(¡OST European cooperation n the field of scientific and technical research

Urban civil engineering

Improvement of buildings7 structural

quality by new technologies
Proceedings of the international seminar
Lisbon, 19 and 20 April 2002

ëDoTÀstim Ê$^
* * *
*

* j.

EUR 20728
EUROPEAN COMMISSION
Directorate-General for Research
European research area: Structural aspects COST
Contact: Dr. Ilias Samaras (Scientific Officer)
European Commission, Office SOME 09/81
B- 1049 Brussels - Tel (32-2) 296 55 12 - Fax (32-2) 296 42 89
E-mail: ilias. samaras @cec. eu. int
Web site: http://cost.cordis.lu
 European Commission

Directorate-General for Research

COST Action C1 2

Improvement of buildings' structural

quality by new technologies

Proceedings of the international seminar

Lisbon, 19 and 20 April 2002

Edited by

Luís Simões da Silva and José F. G. Mendes

Universidade de Coimbra and Universidade do Minho, Portugal

Scientific Committee

Gerald Huber (A)

Jean-Pierre Jaspait (Β)
Federico Mazzolarli (I)
José F. G. Mendes (Ρ)
Christian Schaur (A)
Luís Simões da Silva (Ρ)

2003 EUR 20728

 Europe Direct is a service to help you find answers
to your questions about the European Union
New freephone num»ber:
00 8006 789 1011

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 of the European Union is available on the Internet.
It can be accessed through the Europa server (http://europa.eu.int).

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

Luxembourg: Office for Official Publications of the European Communities, 2003

ISBN 92-894-5684-1

© European Communities, 2003

Reproduction is authorised provided the source is acknowledged.

Printed In Belgium

PRINTED ON WHITE CHLORINE-FREE PAPER

 FOREWORD

The Parque das Nações, in the city of Lisboa, has been chosen to host the first COST
C12 International Seminar, which took place on 19 and 20 April 2002. The
exceptional conditions offered by the former EXPO'98 site, together with the strong
involvement of Portuguese researchers within this COST Action, were the reasons to
bring the organisation of this high level Seminar to Portugal.

Several state-of-the-art papers have been planned, elaborated and presented by the
active experts of the three working groups: Mixed Building Technology (MBT),
Structural Integrity under Exceptional Actions (SI), and Urban Design (UD). The
present proceedings include fifteen papers approved by the International Scientific
Committee.

In two days of presentations, the latest research on the improvement of building's

structural quality by new technologies was discussed. In addition to the working
groups members' presentations, three invited speakers offered very interesting and
updated presentations, as well included in this volume. These are the paper of G.
Olivete on the collapse of the Twin Towers, the paper of V. Gioncu on ductility
design of steel structures, and the paper of L. Caldas on generative systems for low-
energy architecture design.

The Organising Committee wishes to thank Jean-Pierre Jaspart and Gerald Huber,
respectively Chairman and vice-Chairman ofthe COST C 12 Action; the Chairmen of
the Working Groups, C. Schaur (MBT), F. Mazzolani and D. de Matteis (SI) and J.
Mendes (UD); the invited speakers, G. Olivete, V. Gioncu, and L. Caldas; and all the
participants for their helpful assistance, both in the realisation of the Seminar and in
the preparation of this volume. Finally, special gratitude is addressed to Ilias Samaras,
Scientific Secretary ofthe COST C12 Action and to the secretariat staff for all their
efforts.

The realisation of the Seminar and the publication of this volume have been
financially supported by the European Commission.

Luís Simões da Silva

José F. G. Mendes

Local Organizing Committee

Dedicated to Martin Steenhuis

111
 COST C12

CHAIRMAN: Jean-Pierre Jaspart

VICE-CHAIRMAN: Gerald Huber

WORKING GROUPS (Chairmen)

WG1 Christian Schaur

WG2 Federico Mazzolani and Gianfranco de Matteis
WG3 José Mendes

IV
 TABLE OF CONTENTS

Preface

Jean-Pierre Jaspart, Université de Liège, Belgium ...ix

Chairman of COST C 12

WGl - Mixed Building Technology

Structural behaviour in mixed buildings 3

E. FOURNELY, Blaise Pascal University, Clermont-Ferrand (FR), D.
DUBINA, L.A. FULOP, University 'Politehnica' Timisoara, Timisoara
(RO), Z. PLEWAKO, Rzeszow University of Technology, Rzeszow (PL), U.
KUHLMANN, M. SCHAFER, University of Stuttgart, Stuttgart (D), J.
HAMM, Swiss School of Engineering for the Wood Industry (CH) and H.
TRUMPF, University of Aachen, Aachen (D)

Shear transfer in composite members: testing, modelling, standards and damage.. 21

O. BURSI, M. BALLERINI, M. PIAZZA, R. ZANDONINI, University of
Trento, Trento (IT), E. FOURNELY, Blaise Pascal University, Clermont-
Ferrand (FR), U KUHLMANN, K. KÜRSCHNER and J. SCHANZLIN,
University of Stuttgart, Stuttgart (D)

Repairing and strengthening for new requirements: use of mixed technologies .....37
A. MANDARA, Second University of Naples, Napoli (TT), J.P. MUZEAU,
Blaise Pascal University, Clermont-Ferrand (FR), P. PERDIKARIS,
University of Thessaly, Volos (GR), M. PIAZZA, University of Trento,
Trento (TT) and C. SCHAUR, A-MBT Application Centre Mixed Building
Technology Ltd., University of Innsbruck, Innsbruck (AT)

Service life and life cycle of building structures 55

R. BLOK, F. van HERWIJNEN, University ofEndhoven, Eindhoven (NL), A.
KOZLOWSKI, S. WOLINSKI, Rzeszow University of Technology, Rzeszow
(PL), H. GERVASIO, Gipamb Consulting Engineers Ltd., Lisboa (PT) and
L. SIMÕES DA SILVA, University of Coimbra, Coimbra (PT)

WG2 -Structural Integrity under Exceptional Actions

Structural integrity under exceptional actions: basic definitions and field of

activity 67
F. MAZZOLANI, University of Naples "Federico II", Napoli (TT)

Twin Towers: qualitative analysis of collapse behaviour 81

G. OLLVETO and A. MARINETI, University of Catania, Catania (TT)
Recent development in ductility design of steel structures 99
V. GIONCU, University 'Politehnica ' Timisoara, Timisoara (RO)

General methodologies for evaluating the structural performance under

exceptional loadings 113
M. FISCHINGER, P. FAJFAR, M. DOLSEK, University of Ljubljana,
Ljubljana (SL), D. ZAMFIRESCU, Technical University of Civil
Engineering, Bucharest (RO), A. STRATAN, "Politehnica" University of
Timisoara, Timisoara (RO), T. LEINO, VTT Building Technology, Helsinki
(FI), R. VACAREÄNU, T. CORNEA, D. LUNGU, Technical University of
Civil Engineering, Bucharest (RO), S. WOLINSKI, Rzeszow University of
Technology, Rzeszow (PL) and E. ÁLARCON, Polytechnic University of
Madrid, Madrid (ES)

Structural integrity of buildings under exceptional earthquakes 127

ESMISTAKIDIS, University of Thessaly, Volos, (GR), V. BOSILJKOW,
Slovenian National Building and Civil Engineering Institute, Ljubljana (SL),
G. de MATTEIS, University of Naples "Federico ΙΓ, Napoli (IT), F. DINU,
Romanian Academy, Timisoara (RO) and M FISCHINGER, University of
Ljubljana, Ljubljana (SL)

Structural integrity of buildings under exceptional fire 143

F. WALD, Czech Technical University in Prague, Prague (CZ), V.
BOSILJKOW, Slovenian National Building and Civil Engineering Institute,
Ljubljana (SL), G. deMATTEIS, University of Naples "Federico II", Napoli
(IT), P. HALLER, Technical University Dresden, Dresden (D), A.
SANTIAGO, L. SIMÕES DA SILVA, University of Coimbra, Coimbra (PT)
and P. VILA REAL, University of Aveiro, Aveiro (PT)

Evaluation of structural robustness of members and connections 155

L. SIMÕES DA SILVA, L. NEVES, University of Coimbra, Coimbra (PT), L.
BANIOTOPOULOS, P. PERDIKARIS, M. ZYGOMALAS, Aristotle
University of Thessaloniki, Thessaloniki (GR), V. BOSILJKOW, Slovenian
National Building and Civil Engineering Institute, Ljubljana (SL), H.
BOUCHAIR, Blaise Pascal University, Clermont-Ferrand (FR), G. de
MATTEIS, University of Naples "Federico II", Napoli (LT), D. DUBINA,
"Politehnica" University of Timisoara, Timisoara (RO), P. HALLER,
Technical University Dresden, Dresden (D), U KUHLMANN, University of
Stuttgart, Stuttgart (D), H. STANGENBERG, University of Aachen, Aachen
(D), G. HUBER, Aste Engineering, Innsbruck (AT), A. KOSLOWSKI,
Rzeszow University of Technology, Rzeszow (PL) and F. WALD, Czech
Technical University in Prague, Prague (CZ)

WG3 - Urban Design

A generative design system for low-energy architecture design 177

L.G. CALDAS, Instituto Superior Técnico, Lisboa (PT)

The existing buildings in the future 195

J. ENGELMARK, Technical University of Denmark, Kopenhagen (DK)

VI
Large-panel dwelling building, current state - ways of construction and
architectural modification 207
A. RYBKA, Rzeszow University of Technology, Rzeszow (PL)

Comparative response of lightweight and traditional building solutions with

respect to thermal, acoustical and visual comfort functional demands 217
L. BRAGANÇA, M. ALMEIDA, J. MENDES, Universidade do Minho,
Guimarães (PT) and L. SIMÕES DA SILVA, Universidade de Coimbra,
Coimbra (PT)

vu
 PREFACE

It is now recognised throughout Europe that there is a special need for higher quality
in the field of urban buildings. This requires the development of new and suitable
strategies for architects, sociologists, urban planners, local authorities and engineers.

