predesign tables are given hereafter for steel cellular beams.
For steel cellular beams, only doubly symmetric cross-sections are considered : same basic profile
for the upper chord and the lower chord of the cellular beam, with the same steel grade.
For a given basic profile, an infinite number of geometrical possibilities (diameter a
0
and width w of
intermediate posts) exist even if some geometrical requirements must be fulfilled. Therefore some
specific limits have been proposed.
44


a) Geometrical conditions
The diameter a
0
and the width w of the posts are chosen in such a way that :

a
0
100 mm
w 70 mm
0.2 w/a
0
0.5
H
t
/a
0
2.5

w
a
0
a
0
H
t

Figure 3-1: Geometry of the openings
In the tables, the diameter varies from the minimal diameter to the maximal diameter with a pitch of
50 mm. For each diameter, the width of the web posts is taken equal to 70 mm, 100 mm, 150 mm,
200 mm up to the maximal value w
max
.
The basic profiles with a total depth lower than 190 mm are excluded.

b) Basic profiles and steel grades
The following basic profiles are considered : IPE, HEA, HEB, HEM
The following steel grades are considered : S235, S275, S355, S420, S460

c) Span length
The span length varies from 5 m to 10 m with an increment of 1 m and from 12 m to 24 m with an
increment of 2 m. In order to propose realistic solutions, it has been decided to keep the solutions
which approximately fulfil the following conditions :
5 L/H
t
50
However the solutions which do not fulfil :
10 L/H
t
30
are presented with a grey background colour.

d) Supports and lateral restraints
The beam is assumed to be simply supported and fully restrained against lateral displacement.
Therefore no lateral torsional buckling verification is performed.
45


e) Load combinations
Combination for ULS checks :
G
G +
Q
Q
Combination for SLS check : G + Q
where :
G
= 1.35

Q
= 1.50
G is only the self-weight of the cellular beam
Q is the variable load (uniformly distributed)
The predesign tables supply the load capacity Q. This is conservative because the user may include
other permanent loads in the load capacity Q. These permanent loads are then assumed to be
factored by 1.50 instead of 1.35. This may be considered as acceptable at a predesign stage.

f) Limit states
Under the ULS combination, one of the following limit states may be determinant for the load
capacity of a cellular beam :
1 Resistance to axial force (N)
2 Resistance to shear force (V)
3 Resistance to bending moment (M)
4 N-V interaction
5 M-N interaction
6 M-V interaction
7 M-N-V interaction
8 Horizontal shear resistance of a post
9 Buckling resistance of a post
10 Section resistance
11 SLS deflection limit
Items 1 to 7 are related to the resistance of each chord. These limit states include Vierendeel effects.
All the verifications are performed according to Eurocode 3 part 1-1 "General rules and rules for
buildings" and according to the specific rules from Annex N "Openings in webs".
The partial safety factors are taken from the ENV version of Eurocode 3 :

M0
= 1,10

M1
= 1,10
Item 11 is related to the maximal deflection of the beam under the SLS combination :
v
max
L/250
where : v
max
is the maximal deflection,
L is the span length of the beam.
46



3.2.2.2 Examples of predesign tables for steel cellular beams
The Table 3-1 is an example of predesign table for steel cellular beams. For various solutions (basic
profile, opening diameter and intermediate post width) and for various span lengths, it gives :
the number n of openings along the beam
the width we in mm of the end posts
the loading capacity q in kN/m (uniformly distributed)
a code for the determinant limit state as above mentioned

Table 3-1: Predesign tables for steel cellular beams
Steel grade S235 7 7 m 8 8 m 9 9 m 10 10 m
Basic a0 w we q we q we q we q
Profile mm mm
n
mm kN/m
L. S. n
mm kN/m
L. S. n
mm kN/m
L. S. n
mm kN/m
L. S.
IPE 330 300 100 17 150 16,8 9 19 250 14,8 9 22 150 12,5 9 24 250 11,2 7
IPE 330 350 70 16 175 10,6 9 18 255 9,3 9 21 125 7,7 9 23 205 7,0 9
IPE 330 350 100 15 175 15,4 9 17 225 13,2 9 19 275 11,6 9 22 100 9,8 9
IPE 330 350 150 13 325 19,9 7 15 325 15,8 7 17 325 12,8 7 19 325 10,5 7
IPE 330 400 100 13 300 14,6 9 15 300 12,3 9 17 300 10,6 9 19 300 9,3 9
IPE 330 400 150 12 275 17,8 7 14 225 14,5 7 16 175 11,9 7 17 400 9,9 7
IPE 360 200 70 25 160 18,1 9 29 120 15,3 9 33 80 13,3 9 36 175 12,1 9
IPE 360 250 70 21 175 15,5 9 24 195 13,3 9 27 215 11,7 9 31 75 10,0 9
IPE 360 250 100 19 225 22,0 9 22 200 18,6 9 25 175 16,1 9 28 150 14,0 7
IPE 360 300 70 18 205 13,7 9 21 150 11,5 9 24 95 9,8 9 26 225 9,0 9
IPE 360 300 100 17 150 19,2 9 19 250 16,9 9 22 150 14,3 9 24 250 13,0 9
IPE 360 350 70 16 175 12,1 9 18 255 10,5 9 21 125 8,8 9 23 205 7,9 9
IPE 360 350 100 15 175 17,6 9 17 225 15,1 9 19 275 13,3 9 22 100 11,2 9
IPE 360 350 150 13 325 25,5 9 15 325 20,5 7 17 325 16,5 7 19 325 13,5 7
IPE 360 400 100 13 300 16,7 9 15 300 14,1 9 17 300 12,1 9 19 300 10,7 9
IPE 360 400 150 12 275 23,6 9 14 225 19,3 7 16 175 15,8 7 17 400 13,1 7
IPE 360 450 100 12 250 15,0 9 14 200 12,5 9 16 150 10,6 9 18 100 9,2 9
IPE 360 450 150 11 275 21,0 5 13 175 17,3 7 14 375 14,5 7 16 275 12,1 7
IPE 400 200 70 25 160 21,1 9 29 120 17,9 9 33 80 15,5 9 36 175 14,1 9
IPE 400 250 70 21 175 18,0 9 24 195 15,5 9 27 215 13,5 9 31 75 11,6 9
IPE 400 250 100 19 225 25,7 9 22 200 21,8 9 25 175 18,9 9 28 150 16,6 9
IPE 400 300 70 18 205 15,9 9 21 150 13,3 9 24 95 11,4 9 26 225 10,4 9
IPE 400 300 100 17 150 22,5 9 19 250 19,8 9 22 150 16,7 9 24 250 15,1 9

The Figure 3-2 is an example of graph for the predesign of steel cellular beams. The Table 3-2 gives
the main section properties of the cellular beams contained in the predesign table.

47


Sections "IPE" - Steel grade S235
0
5
10
15
20
25
30
35
40
45
50
6 8 10 12 14 16 18 20 22 24
Span length (m)
L
o
a
d

c
a
p
a
c
i
t
y

(
k
N
/
m
)
I PE 200 (200 - 70)
I PE 220 (250 - 70)
I PE 240 (250 - 70)
I PE 270 (300 - 70)
I PE 300 (350 - 70)
I PE 330 (400 - 100)
I PE 360 (450 - 100)
I PE 400 (500 - 100)
I PE 450 (550 - 150)
I PE 500 (600 - 150)
I PE 550 (700 - 150)
I PE 600 (750 - 150)

Figure 3-2: Predesign of Steel Cellular Beams (IPE S235)

Table 3-2: Section properties
IPE Ratios Gross Section Net Section Tee Section
Basic a0 w Ht w/a0 Ht/a0 A Iy Anet Iy.net Mass Am zG.m Iy.m
Profile mm mm mm cm
2
cm
4
cm
2
cm
4
kg/m cm
2
mm cm
4

IPE 330 300 100 462,90 0,333 1,54 72,57 25368 50,07 23681 46,57 25,04 66,63 91,30
IPE 330 350 70 493,29 0,200 1,41 74,85 29346 48,60 26667 45,27 24,30 58,69 62,37
IPE 330 350 100 489,33 0,286 1,40 74,56 28810 48,31 26130 45,94 24,15 57,07 57,41
IPE 330 350 150 479,21 0,429 1,37 73,80 27461 47,55 24781 46,60 23,77 52,87 45,92
IPE 330 400 100 515,37 0,250 1,29 76,51 32454 46,51 28454 45,26 23,25 47,06 32,95
IPE 330 400 150 506,74 0,375 1,27 75,86 31218 45,86 27218 46,10 22,93 43,38 26,33
IPE 360 200 70 445,08 0,350 2,23 79,54 26294 63,54 25761 55,13 31,77 98,79 326,59
IPE 360 250 70 471,64 0,280 1,89 81,66 30003 61,66 28961 54,47 30,83 89,90 243,05
IPE 360 250 100 465,77 0,400 1,86 81,19 29159 61,19 28118 54,93 30,60 87,64 224,56
IPE 360 300 70 497,61 0,233 1,66 83,74 33908 59,74 32108 53,74 29,87 80,58 173,13
IPE 360 300 100 492,90 0,333 1,64 83,36 33179 59,36 31379 54,34 29,68 78,73 161,17
IPE 360 350 70 523,29 0,200 1,50 85,79 38046 57,79 35187 52,96 28,90 70,93 117,18
IPE 360 350 100 519,33 0,286 1,48 85,48 37391 57,48 34532 53,67 28,74 69,34 109,40
IPE 360 350 150 509,21 0,429 1,45 84,67 35742 56,67 32884 54,38 28,33 65,23 91,06
IPE 360 400 100 545,37 0,250 1,36 87,56 41831 55,56 37564 52,95 27,78 59,54 69,50
IPE 360 400 150 536,74 0,375 1,34 86,87 40327 54,87 36060 53,84 27,43 55,95 57,98
IPE 360 450 100 571,16 0,222 1,27 89,62 46521 53,62 40446 52,19 26,81 49,37 40,67
IPE 360 450 150 563,62 0,333 1,25 89,02 45120 53,02 39045 53,23 26,51 46,14 33,80
IPE 400 200 70 485,08 0,350 2,43 91,78 35823 74,58 35249 64,19 37,29 114,00 547,09
IPE 400 250 70 511,64 0,280 2,05 94,06 40453 72,56 39333 63,48 36,28 105,28 425,83
IPE 400 250 100 505,77 0,400 2,02 93,56 39403 72,06 38283 63,98 36,03 103,08 398,46
IPE 400 300 70 537,61 0,233 1,79 96,30 45303 70,50 43368 62,70 35,25 96,18 320,90
IPE 400 300 100 532,90 0,333 1,78 95,89 44399 70,09 42464 63,35 35,05 94,38 302,51

48



3.2.3 Composite cellular beams
3.2.3.1 Basic assumptions
The number of basic parameters for composite cellular beams is far greater than for pure steel
cellular beams. So it has not been possible to deal with the whole field of possibilities including
various classes of concrete, various types of slab (plain slabs, slabs with profiled steel sheeting),
various steel grades Moreover it is more efficient for composite cellular beams, to use a mono-
symmetric steel section (lower chord and upper chord are cut from different I-rolled profiles). This
leads to a greater number of solutions.
So it has been decided with the partners to limit the number of loading tables to some very common
solutions. The main assumptions for these solutions are presented hereafter.


Profiled steel sheeting
B
a
0

Upper chord
Lower chord
Ht
ES

Figure 3-3: Composite cross-section
The basic assumptions are the following :
- Slab with profiled steel sheeting perpendicular to the beam
- Total depth of the slab 12 cm
- Participating width of the slab : 2 x L/8
- Concrete class : C25/30
- The beam is assumed to be fully propped during concreting
(no check at this stage)
- Profiled steel sheeting : PAB Cofradal 60 (open profile)
- Connector : Welded stud 100x19 mm
The calculations are similar to those performed for steel cellular beams, except that the composite
effect significantly improves the loading capacity of the beam. The self weight of the slab is
automatically included in the permanent loads, factored by 1.35 for the ULS verifications.
49



3.2.3.2 Examples of predesign tables for composite cellular beams
Tables are available for 2 cases of association of basic profiles :
- IPE serial for upper chord and HEA for lower chord,
- IPE serial for upper chord and HEB for lower chord,
for 3 steel grades : S235, S355 and S460,
and for 2 ratios : w/a
0
= 0.2 and w/a
0
= 0.4.

Table 3-3: Predesign tables for composite cellular beams
Steel Grade S235 6 m 7 m 8 m
Upper Lower a0 w Ht we q we q we q
Chord Chord mm mm mm
n
mm kN/m
L. S. n
mm kN/m
L. S. n
mm kN/m
L. S.
IPE 300 HEA 260 240 50 384 20 125 5,9 9 24 45 3,3 9 27 110 2,6 9
IPE 330 HEA 280 270 54 424 18 111 6,8 9 21 125 4,9 9 24 139 3,4 9
IPE 360 HEA 300 300 60 464 16 150 8,8 9 19 110 6,3 9 22 70 4,4 9
IPE 400 HEA 340 350 70 528 14 95 11,6 9 16 175 9,2 9 19 45 6,3 9
IPE 450 HEA 400 400 80 608 12 160 16,9 9 14 180 13,0 9 16 200 10,3 9
IPE 500 HEA 450 450 90 682 11 75 20,8 9 12 305 18,3 9 14 265 14,3 9
IPE 550 HEA 550 500 100 782 10 50 26,2 9 11 250 22,7 9 13 150 17,3 9
IPE 600 HEA 700 600 120 931 8 180 39,0 9 9 320 32,4 9 11 100 23,7 9
IPE 300 HEB 240 240 50 379 20 125 5,7 9 24 45 3,1 9 27 110 2,4 9
IPE 330 HEB 260 270 54 419 18 111 6,6 9 21 125 4,6 9 24 139 3,2 9
IPE 360 HEB 300 300 60 469 16 150 8,8 9 19 110 6,3 9 22 70 4,4 9
IPE 400 HEB 320 350 70 523 14 95 11,4 9 16 175 9,1 9 19 45 6,1 9
IPE 450 HEB 360 400 80 593 12 160 16,2 9 14 180 12,5 9 16 200 9,8 9
IPE 500 HEB 450 450 90 687 11 75 21,0 9 12 305 18,5 9 14 265 14,5 9
IPE 550 HEB 600 500 100 812 10 50 27,8 9 11 250 24,2 9 13 150 18,5 9
IPE 600 HEB 800 600 120 986 8 180 42,7 9 9 320 35,6 9 11 100 26,1 9
IPE 300 HEA 260 240 96 376 17 192 16,2 9 20 188 12,5 9 23 184 9,9 9
IPE 330 HEA 280 270 108 415 15 219 20,2 9 18 152 15,1 9 21 85 11,6 9
IPE 360 HEA 300 300 120 454 14 120 22,4 9 16 200 18,7 9 19 70 13,9 9
IPE 400 HEA 340 350 140 517 12 130 28,4 9 14 140 22,7 9 16 150 18,5 9
IPE 450 HEA 400 400 160 595 10 280 40,8 9 12 220 31,2 9 14 160 24,9 9
IPE 500 HEA 450 450 180 667 9 255 50,5 9 11 125 37,1 9 12 310 33,1 9
IPE 550 HEA 550 500 200 765 8 300 63,2 9 10 100 45,5 9 11 250 39,9 9
IPE 600 HEA 700 600 240 911 7 180 78,8 9 8 260 64,5 9 9 340 54,4 9
IPE 300 HEB 240 240 96 371 17 192 15,9 9 20 188 12,2 9 23 184 9,6 9
IPE 330 HEB 260 270 108 410 15 219 19,9 9 18 152 14,8 9 21 85 11,3 9
IPE 360 HEB 300 300 120 459 14 120 22,7 9 16 200 18,9 9 19 70 14,1 9
IPE 400 HEB 320 350 140 512 12 130 28,0 9 14 140 22,3 9 16 150 18,3 9
IPE 450 HEB 360 400 160 580 10 280 39,8 9 12 220 30,4 9 14 160 24,1 9
IPE 500 HEB 450 450 180 672 9 255 51,2 9 11 125 37,7 9 12 310 33,6 9
IPE 550 HEB 600 500 200 795 8 300 66,9 9 10 100 48,0 9 11 250 42,3 9
IPE 600 HEB 800 600 240 966 7 180 85,9 9 8 260 70,4 9 9 340 59,5 9

50




Srie "IPE-HEA" - Steel Grade S235
w/a0 = 0,2
0
5
10
15
20
25
30
35
40
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Span Length (m)
L
o
a
d

c
a
p
a
c
i
t
y

q

(
k
N
/
m
)
IPE 300-HEA 260(240, 50)
IPE 330-HEA 280(270, 54)
IPE 360-HEA 300(300, 60)
IPE 400-HEA 340(350, 70)
IPE 450-HEA 400(400, 80)
IPE 500-HEA 450(450, 90)
IPE 550-HEA 550(500, 100)
IPE 600-HEA 700(600, 120)

Figure 3-4: Predesign of composite cellular beams (IPE-HEA S235)

3.3 Fire behaviour of cellular beams
A test on cellular beams exposed to fire was performed in the CTICM Test Station (at Maizires-
les-Metz) on Friday 4
th
October 2002.
The testing elements were composed of two beams:

P1 : a loaded protected cellular beams of 7.20m span, same as the Test 3 beam tested at cold
temperature
The beam is composite. It is composed of a steel profile, based on HEB 340 for the lower flange
and IPE 300 for the upper flange, and a 130 mm thick concrete slab.
The beam length is 7200 mm and it is composed of 12 openings of 570 mm length. The 1
st
and the
12
th
cell are full. The distance between supports is 6840 mm. There are 3 stiffeners. The slab
includes a HI-BOND 55 steel deck. The slab is connected to the steel profile with one file of 19
connectors, spaced out each 150 mm. The yield stress of steel was measured on different points and
is resumed in the Table 3-4.
51



Table 3-4: Yield stress of steel (MPa)
Web
Lower flange
Lower part Upper part
Upper flange
484 390 456 400

The yield stress in compression of concrete is 40 MPa. The load is distributed on 3 times 100 kN, at
the level of the 3 central stiffeners (see Figure 3-5).


1 2 3 4 5 6 7 8 9 10 11 12
east side west side 100 000 N
100 000 N 100 000 N

Figure 3-5: Loading

P2 : a cellular beam based on IPE 400, with some openings are filled, same as Test 1 beam
tested at cold temperature, protected with some unprotected edges of opening
The beam P2 was unloaded with three different part for which the details are given in the following
figure.
Protected
edge of openings
Unprotected
edge of openings
Protected
edge of openings
Unprotected
edge of openings


1 9 0 1 9 0 1 9 0 5 7 0 5 7 0 5 7 0 5 7 0 5 7 0 5 7 0 5 7 0 5 7 0 5 7 0 5 7 0 2 1 5 2 1 5

250


250

7200
3
8
0
7 9
5 1
5 5 5 . 2
I P E

4 0 0
3 8 0
1 8 0 0

Figure 3-6: View of the beam P2
More details about the numerical and testing conditions and results are given in the plenary report.
52



3.3.1 Experimental results on the unloaded beam
For the unloaded beam P2, the protection thickness was applied with the following conditions :

Table 3-5: Protection thickness
Position Measured thickness (mm)
Min Average Max Variation
total 10 21 28 3
Opening edge 10 18 25 6

The temperature was recorded on different sections that correspond to the following conditions :
o Sections S1, S2 and S3 : sections with openings but without protection on the opening edges
o Sections S4 and S5 : sections without opening
o Sections S6, S7 and S8 : sections with openings and with protection on the opening edges
In that respect, the effect of the opening and the effect of the protection on the opening edge can be
studied by comparison of the measurements on the three types of sections.
The following figures show the comparison of the average temperatures on the three types of
sections at the centre of the web, near the opening in the web, at the lower flange and at the upper
flange. For more details, one can refer to the test report.
After about 90 minutes, a part of the protection localised on the no opening part (mid span of the
beam) on the lower face of the lower flange (a length of about 50 cm) has fallen down, leading to
higher temperature of this part (Sections S4 and S5).
0
100
200
300
400
500
600
700
800
900
1000
0 20 40 60 80 100 120
Time (min)
T
e
m
p
e
r
a
t
u
r
e

(

C
)
web center at sections S1-S3 - without
protection on the opeining side
web center at the sections S4-S5 - without
opening
web center at the sections S6-S8 with
protection on the opening side

Figure 3-7: Average temperature in the centre of the web at mid height
53



0
100
200
300
400
500
600
700
0 20 40 60 80 100 120
Time (min)
T
e
m
p
e
r
a
t
u
r
e

(

C
)
average value on the upper flange without
protection on the opening side (sections : S1-S2)
average value on the upper flange without
opening (sections :S4 )
average value on the upper flange without
protection on the opening side (sections: S6 S7)

Figure 3-8: Average temperature in the upper flange

0
100
200
300
400
500
600
700
800
900
1000
0 20 40 60 80 100 120
Time (min)
T
e
m
p
e
r
a
t
u
r
e

(

C
)
average value on the lower flange without protection
on the opening side (sections : S1-S2)
average value on the lower flange without opening
(sections S4 S5)
average value of the lower flange with protection on
the opening side (sections :S6 S7)

Figure 3-9: Average temperature in the lower flange

These results show that :
- the lower flange temperature is not modified by the opening (with or without opening, or
with or without protection on the opening edge) Figure 3-9
- for the upper flange temperature, the non protection of the opening edge leads, after 20
minutes of fire exposition, to higher temperature, that can be more than 100C compared to
the case with protection on the opening edge or more than 80C compared to the case
without opening. The opening with protection on the opening edge, leads to lower
temperature than the case without opening. This can be explained by the heat flux from the
web to the flange that is reduced due to the reduction of the width of the web Figure 3-8.
54


- For the web, the average or localised temperatures lead to the same analysis : the opening
increases a little bit the temperature on the web when all is protected, of about 30C, but this
temperature increase is higher without protection on the opening edge of more than 100C
Figure 3-7.
In conclusion, it is clear that if there is no protection on the opening edges of cellular beams,
specific assessment of the temperature field in the cellular beams has to be made in order to study
the fire stability of the cellular beam. When protection is applied on the opening edge, compared to
the temperature without opening, higher temperature (about 30C) is recorded. This needs to be
taken into account in the methodology to be developed for calculating the fire stability duration of
cellular beams. Generally, the simple approach of massivity factor differentiating lower flange,
upper flange and web part leads to safe approach of temperature distribution in cellular beams,
when protection is applied on the opening edges.

3.3.2 Experimental results on the loaded beam
The loaded beam has been protected by an average thickness, of Newspray insulation Product,
given by the manufacturer Projiso that is thanks for its contribution to the present research, of
24 mm. Displacements, rotation and many temperatures were measured in some locations. The fire
stability was 108 minutes.

3.3.2.1 Displacements
The defection was measured at , and of the beam length. The recorded measurements are
shown in the following figure.
0.00
50.00
100.00
150.00
200.00
250.00
300.00
0.00 20.00 40.00 60.00 80.00 100.00 120.00
Time (min)
D
e
f
l
e
c
t
i
o
n

(
m
m
)
D3 (Est) (mm)
D2 (Lp/2) (mm)
D1 (Ouest) (mm)

Figure 3-10: Deflection of the beam
55



3.3.2.2 Temperature measurements
About 172 Thermocouples recorded the temperature on the steel sections as a function of time.
In order to show that the temperature distribution along the beam length was not constant, the
following figure shows the temperature at three same point on the web position but at , and
of the beam length.
The temperature reached at the failure time is given in the following table.

Table 3-6: Temperature at the failure time
Temperature (C)
Min Average Max
Web 286 545 663
Lower Flange 326 517 590
Upper Flange 263 317 374

The global analysis of the test results for the two beams by numerical simulations describes the fire
behaviour of the beam. Finite element calculations were made for comparison and understanding of
the thermo-mechanical behaviour of these beams.

0
100
200
300
400
500
600
700
800
900
1000
0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00
Time (min)
T
e
m
p
e
r
a
t
u
r
e

(

C
)
TC 5 (C)
TC 30 (C)
TC 47 (C)

Figure 3-11: Web temperature along the beam versus time
56



3.3.2.3 Observations
During the tests, no fall or degradation of the insulation was observed up to failure time.
The following photo shows the state of the beam after the tests.


Figure 3-12: View of the failed part of beam in the furnace after the test
The results of the test of the loaded beam have shown that the failure had occurred firstly by web
instability at the first exposed web element (between opening 2 and opening 3), but we can observe
the effect of failure by the instability of the second web post (between opening 3 and opening 4).
The numbering of openings is given in Figure 3-5.

3.3.3 Numerical mechanical behaviour of the loaded beam
3.3.3.1 At ambient temperature
The test has been performed taken into account as a reference the Test 3 of the test campaign
realised at ambient temperature. Nevertheless, in order to adapt the beam to the furnace of CTICM,
the beam tested at high temperature has been modified reducing a few the length, but closing the
two extreme opening on the web, decreasing the number of openings from 12 openings to 10
openings. The test 3 has leaded to failure load of 656 kN.
The maximum load level reached by the FEA has been 628.2 kN with a corresponding deflection of
43.73 mm. The load-deflection curve is displayed in Figure 3-13.
57



Load-Deflection Diagram
0
100
200
300
400
500
600
700
0 10 20 30 40 50 60
Deflection z [mm]
L
o
a
d

[
k
N
]

Figure 3-13: Load-Deflection Curve of the FEA
The indicated failure mechanism has been web post buckling occurring between the 9
th
and 10
th

opening.

3.3.3.2 At high temperature
The tested beam is not totally symmetric, at the level of the slab and the connectors, as shown on
the Figure 3-5. However only half the beam was modelled, in order to minimize the calculation
time. The modelled side is the east side, which is the weakest.

3.3.3.2.1 Heating up
The heating up of the beam used in the calculation comes from the measured temperature. For each
point, an average was done on the 3 sections. This average was used in the calculation. Also the
temperature field used in the calculation is uniform on the length. During test, the part of beam
which is at the extremity, after the support, is outside the oven. This part of beam remains cold.
The Figure 3-14 shows the heating up of the beam used in the calculation.
58




Figure 3-14: Heating up of the beam

3.3.3.2.2 Results
The failure mode of the beam, observed during test, is web post buckling, between openings 2 and
3, and then between openings 3 and 4. The calculation allows to find this failure mode and gives the
horizontal displacement of web. As the simulation used a symmetrical beam, the numerical failure
occurs in a same way between openings 2 and 3 or openings 10 and 11. During test, the vertical
displacement was measured at quarter of span, at mid-span and at three quarters of span. The
rotation on the supports was measured too. The Figure 3-15 shows the comparison between test and
calculation for these results.
The Figure 3-15 shows a very good correlation between calculation and test. Indeed the
displacement value before collapse, as the instant value of collapse, are very well found by
calculation.
The simulation, as it has been observed during the test, leads to the conclusion of a weak impact of
the reduction of connection between the beam and the slab at the extremity of the beam. The term
weak is used in order to show that the small difference between the failure time (2 minutes) may
be due to that, but this should be shown only by simulation of the global beam. It appears also that
temperature during the test was not exactly symmetrical, and adaptation for simulation may lead to
small differences between test and numerical results.
59



0.00
50.00
100.00
150.00
200.00
250.00
300.00
0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00
time (min)
test
calculation

Figure 3-15: Numerical and experimental vertical displacement at mid-span

3.3.4 Analysis
The numerical model leads to a very good correlation with the main experimental results :
- the failure mode (web post buckling)
- the time of failure
- the vertical displacement at mid-span.

3.3.4.1 Thermal analysis
The thickness of the protection applied on the beam varied from 11 to 45 mm, with an average
value of 24 mm.
This leads to very difficult comparison between calculations and measurements for the thermal
transfer. So it has decided to not compare calculated temperature with measurements.
A simple calculations considering a 3 different part of the beam with massivity factors protected by
the average value of Newspray of 24 mm, leads to a high temperature compared to measured one :

60


Table 3-7: Comparison of measured and calculated temperatures
Thickness
(mm)
Massivity
factor
Mean measured
temperature (C)
Calculated
temperature (C)
Web HEB 240 10 200 576 694
Flange HEB 240 17.5 114.285714 517 570
Web IPE 300 7.1 281.690141 521 750
Flange IPE 300 10.7 93.4579439 317 514

It can be observed the effect of the slab by a lower temperature on the IPE 300 upper flange, and on
the web IPE 300 that are lower than the calculated one. This leads to a large safety factor when
using simple calculation model. A large effect is also shown on the lower part of the web (web of
HEB 240), but a weak effect on the lower flange of the HEB 240 part.
The temperature distribution on the unloaded beam should lead to better understanding of the
thermal behaviour of such a beam.