New technologies have been developed in many countries in Europe and these have
not yet been transferred into practice to achieve higher quality, economy and safety in
urban buildings as well as a lower impact on urban environment.
A number of technical fields of development can be identified as being appropriate to
this transfer, but the two following ones are nowadays mainly receiving the attention
of engineers and researchers:
the Mixed Building Technology (MBT);
the Structural Integrity of urban buildings under exceptional actions (SI).

The situation for the development of MBT in Europe is now excellent. Fundamental
connection problems have been successfully solved, both for different material
technologies in isolation and in combination. Evidence for this can be found by way
of the deliverables following the COST CI project on "Control of the Semi-rigid
Behaviour of Civil Engineering Connections". Furthermore, the unified safety and
action concept for most construction methods has been agreed at the European level
and implemented in the structural Eurocodes, published by CEN.
The effect of normal actions on buildings is extensively covered by the existing
Eurocodes. Many structures are, however, exposed to certain extreme actions arising
from natural and human-made hazards, such as: fire, earthquakes, wind storms,
explosions, exceptional snow loading. Robustness is the capability of the building to
remain stable even when one of its parts is heavily damaged, for instance because of a
local explosion on one storey or the impact of a truck against the structure.
Robustness is a strong requirement as far as human integrity of inhabitants and
neighbours is of concern, and presently no design guidelines are provided by the
structural Eurocodes.

But the smooth and efficient transfer of these new technologies within the specific
field of urban habitat requires a strong link between all the partners involved in the
construction business, such as urban planners, local authorities, architects, engineers,
. . .one could speak about Urban Design (UD) using MBT.

Expertise on each of these topics exists locally in many European countries, but till
now no integrated knowledge combining MBT, SI and UD was available, which
could lead to new urban structures meeting the imposed structural requirements and
the highest non-structural demands ofthe people using them. In this context COST is
the ideal platform to have a multidisciplinary Action allowing to integrate these new
technologies in the field of urban buildings and to disseminate understanding,
knowledge and a unified technical vocabulary throughout Europe.

The main objective of the four-year COST C12 Action is therefore to develop,
combine and disseminate new engineering technologies, to improve the quality of
urban buildings, to propose new technical solutions to architects and planners, to

IX
reduce the disturbances of the construction process in urban areas and finally to
improve the quality of living in the urban habitat.

According to the Memorandum of Understanding of the COST C12 Action, a first

state-of-the-art seminar had to be organised after the first one and a half year of
activities in order to discuss the research progress in each of the three Working
Groups (MBT, SI and UD) and to organise the planning of activities for the following
research steps. In the present proceedings, the reader will find the text of all the
presentations made during this first seminar.

Portugal, and more especially Lisbon, has been chosen for the location of this seminar
because of the possibilities offered by the Expo'98 site but also due to the strong
involvement ofthe Portuguese representatives within the COST C 12 Action.

I wish to thank warmly all the participants in this seminar and to

express my extreme
gratitude to L. Simões da Silva and J. Mendes for the successful organisation of the
seminar, as well as to the chairmen of the Working Groups for the perfect
management of their sessions: C. Schaur (MBT), F. Mazzolani and G. De Matteis (SI)
and J. Mendes (UD) respectively.

Jean-Pierre Jaspart
Chairman COST C 12
WGl - Mixed Building Technology
Structural Behaviour in Mixed Buildings

E. Fournely J. Hamm
Lermes Laboratory, CUST Clermont-Ferrand (F) Swiss School of Eng. for the Wood Industry (CH)

D. Dubina, L.A. Fulop H. Trumpf

Department of Steel Structures and Structural Mechanics Institute of Steel Construction, RWTH Aachen
(GER)
Politehnica University of Timisoara (RO)
Z. Plewako
Dep. of Building and Enviromental Engineering
Rzeszow University of Technology (POL)

U. Kuhlmann, M. Schäfer
Institute of Structural Design, University of Stuttgart (GER)

KEYWORDS: sway-frames, ductile, semi-rigid, timber-glass, glue, composite, hybrid connections,

temperature, FRP, bonded connections, concrete structures, FRP reinforcement, Timber-concrete, static, cycle
behaviour, shear specimens, light-weight, mixed shear wall, cold-formed, sheeting, gypsum board, earthquake

1. INTRODUCTION New local and global approaches for timber-

This chapter presents the contributions of six
universities investigating the structural behaviour of
mixed building technologies. These research
programs show a big variety of possible combination
of two or more materials.
Traditional combinations, e.g. steel and concrete,
are improved and supplemented by new intelligent
solutions. In general new sophisticated systems, such
as timber and glass or timber and concrete, offer
advantages for specific applications with a
characteristic load-bearing behaviour.
New materials like FRP, reinforced or pultruded,
change traditional designs and need new solutions
for calculation.
Special regard in these investigations is taken to
dynamic loading for lightweight steel structures in
seismic regions.
concrete structures are necessary to refurbish and
design new buildings.
Lightweight mixed shear wall solutions using
corrugated sheets or OSB-panels instead of bracing
present good seismic behaviour and ductility.
Recent applications show that these mixed
building technologies are practicable, economic and
successful in their performance.
Finding the best combination of materials for the
structural behaviour and building physics with a
special regard to the boundary conditions and
demands is the main challenge for the modern
innovated engineer.
2. RESEARCH ON NEW MTXED BUILDING
TECHNOLOGIES

By combining the benefits of semi-rigid joints the 2.1 Steel and concrete (sway buildings and semi¬
fabrication costs are reduced with the advantage of rigid composite joints)
plastic design for frames.
Composite systems like timber and glass offer 2.1.1 Introduction
light structures and a sufficient load-bearing
capacity.
Using pultruded FRP-profiles high strength and The idea is to combine the benefit of semi-rigid
lightweight structures are obtained with superior joints in reducing fabrication costs by omitting
corrosion resistance and thermal insulation. FRP additional stiffeners and web-plates (see Fig. 1) with
reinforcement out of glass, aramide or carbon makes the advantage of plastic design for frames.
it possible to create structures with specified
material properties.
 column composite sway-frames. Provided that the elastic
HE 600 8
sagging moment resistance ofthe composite girder is
bolts 3x4 M 24 10.9
, ^.14 .14 not exceeded, the required rotational capacity
¿±ii 5
V -V
reaches about 30mRad. High ductile semi-rigid
f
30
composite joints are developed to fulfill these
additional
web-plate
rotational requirements.
beam HE 800 A
.5 t = 15 S. N y 2. 1.3 Ductile semi-rigid composite joints
E>.5-H^S j

f
30
^"6 Afl h2n
2.1.3.1 General considerations
A new ductile composite joint is tested, see Fig 2
(Kuhlmann, Schäfer) and guidelines how to create
ductile semi-rigid composite joints are developed.
For the rotational capacity of a joint the
deformability of the yielding components is of
crucial importance. One possibility is to design the
joint in order to provoke the failure of a ductile
bolts 3x2 M 30 10.9 component, e. g. by increasing the reinforcement so
that the column web in compression yields instead of
the reinforcement. On the other side, a higher
rotational capacity of the joint can be achieved by
beam HE 800 A
increasing the ductility ofthe yielding component (e.
> g. using ductile reinforcement), see the following
examples:

ch en to 0.7 m
Tension tie 1.8m

'-. a
Figure 1. Full strength joint and semi-rigid joint EPT

Especially composite girders profit of the plastic

moment redistribution from the hogging moment at
the column to the sagging moment within the girder.
However this plastic moment redistribution requires
a specific rotational capacity of the joints, where the
0.7
plastic hogging moment primarily develops. The m
required rotation capacity is especially high for
sway-frames due to their relatively high deformation.
IS. ,8» 4& ¡SS" 'ìx^£^ ί^
Therefore some special considerations have to be
given to the increase of ductility in semi-rigid joints
in composite sway-frames.

2.1.2 Composite sway-frames

The elastic design of frames requires joints that
possess a large moment resistance due to the high
hogging moments. Adopting plastic design for Figure 2. High ductile and standard composite joint
systems with partial-strength joints, a moment
redistribution from the hogging moment to the
sagging moment occurs, resulting in a necessary 2.1.3.2 Reinforcement in tension (RFT)
plastic rotation in the joint. It has to be ensured that The length of the component "Reinforcement in
the rotation that is required by the system is smaller tension" is expanded by arranging a space between
than the available rotation of the joint. A parameter the first shear stud and the column, see Fig. 2. This
study on three story sway-frames with different is studied in nine tests (Kuhlmann, Schäfer). Using
spans and loadings is carried out to calculate the reinforcement with a strengthening factor fu/fy>l,14
required rotational capacity of semi-rigid joints in
also increases the plastic deformability of the
reinforced concrete.
2.1.3.3 Endplate in bending (EPT), Column flange in
bending (CFT)
Large bolt diameter in comparison to the endplate
should be chosen so that no early failure of the bolt
occurs. The provisions according to Annex J (J.49)
(Eurocode 3 1988) are sufficient to reach the plastic
moment ofthe joint but may not be enough to assure
large deformations, due to a high strengthening
factor f/fy of the endplate in comparison to the bolt,
and due to membrane effects forcing the bolt to
bend. Arranging the bolts not too close to the beam
flange and to the beam web, increasing the width of
the endplate and larger bolt diameter lead to much
higher ductility, see Figure 3.
·
·
·
Figure 3. High ductile and standard endplate connection
2.1.3.4 Other components (CWT, BFT, BFQ CWC)
Other components as column web in tension (CWT),
beam flange in tension (BFT), beam flange in
compression (BFC) are ductile components
(Kuhlmann et al. 2001), (Kühnemund 2002). The
component column web under compression (CWC)
may be classified ductile for HEA and HEB profiles.
HEM profiles provide less ductility but a sufficient
strength so that no early failure of the component
CWC occurs (Kuhlmann et al. 2001), (Kühnemund
2002).
2.1.4 Results
The tests results have shown that it is possible to
create high ductile semi-rigid composite joints with
rotational capacities larger than 60mRad. These
composite joints fulfill the rotational requirements of
the analysed composite sway-frames.
2.2 Timber and glass
_ _ , _ ,
2.2.1 Introduction
Timber and glass are often used in combination in
greenhouses, facades and other structures. In
traditional (not composite) timber and glass
construction, the beam-sections are often so large
that, they may create problems with regard to the
aestetics and natural lightning. A timber-glass
composite element has been developed where the
wooden frame is directly glued to the glass-plate.
This kind of composite structure has the following
advantages:
light structure:
A comparison between the traditional and the
composite structures shows that the wooden
sections are much smaller with timber-glass
composite structures.
prefabricated elements:
Timber-glass composite elements can be
mounted quickly on the building site.
an optimum use of material:
timber under tension, glass under compression
constructive wood preservation
Timber is protected by the glass-plate on the
outside.
The timber-sections prevent the brittle-failure of
the composite section
The work has been carried out in the following
stages:
tests
After the suitable glue was found, 4-point
loading tests on timber-glass composite
structural elements were carried out.
different methods for the design
design method based on safety concepts
Design methods based on EC 1 are proposed for
the different composite structures.
2.2.2 Tests
2.2.2.1 The glue
To find the right glue for the timber-glass composite
structure, several small specimens were tested under
shear compressive loading. Because the composite
elements are normally exposed to climatic
variations, the test specimens were submitted to
several climatic cycles. One cycle consisted of 4
hours at -30°C, 4 hours at +70°C, 16 hours at +30°C
and 80% humidity. This cycle was repeated up to
seven times and specimens were tested after 0-cycle,
2-cycles, 5-cycles and 7-cycles to evaluate the
possible degradation. Of the four glues which were
tested, only one exhibited adequate resistance.
2.2.2.2 I-beam 2.2.3 Final remarks
The I-beams were also tested under 4-point loading.
All test specimens showed a similar behavior. Long An adhesive has been found which
before the failure ofthe test beam, the glass began to withstands both temperature cycles and dampness
crack in several places. The cracking occurred in variations in the environment of the timber-glass
regular time intervals as the load increased. The first composite structure.
cracks were registered at a load of approx. 4 kN. An The final failure of the timber-glass
average of nine successive cracking steps developed composite structure announces itself well in
before failure occurred. The regularity of the failure advance, which represents a substantial safety
mode was uncanny (fig.4) aspect.
The research work has shown the possibility
for a practical application and demonstrated the
^
Jf
+ efficiency of this construction.

)
_ :
\ ,
ι
/ #^ íK "Ç^^^f ψ Λ
y \ * * " 2 2»ν '
ψ
.* 1

/ 2.3 Hybrid Connections out of Fibre-Reinforced-

it Polymer and Steel
» î»i
_E;1_>

Jr 131.7 j ni.,
t uu y 2.3.1 Introduction

Fig. 4 Groups of cracks in a tested I-beam FRP-profiles comprise a large amount of advantages
such as longevity, minimal maintenance problems,
The first cracks developed between the places where inhibited fire properties, thermal insulation and a fast
the forces were applied. Only later did cracks and flexible erection by using these light-weight
develop outside of this area. Towards the end of the elements. Taking into account the durability and
loading tests the groups of cracks had an average maintenance FRP structures are even more economic
distance of 25 cm from each other. than conventional designs. For business- or
The force-deformation curve shows a kink when the residential buildings the facade can be integrated in
cracks occurred. This phenomenon can be attributed the bearing structure by reaching optimized heat
to the fact that with the development of cracks the insulation.
total flexural rigidity of the laminate beam decreases For transfering high internal forces in multi-storey
(fig.5). buildings or halls, the FRP-profiles have to be
strengthened in the area ofthe bolts with e.g. bonded
steel plates due to the lower shear resistance of the
matrix.
11000
9 For these hybrid connections stainless steel, material
10000 . 8 , number 1.4571, has been used to guarantee the
7 rS durability and reliability against corrosion. The
9000 .
6 a/ applied bolts M16 and M20 are also made of
8000 .
5 f^ stainless steel, material number 1.4571, with a
7000 .
4 y^ material strength of 8.8.
3 /
Force 6000 .
/
2.3.2 Hybrid connections
(N) 2
5000 -

1 / Beginning of the cracks Due to the low shear resistance of pultruded FRP-
4000 profiles and the high shear forces in the connection
3000 . areas the ends of the profiles are reinforced by
bonded stainless steel plates. Bonded connections
2000 .
are well known and a wide variety of different
1000 . adhesives based on 1 or 2-component epoxy resins,
Polyurethans, acrylats, silicones etc. exist on the
0
market. Furthermore a big amount of papers have
,]
() 10 20 30
Deformation(mm) been published in the last decades mainly about the
chemical composition and their modes of action,
linkage forces (e.g. adhesion, cohesion, etc.),
calculation models and test specimen, see e.g. (1)
Fig. 5 load-deformation curve of a timber-glass composite
Habenicht.
I-beam
For simple overlapping joints with hybrid materials
under shear forces a basic approach to determine the
shear stress distribution in the joint is the equation
by Volkersen (2): The internal forces in the bolts are transfered by
shear and bearing in the stainless steel plate. The
adhesive joints transmit these forces by means of
shear stresses into the FRP-profile. The maximum
cosh·* (l-»/)-sinh -jc peak of the strain difference between both materials
K*) _ Ρ
in the joint is the maximum shear strength of the
7- ~2[ sinh(j] (l + r)-cosh^j adhesive. These simplified shear stress distributions
can also be determined by FE-analyses. Such
withp=W±^KdL andy/=E^ numerical investigations give good accuracy when
Et-tt-h E2-t2 the plastic restoring stresses which compensate the
parabolic stress distribution are considered.

2.3.3 Test programme

2.3.3.1 Connections under static loading

Various tests have been performed to check the

A ultimate resistance of the different connections and
τ(χ)
the applicability of design rules from standards and
literature. Examples of test specimen may be taken
from Figure 8.
> χ

H. SO 100 BO 412

Figure 6. Notation by Volkersen

For combined bolt-adhesive connections the stress

distribution cannot be determined by such equations
because of the complex three-dimensional effects. In
Figure 7 a simplified shear stress distribution
including the strain in the steel plate and in the FRP
310 310
ZE
S

8 8
3M20x300SJSU>

+
6M3DX3O0 SU i

'I
β
S §
is shown. + 11 + 11+ 1

s
S JS

za
F 8
Stainless steel plate >. ^ >
Adnesi»/e- -Bolts- g
4M2IXS28 8J SJ»· l7 | 00 1

FRP- FX
Y 230 | 100 1

2F

Figure 8. Examples oftest specimen

Steel

All connections have been tested with different

stainless steel plate thicknesses t = 2, 3 and 4 mm.
-FRP | ©
After several pre-loading steps up to 50 % of the
calculated ultimate load the connections have been
tested until failure. An example that demonstrate the
load-deflection characteristics of these hybrid-
Rash line
*
connections is given in Figure 9.
All specimen have shown an elastic-plastic load-
displacement curve. The displacements depend
Figure 7. Simplified shear stress distribution for combined mainly, in particular in the first loading steps, on the
bolt-adhesive connections
slip in the clamping devices. The stiffness of the
connection is decreasing up to a specific load
depending on the plate thickness (e.g. above 250 kN
for a plate thickness of t = 2 mm).
 6-8 temperature cycles between -25°C and
+40°C (under the static pre-loading)
V1: stainless steel plates t = 2 mm 30 min. loading time for each maximum
temperature limit (under the static pre-loading)

V 1-1-2
V1-M Constant Temperature FRP / stainless steel
V1-M 4000
ο,

X
3& V ν
iC χ χ
Displacement [mm] 41 χ
O
γ ν
y X
χ
Figure 9. Example for a typical load-displacement curve χ χ
en
X

The failure always occurred by delamination of the -20 -15 -10 -5 0 β 10 18 20 2S 30 3S 40 50 SS M