3.3.4.2 Mechanical behaviour
The numerical thermo-mechanical analysis of the tested loaded beams leads to a very good
correlation with the test. The failure occurs numerically and experimentally at about 108 min, by
web buckling.
At this time, the average temperature of the web was 545C, with maximum localised temperature
of 663C. The calculations have shown, without horizontal gradients in the beam, that the
maximum temperature in the web (near the centre) was 600C at the failure time. This confirms the
global failure temperature of the web.
Due to similar failure mode at high temperature and at ambient temperature, the mechanical load
factor is determined from mechanical behaviour at ambient temperature. The failure load is
628.2 kN. So a load factor of 0.477 has been applied at high temperature. The maximum web
temperature at the failure time, so 600C, leads to a reduction of effective yield strength to 47%
according to the Eurocode 3 part 1.2.
So it can be thought that a simple approach considering the reduction of the effective yield strength
reduction from the load factor could be used for the determining the critical temperature of the
beam.
Nevertheless, a single test may not generalised this simple approach to design of protection
thickness to all cellular beams.
61



3.3.5 Conclusion
First, the experimental study of the fire behaviour of the cellular beams has allowed to give
referenced value for the scientific world in the structural fire resistance world, that did not exist
previously.
The analysis has been performed in the two domains :
- thermal transfer
- mechanical behaviour at high temperature

Thermal transfer :
Concerning the thermal transfer, the temperature measurements on the unloaded beam leads to the
following conclusions :
- if there is no protection on the opening edges of cellular beams, specific assessment of the
temperature field in the cellular beams has to be made in order to study the fire stability of
the cellular, because of higher temperature can be encountered.
- When protection is applied on the opening edge, compared to the temperature without
opening, higher temperature (about 30C) is recorded. This needs to be taken into account in
the methodology to be developed for calculating the fire stability duration of cellular beams.
Generally, the simple approach of massivity factor differentiating lower flange, upper flange
and web part leads to safe approach of temperature distribution in cellular beams, when
protection is applied on the opening edges.
This last conclusion about the simplified model considering the massivity factor if the flanges and
the web has been confirmed by the temperature measurement in the loaded beam.

Mechanical behaviour :
First the numerical thermo-mechanical analysis of the tested loaded beams leads to a very good
correlation with the test, confirming the ability of such FEM codes with Eurocode 4 part 1.2
materials properties at high temperature to determine the fire behaviour of cellular beams.
The mode of failure of the cellular beam was the buckling of the web, similarly to the failure mode
at ambient temperature.
The load factor from the failure load at ambient temperature is about 0.48 that is equal to the
reduction of the effective yield strength at the maximum web temperature calculated (600C). So it
can be thought that a simple approach considering the reduction of the effective yield strength
reduction from the load factor could be used for the determining the critical temperature of the
beam.
Nevertheless, a single test may not be generalised this simple approach to design of protection
thickness to all cellular beams. Developments by calculations and verification on additional tests are
necessary to develop a simplified thermal and mechanical model for design of protection thickness
of cellular beams.
62



General :
The development of a more specific simplified method for the design of protection thickness of
cellular beams needs more work. This future work should be based in a large part on numerical
tools for which the present calculations have again shown their ability to model thermo-mechanical
behaviour of steel beams, even if the particular situations (cellular beams) or with specific failure
mode (buckling of the web). But it will need additional test to validate it.

3.4 References
[1] Eurocode 3 : Design of steel structures - Part 1.1 : General rules and rules for buildings.
ENV 1993-1-1. European Committee for Standardisation.
[2] Eurocode 3 : Design of steel structures - Annexe N : Openings in webs. ENV 1993-1-1 :
1992/A2. European Committee for Standardisation.
[3] Eurocode 4 : Design of composite steel and concrete structures Part 1.1 : General rules
and rules for buildings. ENV 1994-1-1. European Committee for Standardisation.
[4] J. K. Ward. Design of composite and non-composite cellular beams. SCI Publication n100.
[5] Catalogue ProfilARBED : Programme de vente, poutrelles, profils U et cornires. Edition
octobre 1995.
[6] ARCELOR Cellular Beams Version 1.33. User's manual. February 2003.

63





4 Contribution from PROFILARBED-RESEARCH
4.1 Summary
The tasks of PROFILARBED-Research were mainly focused on the WP2 "Improved manufacture
of beams with web openings". Nevertheless cellular beams as well as built up sections have been
realised in the scope of the different working packages. The 3 different steps can be pointed out:
ACB ARCELOR Cellular Beams fabrication ;
Fabrication of the test specimen for the different universities ;
Improvement of the welding procedures for cellular beams.

4.2 ACB ARCELOR Cellular Beams fabrication
ARCELOR Cellular Beams are rolled reconstituted beams with circular periodic web openings.
These beams are produced by oxycutting the web in two passes (one pass consists in a repetitive
series of a continuous horizontal line and a half circle line (see Figure 4-1). After having separated
them the two half beams are shifted until the 2 straight parts are lined up (see Figure 4-2). The webs
between two openings are then welded together.



Figure 4-1: ACB oxycutting Figure 4-2: ACB reassembling

4.3 Fabrication of the test specimen
During the second project meeting that was held in Paris in February 2002, the final configuration
of the beams to test at the University of Lule in the scope of WP5, at the University of
Kaiserslautern for RWTH in the scope of WP1 and at CTICM in the scope of WP3 has been
discussed. The final drafts have been sent after this meeting by the different partners to
PROFILARBED-Research by March 2002. The beams have been ordered and characterisation tests
have been performed in order to know the real steel properties. The beams have been analysed at
following specific points (see Figure 4-3).

65


H
A2
B
1/2
1/6

Figure 4-3: Outcut for characterisation
After confirmation of the characterisation results, the beams have been assembled and sent with
their certifications to the different partners.
The next figure shows a fabricated specimen for the University of Lule.


Figure 4-4: Test specimen for the University of Lule

4.4 Improvement of the welding procedures for cellular beams
4.4.1 Actual assembling procedure for cellular beams
Following the shifting step in chapter 4.2, two configurations are then possible; if the weld throat
thickness to apply is less than two times six millimetres a 26mm, the chamfering of the two parts
to weld isnt necessary. This value represents the upper weld throat thickness limit that can be
achieved without chamfering. In case that the weld to apply is more important than two times six
millimetres, a > 26mm, a chamfering of the two parts to weld isnt avoidable.
66


In order to simplify the welding procedure, every web post is foreseen with a gap of two millimetres
(see Figure 4-5). This allows to start welding without any kind of grinding.

Web post width
2 mm
2 mm
< 6 mm
< 6 mm
Web post thickness


Figure 4-5: Welding of the web post

4.4.2 Small scale tests
4.4.2.1 General
In order to improve the welding procedures for cellular beams, it has been decided to compare the
actual welding configurations with new possibilities. The aim is as required into the WP2 to achieve
an improved welding procedure for the rewelding of oxycutted beams.
At the beginning only the welds were to be analysed. As real scale prototypes are very expensive,
small scale tests have been performed. These tests are resumed in welding two plates together.
These plates are supposed to represent the web post.
Different welding procedures, configurations and parameters have been analysed. The goal was to
determine the deformation in case of a one side welding after taking away of the restraints.
67


Welded
restraint
L
B
x
t

Figure 4-6: Plates with supposed deformation after taking away of the restraint

4.4.2.2 Test configuration and observations
The whole set of the tests had total length of 600mm, a width varying from 60mm to 150mm and a
thickness from 8mm to 20mm. These parameters represent the general geometrical web values of
cellular beams.
13 specimens have been tested, the deformation x (see Figure 4-6) has been measured and a
comparison in the welding configuration has been made. A small space between the two plates is
foreseen for the specimens, in order to consider the effective oxycutting method for cellular beams
as shown in chapter 4.2.
The following shows only a few tested configurations. The results obtained for the whole
configuration can be found into the large report under chapter "Small scale tests".
68



Specimen 2


B x L x t
Dimensions: 150x600x12

Welding on one side with chamfering
Welding thickness a = 8mm (3 passes)
x = 0

Specimen 3

B x L x t
Dimensions: 150x600x12

Welding on two sides without chamfering
Welding thickness = 4+4mm (1 pass each)

x = 0

The specimens 2 and 3 which have exactly the same configuration achieving a total weld thickness
of 8mm. Two execution configurations are possible:
a one side welding of 8mm with chamfering avoiding to turn the beam upside down
two weldings of 4mm (1 pass for each side) avoiding the chamfering which is an annoying step,
but with a supplementary manipulation step. The comparison for the welding passes number is
increased from 2 to 3 in case of a one side welding. The measured deformation x is in both
cases equal to 0.
Specimen 9

B x L x t
Dimensions: 150x600x20

Welding on one side with chamfering
Welding thickness = 10mm (5 passes)

x = 2.0mm

a
a a
a
69



Specimen 10

B x L x t
Dimensions: 150x600x20

Welding on two sides without chamfering
Welding thickness = 5+5mm (1 pass each)

x = 0mm

The specimens 9 and 10 have again exactly the same configuration as for specimens 2 and 3. Here
again the comparison for the welding passes number is increased from 2 to 5 in case of a one side
welding. In this case, a small deformation has been highlighted; a 2.0mm deformation over a length
of 300mm corresponding to a slope of 0.67%.

Specimen 11

B x L x t
Dimensions: 150x600x20

Welding on one side without chamfering
Welding thickness = 5mm (1 pass)

x = 3mm

For the specimen number 11, a deformation of 3mm has been measured over a length of 300mm
corresponding to a slope of 1.0%, which is the maximum highlighted deformation (see Figure 4-7).

Specimen 12

B x L x t
Dimensions: 150x600x20

Welding on one side with chamfering
Welding thickness = 16mm (6 passes)
x = 1.5mm


a a
a
a
70



Specimen 13

B x L x t
Dimensions: 150x600x20

Welding on two sides with chamfering
Welding thickness = 8+8mm (5 passes each)
x = 0


The specimens 12 and 13 have again exactly the same configuration. In this case the comparison of
the welding passes number is increased from 6 to 10 in case of a two side welding. The reason is
that in both cases, a chamfering has been necessary as the welding thickness is more important than
6mm. A deformation x of 1.5mm for specimen 12 corresponding to a slope of 0.5% has been
highlighted.

Figure 4-7: Deformation (3mm) after taking away of the restraint for the
specimen 11
4.4.2.3 Conclusions
The results show that a deformation occurred for the one side welding specimens only in some
cases. The most important deformation has been highlighted for test specimen 11 with 3mm
deformation giving a slope of 1.0%. This can be accepted and the conclusion is that a one side
welding applied for cellular beams is possible without important deformations.
The esthetical aspect isn't considered until now. Due to the small space that will remain in case of a
one side welding on the not welded side (see specimen 11), the fact is that an oxidation could be
possible in this zone. The second aspect is: would a person feel insecure in case of standing under a
cellular beam that hasn't been welded on one side?
The second goal of this analyse was to have a direct comparison between
a one side welding with chamfering,
and a standard 2 sides welding without chamfering but with a supplementary manipulation of
the beam.
a a
71


The first observation is that the chamfering step is to be avoided as it is a very annoying and tiring
step for the welder. In second, the welding passes generally increase for a one side welding as the
chamfer shape becomes wider with an increasing in depth. This means that more weld material has
to be deposed for a one side welding, for example a throat depth of 10mm has a bigger surface
compared to a two sides welding with a throat depth of 2 times 5mm.
By the other hand a manipulation step, this means turning the beam upside down one more time, is
necessary in case of a two sides welding.
The conclusion is that a two sides welding will be more efficient in case that the welding
configuration allows to avoid the chamfering step. This is possible until the welding thickness
doesn't exceed 12mm as a two times 6mm weld is possible without chamfer (see specimens 5 and
6). In case that a two sides chamfering is necessary (a>6mm for both sides) as it was done for the
specimens 12 and 13, a one side welding is usable.
The next step will consist on performing real scale tests with welding thicknesses lower than 2
times 6mm subjected to horizontal shear failure mechanism. The goal will be to prove that there is
no failure into the weld itself.

4.4.3 Large scale tests
4.4.3.1 Summary
Following the previous realised small scale welding tests during the year 2002 and according to the
explanation of the technical report N3, it has been decided to perform real scale tests on cellular
beams and test different welding configurations in order to optimise the actual procedure. As
PROFILARBED itself isnt able to perform such tests, a subcontracting has been requested to the
COMMISSION. Here the University of Kaiserslautern has been chosen, as PROF. FELDMANN
and his staff have already performed cellular beam tests in the scope of this project for WP1.
The report presents the results of short steel beams with large web openings gained from
experimental and numerical investigations. The results were evaluated in terms of load deformation
behaviour, failure modes and strain development. The comparison to the available prediction model
shows the large safety of the model.

4.4.3.2 Objectives
As explained in the conclusions of the small scale tests (see chapter 4.4.2.3), a two sides welding
without weld preparation is more efficient if the welding configuration allows to avoid the
chamfering step. This is possible until the welding thickness doesn't exceed 12mm resulting from
the consideration that a 2 x 6mm weld is feasible without chamfer. The objective is to
demonstrate that a full penetration should only then be necessary if the calculation method requires
it for the throat thickness. A more realistic and economic design approach is to provide a partial
penetrated weld being determined to be fully exploited.
In observing the load deformation behaviour, the stress capacity and the reserves of the welds while
loading the beam up to failure the task is to validate the method.

4.4.3.3 Background
Although normal practice in the industry has been to require full penetration welds at the web posts,
it has been recognised that this full penetration is often not necessary. A consistent standard
minimum thickness required by the loading and design is often satisfactory. The requirement for
full penetration butt welds can increase significantly the costs of production, thus it should be
avoided.
72


Following the previous performed calculations of the project it has been confirmed that for many
geometries, the standard design procedures indicate that other modes of failure intervene before
fully penetrated butt welds of the web post fail by horizontal shear. The weld requirements for butt
welds are defined by the magnitude of the horizontal shear forces in the web posts.
The minimum throat thickness is calculated considering two continuous welds on a web post with a
width w. The minimum throat thickness is calculated as follows:

Mw u
w h.Sd
f w 2
3 V
a

= (41)
with
V
h,Sd
= the acting horizontal solicitation into the web post

w
and f
u
defined as follows:

Table 4-1: Weld throat thickness parameters
Steel grade S235 S275 S355 S420 S460
f
u
(N/mm
2
) 340 370 470 520 550

w
0,80 0,85 0,90 1,00 1,00

Mw
= 1.25 following EC3 Part1.1

2 mm
a a

Figure 4-8: Weld throat thickness definition
It is important to recognise however that failure may occur by other mechanisms and limit states
before the limiting condition for horizontal shear in the web post is reached.

4.4.3.4 Testing
4.4.3.4.1 Test specimen
Following the different communications that took place between the fabrication shop, the
University of Aachen, PROFILARBED-RECHERCHES and the University of Kaiserslautern, a
configuration for 10 tests has been defined; the idea is to cover the whole web thickness' range of
all available PROFILARBED profiles for cellular beams. Therefore the following solutions have
been retained (nominal values). The nominal steel grade is S235 for the whole sections.
The detailed view of the test specimen configuration is given in Annex A of the contribution to
Part 2 of this report.