Temperature PÇ]
FRP-Profile only 1-2 mm below the adhesive joint
surface. This failure mode confirms the results from
Figure 10. Ultimate loads for bonded FRP-stainless steel
the pre-test series and proves sufficient ultimate connections under constant temperature
resistance of the adhesive. In accordance with the
theory, e.g. Volkersen in Figure 6, the highest
In Table 1 an extract of the alternating test results
ultimate load was reached when bonding a thin
stainless steel plate (t = 2 mm) on the FRP-profiles and the constant temperature test results is shown.
with a ratio of: Ψ = (Ei * ti)/ (E2 * t2) = 2,28
Table 1. Ultimate loads in alternating and constant
temperature tests: Bonded connection FRP/FRP
2.3.3.2 Connections under temperature behaviour
Specimen FRP/FRP Max, load [N]
Ixi order to verify the temperature behaviour of the
V 3-2-1 (-25°C- + 40°C) 2704
bonded connections two test series have been
performed. In accordance with the standards, e.g. V 3-2-2 (-25°C- + 40°C) 2269
(3), the ultimate resistances of FRP-FRP and FRP- V 3-2-3 (-25°C- + 40°C) 2489
stainless steel specimen under constant and Average load (-25°C - + 40°C) 2487
alternating temperatures were tested. V 2-2-1 (+23°C) 2870
For the two component epoxy resin it is known that
V 2-2-2 (+23°C) 2401
the strength decreases for temperatures above 50°C.
V 2-2-3 (+23°C) 1578
For the specified temperature range from -37°C up
to +44°C the following test programme was Average load (+23°C) 2283
developed:
The results show that there is no significant
- dimensions ofthe FRP-stainless steel specimen in
influence of the cycled temperature on the ultimate
accordance to DIN EN 1465 (3)
resistance of bonded connections.
- constant temperatures: -40°C, -25°C, -10°C, For constant and alternating temperatures the
+23°C,+40oCand+55°C ultimate resistance of bonded connections is
- three specimen for each constant temperature sufficient for the postulated temperature range.
The specimen have been produced and storaged with Further tests have to be performed to determine the
a temperature of 15°C and a curing time of seven adhesive strength under the combination of .

days. The temperatures in the adhesive joint have
been measured by special temperature sensors. The
results may be taken from Figure 10.
Most of the failures occurred due to delamination of
the FRP. Only two specimen have shown a
combined delamination with a cohesive failure.
Although the test results scatter the diagram shows a
tendency of a decreasing strength above 50°C.
The same test set up has been used for the tests with
alternating temperature. The following load
histogram has been applied to the specimen:
- pre-loading by 20% of the calculated ultim, load
temperature, static- and dynamic loads.
2.3.4 Results
The paper presents the design and test programme
hybrid connections made out FRP and strengthened
by stainless steel plates.
The test programme shows that the existing rules
and publications are not sufficient to determine the
ultimate resistance for hybrid connections designed
as combined bolt-adhesive connections. The test
programme proved sufficient ultimate resistance of
all connections.
8
in order to verify and to extend the rules for FRP-
connections further tests and numerical simulations em = Sfu =
would be useful taking into account various cross-
sections geometries and FRP-qualities. Basic precondition of co-operation between fibres
and matrix in FRP products is, that Smu > Efu .

2.4 FRP Reinforcement Given formulas shows clearly, that engineers can
create FRP product with required mechanical
Fibre Reinforced Plastic (FRP) reinforcing bars, properties (it is a basement of material engineering
tendons and strips, which recently have been used in science and practice). This gives new possibilities in
particular numbers of concrete structures, show theory and technology of prestressed structures.
significant differences in mechanical properties
comparing to steel elements. Opposite to steel, FRP
products have anisotropic mechanical properties, and 3500
different resistance to environmental media. Basic υ [MPa] 7
mechanical differences are: modulus of elasticy 3000
constant in full range of strains and various values of 2500
ultimate strength.
2000 4
/, 6 8
Table 2. Basic properties of reinforcement products 1500 3 ltf^f 1
øs****! s I
1000
9
Fibre Reinf/
500
Property Prestr
ε !%.]
Glass Aramide Carbon 0
Steel
10 20 30 40
Tensile strength + + ++ +

Long-term strength 0 O + + Legend

++ + + Producer Name/Country Type E[GPa] flMPa]

Fatigue strength -

Multiaxial strength - - 0 + 1 POLYSTAL GER GFRP 51 1520

2 KHSKR0SN0 POL GFRP 40 900

Durability in alkaline I
3 FIBRA JAP AFRP 70 1400
Environment I 0 + ++
4 TECHNORA JAP AFRP 54 1800
(like concrete) I 5 ARAPREE NED AFRP 52 1250
Durability in acidic + + ++ - 6 CFCC USA CFRP .137 1800

Environment 7 BBRCFP SUI CFRP 165 3300

Durabilityin carbonated ++ ++ ++ - 8 Reinf. Steel 200 500

Concrete
9 Prestr. Steel 190 1800

Weight ++ ++ - Tested reinforcement with bolded numbers

+
Figure 11. Range of steel and FRP Reinforcement elasic
behaviour
Axial strength fp and modulus of elasticy Ep of FRP
fulfils "rule of mixtures" given by formulas: A main aim of presented project is to determine
fp=ffX<P + omx(l-<p) influence of various reinforcement strain
Ep=Efx<p + Emx(l-<p), (Eq.l) characteristics on main properties of reinforced and
prestressed beams, expressed by Ultimate Limit
where: States. In addition, conclusions related to design
subscripts p, f, m denotes respectively: application of bent beams with FRP reinforcement
composite, fibre, matrix will be drown.
Programme of this project is:
1. Models of materials and elements
φ - relative volume (cross-section area) of fibres 2. Theoretical calculations based on relative values
of material and element properties
am - stresses in matrix with strains equal to: 3. Numerical calculations
4. Experimental tests
5. Analysis of calculation results
6. Conclusions, relations and guidelines
9

Fig. 1 1shows preliminary results of experimental
tests of concrete beams reinforced/prestressed with
various types of materials. Correlation between
modulus of elasticity of reinforcement and flexural
stiffness of cracked beams is clearly seen.
Reinforced
35
Q[kN] . -o- - (3FRP-0
30 . -a- -AFRP-0
-o- -CFRP-0
25
.
20
15
+ i
10
5
*··
0
12 16 20 24 28
a (mm]
Figure 12 Deflection vs load curves for
reinforced/prestressed with different type of tendons
2.5 Timber and concrete
2.5.1 A set of techniques, a set of responses
For timber-concrete composite elements, the main
observed solutions are connections slab to slab and
slab to beam. A lot of various techniques are
available or possible. A lot of mechanisms in order
to transfer connection force are also possible and a
large set of local behaviour (rigidity, ductility or not,
reversibility, dissipation or not) can be observed.
The types of behaviour are often different among
Service Limit States and Ultimate Limit States. The
concreting technology is quite important on the
global characteristics of a timber-concrete element.
The possible semi rigidity for one connection can be
differently integrated in global behaviour. The
number of points of connection or its continuity can
induce various types of global mechanical response.
For a mixed floor, it is important to have a global
approach, without a unique limitation to a regular
bending problem. Floors often present hoppers of
various sizes. Bending is only one ofthe functions of
a floor. Out of and in plan behaviour have to be
taken into account. The integration of a composite
element in various buildings, different technologies
(reinforced concrete, steel or timber construction...)
and refurbishment or new construction work is an
important point to develop new mixed building
technologies.
2.5.2 Local single and multiple connection
behaviour
In order to illustrate some of the points called back
above, a recent study carried out in GEN-' and
Leimes laboratories in Clermont-Ferrand [1], on
"Paris-Ouest" process is going to be briefly
presented here. An illustration is shown on figure 13.
Results of shear tests are illustrated on figure 14.
Cyclic and static behaviours are in a good accuracy
regard to strength and rigidity, it is quite good in the
first part of behaviour until about yield force. After
that, damage appears progressively for cyclic loads
(mainly in concrete) with a reduction of ductility.
Energy dissipation exists, but reaches only a low
level. The comparison between 1 and 4 connector
results exhibits a beginning of interaction effect.
32
beams
Qff
-%:
Figure 13. Paris Ouest process, shear specimen and bending
30m2 floor tests
-(1-b)-2
-(1-t>)-3
-Stat2a2
-Stat2a3
max
rrin
/ >
-(M»)-1
Slip (mm)
Slip (mm)
Figure 14. Static and cyclic behaviour of shear specimens, for 1
and 4 connectors
2.5.3 Global structure out of plan behaviour
Large-scale bending test results allow us to compare
deflection and, on the extremities of the floor,
corresponding connection slip (cf. figure 15). This
slip can be also compared with records for 1 and 4
connector shear tests. Deflection appears to be less
non-linear than connector slip. Nevertheless,
tendencies are well respected and a slip in the
connection induces a deflection increase. Hysteretic
loops are wider for connection slip than for
deflection. Non-linearity appears earlier in
connection slip than for the two configurations
10
presented for shear tests. The global deflection is a
semi rigid behaviour between non-connected
components and total rigid connection. Modelling
with a double beam and connectors [2] gives values
and distribution of connection force along jc-axis. It
is different if the connection is rigid or semi rigid; an
equivalent shear force can be found on the extreme
semi-rigid connectors (1/6 to 1/4 ofthe span on each
side).
Slip (mm)
0.5 1,0
Figure 15. Results for large scale bending test, deflection and
connection slip
2.5.4 Inclusion of mixed element technologies in
global structure
A floor has to be included in a whole structure. Out
of plan behaviour is a specific function of the floor,
but in plan behaviour is in interaction with vertical
elements. Diaphragm component is concrete one. If
vertical structure is concrete, direct connection is
possible as for a whole concrete structure. If vertical
structure is timber framework, peripheral connection
must also be specific. This function is particularly
important for structures built in seismic zone. We
have to found an anchoring flat able to guarantee the
strength in plane (warping also) and a solution for
the anchorage of the resistant diaphragm on the
vertical walls. These two functions can also be
realized with a composite solution, which can
assume a transition with a framework structure.
These in plan actions are very important to develop
mixed solutions and cyclic behaviour parallel and
perpendicular to beam gives interesting information.
2.6Ligh- weight mixed shear wall solutions
2.6.1 Introduction
Steel houses are usually built of light cold-formed
load bearing structure and different interior and
exterior cladding. This technology is very popular
and accounts for an important and increasing market
share in the US, Japan, Australia and Europe. The
same method can be used for non-residential small
or moderate size buildings.
In such buildings shear walls, which are an example
of MBT, are made of light steel wall stud systems
with sheeting, sandwich panels, gypsum board or
OSB cladding and they are designed to act against
horizontal loads, e.g. wind and earthquake. Even if
widely used in practice, the behaviour of shear walls
subjected to earthquake is not fully clarified by
research. In recent years important progress has been
made to understand certain aspects related to shear
strength, stiffness and ductility.
2.6.2 Review of actual research advances
1.5
Research in the US has been focused mainly towards
experimental testing, both monotonie and cyclic, of
shear walls typical to their home practice in order to
produce practical racking load values for the
manufacturing industry. Load bearing capacities
were derived from static push-over curves, envelope
and stabilised envelope curves. Elastic behaviour is
suggested to be obtained up to 0.4 of the ultimate
load. Different frame typologies with various
claddings were tested determining influence of
length/height ratios and effect of openings. The
supplementary gypsum wallboard was identified to
increase strength, stiffness, ultimate load being
increased by 30%, compared to the basic case of
only external sheeting. However, many studies do
not address the aspect of energy dissipation capacity
due to cyclic characteristics.
Testing and numerical simulation was combined in
Japanese research to account for real hysteretic
characteristics in an attempt to provide evidence on
the possible values of response modification factors
(q) to be employed in practice. Vibration tests of
actual steel-framed houses was conducted and
damping ratios of relatively large values were found
due to interior and exterior finishes. According to
tests damping ratio of 6% was accepted for seismic
analysis. A maximum 2% (1/50 rau) story drift limit
is also suggested during severe earthquakes. In the
FE analysis a steel-framed house was subjected to
two levels of seismic waves, exhibiting very good
performance and reaching a maximum drift of
0.33% (1/300 rad). Even when minimum required
wall length was provided the maximum drift did not
exceed 1.67% (1/60 rad).
The same issue is analysed by Gad & all., who
proposed an analytical approach to evaluate ductility
parameter (R^), and found values between 1.5 and
11
3.0. The same research briefly assesses inherent testing was completed with tensile tests to determine
structural over-strength and finds it to be important material and connection behaviour.
factor for earthquake performance.
Experiments and FEM was employed by Italian Table 3. Description of wall specimens
researchers to asses shear behaviour of sandwich
panels. A number of monotonie and cyclic tests were Exterior
Ser. Open. Brae.
performed on full-scale sandwich panel specimens.
These tests were completed with tests on behaviour O - - -

of connections and numerical models to describe

Corr. Sheet
monotonie response of the panels. In the final stage I - -
LTP20/0.5
dynamic modelling on panels integrated in building
Corr. Sheet
structures under earthquake records were performed. II - -
LTP20/0.5
According to the results, diaphragm action can
replace classical bracing solutions only in low rise III - Yes -

buildings and in areas of low seismicity. For multi¬ Corr. Sheet

IV Door -
story frames shear cladding panels can only be used LTP20/0.5
in an integrated system, sharing horizontal force with OSB 10 mm
frame effect. I OSB
OSB 10 mm
Door -
II OSB
2.6.3 Description ofthe testing programme at PUT

Interior Testing Load Vel No.

The experimental program is based on six series of Ser.
Cladding Method (cm/min) Test
full-scale wall tests with different cladding Ο - Monotonie 1 1
arrangements (Table 3). Each series consist of Monotonie 1 1
identical 2440x3600mm wall panels, one being I -
Cyclic 6;3 2
tested monotonically and one or two cyclically. Top Gypsum Monotonie 1 1
and bottom tracks ofthe main frame were U154/1.5 II
Board Cyclic 6;3 2
and studs C150/1.5 profiles, fixed using 4.8mm Monotonie 1 1
diameter self-drilling screws. Different cladding III -
Cyclic 3 1
arrangements were used based on common solutions Monotonie 1 1
IV -
in both housing and small industrial buildings (Table Cyclic 6;3 2
)· OSB Monotonie 1 1
-
In Series Ι, Π and IV, sheets were placed I Cyclic 3 1
horizontally, fixed to wall frame using self-tapping OSB Monotonie 1 1
-
screws in every trough at sheet ends, and every other Π Cyclic 3 1

on intermediate studs. Seams fasteners were 4.8mm Total Number of Specimens 15

screws at 200 mm intervals. In addition for each
specimen in Series Π three, vertically placed
1200x2440mm, 12.5mm thick gypsum boards were 2.6.3.1 Comparative analysis of experimental results
used, fixing being at 250 mm intervals on studs.
Series I - Series Π: Differences can be attributed to
Panels in Series ΙΠ were equipped with strap bracing
the effect of gypsum board. Initial stiffness is not
on both sides, being fixed at the ends with enough
affected, but an increase in ultimate load of 16-18%
self-drilling screws to facilitate yielding of the brace
can be observed together with a slight increase of
itself. For Series OSB I and OSB Π, 10mm thick
ductility.
OSB panels were fixed to a frame in vertical
Series I - Series IV: There was significant decrease
position using bugle head self-drilling screws
of initial rigidity (50-60%), and for lesser degree of
(4.2mm) at 10.5 mm intervals.
ultimate load (16-20%), but ductility values were
A preliminary monotonie test was performed for
similar.
each series and conventional elastic limit
Series I - Series ΓΠ: Comparisons are to be
displacement (Aei) was determined. Loading
understood more qualitatively, as sheeting systems
amplitudes for cyclic tests were determined using Δ^
are different. Initial rigidities were similar and an
and following ECCS guidelines. The full-scale
increase of ductility was expected, which did not
materialise, as failure for strap-braced panels was
12
due to corner fixing. It can also be noted that strap should directly be transmitted from the corner stud
braced wall panels have the advantage of stable to the anchoring bolt. Seam fastener represented the
hysteretic loops, but disadvantage of higher sensitive part of the corrugated sheet specimens;
pinching. damage has gradually increased in seam, until their
Series I - Series OSB I: Comparison is also more failure causes the overall failure ofthe panel..
qualitative, due to different sheeting arrangements. Steel-OSB panels provided comparable performance
Initial rigidity was of similar magnitude, with an as steel-sheeting ones, while the presence of gypsum
increase of ultimate load. Failure of OSB specimens board improved ultimate capacity and ductility of
under cyclic loading was sudden, unlike for sheeting panels.
corrugated sheet specimens where degradation is For corrugated sheet specimens, the following
gradual, reflected by lower ductility. performance design criteria can be suggested, based
Series OSB I - Series OSB Π: The effect of on connection slip values, and with corresponding
opening produced similar results as in case of Series drift estimates.
I - Series IV. Initial rigidity decreased sharply with - fully operational: elastic deformations and very
60-65%, while ultimate load decreased with 30-35%. small plastic elongation of fastener holes, which do
not affect the water proof capacity of cladding
(0.3%).
Top Dlsp. - Comp.
- partially operational: plastic deformation of
I Top Dlsp. - Comp. II
80000 η 80000
fastener holes are accepted in given limits, local
repairing and water proofing necessary, and possible
(1%)
-150 - safe but major repairs required: damages of
-80000 J cladding impose its replacement, drift of the main
-80000
Displacement (mm) Displacement (mm) frame structure being the limiting factor in this case
Top Dlsp. · Comp. Ill (2.5%)
Top Disp. - Comp. IV
80000 The first performance level does not provide
80000
ductility, because it means the shear panel work is
limited to the elastic domain. This could be the
£-150 150 £-15 IV-1 150 design criteria for frequent, but low intensity
111-1 IV-2
1I1-2
-80000
IV-3 earthquakes. In case of rare but severe earthquakes,
-80000
Displacement (mm) Displacement (mm) the last two design criteria can be used, and in such
case some ductility will be available.

Top Dlsp. · Comp. OSB I Top Disp. · Comp. OSB II

80000 80000
3. RECENT APPLICATIONS OF MIXED
BUILDING TECHNOLOGIES
-150 150 0.15fJ 150
O. 1-1 O.II-1
3.1 Residential apartments and Office Buildings
-80000 J °- '~2 -80000 J O· ""2
Displacement (mm) Displacement (mm)
3.1.1 Steel framed house(D. Dubina, L:A. Fulop)
Figure 16. Summary of eExperimental curves The structure is a private house built on 1999 in
Timisoara, identified as medium seismic region in
Romanian. It is of "wall stud" type, with
2.6.3.2 Conclusions corresponding floor and roof joists of cold-formed
Shear-resistance of wall panels is significant both in
lipped channel steel sections and fastening
terms of rigidity and load bearing capacity, and can technology being based on self-drilling screws.
Four bracing systems have been studied in order to
effectively act against lateral load. The hysteretic
select cross bracing made of straps as design
behaviour is characterised by strong pinching and
solution, due to large openings on walls, which made
therefore reduced energy dissipation. inefficient any attempt to use wall diaphragm.
The firsts signs of damage appear at the bottom track Contrary, three alternative floor solutions were
in the anchor bolt region, therefore strengthening of considered and floor diaphragm effect proved to be
the corner detail is very important. The uplift force significant in each case.
13
 -hard-wood parquet
-ISOVER phonic
isolation
-wood decking
-C200/2 joists
-gypsum board

-hard-wood parquet
-ISOVER phonic
isolation
-wood decking
-sheeting diaphragm
-C200/2 joists
-gypsum board

-hard-wood parquet
-light-weight concrete
slab with mash v *
-sheeting (LLP
20/0.5)
-C200/2 joists »V
-gypsum board

Figure 17. Variations of hard-wood parquet

Seismic load was the design criteria mainly for both

Figure 18. Presented steel framed house
bracing and anchor bolts. The main structure has
been basically designed from fundamental load
combination (ie. dead load, live load and snow).
Interior finishing was classical gypsum board
solution, providing clean and pleasant internal space.
External cladding was based on soft, thermally
insulated wood Heraklith panels to eliminate thermal
bridging in vicinity of studs. As seen in the picture
with external finishing the building has the look of a
classical brick house.