73


Table 4-2: List of specimens for testing
Web
thick.
Beam
depth
Span
length
Web post
width
Diameter
a
0
w/a
0

ratio
S
p
e
c
i
m
e
n

n

Upper
profile
Lower
profile
[mm] [mm] [mm] [mm] [mm] [-]
1-ss

IPE 400 IPE 400 8.6 558,0 1910 122 360 0.34
2-hs HE 300 A HE 300 M 8.5/21 477,0 1800 95 350 0.27
3-ss HE 400 B HE 400 B 13.5 599,0 1990 63 425 0.15
4-ss IPE 400 IPE 400 8.6 600,0 1990 55 430 0.13
5-hs IPE 300 HE 300 M 7.1/21 482,0 1690 55 355 0.15
6-ss HE 300 M HE 300 M 21 495,0 1590 50 335 0.15
7-hs HE 300 B HE 300 M 11/21 482,0 1870 123 350 0.35
8-hs HE 300 A HE 300 B 8.5/11 415,0 1800 147 300 0.49
9-ss HE 400 B HE 400 B 13.5 524,0 1846 150 310 0.48
10-ss HE 300 M HE 300 M 21 462,0 1800 147 300 0.49
ss corresponds to symmetric section; hs corresponds to hybrid section

To ensure proper load introduction for supports and centre-load, introduction stiffeners are applied
across the total depth of the beam.
As mentioned two failure modes can occur: web post buckling and shear failure in weld. These
failure modes occur to different load levels and at different locations. For this reason one half of the
beam was stiffened such that the web post should not buckle any more and thus failure due to
excessive plastic strains can develop (see Figure 4-9 that shows a photo of a specimen during
testing).


Figure 4-9: Web post stiffeners
The dimensions of the stiffeners depend on the prevailing profile. The thickness of all stiffeners is
constant with 20mm. The stiffeners are only welded at the flange (and not at the web) such that a
lateral support of the web will be created without having an in-plane constraint in the web.

4.4.3.4.2 Test setup and test procedure
The statical system was a single-span beam under three point bending, both ends were simply
supported so that the system was statically determinated, (see Figure 4-10). A concentrated load
was introduced by a hydraulic jack onto the top flange in the middle of the beam.
74


1020
20 310
423.7-566.6
P
P

1700 - 2000

Figure 4-10: Test arrangement
The load application was deformation controlled by a low displacement rate. Furtheron the load
was applied stepwise by holding the displacement at these steps to record the load relaxation
effects. The tests were performed until not only the maximum load has been reached but also a
sufficient branch of the descending load deformation curve has been recorded.
The complete test arrangement and loading data are given in Annex A of the contribution to Part 2
of this report.

75


4.4.3.5 Measurement Devices
4.4.3.5.1 Strain Gauges
Strain gauges are located on both halves of the beam and on both sides of the web. It was intended
to gain most exact overview of the strain distribution in the web post. The exact position can be
taken from the following drawings.

Forefront

G10
G9
G8
G7
G14
G11
G12
G13


Back-side

G18
G15
G16
G17
G19
G22
G21
G20

Figure 4-11: Strain gauge distribution
The four vertical gauges in vertical line are arranged to observe the influence of the weld geometry
on the stress distribution. The gauges G7 and G21 as well as G13 and G15 are at the same position
from the weld to control the results of the gauges of the other side of the weld. By means of the
rosettes G10 and G18 the two dimensional stress-state can be fully recorded and thus the shear
stress can be determined.

4.4.3.5.2 Displacement Transducers
To measure the vertical deflection of the beams three inductive displacement transducers have been
used. They are positioned underneath each web post in the middle of the lower flange. The range of
each transducer was 50mm, see Figure 4-12.
76





Figure 4-12: Displacement Transducers
4.4.3.6 Test results
As agreed between both parts only eight specimen were carried out including all hybrid beams.
Instead of both tests, not foreseen F.E. simulations were carried out for the whole configuration. In
the following the plain data obtained from the experiments are given.

4.4.3.6.1 Actual dimensions
In Table 4-3 the actual measured dimensions of the test specimen are given.

Table 4-3: Actual dimensions of specimen
Specimen
n
Upper
Profile
Lower
Profile
Span Length
[mm]
Total Depth
[mm]
Web post
width
[mm]
Opening
Diameter
[mm]
1-ss IPE 400 IPE 400 1897 558 122 358
2-hs HEA 300 HEM 300 1800 475 95 350
3-ss HEB 400 HEB 400 1940 599 63 422
4-ss IPE 400 IPE 400 1944 600 55 430
5-hs IPE 300 HEM 300 1644 482 55 354
6-ss HEM 300 HEM 300 1559 495 50 334
7-hs HEB 300 HEM 300 1866 472 123 350
8-hs HEA 300 HEB 300 1800 415 147 300

4.4.3.6.2 ACB design results with actual dimensions
Following the design calculations performed with ACB, although considering the real geometrical
and mechanical properties, the failure load as well as the weld throat thickness have been
determined for the different configurations. The theoretical failure occurs by web post buckling for
all cases, except for test 8 where a Vierendeel mechanism should occur. The theoretical second
failure mechanism is the horizontal shear failure. Following the 1
st
failure mode, the minimum
throat thickness has been calculated according to equation (41), see Table 4-4.

77



Table 4-4: Theoretical failure load and throat thickness values
S
p
e
c
i
m
e
n

n

Upper part
of the
profile
Lower part
of the
profile
Theoretical web
post buckling
failure load
*

[kN]
Theoretical
Horizontal
shear failure
load
*

[kN]
Calculated
minimum weld
throat thickness
for the 1
st
failure mode
[mm]
Theoretical
applied weld
throat thickness
a
[mm]
Measured
average weld
throat
thickness a
[mm]
1-ss IPE400 IPE400 357 465 2.8
**
3.0

3.3
2-hs HE300A HE300M 245 320 2.7
**
3.0 4.2
3-ss HE400B HE400B 260 390 3.7 4.0 4.5
4-ss IPE400 IPE400 138 227 2.2
**
3.0

3.2
5-hs IPE300 HE300M 110 169 1.9
**
3.0

3.9
6-ss HE300M HE300M 331 486 5.9 6.0 6.0
7-hs HE300B HE300M 417 507 3.7 4.0 4.8
8-hs HE300A HE300B 249
Vierendeel
436 2.0
**
3.0 4.1
* According to ACB-software
** According to EC3, a minimum throat thickness of 3.0mm has been retained


4.4.3.6.3 Macrography
After having performed the tests, small pieces of the web post have been cut out in order to perform
macrographies and determine the real weld throat thickness. Both welds have been measured and
the average value has been retained.










Figure 4-13: Specimen 1 Figure 4-14: Specimen 2

a
a
a
a
78











Figure 4-15: Specimen 3 Figure 4-16: Specimen 4









Figure 4-17: Specimen 5 Figure 4-18: Specimen 6









Figure 4-19: Specimen 7 Figure 4-20: Specimen 7 with weld
failure

a
a
a
a
a
a
a
a
a
a
79



a
a

Figure 4-21: Specimen 8
4.4.3.6.4 Failure loads and mechanisms
In the following Table 4.5 the designed (= theoretical by ACB-software) limit loads and the
measured limit loads are given.

Table 4.5: Failure loads and mechanisms
as designed
(= theoretical, see
Table 4-4):
as observed:
S
p
e
c
i
m
e
n

n

F
u,1,th
[kN]
F
u,2,th
[kN]
F
u,1,ex
[kN]
1. Failure Mechanism
F
u,2,ex
[kN]
2. Failure Mechanism
F
u,i,ex
/
F
u,i,th

1-ss 357 465 500
WPB of the unstiffened
post
537 WPB of the stiffened post 1.50
2-hs 245 320 660
1)
LSF at the unstiffened side - - 2.06
3-ss 260 390 627 WPB of the stiffened post - - 2.41
4-ss 138 227 252
WPB of the unstiffened
post
264 WPB of the stiffened post 1.91
5-hs 110 169 393 WPB of the stiffened post - - 3.57
6-ss 331 486 971
Stop of the test due to reaching of the limit load of the dynamometer
no failure
-
7-hs 417 507 905
2)
LSF at the stiffened side - - 1.78
8-hs 249 436 685 WPB of the stiffened post - - 2.75
1)
The maximum load was reached at 660 kN, the failure occurred in the descending load deformation curve at 606 kN.
2)
The maximum load was reached at 905 kN, the failure occurred in the descending load deformation curve at 880 kN.

With F
u,1,th
= designed limit load for web post buckling (WPB)
F
u,2,th
= designed limit load for longitudinal shear failure (LSF)
F
u,1,ex
= Load at which the first failure mechanism in the test occurred
F
u,2,ex
= Load at which the second failure mechanism in the test occurred

Some explanations and the reasons for the big (safe-sided) differences of the theoretical and
experimental loads are indicated under chapter 4.4.3.8 Conclusions.

80


min
- +
max
F
4.4.3.6.5 Strains
4.4.3.6.5.1 Strains in the midpoint of the web post
The characteristic (typical) load-strain development in the centre of the web post is given in Figure
4-22.








Figure 4-22: Characteristic load-strain development

Rough interpretation of this behaviour can be the following:
Phase 1: linear elastic behaviour according to theory, no normal strains occur
Phase 2: change to plasticity and in-plane membrane effects lead to a developing of normal strains
Phase 3: Overlaying of instability effects such as local buckling and change of sign of the strains
can finally occur.

Vertical strains
The results of the gauges G2 and G5 on the unstiffened half and gauges G10 and G18 (vertical part
of the rosette) on the stiffened half are given in Table 4-6 (in [m/m]).

Table 4-6: Vertical strains in the midpoint of the web
G2 (upper profile) G5 (lower profile) G10 (upper profile) G18 (lower profile) Specimen
n min max min max min max min max
1-ss 0 2781 -202 1303 -1724 0 0 1343
2-hs -112 5413 defect 0 6576 defect
3-ss -479 4835 -690 886 -1044 7409 0 7984
4-ss -1278 0 -189 0 -506 379 0 649
5-hs 0 5292 0 896 -2133 1580 0 397
6-ss -2641 6611 -2239 4816 -1677 6259 -1379 3024
8-hs 0 5524 -538 4094 -1746 1639 -211 8065

Three phases can be distinguished:
1. linear load strain behaviour within a
range of +/- 100 m/m
2. increasing negative strains
3. reversal and increasing positive
strains
81


Horizontal strains
The measured strains of the gauges G1 and G4 on the unstiffened half and gauges G10 and G18
(horizontal part of the rosette) on the stiffened half are given in Table 4-7 (in [m/m]).

Table 4-7: Horizontal strains in the midpoint of the web
G1 (upper profile) G4 (lower profile) G10 (upper profile) G18 (lower profile) Specimen
n min max min max min max min max
1-ss -914 0 -689 1085 -1421 0 -854 0
2-hs -1728 1829 -582 0 defect -2098 434
3-ss -376 1442 -435 674 -359 675 -1269 470
4-ss -544 297 0 635 0 1402 -816 156
5-hs -218 2887 -521 0 0 5164 -770 0
6-ss 0 2494 -427 3351 0 5813 0 2779
8-hs -1928 1047 -1254 434 0 4869 -2301 0

The corresponding load strain diagrams are given in Annex B of the contribution to Part 2 of this
report.
Some explanations of the strain development are given in chapter 4.4.3.8.

4.4.3.6.5.2 Strains at the edge of the web post
The strain development at the edge of the web post is characterised by a slow linear increase over
50-70 per cent of the limit load, a sharp bend with a following excessive strain increase.

Vertical strains
The results of the gauges G3 and G6 on the unstiffened half and gauges G14 and G22 on the
stiffened half are given in Table 4-8 (in [m/m]).

Table 4-8: Vertical strains at the edge of the web
G3 (lower profile) G6 (upper profile) G14 (upper profile) G22 (lower profile) Specimen
n max max max max
1-ss 7151 6820 17578 21537
2-hs -1699 6351 10087 15989
3-ss defect 7636 4477 5217
4-ss 1926 1828 3057 1625
5-hs 3762 -4041 2028 1513
6-ss defect -5307 - -
7-hs - - 19397 19798
8-hs -2128 3564 9878 19463

The corresponding load strain diagrams are given in Annex B of the contribution to Part 2 of this
report.
Some explanations of the strain development are given in chapter 4.4.3.8 .

82


4.4.3.7 Finite Element Analysis (FEA)
For the verification of the test results and for clarifying the internal forces (in particular in the
horizontal weld) a finite element analysis was carried out using the FE-program ANSYS, version
7.0. Meshing and particular parameters can be taken from Annex D of the contribution to Part 2 of
this report.
In the following figure (Figure 4-23) the stresses due to a unit load according to the first critical
mode are given as a representative for the imperfection model of the FEA. It becomes clear that the
buckling behaviour of the web post can be simulated fairly exact.


Figure 4-23: Imperfection acc. to first critical mode
The nonlinear buckling analysis was carried out using the following assumptions:
- Use of 4 nodes shell element with quadratic deformation approach.
- Big distortion approach.
- Geometrically as well as materially non-linear calculation.
- Solver by iterations according to Newton-Raphson.
- Ideal elasto-plastic material behaviour using actual (measured) values.
- The full (real) stress strain curve was not included. The error that was induced by this
simplification however is negligible.
- The beam geometry is based on the measured dimensions of tested beams.
- The little stiffeners at the outer web post are not considered.
- The imperfection of the web is based on the first critical mode (eigenvalue) that was
obtained from a linear buckling analysis in a first step, see Figure 4-23.
- The load is applied stepwise on one node in the middle of the upper flange.
- The influence of meshing and its refinement has been tested out in advance, such that the
presented FE-results can be regarded as consolidated.

83



Table 4-9: Results of the FEA
Limit load [kN] Max shear force in the weld V
h
[kN]
Specimen
n
ACB Test ANSYS ACB*

ANSYS ANSYS**
Failure mode
1-ss 357 500 480 195 184 157 Web post buckling
2-hs 245 606 565 193 242 145 Web post buckling
3-ss 260 627 623 137 120 98 Web post buckling
4-ss 138 252 261 85 61 51 Web post buckling
5-hs 110 393 340 104 114 62 Web post buckling
6-ss 331 971 1072 145 147 105 Web post buckling
7-hs 417 905 815 285 363 249 Web post buckling
8-hs 249 685 601 291 321 144 Web post buckling
*
according to the first failure mode for ACB, according to eq.(41) and using actual dimensions
(see Table 4-4), and material properties
* *
at the ACB limit load

To show the good congruence of the global beam behaviour between the FEA and the real tests one
of the load-deflection-curve is compared in the following.