3.1.2 FRP-Building (Heiko Trumpf)

/
On the Swissbau 99 a five storey building called
Eyecatcher out of FRP has been built with a floor
space of 330m2 for residential apartments and
offices. The main advantages are a low energy
demand by including the FRP structure in the facade,
short construction time by using a high grade of
prefabricated pultruded profiles (3 days) and flexible
ñ
partition walls for variable rooms.
Figure 19. Eyecatcher Building, Swissbau '99

3.1.3 New development for wooden windows

(Jan Hamm)
The glass is glued on the outside of the framework
and is stressed as part of the composite glass-wood

14
Recently SWOOD started a project in COST E13
with the title: "Strengthening of Timber Beams with
Prestressed Fibres: Development of a Suitable
Adhesive and Technology". The project is concerned
with the danger of delamination of the prestressed
fibres. The aim is to develop suitable adhesive for
connect the wood to the prestressed fibres, in order
to prevent delamination.
3. 1 .7 Single story office building
(D. Dubina, Viorel Ungureanu)
The structure is a single storey office building of the
ROMRECYCLING Company built in Jilava -
Bucharest, identified as strong seismic region of
Romania. The structure is composed of longitudinal
"wall stud" structure combined with transversal
frames made by cold-formed lipped channel steel
sections. Sandwich claddings of outer steel sheeting,
TERWOOLIN insulation and inner gypsum board
were used for walls. Roof panels are made of steel
sheeting, TERWOOLIN and a plastic film for
protection. A ceiling made by gypsum board was
placed at the bottom flange ofthe trusses.
C.
*· -r**%
***i*Häj|f-
Figure 24. Single storey office Building
Horizontally placed external corrugated-sheeting
emphasizes the non-industrial role of this building,
while gypsum domestic-like interior finish makes it
suitable as an office or headquarters of a small firm.
For the design diaphragm effect of the roof sheeting
based sandwich panel was considered, the capacity
of the diaphragm being governed by the sheeting to
purlin fasteners. For finite element modelling
diaphragm effect of the roof was replaced by an
equivalent cross bracing system. Similarly, wall
bracing can also be replaced by diaphragm effect,
especially for light structures when earthquake
loading is absent.
Table 4. Diaphragm flexibility and strength
Equivalent
Flexibility Capacity
bracing diameter
(mm/kN) (kN/mm)
(mm)
Roof: 1.151 6.5 22.30
3.2 Industrial Constructions
3.2.1 Over-claddingfor Alcatel-Datatim
(D. Dubina, Viorel Ungureanu, M. Georgescu)
The structure described represents an extension of
production capacity of ALCATEL-DATATIM
Company of Timisoara. An extension ofthe existing
production space was required by erection of a steel
penthouse superstructure over the existing RC
building.
C
Due to the fact that Timisoara is a seismic territory
the penthouse structure had to be light enough to
avoid an expensive strengthening of the existing
ground structure.
A cold-formed structure connected exclusively by
bolts was proposed the penthouse superstructure.
The roofing and cladding panels were trapezoidal
steel sheet, with suitable thermal insulation. As
interface structure, a metallic grid built of castellated
steel beams was connected to the RC structure by
cruciform steel elements and the floor of the
penthouse is a composite steel-concrete deck.
An important advantage of this structural solution
was the easy erection, in a particular narrow building
site. The structure was raised, stick by stick, on the
deck of new floor, and there the trusses and frames
were assembled and lifted with a light crane.
16
system. Both the gluing technique and the window 3.1.5 Joint connection with welded thermoplastic
construction were developed and production was dowels & wood welding technologies
planned at SWOOD. (Jan Hamm)
This combination of painted wood and common
floating glass is an efficient solution for window and Since 1997, thanks to the good results obtained for
façade systems. the tension and shear resistance of welded wood
joints, SWOOD obtained opportunities to extend and
apply WWT to timber construction. In mid 1999,
SWOOD launched a new project to study the
r
s .
.) ··"* functionality and adequacy of alternative
4 <[* yç** thermoplastic welding technologies such as vibration
*
and spin processes.
I
r

ί .,· .C* '

f"

ΐ-ί Ehrnctr- '

Figure 20. New composite glass-wood window

Figure 22. Assembly before welding (right), section of welded
dowels perpendicular to grain (left)

3. 1 .4 Prefabricated wood-glass structural elements

(Jan Hamm)
3.1.6 Composite decks of concrete glued to timber
A wood-glass composite element has been (Jan Hamm)
developed at EPF Lausanne where the wooden frame
In 1998, together with partners from the wood and
is directly glued to the glass-plate, with both adhesive industry, SWOOD launched a research
components acting as the supporting structure. program to investigate timber-concrete-composite
Advantages: decks connected by glue.
Light structure
Prefabricated elements The structural advantages seemed obvious:
An optimum use of material No concentration of the forces as with
Constructive wood preservation mechanical connectors.
100% connection by fully stopping the
relative movement between the wood and the
concrete: the increased bending stiffness (ΕΓ)
^7 would reduce deflections.
Economic feasibility because of the
simplification ofthe work.
7 '

A*.
t .-«f>
' Λ--·1"*
. Λ is.;

\ ·.·.. .

Figure 21. Handrail with load-bearing function

Figure 23. Test specimen: concrete glued on timber

15
 3.2.2 FRP-Cooling Towers (Heiko Trumpf)

The energy-industry is recommending FRP as

structural material for small and large wet cooling
towers to ensure longevity, minimal maintenance
problems and a fast and flexible erection by using
these light-weight elements. In comparison with
timber, steel and concrete FRP-profiles comprises
weathering-, corrosion- and decay resistance as well
as fire inhibited fire properties. Taking into account
the durability and maintenance FRP structures are
even more economic than conventional designs.

-'-": ;;

" I»

{SSïiiï^U «f '. E V ;

J**ri
Λ-.·»·
^ ,-a >n

Figure 25. Structure during the erection and after the roof
envelope was completed

Figure 27. Mechanical draught Cooling Tower (Photo BDT

Engineering)

Figure 28. Electric power Station in Kansas (Photo

BDT Engineering)

3.3 Bridges

3.3.1 Light-Weight Bridges

(Heiko Trumpf, Eric Knudsen)
&
There is a demand for light-weight bridges that are
Figure 26. View ofthe erection procedure and corresponding
suitable for air-transport, hand erection by unskilled
building facade workers and which may be used for light support
vehicles for humanitarian services. The Institute of

17
Steel Construction at the RWTH Aachen, Germany,
has been awarded project to develop an
a of composites. Composites are superior to traditional
economical solution for such light-weighted single
lane bridges for a span length of 20,0 m and MLC-
12 loads.
This light-weight emergency bridge has been
designed with particular regard to the mobility and
cost effectiveness:
- Low weight and small volume for ease of
transportation and erection (e.g. transportation by
air, erection by hand) by using pultruded FRP-
profiles and stainless steel
- Total dead weight for the structure 3,5 t, total
dead weight for the deck 2 1
- Development of dismountable bolted-connections
using bonded steel reinforcements
- Low production costs by using standard profiles
commonly available in the market
- Modular design to be flexible in use with a modular
length of 4000 mm
- Development of FPR-deck elements coated with a
primer and finished with a mineral layer
- sufficient lifetime (fatigue) and low maintenance
costs
A large test programme has been performed to
verify the best application of adhesives and
sufficient bearing capacity as well as the
determination of the failure criteria and temperature
behaviour of hybrid-connections.
The first prototype of the bridge made of FRP
with a length of 20 m has been built in 2001 which
will be followed by a test programme under realistic
conditions in this summer.
Ζ~3ί (B; ¡TFT? Γ*'
Figur 29. FRP-Light-Weight-Emergency Bridge (Photo RWTH
Aachen)
Scandinavia's first composite bridge - the Fiberline
Bridge in Kolding, Denmark, was officially opened
on 18th June, 1997. The new bridge, matching the
strictest safety standards, has that FRP composites
may represent the building material of the future for
lightweight, maintenance-free bridges.
The busy railway line restricted installation work to
only a few hours during nights between Saturdays
and Sundays, and after only 3 such nights in May
and June, the bridge was ready. The short
18
installation time has illustrated the clear advantages
materials like steel and concrete in a number of
ways:
- With a span of 40 metres and a load bearing
capacity of 5 kN/m2 the bridge weighs some 12
tons only - less than half the weight of a similar
steel construction, making foundation, assembly
and transport both simpler and cheaper.
- The composite bridge has an important
environmental advantage. The total consumption
of energy required for raw materials, production
and assembly accounts for only one quarter ofthe
would-be energy consumption for a similar
construction of steel or concrete.
- The bridge, which is uninfluenced by rain, salt
and sub-zero temperatures, is expected to require
only purely cosmetic maintenance over the
coming 50 years or more, reducing maintenance
costs considerably compared to traditional bridges
of steel, concrete or wood.
v ^
Figure 30. FRP-Bridge, Kolding (Photo Fiberline Composites
A/S)
The Fiberline Bridge is constructed by Fiberline
Composites A/S in close co-operation with
RAMBOLL and the City of Kolding.
Sponsors behind the Fiberline Bridge include also
Jotun Polymer AS, Norway, who supplied the
polyester and the Toledo, Ohio- based Owens-
Corning, who supplied the fiberglass.
3.3.2 Bridges with FRP-Reinforcement
(Zbigniew Plewako)
A particular number of concrete bridge structures
reinforced with FRP elements are under service and
constant monitoring. They all are treated as
experimental structures, where in alert situation
replacement of reinforced is possible. Interesting
examples from Europe are described in following The American Iron and Steel Institute "Shear Wall Design
publications. Guide", Publication RG-9804, , February 1998
Volkersen, Die Schubkraftverteilung in Leim-, Niet- und
Bolzenverbindungen, Energie und Technik, 1958
Wolff R., Mießeler H.-J., Die Brücke Schießbergstraße in
Leverkusen, Beton 4/91
REFERENCES Y. Kawai, R. Kanno, N. Uno, Y. Sakumoto, "Seismic
resistance and design of steel framed houses", Nippon Steel
Technical report, No. 79, July 1999
A. J. Salenicovich& all., "Racking Performance of Long Steel-
Frame Shear Walls" »Fifteenth Int. Speciality Conference on
Cold-Formed Steel Structures, St. Louis, Missouri, Oct. 19-
20, 2000
Casas J., Aparicio A. C, A full-scale experiment on a
prestressed concrete structure with high strength fibres: the
North Ring-Road in Barcelona, FIP Notes 1991/4
CEN/TC 250/SC5 N158, EUROCODE 5 Design of timber
structures Part 1.1 General rules and rules for buildings,
final draft stage 34, 2001, 121 pages
DIN EN 1465, Adhesives - Determination of tensile lap-shear
strength of rigid-to-rigid bonded assemblies (ISO 4587:
1979 modified), German version EN 1465: 1994
E.F. Gad, A.M. Chandler, CF. Duffield, G. Hutchinson,
"Earthquake Ductility and Overstrength in Residential
Structures", Structural Engineering and Mechanics, Vol.
8,No. 4 (1999) 361-382
ECCS "Recommended Testing Procedure for Assessing the
Behaviour of Structural Steel Elements under Cyclic
Loads", September 1985
ENV 1993-1-1/A2, Eurocode 3 1988: Design of steel
structures, Part 1-1: General rules and rules for buildings,
Amendment A2, Annex J.
Fournely E., Lecrompt Ch.,Racher P., Sauvat N.
"Comportement dynamique de plancher bois-béton du
système Sylvabat" Report of contract 2001-6, GEN-MAT'
Leimes laboratory, 2001, 17 pages
Gianfranco de Matteis, "The Effect of Cladding in Steel
Buildings under Seismic Actions", PhD Thesis, Università
degli Studi di Napoli Federico Π, 1998
Habenicht, Kleben - Grundlagen, Technologie, Anwendung,
3rd edition, Springer Berlin Heidelberg New York 1997,
Khalifa Μ. Α., Kuska S. S. B., Krieger J., Bridges Constructed
Using Fiber Reinforced Plastisc, Concrete International,
June 1993
Kuhlmann, U.; Schäfer, M. (in prep.): Innovative
verschiebliche Verbundrahmen mit teiltragfähigen
Verbundknoten. Project-Nr: P505, supported by
Studiengesellschaft für Stahlanwendung e. V.
Kuhlmann, U.; Sedlacek, G.; Kühnemund, F.; Stangenberg, H.
2001: Verformungsverhalten der Komponenten von
wirtschaftlichen steifenlosen Anschlußkonstruktionen für
die Anwendung plastischer Bemessungskonzepte im
Stahlbau. Forschungsbericht 4/2001 Deutscher Ausschuß
für Stahlbau. Stahlbau Verlags- und Service GmBH.
Kühnemund, F. 2002: Zum Rotationsnachweis nachgiebiger
Knoten im Stahlbau. Thesis, Institute of Structural Design,
University of Stuttgart.
Meier H., Meier U., Brönnimann R., Zwei CFK-Kabel für die
Storchenbrücke, Schweizer Ingenieur und Architekt, No 44,
24. Okt. 1996
R. Serrette, G. Hall, J. Nygen, "Shear Wall Values for Light
Weight Steel Framing", AISI, January 1996
R.L. Serrette, "Seismic Design of Light Gauge Steel Structures:
A discussion", Fourteenth Int. Speciality Conference on
Cold-Formed Steel Structures, St. Louis, Missouri, Oct. 15-
16, 1998
The American Iron and Steel Institute "Monotonie Tests of
Cold Formed Shear Walls with Openings", Prepared by
NAHB research Center. Inc., 1997

19
Shear transfer in composite members: testing, modelling, standards and
damage

O.S. Bursi, M. Ballerini, M. Piazza & R. Zandonini

University of Trento, Italy
E. Fournely
University Blaise Pascal, France
U. Kuhlmann, K. Kürschner & J. Schänzlin
University of Stuttgart, Germany

ABSTRACT: The shear connection between steel and concrete, timber and concrete and more in general
between other materials is one ofthe fundamental components in any composite member in which it is used.
Thereby, it can be argued that a thorough understanding of the performance characteristics of the shear
connection is essential for a reliable analysis and design of composite members. The contents reported herein
address these issues. More specifically, this review clarifies the basic concepts involved in the shear transfer
mechanism and introduces different shear connector types. Typical examples are provided for each class of
connection exploited for monotonie, low and high-cycle fatigue loading, with emphasis on more recent studies.
The available test procedures for the determination of the shear force-slip behaviour of connectors are
evaluated and the associated problems are discussed. A reference is also made to the modelling of connectors
under monotonie and cyclic loading. Moreover, the most commonly used design provisions are listed and
commented upon. Reference is also made to procedures for damage assessment of headed stud connectors
under seismic loading and some damage indices are evaluated in quantitative terms, using data from very
recent studies. Finally, conclusions are drawn and some directions for future research efforts are identified.