Beam 4: Load-Deflection-Diagram
0
50
100
150
200
250
300
0 2 4 6 8 10 12 14 16 18
Deflection [mm]
L
o
a
d

[
k
N
]
ANSYS
TEST

Figure 4-24: Load-Deflection Diagram (Beam 4)
Further FE-results are given in Annex D of the contribution to Part 2 of this report.
From the comparison between the maximum shear force in the weld at ACB load level obtained by
FE-calculations and the design shear force at the first instability mode obtained by ACB-software it
can be seen that the horizontal acting loads predicted are greater than in the simulated reality.
Therefore the resulting weld-dimensions will be overestimated. Consequently the welds can be
minimized and performed mostly without chamfering if the weld throat thickness to apply is less
than two times six millimetres a 26mm.

84


4.4.3.8 Conclusions
The load-deformation behaviour of short beams with large web openings has been investigated. In
particular the welds of the web posts along the neutral axis have been thoroughly considered by
means of experiments and Finite Element simulations. The research results have been compared to
the results from the ACB-software. Furthermore very important conclusions for the preparation,
design and dimension of the welds could be drawn.
From the experiments it came out, that in all cases the obtained experimental ultimate load
exceeded the predicted loads. This shows the safe sided approach of cellular beams behaviour and
the reserves into the welds that are available before failure occurs. Although having reached the
horizontal plateau into the load deflection curve (see Annex B of the contribution to Part 2 of this
report and chapter 4.4.3.7), meaning that all redistribution mechanism has been activated and that
the welding thickness corresponding to the 1
st
redistribution mechanism has been exceeded by far,
only one horizontal shear failure into the weld itself (test 7) could be provoked, in test 2, a failure
into the base material occurred .
As far as the welds are concerned it can be noted that predicted shear stress values by ACB-
software are safe sided higher as the calculated ones from the simulations. Together with the
experiments it can be considered, that the theoretical shear stresses and as result of the calculated
weld throat thicknesses are overestimated, taking into account the whole stress state at the weld in
longitudinal direction and perpendicular to the weld. So the model used in the ACB software for the
prediction of the weld throat thickness is on the safe side and it reproduces the behaviour with an
acceptable reliability. In many cases a full penetration weld is not necessary. Even chamfering can
be avoided in a lot of cases.
It should be pointed out again that the weld throat thicknesses determined by the means of ACB
correspond to the stress level at the first theoretical instability mode. Therefore the effects of the
further developing redistribution mechanism and their complexity are not considered for the
determination of the theoretical weld throat thickness by ACB.

85





5 Contribution from SCI
This section summarises the activities of the Steel Construction Institute (SCI) in the ECSC project
7210-PR-315 entitled Large Web Openings for Service Integration in Composite Floors for the
period 1 July 2001 to 31 December 2003.
In the framework of this project, SCI has focused its efforts in developing design guidance and
software. An overview of the work is given below, which describes the following:
1. A summary of the results of a series of Finite Element Analyses (FEA) of cellular beams
comprising regular circular openings in which the openings were placed non-centrally in the
beam depth, and unequal web areas for the top and bottom Tees.
2. A generalised design method that may be applied to hot rolled and fabricated steel sections, and
to beams with discrete and regular circular openings.
3. The back-analyses of eight Arcelor tests on cellular beams for both symmetric and asymmetric
cross-sections, which demonstrate the level of accuracy inherent within the proposed design
guidance.
4. A full account of the engineering logic used in the design software for the purpose of
representing and testing the proposed design guidance.

5.1 Finite Element analysis of web post buckling
The results of a series of Finite Element Analyses (FEA) were collated for a cellular beam with
equally spaced circular openings. The beam was analysed as a cantilever in order to examine the
shear transfer around the opening and through the web-post between the openings. The FEA was
carried out using ANSYS, a generalised FEA program. The openings were 60% of the depth of the
beam and were placed at 1.25, 1.5 and 1.8 opening diameter (centre-centre) along the beam.
Three cases of asymmetry were considered:
Centrally placed openings in the beam depth.
Non-centrally placed openings (ratio of depth of Tees of 3 : 1).
As the cases above, but with different web thicknesses of the top and bottom Tees (ratio of
thicknesses of 2 : 1).
The FEA considered plasticity, buckling of the web-post, and other failure modes, dependent on the
geometry of the cross section and opening. The following conclusions were drawn from this study:
The effect of asymmetry of opening position makes a small effect on the failure load (by 2 to
10%).
The effect of differential web thickness is beneficial and increases the failure load when
compared to a section of uniform thickness. The level of shear utilisation is similar in all cases
of the same web-post width.
The level of horizontal shear utilisation exceeds 1.0 for some asymmetric openings with
differential web thickness indicating a highly complex stress state. The current design approach
is conservative for asymmetric sections.
Web-posts of width of 0.8 opening diameter do not fail by web-post buckling or horizontal
shear.
Ring stiffened openings increase the buckling resistance of the web-post, but the effect
diminishes for alternate ring stiffened openings with wider web-posts.
Based on these FEA results, a simplified theory of web-post bending has been proposed, which may
be combined with horizontal shear to assess the buckling resistance of the web-post. An alternative
87


compression field or strut model is also proposed, in which compression and bending in the web-
post may be combined.
It is concluded that the effect of asymmetry in opening position and web thickness is not as critical
as first thought, and the tendency for web-post buckling is not increased significantly.

5.2 Design guidance for composite beams with large openings for services
The design of composite beams with large rectangular or circular openings is a practical problem
which is caused by the need to pass service ducts through the structural zone of the beams. Careful
sizing and positioning of these openings can minimise their adverse effects on the shear and
bending resistances of composite beams. The design guidance extends that given in SCI P068 [1]
and P100 [2], and presents a generalised design method that may be applied to hot rolled and
fabricated steel sections, and to beams with discrete and regular circular openings.
The new aspects of the design approach that are covered in detail are: bending and tension
interaction effects in the bottom Tee of the section; treatment of highly asymmetric sections; and
local composite effects at large openings. A step-by-step design procedure is presented, which is
consistent with BS 5950-1 and -3 and with Eurocodes 3-1 and 4-1.
General guidance on positioning of openings is presented. The same general principles may also be
extended to notched beams and to tapered beams.
Large openings cause a significant reduction in the shear resistance of beams, due to the loss of a
major proportion of the web but a smaller reduction in bending resistance. Therefore, the shear
transfer across large openings is the most important design requirement, and it is good practice to
locate large openings remote from the high shear zones of a beam in order to minimise their effect.
Shear transfer occurs by Vierendeel or four corner bending in the web flange Tee sections,
which leads to a complex interaction of forces at the corners of the opening. Horizontal stiffeners
improve the transfer of shear by this local bending action.
Fabricated beams may also have slender webs, which may require stiffening to prevent buckling of
the web close to the opening. However, the use of stiffening should be minimised by choosing a
thicker web to avoid vertical stiffeners, or by careful positioning of the openings away from point
loads.

5.2.1 Existing publications
SCI Publication 068 Design of openings in the webs[1] was prepared in 1988 to present a design
method for discrete openings in hot rolled steel beams in composite construction. The methodology
was calibrated against tests on a 10 m span composite beams carried out at the University of
Warwick. These tests used 533 210 82 kg/m UB sections (d/t 50) with a range of opening
sizes of 700 350, 600 200 and 450 250, and 350 mm square.
SCI Publication 100 Design of composite and non composite cellular beams[2] was prepared in
1990 to extend the design methodology to cover beams with regular circular openings. The new
aspect of SCI P100 was the design method for the interaction of forces between adjacent openings
and its effect on the stability of the web post.
Modern design and construction of composite beams with web openings differs in important
respects from the scope of SCI P068 and P100, and from the data base of existing tests:
The steel sections are often highly asymmetric in terms of their flange area.
The webs are often relatively slender, so that the effect of buckling is increased.
Openings are relatively long in terms of their aspect ratio (length : depth).
Openings may be asymmetric in the depth of the section and are often close to the flanges.
88


Elongated openings can be formed by removing the web post between adjacent circular
openings.
Deep beams may be used as transfer structures supporting columns or heavy loads, and can be
perforated for doors or large ducts from plant rooms.
These changes in practice have necessitated a re appraisal of these existing publications, so that they
can be extended to cover a greater range of applications to both hot rolled and fabricated steel
sections. However, general structural design principles should be observed for unusual design
cases, and it is not possible to give definitive guidance for all cases. This is particularly true for
fabricated sections, where some stiffening is generally required to resist local forces and to control
the effects of buckling.

5.2.2 Scope of publication
The guidance offered in SCI P068 and P100 has been extended for both hot rolled and fabricated
sections. The design method is common to all these applications and differs only in:
The depth of web that is used in the definition of the shear area to BS 5950-1.
The classification of the cross section for local buckling.
The control of web buckling by use of stiffeners.
SCI P068 is still valid for hot rolled steel beams with discrete openings, and SCI P100 is still valid
for cellular beams of approximately symmetric cross section. However, the scope of new design
guidance has been generalised to cover the cases of:
Highly asymmetric sections.
Openings placed non centrally in the depth of the section.
Slender webs.
Narrow web posts.
Fabricated sections.
Notched beams.
The publication does not cover:
Tapered fabricated sections.
Fire engineering design.
A worked example, is provided which illustrates the use of these methods.

5.2.3 Design codes
This guidance is prepared in general analysis terms and cross refers to BS 5950-1 and -3, and
Eurocode 3-1-1 and Eurocode 4-1-1, as appropriate. The principal difference between these design
codes is in terms of the partial factors for loads and materials. However, some other limits exist
regarding partial shear connection and shear connection resistances.
Both codes are expressed in terms of limit state principles.

5.2.4 Material strengths
The strength of steel is expressed in terms of its yield strength in N/mm
2
(f
y
to EC3 1-1 or p
y
to
BS 5950-1). Standard grades are S275 and S355. A partial factor of 1.0 applies to steel.
89


The strength of concrete is expressed in terms of its cube strength, f
cu
, to BS 5950-3, or its cylinder
strength, f
ck
, to EC4 1-1. The design strength of concrete is given by 0.45 f
cu
or 0.85 f
ck
/
c
,
respectively according to the two codes
c
is the partial strength for concrete (
c
= 1.5). The highest
concrete grade that may be used in composite design is C40/50.
The shear resistance of the shear connectors is dependant also on the concrete grade. The design
resistance is taken as 0.8 characteristic resistance (the 0.8 factor representing the inverse of a
partial safety factor of 1.25). A further reduction in resistance may be required due to the shape of
the deck profile and for the use of lightweight concrete (LWAC).
For cases when LWAC is used, the stud strength should be multiplied by 0.9 in accordance with
BS 5950-3. Whereas for EC4-1-1 the secant modulus, EcM should be multiplied by (/2200)
2
in
accordance with EC2-1-1 (where is the dry density of concrete in kg/m
3
).

5.3 ARCELOR web post buckling tests
The following tables present the analysis of eight Arcelor tests on Cellular beams for both
symmetric and asymmetric cross-sections. Most beams failed by web-post buckling or Vierendeel
bending often at large displacements. The web-post buckling analysis is based on the models in the
Design Guide. The Model Factor defines the conservatism that exists in the analysis model. *.pdf
documents from software analysis of Tests 1, 3 & 4 have been created, see contribution to Part 2 of
this report.

5.3.1 Symmetric Beams
Tests 1, 3 and 4 failed by web-post buckling. The model factor for these tests exceeds 1.0 by a
significant margin, indicating that the design model is conservative; particularly for narrow web-
posts. Test 6 failure by plastic deformation, involving some tension field action in the web-posts.

Table 5-1: Model factors for symmetric beams
Test Beam

V
test

kN
V
h

kN
V
h,Rd

kN
Model
factor
Failure mode
1 IPE 400 0.95 268 240 197 1.23
Web-post
buckling
3 HEB 400 0.72 313 269 172

1.73
Web-post
buckling
4 IPE 400 1.14 132 110 77 1.54
Web-post
buckling
6 HEM 300 0.38 350
*
311 233
*
1.33
Web-post
shear/rotation
Assumed yield strength = 350 N/mm
2


effective slenderness of web-post
V
h
is horizontal shear in web-post
V
h,Rd
is web-post shear or buckling resistance
*
failure load taken at point of plasticity in test

failure based on pure shear of web post

90


5.3.2 Asymmetric Beams
Tests 2, 5 and 7 failed by Vierendeel bending of the top Tee at the opening together with web-post
buckling or shear. The distribution of shear between the top and bottom Tees is unknown, and the
difference between the shear forces in the Tees determines the moment developed in the web-post
for equilibrium between the Tees.
An upper bound to the shear force in the top Tee is its Vierendeel bending resistance or its shear
resistance. The Vierendeel bending resistance is calculated for an effective opening width of
0.45 d
o
and opening height of 0.9 d
o
(d
o
= opening diameter). The remaining shear force is resisted
by the bottom Tee. The web-post moment M
h
is calculated from equilibrium, and is generally
higher than the elastic bending resistance of the web-post M
h,el
. This indicates that plasticity or
tension yielding develops in the web-post, and so the use of its elastic bending resistance is
conservative.
The effective horizontal shear force V
h,eff
acting on the web-post takes into account the combined
effect of shear and moment (which reduces the effective shear in the top Tee but which increases
the effective shear in the bottom (heavier) Tee). The model factor in tests 7 and 8 was reasonable,
but was very high for radically different Tee sizes (e.g. IPE300/HEM300) and a narrow web-post,
indicating that the method is very conservative for this case. Test 8 failed more by vertical and
horizontal shear with a model factor of 1.19.

Table 5-2: Model factors for asymmetric beams
Test
Beam
(top/
bottom)

V
test

kN
V
h

kN
V
h,Rd

kN
M
h

kNm
M
h,el

kNm
V
h,eff

kN
Model
factor
Failure mode
2
HEA300
HEM300
0.97 303 303
158
403
3.0
4.5
11.1
285
320
1.80
Vierendeel bending
and web-post buckling
5
IPE300
HEM300
1.14 196 174
63
233
9.4
1.3
3.9
120
226
1.90
Vierendeel bending
and web-post buckling
7
HEB300
HEM300
0.76 452 485
273
522
15.6
9.8
18.7
401
568
1.47
Vierendeel bending
and web-post shear
8
HEA300
HEB300
0.89 392 372
262
322
8.8
10.7
13.8
313
430
1.19
Vertical shear and
web post shear
Assumed yield strength = 350 N/mm
2

Notes to table:
Failure occurs by buckling or yielding in the top part of web-post or by Vierendeel bending in the
top chord.
M
h
is the moment in the web-post which is calculated assuming shear is resisted by the top Tee
according to its maximum shear or Vierendeel bending resistance and the remaining shear is
resisted by the bottom Tee. M
h
is allowed to exceed M
h,el
in this analysis.
M
h,el
is the elastic bending resistance of the web post
Effective horizontal shear force: V
h,eff
= V
h
2 M
h
/d
o
(top) or V
h,eff
= V
h
+ 2 M
h
/d
o
(bottom)

5.3.3 Conclusions
The average model factor for the symmetric beams was 1.46, whereas for the asymmetric beams
this was 1.59; indicating that the proposed design method for web-post buckling is conservative,
and becomes increasingly conservative when the relative sizes of the Tee increases. This is
probably because of tensile yielding effects in the web leading to additional post-buckling
resistance. This effect was apparent in some tests on beams with thicker flanges.
91


5.4 Engineering logic for the analysis and design software
The following application rules are consistent with the principles of Eurocode 4 and Eurocode 3
Annex N, and apply to cellular beams with regular circular openings and beams with isolated
openings. The rules apply to composite beams and may be adapted to steel beams (by ignoring the
contribution of the concrete slab). They are suitable for hand calculations.

1 Effective length of opening
1.1 Effective length
For Vierendeel bending, the effective length of the opening is given by:
o
l = 0.5 d
o
- circular openings
o
l = s + 0.5 do - elongated openings
o
l =
o
l - rectangular openings
where:
d
o
= the opening depth (or diameter).
s = centre-centre spacing of openings
The minimum depth of the tees is used in subsequent checks for circular and elongated openings.

2 Shear resistance of perforated composite beam
2.1 Shear resistance of tees
For hot rolled steel sections, the shear resistance of the top and bottom tees is given by:
V
t,Rd
= 0.9 [(2r + t
wt
) t
ft
/2 + d
t
t
wt
] f
y
/ 3
and:
V
b,Rd
= 0.9 [(2r + t
wt
) t
fb
/2 + d
b
t
wb
] f
y
/ 3

For fabricated steel sections:
V
t,Rd
= d
t
t
wt
f
y
/ 3
V
b,Rd
= d
b
t
wb
f
y
/ 3
where:
t
wt
= thickness of top web
t
wb
= thickness of bottom web
d
t
= depth of web of top Tee (below flange)
d
b
= depth of web of bottom Tee (above flange)
t
ft
= thickness of top flange
t
fb
= thickness of bottom flange
r = root radius of a rolled section
f
y
= design strength of steel

92


2.2 Shear resistance of concrete
At an opening, the shear resistance of the concrete slab is considered over an effective width of 3
effective slab depth plus the flange width according to:
V
Rd,c
= (1.5 (h
s
+ h
p
) + b
f
) (h
s
h
p
) v
cd

where:
h
s
is the slab depth
h
p
is the deck profile height
b
f
is the width of the steel flange ( 300 mm)
v
c
is the shear strength of the concrete to EC2-1-1, depending on the percentage reinforcement in
the slab

2.3 Pure shear resistance of perforated section
The pure shear resistance of the section at an opening is given by the sum of the shear resistances of
the steel section and the slab:
V
Ed
V
b,Rd
+ V
t,Rd
+ V
Rd,c

However, the shear force that can be transferred by the bottom Tee is reduced due to Vierendeel
bending and is given by:
V
b
= 2 M
b,red
/
o
l V
b,Rd

where:
M
b,red
is the reduced bending resistance of the bottom Tee due to axial tension (see Step 5)

2.4 Shear resistance of highly asymmetric sections
For design of highly asymmetric composite sections, the shear resistance is based on equilibrium of
the top and bottom tees, as influenced by the width of the web post between adjacent openings. For
narrow web posts, the ability to re-distribute shear between the tees is limited. As a first
approximation for a composite beam, the shear transferred by the bottom Tee controls the overall
shear resistance of the beam, as follows:
V
Rd
= 2V
b
/(1 + 2e/h
eff
)+ V
c,Rd
V
Ed

where:
e is the eccentricity of the centre-line of the openings above the mid-height of the beam
(negative below mid-height)
h
eff
is the distance between the centroids of the tees of the steel section
V
Rd,c
is the shear resistance of the slab (see 2.3)
and:
V
Ed
is the applied shear force at the centre line of the opening
For wider web posts, the shear forces can be redistributed from the bottom Tee to the stronger top
Tee by development of an in-plane moment in the web-post, M
h
. A closed solution for this case is
given in Step 6.4:

93


2.5 Effective web thickness due to shear
The effective web thickness of the tees is reduced due to the presence of high vertical shear
according to:
t
w,eff
= t
w
[1 (2p 1)
2
] for p > 0.5
or:
t
w,eff
= t
w
for p 0.5
where:
t
w,eff
= effective web thickness
t
w
= actual web thickness
p = utilisation in shear
= V
Ed
/(V
b
+ V
t,Rd
+ V
c,Rd
)
This effective thickness of the web is used in calculating the reduced section properties of the tees
for global and Vierendeel bending.
Note: A process of iteration if required because V
b
depends on p. Conservatively V
b
can be first
set to zero to calculate p.

3 Section Classification
The section classification of the composite section depends on:
The top flange, as influenced by its attachment to the slab
The web in compression, depending on the asymmetry of the section
The depth of the web of the tees
The length of the opening
The first two criteria are defined in EC3 and EC4. The third and fourth criteria depend on the
proportions of the opening, as follows:
For Class 1 or 2 sections, plastic section properties may be used.
For Class 3 or 4 sections, elastic section properties may be used.

3.1 Classification of flange
The classification of the flanges may be taken as in EC3 1-1 and EC4 1-1. Generally the top flange
may be treated as Class 2 because of its attachment to the slab.

3.2 Classification of the web of the Tee (ENV 1993 1-1, Clause N.1.7.2)
The outstand of the web of the Tee may be classified, depending on the ratio of the length of the
opening,
o
l , to the outstand depth d
t
, as follows:
Class 2 webs: >
|
|
.
|

\
|

t 32 if
t 32
1
t 10
d
o
2
o
w
w
t
l
l

94


All Class 3 webs may be treated as Class 2 if
o
l 32t
w

For a Class 2 section, the plastic properties of the Tee may be used.
Class 3 webs: >
|
|
.
|

\
|

w o
2
o
w
w
t
t 36 if
t 36
1
t 14
d l
l

All Class 4 webs may be treated as Class 3 if
o
l 36t
w

For a Class 3 section, the elastic properties of the Tee should be used.
Class 4 webs: Calculate the effective section properties using the limiting value of d
t
for a Class 3
web.
Webs that satisfy the lower limit on
o
l are classified independently of d
t
.
The effective length of the opening may be treated as follows for stability of the web above or
below the opening:
o
l =
o
l - rectangular openings
o
l = 0.7d
o
- circular openings
o
l =
o
l 0.3d
o
- elongated openings
where:
o
l is the actual length of the opening.
The d
t
/(t
w
) limits are presented in Table 1 for different proportions of openings. A second check
on the length of the opening should also be made, as in Table 2, as the web classification is
automatically satisfied for small openings.

Table 1: Limit on maximum d
t
/(t ) for webs above openings
o
l /d
t
(or d
o
/d
t
for circular openings) Type of
opening
Class limit
> 10 10 5 3 2
Rectangular 2 10 10.5 11.9 14.6 18.9
Rectangular 3 14 14.9 17.2 21.9 28.8
Circular 2 10 12.1 13.5 18.2 22.8
Circular 3 14 15.7 20.1 27.8 38.6
d
t
= depth of web of Tee

Table 2: Lower bound of
o
l below which the web class is satisfied for all
values of d
t

Type of opening Class limit Web class satisfied if:
Rectangular 2
o
l 32t
w

Rectangular 3
o
l 36t
w

Circular 2 d
o
46t
w

Circular 3 d
o
51t
w


95


3.3 Effect of axial tension on section classification
Tension in the bottom Tee modifies the section classification for unstiffened tees.
Class 3 webs may be treated as Class 2 if:
y b
y
2
w
y b
f A
f t 20
1
f A
) x ( N

where:
N(x) is the axial tension in the bottom Tee at point x in the span
A
b
is the cross sectional area of the bottom Tee
In this case, the plastic stress blocks for a Tee section subject to bending and tension are such that
the depth of web in compression does not exceed 10t
w
.
and:
All Class 4 webs may be treated as Class 3 when also subject to axial tension, provided:
d
b

2
o
w
w
' t 36
1
' t 14
|
|
.
|

\
|

l
if
o
l > 36t
w

where:
= /(1 N(x)/(A
b
p
y
))
0.5

and:
d
b
is the depth of the web of the bottom Tee above the flange

4 Global Bending Action
4.1 Compression in slab
The bending resistance of a perforated section of any position x along the beam depends on the
degree of shear connection at that position.
Compression force in slab
N
c
(x) = n
sc
(x) P
d

where:
n
sc
(x) is the number of shear connectors from the support to point x
P
d
is the design resistance of a shear connector
x is the distance from the support to the closer edge of the opening

4.2 Tensile force in bottom Tee
Equilibrium of the cross-section leads to the following tensile force in the bottom Tee due to global
bending:
N(x) =
( )
b e et
et p s c
y y h
y 2 / ) h h ( ) x ( N ) x ( M

+ +

where:
N(x) is the global moment acting at point x
96



4.3 Force in top Tee
The tension or compression force in the top Tee depends on the difference between N
c
(x) and N(x).
A further check is required on the absolute value of the tension or compression force in the top Tee,
according to:
|N(x) N
c
(x)| A
t
f
y

where :
A
t
is the cross-sectional area of the top Tee

4.4 Minimum degree of shear connection
The minimum degree of shear connection that should be achieved is given by N
c
(x) 0.5 T(x) at
the opening.
Note: No limit on the degree of shear connection is made in the region close to the supports,
provided the minimum degree of shear connection is achieved in mid-span. The shear
connectors in the last 300 mm should be ignored because of the unknown detailing of the
connection at the stage when the beam is designed.

5 Bending Resistances of tees
The bending resistance of the tees depends on the section classification and is given by the
following equations:

5.1 Plastic Resistance (Class 2)
M
pl,T
= A
w
f
y
(d
t
/2 + t
f
y
p
) + A
f
f
y
(y
p
t
f
/2) + A
s
f
y
(d
t
- d
s
+ t
f
t
p
)
where the plastic neutral axis position is given by:
y
p
= (A
f
+ A
w
+ A
S
)/(2A
f
) for A
s
A
f
- A
w


5.2 Elastic Resistance (Class 3)
M
el,T
=
e f t
e f s s s
2
f e y f
2
e f t y w
y t d
) y t e h ( A ) 2 / t y ( f A ) y t 2 / d ( f A
+
+ + + +

where the elastic neutral axis position is given by:
y
e
=
) A A A (
2 / t A ) t 2 / d ( A ) t e d ( A
s w f
f f f t w f s t s
+ +
+ + + +

where:
A
w
= cross sectional area of web of the Tee = d
t
t
w,eff

A
f
= cross sectional; area of flange
A
s
= cross sectional area of horizontal stiffener (if no stiffener; A
s
= 0)
d
t
= depth of web of Tee excluding the flange
e
s
= offset distance of centre of stiffener from tip of web ( 10 mm)
t
f
= thickness of flange
97


A further limit is that A
s
0.5 d
o
t in order to ensure that the adjacent solid web is strong enough to
resist the forced transferred from the stiffener.

5.3 Reduced bending resistance due to axial force
The bending resistances of the tees is reduced due to axial tension, according to the following
formula:
Plastic resistance:
No stiffeners: M
pl,T,red
= M
pl,T
(1 (N/N
pl
)
2
)
With stiffeners: M
pl,T,red
= M
pl,T
(1 N/N
pl
)
Elastic resistance:
All cases: M
pl,T,red
= M
el,T
(1 (N/N
el
)
2
)
where:
M
pl,T,red
is the reduced plastic resistance of the Tee section
M
el,T,red
is the reduced elastic resistance of the Tee section corresponding to the Class 3 limit
N
pl
is the plastic compressive resistance of the Tee including the depth of web corresponding
to the Class 2 limit
N
el
is the elastic compressive resistance of the Tee including the depth of web corresponding
to the Class 3 limit
N is the axial compression or tension force due to the global moment action at any position
(x)

6 Vierendeel Bending
6.1 Applied Vierendeel moment
The Vierendeel bending moment due to transfer of shear across the opening is given by:
M
v
= V
Ed
o
l
The effective length of the opening is defined in 1.1.

6.2 Local composite action
The Vierendeel bending resistance is due to both local composite action and the reduced bending
resistance of the tees. The local bending resistance due to composite action across the opening is
given by:
M
vc,Rd
= n
sc,o
P
d
(y
et
+ h
s,eff
)
where:
y
e,t
is the distance of the elastic neutral axis of the top Tee below the top of the flange.
n
sc,o
is the number over shear connectors over length,
o
l .
h
s,eff
is the effective depth to the centre of compression of the slab.
l
k is a reduction factor due to the length of the opening given by:

l
k = 1 -
o
l /(25 h
t
) for unstiffened openings

l
k = 1 -
o
l /(35 h
t
) for stiffened openings
98


where:
h
t
is the depth of the top Tee.
l
k can be taken as 1.0, provided
o
l 5 h
t
for both unstiffened and stiffened openings.
The maximum value of this local composite resistance that may be included is dependent on
possible pull out failure of the shear connectors. The upper bound is given by:
M
vc,Rd

o
l P
d
for single shear connectors in line
1.5
o
l P
d
for shear connectors in pairs

6.3 Total Vierendeel bending resistance
For widely spaced openings, the shear transfer across the opening must satisfy the combination of
the Vierendeel bending resistance of the steel sections and local composite action:
V
Ed
(2M
b,red
+ 2M
t,red
+ M
vc
)/
o
l

6.4 Closed solution Circular openings
For closely spaced openings, the Vierendeel bending resistance is potentially reduced due to the
bending resistance of the adjacent web post.
As a closed solution, the maximum shear transfer across a beam with circular openings controlled
by web post bending should satisfy:
( )
s / M
h / e 2 1
s / M 2 / M 4
V
s , c
eff
e h, o red b,
Ed
+
+
+

l

where:
M
h,e
is the elastic bending resistance of the web post given by:
M
h,e
= s
o
2
t
w
f
y
/6
M
c,s
is the moment developed due to local composite action between the top Tee and the slab over
a length s, given by:
M
c,s
= n
sc
P
d
(y
c,t
+ h
s,eff
)
h
eff
is the effective depth of the steel section between the centroids of the tees (= h
s
y
et
y
eb
)
n
sc
is the number of shear connectors in length s between the centre of adjacent openings

6.5 Elongated openings
For analysis of elongated openings in beams with otherwise regular circular openings at spacing s,
the term is M
h,e
should be reduced by a factor of:
0.67 for adjacent circular and elongated openings
0.5 for adjacent elongated openings.
Note: An elongated opening is formed by cutting away the web between adjacent openings.

99


6.6 Closed solution Rectangular openings
For closely spaced rectangular openings, the shear transfer should satisfy the conditions for web
post bending at the top of the opening, given by:
V
Ed

( )
( ) ( )
eff o
e h, o red b,
h / d e 2 1
s / M 2 / M 4
+ +
+ l
+ M
c,s
/s

7 Web post resistance
7.1 Horizontal shear force
Web post shear occurs due to the development of longitudinal forces in the bottom Tee necessary to
resist global bending. The horizontal shear force acting on the web post is given by:
V
h
= N =
s
M
h
s V
s c,
eff
Ed

where:
N is the increase in tension in the bottom Tee over a length, s

7.2 Horizontal shear resistance
The horizontal shear resistance is obtained by considering a shear area of the web post (A factor
may be introduced due to the possible coincident moment acting on the web post).
V
h,Rd
= s
o
t
w
f
y
/ 3
where:
s
o
= edge to edge spacing of openings
and:
V
h
V
h,Rd

Note: For asymmetric beams, t
w
is the minimum web thickness. Web-post buckling need not be
checked if d
o
20t
w
for circular openings.

7.3 Effective horizontal stress
Web post buckling may be analysed by considering the compressive stress due to an effective
horizontal force in the web post given by:
V
h,eff
= V
h
2M
h
/d
o
on the bottom Tee
or:
V
h,eff
= V
h
+ 2M
h
/d
o
on the top Tee
where:
V
h
is the horizontal shear in the web post
and:
M
h
is the moment acting on the web post, with a maximum value as in 2.4 (This is positive when
moment is transferred from the bottom Tee to the top Tee).
100


The compressive stress acting on the web post is given by:
s
c
= V
h,eff
/(s
o
t) for s
o
d
o
(circular openings)
for s
o

o
l (rectangular openings)
or:
s
c
= V
Ed
/(d
o
t) for s
o
> d
o
or
o
l (widely spaced openings)
Generally, it is the shear resistance of the bottom Tee which is critical and therefore t is the
thickness of the bottom Tee.
The compressive strength,
c,Rd
, is established from an effective length of the web post, given by:
e
l = 0.5
2
o
2
o
d s + 0.7d
o
for circular openings
e
l = 0.7
2
o
2
o
d s + d
o
for rectangular openings
The effective slenderness is t / 12
e
l = , and the compressive strength
c,Rd
is established from
buckling curve c of Eurocode 3-1-1.
Web post buckling does not occur if:

c

c,Rd


7.4 Web post bending for rectangular openings
The bending resistance at the top or bottom of a rectangular or square opening should also satisfy:
V
h,eff
d
o
/2 M
h,e

This is satisfied by the linking equation for V
Ed
in Section 6.4.
No further web post bending check is required for circular or elongated circular openings, provided
M
h
M
h,e
.

7.5 Closed solution for web post buckling
A closed solution for the maximum shear force controlled by web post buckling should satisfy:
V
Ed

( )
( ) ( )
eff o
o w Rd c, o red b,
h / d e 2 1
s t / M 4
+ +
+ l
+ M
c,s
/s
This applies to rectangular openings and conservatively to circular openings.

101


8 Serviceability Performance
8.1 Additional deflection
The additional deflection due to a simple opening at position x may be given by:
|
.
|

\
|
|
.
|

\
|
|
.
|

\
|
=

L
x
1
L
d
L
k
o o
o
b
add
l
for x 0.5L
where:

b
is the pure bending deflection of the beam (steel or composite beam as appropriate)
and
k
o
is the coefficient obtained from the following table:
Opening type
Beam type
Unstiffened Stiffened
Composite 1.5 1.0
Steel 2.0 1.5

Coefficient, k
o
, used in serviceability calculation
For circular openings:
o
l = 0.5d
o

For multiple similar sized openings: |
.
|

\
|
|
.
|

\
|
=

L
d
L
k N 5 . 0
o o
o o
b
add
l


8.2 Natural frequency
The natural frequency of the beam may be calculated ignoring the influence of the openings.

5.5 Design Software
The engineering logic described above has been implemented within the software provided on a CD
with this report. This software designs and analyses composite and non-composite beams, under
distributed and/or point loads. It also covers web openings which may be circular, rectangular or
elongated oval-shaped. Both internal beams and edge beams in general building construction are
covered.
The program covers the use of the following hot rolled I-sections: UB, UC, IPE and HE (HEA,
HEB and HEM). The beam is assumed to be single spanning and simply-supported at both ends.
Cantilever and continuous construction are not covered in the present version of this program. The
program incorporates options for the use of a solid concrete slab or steel decking acting compositely
with the beam. The floor slab may span parallel to or perpendicular to the beam on one or both
sides. For edge beams ONLY, the beam may also be loaded by a cladding element on the other side.
In all cases, the self weight of the beam and the supported floor is automatically calculated by the
program from the floor layout and cross-section geometry.
102


The program performs a design optimisation on all the sections in the program database for the
selected section type, and lists (in weight order) the lightest 10 sections that satisfy all design
criteria. The software covers:
a) Composite and non-composite construction
b) Use of standard deck profiles, user defined decks or a solid slab
c) Propped and unpropped construction (of both the beam and the decking)
d) Use as an edge or an internal beam
e) Circular, rectangular and elongated openings, with or without stiffeners
f) Additional point loads and additional uniformly distributed loads
g) Use of normal weight and light weight concrete
h) Use of headed shear connectors (including Hilti HVB shear connectors)
i) Groups of shear connectors along the beam
Some selected screen shots from the software are shown in Figures 5-1 to 5-4 inclusive:


Figure 5-1: Title screen


103



Figure 5-2: Construction details

Figure 5-3: Loading data





104



Figure 5-4: Definition of opening positions

5.6 References
1. R M Lawson
Design for openings in the webs of composite beams (SCI P068)
The Steel Construction Institute, 1990
2. J K Ward
Design of composite and non-composite beams (SCI P100)
The Steel Construction Institute, 1990

105





6 Contribution from UNIV LULEA
6.1 General
The work at the Division of Steel Structures, Lule University of Technology (LTU), within this
project concerned mainly Work Packages (WP) 1 and 4:
WP 1: Tests on composite beams with web openings.
Slender web beams with single large web openings subjected to shear.
WP 5: Special cases.
Design of slender web beams
Design of elongated circular openings
Beams in compression including lateral restraints
Design of semi-continuous joints

6.2 Main objectives
Tests done at LTU constitute the basis for the study of slender web beams subjected to shear.
Different sizes of an unstiffened single web opening and the addition of horizontal and vertical
stiffeners has been tested at LTU [10].
The expected buckling phenomena for a slender web beam subjected to shear are;
Local buckling from Vierendeel bending, Figure 6-1A)
Shear buckling, Figure 6-1B)
Web-post buckling, Figure 6-1C)


V
V
V
V V
V
V
V
V
V
A)
B)
C)

Figure 6-1: Local buckling from Vierendeel bending, B) shear buckling,
C) web- post buckling



107


Several factors effect local buckling due to Vierendeel bending of the outstand of the web, Figure
6-1A). The Vierendeel bending moment has a large gradient over the length of the opening; the
flange of the top Tee is connected to the concrete slab and the bottom Tee is subjected to tension
from global bending action. The latter assuming a simply supported beam. In design these effects
should be included in the classification of the unstiffened web outstands. The phenomenon is
prevented by horizontal stiffeners around opening.
Shear buckling, Figure 6-1B), with a web opening is described for bare steel sections in prEN 1993-
1-1 (former ENV 1993-1-1, Annex N). This is too conservative for large web openings. The
suggestion here is that the concrete adds little to the shear buckling resistance and that the same
approach used for bare steel beams should be used for composite cross-sections.
Web-post buckling is a result of Vierendeel bending, Figure 6-1C). The phenomenon is prevented
by vertical stiffeners at opening.
When the floor is in place it acts like a bracing of the top flange that prevents lateral deflection. For
a simply supported beam this means that lateral buckling cannot occur provided that the supports
are sufficiently prevented from rotation. This can normally be assumed to be the case in common
frames. For a continuous girder the bottom flange will be in compression close to the supports and
hence it may buckle laterally.
Semi-continuous joint are described in EC3 1-8 [9]. The main question is at what distance from the
end-plate a large web opening can be placed without interfering with the connection.

6.3 Slender web beams
6.3.1 Test results
Tests 1- 4 consisted of a bare steel section and Vierendeel bending was identified as failure mode
for most. Lder strains were observed at each corner of the openings, i.e. plastic strains suggesting
plastic resistance despite a very slender web. As expected the resistance for Test A1-1 was ca twice
that for Test A1-2, given Vierendeel bending as failure mode with twice the length for the opening.
With horizontal stiffeners the registered resistance increased compared to unstiffened opening but
not in correspondence with the area of the stiffener in Vierendeel bending. Shear buckling and web-
post buckling was not prevented by horizontal stiffeners only. Adding vertical stiffeners prevented
all web buckling.
Tests 5- 8 were composite cross-sections. Tests A2-5 and A2-6 correspond to Tests A1-1 and A1-2
in opening size and steel beam cross-section. Tests B2-7 and B2-8 correspond to Test B1-3 and Test
B1-4. Again Lder strains were observed but also more developed local buckling of the web, shear
buckling, local buckling from Vierendeel bending and web-post buckling. Due to the increased
resistance from the composite slab these local buckling phenomena developed. Table 6-1 shows the
results from the laboratory.







108


Table 6-1: Results from laboratory test at LTU
TEST RESULTS Shear
buckling
Vierendeel
Vl0=2Mvt+2Mvb+Mvc
Web-post
buckling
Test
no.
Applied
V (kN)
Observed
failure mode
EC3
[4]
Elastic
(unreduced)
Plastic
(unreduced)
[5]
A1-1 394 Vierendeel 225.2 212 290 225
A2-5 491 Web buckling 275.2
2)
340 455 225
A1-2 206 Vierendeel 40.1 113 179 275
2)
A2-6 302 Vierendeel
buckling
90.1
2)
238 308 275
2)
B1-4 393 Web buckling 40.1 338 418 225
B2-8 485 Web buckling 90.1
2)
469 593 275
2)
B1-3 443 Vierendeel 713 338 418 -
B2-7 537 Vierendeel 763
2)
468 588 -
2)
Including concrete [5] and [7]

Test B1-4 and B2-8 developed web buckling to the side of the horizontal stiffener. The buckling is
clearly not from Vierendeel but rather a mixed shear and web-post buckling. Calculations for shear
buckling and web-post buckling resistance for the unstiffened webs are conservative for both bare
steel and composite test beams.
Test A2-6 clearly showed local buckling from Vierendeel bending, visible after maximum load was
reached. Test A2-5 buckled locally in a mixed shear/ Vierendeel mode.
Plastic Vierendeel resistance of the Tees was reached for several of the tests. This fact is supported
by the presence of Lder strains during testing.

6.3.2 Shear buckling
The resistance to shear of a web with hole according to Annex N of ENV1993-1-1 is based on
recommendations in [6], which in turn is a development of design rules in [3]. The reduction factor
applied to the shear resistance of the unperforated web
)) 3 , 0 ( 1 (
0 0
w w
h
a
h
h
+ (61)
is based on a model that takes Vierendeel action into account as well as shear buckling. Those
failure modes are interacting in many cases and it may be hard to identify which one is dominant.
For a composite girder the strength for Vierendeel action is increased by the presence of the
concrete slab. This resistance has to be calculated separately and there is accordingly a need for a
separate check of the shear buckling resistance. Equation (61) would be too conservative for this
purpose and computer simulations have been performed in order to identify a more suitable
reduction factor that only takes shear buckling into account. As it is only the steel section that is
considered for the shear buckling resistance the computer simulations were performed without
concrete slab, which makes the calculations much faster. The focus has been on small and medium
size holes because very large holes will always be governed by Vierendeel action. The results are
109


summarized in Table 6-2. V
cR
denotes the shear resistance according to EN 1993-1-5 for the
unperforated web and the resistance considering the hole is given by
) 1 (
0 0
w
cR ocR
h
a h
V V = (62)
This formula has been found by trial and error and it can be seen to give a conservative estimate of
the FE-results [10]. The reduction factor depends on the area of the hole but not on its orientation.
This seems reasonable considering the disturbance of the rotated stress field in the web caused by
the hole. It is clear that the formulae may fail if the size of the hole goes to extremes and the
limitations given in [5] should be respected. A recalculation of the test results, where web buckling
was identified as failure mode, is presented in Table 6-2.

Table 6-2: Test beams
TEST RESULTS Shear
buckling
Vierendeel
Vl0=2Mvt+2Mvb+Mvc
Web-post
buckling
Test
no.
Applied
V (kN)
Observed
failure mode
) 1 (
0 0
w
cR ocR
h
a h
V V =

Elastic Plastic
A2-5 491 Web buckling 410
2)
340 455 225
B1-4 393 Web buckling 155 338 418 225
B2-8 485 Web buckling 205
2)
469 593 275
2)
2)
Including concrete [5] and [7]

Further, Table 6-3 gives V
Vier
which denotes the resistance to Vierendeel action as described in [5].
This clearly depends on the geometry of the hole and it is governing for large holes and also for
long holes. The smallest of V
o,c,R
and V
Vier
will govern the design and represent the resistance. The
last column of Table 6-3 shows the resistance according to FE simulations over the governing
resistance. The failure modes indicated in Table 6-3 have been determined from the deformation
pattern of the FE-model, see examples in Figure 6-2. The top figure is clear case of Vierendeel
failure and the bottom is a clear case of shear buckling. The middle one is deemed as mixed mode.
It can be noted in Table 6-3 that the estimated failure mode does not always coincide with the
governing design criterion. This is not surprising considering the difficulties to judge the failure
mode from the deformation patterns and also the imperfect design models. It should be noted that
the error resulting from mistaking the failure mode will be on the conservative side. If for instance a
failure in Vierendeel action is taken for a shear failure, the actual shear resistance is higher.


110



Table 6-3: Results from parametric study
EC3
ho x lo
(mm)
w
(mm)
V
Vier

(kN)
V
c,R
(kN)
V
o,c,R
(kN)
V
FEM

(kN)
V
FEM

V
o,c,Rd

Failure mode
- 1.1 - 909.3 - Shear buckling
c)

- 1.35 - 610.4 - Shear buckling
c)

- 2.1 - 265.7 - 358.8 Shear buckling
c)

- 3.16 - 130.0 - 182.9 Shear buckling
c)

300 x 150 1.1 779.5 909.3 619.7 - - -
1.35 588.0 610.4 416.0 455.3 1.09 Shear buckling
c)

2.1 344.0 265.7 181.1 234.1 1.29 Shear buckling
c)

3.16 222.7 130.0 88.6 120.0 1.35 Shear buckling
c)

300 x 300 1.1 389.7 909.3 499.7 484.8 1.24 Mixed mode
b)
1.35 294.0 610.4 335.4 368.7 1.25 Mixed mode
b)

2.1 172.0 265.7 146.0 193.8 1.32 Shear buckling
c)

3.16 111.3 130.0 71.4 104.4 1.46 Shear buckling
c)

300 x 450 1.1 259.8 909.3 407.7 397.0 1.52 Mixed mode
b)

1.35 196.0 610.4 273.7 298.3 1.52 Mixed mode
b)

2.1 114.7 265.7 119.1 158.8 1.38 Shear buckling
c)

3.16 74.2 130.0 58.3 88.1 1.5 Shear buckling
c)

300 x 600 1.1 194.9 909.3 330.0 325.6 1.67 Mixed mode
b)

1.35 147.0 610.4 221.6 243.5 1.65 Mixed mode
b)

2.1 86.0 265.7 96.4 131.4 1.52 Mixed mode
b)

3.16 55.7 130.0 47.2 75.1 1.59 Shear buckling
c)

200 x 100 1.1 1617.2 909.3 716.2 - - -
1.35 1258.0 610.4 480.8 - - -
2.1 770.0 265.7 209.3 297.4 1.42 Shear buckling
c)

3.16 512.0 130.0 102.4 155.2 1.52 Shear buckling
c)

200 x 200 1.1 808.6 909.3 636.2 - - -
1.35 692.0 610.4 427.1 - - -
2.1 385.0 265.7 185.9 268.9 1.44 Shear buckling
c)

3.16 256.0 130.0 91.0 142.8 1.57 Shear buckling
c)

400 x 200 1.1 406.2 909.3 523.1 461.2 1.14 Mixed mode
b)

1.35 292.0 610.4 351.2 339.2 1.16 Mixed mode
b)

2.1 158.0 265.7 152.9 183.8 1.2 Mixed mode
b)

3.16 99.0 130.0 74.8 93.9 1.26 Shear buckling
c)

400 x 400 1.1 203.0 909.3 363.2 323.0 1.59 Vierendeel
a)

1.35 146.0 610.4 243.8 234.4 1.6 Shear buckling
c)

2.1 79.0 265.7 106.1 122.3 1.55 Shear buckling
c)

3.16 49.5 130.0 51.9 67.2 1.36 Shear buckling
c)




111





Figure 6-2: Failure modes obtained in the parametric study, the deformations are
enlarged three times. a) Vierendeel bending,
b) combined Vierendeel bending and shear buckling,
c) shear buckling
6.3.3 Vierendeel bending
The classification of the outstand web in [5] is equivalent to EC 3 1-1 [4] outstand flanges:
d < 10t for class 2
d < 13t for class 3
Class 3 and 2 limit originate from buckling factor k

= 0.425, i.e uniform compression. The LTU

test results show this to be a conservative classification of the Tees. There are three positive effects
to consider:
The effect from global moment, tension in bottom Tee.
The effect from concrete slab for the top Tee in a composite cross-section
The effect from a large moment gradient in Tee
The positive effect from tension in the bottom Tee can be considered by calculating the stress
distribution from combined Vierendeel and global bending. The compressed portion of the web-tip
is then used in the classification [5]. The negative effect from compression in the top Tee can be
a)
b)
c)
At ultimate load
After ultimate load
After ultimate load
After ultimate load
At ultimate load
At ultimate load
112


done in the same way resulting in a stress distribution where uniform compression in the out stand
web is a safe assumption.
The restraint of the web-tip by the concrete slab can be included in classification of the Tee by
considering the web as fixed. With this assumption the buckling factor is k

= 1.25 for uniform

compression. With the same slenderness limits the classification becomes:
d < 17t for class 2
d < 22t for class 3
Whether or not fixed conditions apply depends of course on the flange and number of shear
connectors. For the tests here with pairs of connectors in each rib this is clearly the case.
There is also the moment gradient, which is quite large for Vierendeel bending of the Tees. This
gives a positive effect but is not included in the classification.
For the LTU composite beams the classification of the Tees becomes class 3 instead of class 4.

6.3.4 Web-post buckling
From test results it is clear that the buckling curve for resistance is conservative with the approach
described in [5]. Using plate buckling curve on the other hand shows too optimistic results using the
same effective width of the web. Another interesting approach could be to treat the phenomenon as
end patch loading but the effective width of the web needs in this case further studying.
An observation from the parametric study, described for shear buckling, is that the web-post failure
mode does not occur in the computer simulations. Also in [3] this failure mode was not considered
as separate failure mode. It seems likely that this failure can be ignored for isolated holes in thin
webs. In this context isolated may be taken as a minimum distance of 1,5h
w
between adjacent edges
of two nearby holes [3]. However, if the hole is provided with longitudinal stiffeners, the web post
failure mode should be checked because of the increased shear resistance.
There is an effect from asymmetry for a composite beam with respect to applied loads on the web-
post.

6.4 Elongated circular openings
For the design of elongated circular openings in slender web beams the same methods as for
rectangular openings is studied using an effective length, l
o
. Simulations of steel beams with
elongated circular openings are compared to simulations with rectangular openings with the
expression for the effective length described in Figure 6-3 [5].

d
o

l
o

l
o
=s+0.5d
o

s

Figure 6-3: Effective length l
o
of elongated circular openings
The simulations show very good agreement between rectangular and circular openings using this
approach [10].
113


6.5 Beams in compression including lateral restraints - Lateral torsional buckling
of composite beams with web openings
The bottom flange in a continuous girder will be in compression close to the supports and hence it
may buckle laterally. Rules for checking LTB of such beams are found in Eurocode 4-1-1[1] called
the inverted U-frame model. This model includes the favourable effect of rotational restraint of the
top flange. The rules are referred in the text below and modified such that they are applicable also
to webs with openings. Limitations of the applicability of the rules are given in Eurocode 4-1-1 but
in most practical cases they are fulfilled. In the unlikely case of a long opening close to a support
the bottom flange may buckle vertically. This can be checked with the conservative assumption of
buckling length equal the length of the opening.
In checks for lateral stability of beams built unpropped, the bending moment at a composite cross-
section in Class 1 or 2 is the sum of the moment applied to the composite member and the moment
applied to its structural steel component. For cross sections in Class 3 and 4 effects of sequence of
construction should be taken into account. For such beams it is simplest to compare the maximum
compression stress at support caused by the design loads with
0
/
M y LT
f . The compression stress
at support should be calculated considering sequence of construction and for Class 4 sections also
effective areas according to Eurocode 3-1-5. In Eurocode 4-1-1 the rules are given for Class 1-3 but
they may be applied also for Class 4 as described here.
The design buckling resistance moment of a uniform composite beam with Class 1 or 2 cross-
sections is given by
M
b,Rd
=
LT
M
Rd
(63)
where :

LT
is the reduction factor for lateral-torsional buckling depending on the relative slenderness
LT
according to Eurocode 3-1-1. There are two sets of reduction curves at it is a national
decision which to use.
M
Rd
is the design resistance moment under hogging bending determined according to Eurocode
4-1-1, 6.2.1.2 for a beam with full shear connection, or 6.2.1.3 for a beam with partial shear
connection.
The relative slenderness
LT
may be calculated by :

LT

cr
y
cr
Rk
f
M
M

= = (64)
where :
M
Rk
is the resistance moment of the composite section using the characteristic material
properties;
M
cr
is the elastic critical moment of the composite section for lateral-torsional buckling. For a
composite beam assumed continuous at one or both ends, M
cr
should be determined at the
internal support where the hogging bending moment is maximum;

cr
is the stress in the compression flange caused by M
cr
acting on the cracked composite
section.
The calculation of the elastic critical moment, M
cr
, may be based on the "continuous inverted U-
frame" model. As shown in Figure 6-4, this model takes into account the lateral displacement of the
bottom flange causing bending of the steel web, and the rotation of the top flange which is resisted
by bending of the slab.

114



Figure 6-4: Inverted-U frame resisting to lateral-torsional buckling
At the top-steel flange level, a rotational stiffness,

c , per unit length of steel beam may be adopted

to represent the effect of the U-frame:

2 1
2 1
k k
k k
c
+
=

(65)
where:
k
1
is the flexural stiffness of the cracked concrete or composite slab in the direction transverse
to the steel beam, which may be taken as:
k
1
= E
a
I
2
/ a (66)
with = 4 for a slab continuous across the steel beam, and = 2 for a simply supported
slab;
a is the spacing between the parallel beams;
E
a
I
2
is the "cracked" flexural stiffness per unit width of the concrete or composite slab, as
defined in 5.4.2.2 of Eurocode 4-1-1, where I
2
should be taken as the lower of the value at
midspan, for sagging bending, and the value at the supporting steel member, for hogging
bending;
k
2
is the flexural stiffness of the steel web, to be taken as:

hole
s a
w a
h
t E
k
) 1 ( 4
2
3
2

= (67)
L
L
hole
hole
2
1 = for a single opening according to Figure 6-5.
s
d
hole
4
3
1 = for multiple circular openings of diameter d and spacing s.

span L
L
open

Figure 6-5: Continuous composite beam with opening in the web
115


The critical bending moment can be calculated from the following formula (68)

(
(

+ =
eff T
D
z
cr
GJ
L
J E
k
M
,
2
2
) (
1

(68)
where:
T eff T
GI A GJ ) 5 , 0 5 , 1 ( ) ( =
GI
T
St Venant torsional rigidity of the steel section
A, coefficients depending on the moment distribution, see Figure 6-7
EJ
D
= E(J
a
+
2
D
z J
az
)
J
a
the warping section constant of the steel section
J
az
the second moment of area of the steel section around the vertical axis
is given in Figure 6-7 as function of

D
EJ
L c

4
= (69)

y st
ay
Mz D
e
P M D
z
J
J
r z
z
i z z
k
,
2 2
2
) (
(

+
+ +
= (610)

a st
ay
e
A z
J
z = (611)

M
y
Mz
z dA z y z
I
r 2 ) (
1
2 2

|
|
.
|

\
|
+ =

(612)

a az ay p
A J J i / ) (
2
+ =
A
a
is the area of steel section
J
ay
is the second moment of area of the steel section around the horizontal axis
J
st,y
is the second moment of area of the composite section neglecting concrete around the
horizontal axis.
Other notations are defined in Figure 6-6.


116



Figure 6-6: Notations for calculation of M
cr
. S denotes the shear centre and M the
centre of gravity of the steel section
A simplified rule is also given in Eurocode 4-1-1 stating that a check of LTB is not needed if certain
conditions are satisfied. The rule has here been modified to account for openings in the web. The
steel member should be an IPE section or a HE section or another hot-rolled or welded section of
similar shape with A
w
/ A
a
0.45, the same depth h, and satisfy the condition:

4 4
3
10

|
|
.
|

\
|
hole f
f
w
s
b
t
t
h
(613)
where :
A
a
is the area of the structural steel section;
A
w
is the area of the web.
Additional notations are given in Figure 6-4.

117






Figure 6-7: Effective length factor as function of restraint parameter for
various moment distributions




118


6.6 Semi-continuous joints
A semi rigid joint is characterized by its strength, stiffness and deformation capacity. A hole in the
web close to the end of the beam may influence these characteristics, a fact that could be taken into
account. However, this would be a quite elaborate task and the result would most likely be too
complicated for practical use. Instead the aim was finding limitations for the placing of the hole
such that the joint properties are not affected.
Computer simulations of a beam to column connection have been undertaken [10]. The simulations
show no notable influence from the hole on the behaviour of the joint. This would imply that the
hole may be placed as close as 0,75h
w
from the end of the beam without influencing the joint or
even closer if the hole is large. As the number of simulations is limited a safety margin is
reasonable. The recommendation should therefore be that the edge of the hole should be not closer
to the end than h
w
.







Figure 6-8: Limits with respect to joint for an unstiffened opening
If there is a need to place the hole closer to the end a vertical stiffener is suggested at the edge of the
hole closest to the end. As an engineering judgement the stiffener should be taken as single sided
with an area of 0,2A
w
and the distance to the end should not be less than 0,5h
w
.

Figure 6-9: Limits with respect to joint for a stiffened opening



0.5h
w
h
w
Vertical stiffener

h
w
h
119


6.7 References
[1] Eurocode 4-1-1 Design of composite steel concrete structures. Part 1-1: General rules and rules
for buildings, EN 1994-1-1: 2004.
[2] Hanswille, G., Lateral Torsional Buckling of Composite Beams. Comparison of more accurate
methods with Eurocode 4, Proceedings of the Engineering Foundation Conference Composite
Construction IV, Banff, 2002.
[3] Hglund, T., Brfrmga hos tunnvggig I-balk med cirkulrt eller rektangulrt hl i livet.
Meddelande nr. 87, Institutionen fr byggnadsstatik, KTH, Stockholm, 1970.
[4] Eurocode 3-1-1 Design of steel structures. Part 1.1: General rules and rules for buildings,
prEN 1993-1-1 : November 2003.
[5] SCI, Design of composite beams with large openings for services. Document RT959, London,
2003.
[6] Steifenlose Stahlskelettragwerke und dnnwandige Vollwandtrger, Berechnung und
Konstruktion, Europische Konvention fr Stahlbau (EKS), Verlag von Wilhelm Ernst & Sohn,
Berlin 1977.
[7] Eurocode 2-1-1 Design of concrete structures. Part 1: General rules and rules for buildings,
prEN 1992-1-1: July 2002.
[8] Eurocode 3-1-5, Design of steel structures. Part 1-5: Plated structural elements, prEN 1993-1-5:
September 2003.
[9] Eurocode 3-1-8, Design of steel structures. Part 1-8: Design of joints, prEN 1993-1-8:
November 2003.
[10] Part 2 of this final report.


120
European Commission
EUR 21345 Steel products and applications for building, construction and industry
Large web openings for service integration in composite floors
C. Mller, O. Hechler, A. Bureau, D. Bitar, D. Joyeux, L. G. Cajot, T. Demarco, R. M. Lawson,
S. Hicks, P. Devine, O. Lagerqvist, E. Hedman-Ptursson, E. Unosson, M. Feldmann
Luxembourg: Office for Official Publications of the European Communities
2006 120 pp. 21 29.7 cm
Technical steel research series
ISBN 92-79-01723-3
ISSN 1018-5593
Price (excluding VAT) in Luxembourg: EUR 35