1.THE SHEAR CONNECTION embodying ductile connectors or composite dissipative

The behaviour of composite structures represents zones need to be designed, a better understanding
nowadays a many-faceted subject. Indeed, mainly in regarding the monotonic and/or the cyclic behaviour of
the field of steel-concrete composite structures, many the interface is required.
books (Oehler and Bradford 1995, Viest et al. 1997), /. 1 Basic concepts
handbooks (Cosenza and Zandonini, 1997) and a vast
The transfer of forces between steel and concrete
number of articles have dwelt on that topic.
and timber and concrete components is the mechanism
However, the design of composite structures
that renders composite action feasible. Various forms
presents still unresolved issues among which
inadequate design provisions (Leon, 1998). Indeed,
of shear transfer mechanisms can be identified (Oehler
and Bradford, 1995, pp. 52, 79). They can be classified
there are several situations in which the required
schematically in i) chemical adhesion and chemical
composite action between the steel beam and the
bond; ii) interface friction; iii) mechanical interlock;
concrete deck is rather low (Uang et al., 2000, Civjan
et al., 2000) and, thereby, full shear connection is
iv) dowel action. Shear transfer via adhesion and bond
is associated with lack of steel- or timber-concrete slip.
needless. As Eurocode 4 (EC4) (CEN, 2001a) and
Nonetheless, the maximum shear strength of about 0.1
AISCLRFD (1993) specifications permit partial shear
N/mm2 reduces rapidly as soon as slip occurs at the
connection to be adopted, the interface slip between
steel-concrete interface.
the steel beam and the concrete slab cannot be
overlooked. In these conditions, the connector ability
Friction is often associated to bond in resisting
shear and is activated through the normal stresses
to exhibit ductility for plastic analysis or dissipate
applied across the interface and the friction
energy under low-cycle fatigue loading depends
coefficient μ, ; they can reach values of about 0.6
mainly on its characteristics. Thereby, if members
N/mm2. In flexural members, the tendency of steel-
21
 _0
li
ii
//Concrete
i,
PM h η
s
# 10
Figure 1. Typical cross-section; experimental and modeling strain distribution; long-term behaviour
and timber-concrete elements to separate usually
makes frictional actions deteriorate quickly.
Mechanical interlock is obtained, for example, by
embossing the steel decking in such a way that slip
at the interface is resisted by bearing between steel
ribs and the concrete indentation. The effectiveness
of the embossments depends on their geometrical
dimensions, viz. the height, the depth and the shape.
In detail, the interface interlock shear strength can
reach maximum values of about 0.8 N/mim. Similar
%cfraufi¿
for loading
-*-
Figure 2. Details of timber and concrete elements; full-scale floor test; cyclic behaviour of connection
results can be obtained creating grooves in timber
elements. In order to transfer heavy shear forces
suitable mechanical connectors are exploited. More
specifically, the shear is transferred by mechanical
deformations of a steel dowel and hence, a highly
concentrated load is applied into the concrete of the
composite member.
1.2 Examples of shear transfer mechanisms
Examples of shear transfer through chemical
adhesion and chemical bond are shown in Figure 1,
where glulam and concrete are connected by means of
a two-component adhesive (Brunner and Gerber,
2000). Experimental results show that the connection
22
20 30 βο
has a rigid behaviour and that the failure is brittle.
An example of friction interface is shown in Figure
2 where the lateral contact is made between concrete
and vertical nailed timber planks without adhesive.
The concrete shrinkage on camber is more important
in the transversal direction than in the longitudinal
direction, thus creating a high pre-stressed contact.
The friction phenomena are followed by an inelastic
behaviour that provides a high degree of ductility to
this member (Fournely et al. 2001a).
Examples of connections used in timber-concrete
composite constructions are depicted in Figure 3, and
are part of a recent research programme carried out by
Ballerini and Piazza (2000). Several timber-concrete
composite beams have been tested using different
connection systems made with concrete dowels
(Figure 3a); concrete dowels and epoxy-glued steel
rebars (3b); screws (3c); epoxy-glued ribbed rebars
(3d); thin drilled sheet (3e); and shaped epoxy-glued
steel rebars (3f).
A connection realized with steel circular hollow
sections pushed in an annular cavity drilled in a timber
component is illustrated in Figure 4. The connection
shear strength is realized by the strength ofthe circular
hollow section itself (shear strength and deformation
ofthe steel section), the embedment strength in the
load capacity and are exploited mainly for girders with
large spans, especially for bridge engineering.
Concrete dowels going through the perforation
together with a transverse steel reinforcement provide
resistance to shear and uplift forces (Oguejiofor and
Hosain, 1994). Recently, modified perforated
connectors have been proposed for thicker concrete
slabs (Machacek et al. 2000). Comb-shaped strips may
represent a suitable alternative to perforbond strips
mainly for technological reasons because during their
production cutting by flame is continuous with no
material waste (Rovnak et al. 2000).
Rflftflft.. Π8Π8ΠΑ
COiVIB-SHAPEÖ STRIP CONNECTOR
Reinforcing bar
~&?&Z ooooo<
PERFOBOND-STRiP CONNECTOR
Figure 7. Non-traditional shear connectors
The majority of tests regarding mechanical
connectors under reversed loading were carried out
focusing on the performance of dowels subjected to
high-cycle fatigue. Thereby, Log S - Log N limit
domains for headed studs or slip versus cycle number
relationships were defined for bridge applications
(Oehler, 1991, Gattesco and Giuliani, 1996). Other
investigations were conducted in order to estimate the
shakedown behaviour of headed studs (Taplin and
Grundy, 1997), to quantify the stiffness variation of
studs under cyclic loading with limited reversals
(Oehler and Coughlan, 1986) or to calibrate the
Manson-Coffin relationship for the overall fatigue
range, viz. low-cycle and high-cycle fatigue in view of
bridge applications (Dorka and Bode, 2000).
From a seismic standpoint, tests on headed studs
were carried out by Hawkins and Mitchell (1984) both
in a monotonic and low-cycle fatigue regime. Hence,
specimen performances were compared and those
subjected to cyclic loading exhibited a reduction both
of shear strength and of ultimate displacement
ductility. Test results emphasised the advantage of
orienting the troughs of the profiled steel sheeting
parallel to the beam span direction as well as the
importance of having a large stud connector spacing.
Other investigators (Astaneh-Asl, 1995) subjected
headed studs to reversed cyclic displacements in order
to study the effect of a dissipation device, named skirt
(see Figure 8), on the stud connector performance as a
more favourable dissipation mechanism. Along the
same line, Shultz et al. (2000) conceived a polymer
cone placed over the stud before installation, see
Figure 8, in order to increase the shank deformation
ductility. Polymer hardening can retain both strength
and stiffness. In detail, preliminary tests indicate an
increase in ductility of about 60 per cent. Other
connectors recently tested for seismic applications and
depicted in Figure 8 are the spit connector, that is
endowed with a special folded plate pinned to the steel
flange in order to enhance strength, ductility and
economy (Matus and Jullien, 2000). Conversely,
Aribert and Lâchai (1999) tested nailed cold formed
angles or Hilti connectors which are illustrated in the
same figure. They concluded that such connectors
could not be adequate for seismic loading as a
reduction of shear strength of about 20 per cent at ±
2 mm slip was observed.
SKIRT OR
'POLYÍÍtEHCO»«
"Ducme* *spir »nor
WELDED STUD CONNECTOR CONNECTOR
Figure 8. Shear connectors exploited for seismic loading
2. SHEAR CONNECTION BEHAVIOUR
The behaviour of a mechanical dowel depends on
numerous factors that govern the complex mechanisms
by which the shear is transferred. Thereby, the shear
force-slip behaviour of connections is mainly
determined by testing. Besides, an inherent difficulty
exists of deducing the resistance of a connection from
composite beam tests, owing to the relative change in
bending moment versus the relative change in shear
connection. More specifically, a relatively large
change in shear connection results in a much smaller
change in bending moment. Thus, formulating the
shear connector strength by conducting beam tests and
back-computing through the flexural model leads to
difficulty in assessing the fundamental connection
behaviour. As a result, the most common way used to
evaluate the connection behaviour and the associated
strength is the so-called push-type test, by which the
connections are loaded directly. A typical test set-up
for a timber-concrete specimen with a concrete dowel
subjected to monotonic loading is illustrated in Figure
9 (Ballerini, et al. 2002).
ftydraulc
jit*
I Ir I
Teflon
Figure 9. Test set-up of a push-off timber-concrete speciment
with a concrete dowel connector
24
 a "b' »c»

Οτ»Η01γ3

*Á M* '-Φ '·'.

UJ L ι» 1 ι» J
«MXλsaiiraw#Kitat)M

4V/" r

....*»
r^ï* ιψ..

SOKYÍB.UK j.,,1 £ U| i» J
Rr»»VffiV|i

Figure 3. Geometry of some connection types for timber-concrete composite elements

electrically welded to the steel member using an

timber like for dowels (local embedment and shear automatic welding procedure like the TRW Nelson
plus tension perpendicular to grain) and also local welding system. Examples of this and other connectors
embedment in the concrete slab (Foumely et al, 2001b, commonly used in steel-concrete composite
2001c). construction are depicted in Figure 5.
New interesting composite cross sections can lead
i to a horizontally lying arrangement of headed studs
(see Figure 6, Kuhlmann and Breuninger, 2000). In
»? contrast to common shear connections with vertical
studs this innovative shear connection is sometimes
Β not only subjected to longitudinal shear but also to
vertical shear owing to the vertical support of the
reinforced concrete slab. The cracking action of shear
connections with lying shear studs is shown in Figure
PEaiiiiggiEi^lS 6. In practice, a combination of both shear actions
. -s leads to a mutual interference of single resistances.
Figure 4. Failure mode of a "Pans-Ouest" dowel-rype
connection
'//
As far composite structures are
as steel-concrete
Longitudinal and Only Longitudinal
concerned, the headed stud is by far the most popular Vertical Shear Aciion Shear Action
connector adopted and consists of a bolt that is

LcHicmidiixil Shear Vertical Siiear

Figure 6. Composite sections and cracking patterns due to

longitudinal and vertical shear

Figure 7 illustrates non-traditional shear connectors

constituted by comb-shaped strips with small and large
T-CONNECTOR HORSESHOE C-COiNNECTOR BAR CONNECTOR
concrete dowels. Also commonly used perforbond
Figure 5. Commonly used shear connectors strips connectors or rib shear connectors are plotted.
Traditional perforbond strips are endowed with high
23
Conversely, set-ups and steel-concrete specimens for (Bursi and Gramola, 1999). Moreover, the angles
monotonie (push) and cyclic (pull-push) loading are located on top of steel beams equilibrate all the
shown in Figure 10a and 10b, respectively. horizontal forces developed at the base ofthe slab and
.λ*^ angles s. thereby, no compressive force should act on the
LYDT
- J^it dowels owing to these boundary conditions.

uâL Ή Other researches adopted steel plates on the slab

edges aiming at simulating the passive resistance
FASTENING LOCKING
SARS â) DEVICES offered by the continuation of concrete slabs (Taplin
ν ί and Grundy, 1997, Oehler and Coughlan, 1986,
Hawkins and Mitchell, 1984, Astaneh-Asl at al., 1993,
i i i and Aribert and Lâchai 1999). It is evident that the
aforementioned fastening devices create a confinement
effect on shear connectors, thus increasing the ultimate
displacement ductility and the energy dissipation in a
Figure 10. Test set-up and boundary conditions:
cyclic regime.
a) Push-off specimen; b) Pull-push specimen
The instrumentation is usually designed to monitor
the behaviour of the specimens during testing, by
providing a continuous time record both of slips and of
2.1 Push-type tests shear reaction forces. In detail, the reaction force can
Push tests do not entirely reproduce in terms of be measured with a load cell connected to the end of
loading and slab confinement, the most favourable the actuator. The interface slip e between the steel
connector's condition in a composite beam. Moreover, beam and the concrete slabs shown in Figure 10 can be
a problem arises from the lack of consistency among detected by means of linear variable differential
the geometrical and detailing features of push-type transformers (LVDTs).
specimens designed differently in different research
studies. Indeed, different test set-ups embodying 2.2 Behaviour under monotonie loading
consistent boundary conditions were adopted around
Welded headed stud shear connectors are by far the
the world in order to carry out push and/or pull-push
most common type of connector used in the design of
tests. Only Gattesco and Giuliani (1996) used
composite members in steel frame construction today.
boundary conditions able to reproduce a stress state of
Indeed, this has been the case for the last 30 years. The
direct shear onto the connectors, which turns out to be
failure modes relevant to shear connectors embedded
appropriate for connectors close to ends in simply
supported beams.
in standard and high performance concrete are
highlighted in Fig. 11a and 1 lb, respectively. A ductile
In order to standardize testing, EC4 suggests the
failure mechanism caused by bending of the stud as
execution of standard or specific push tests (CEN,
well as local concrete crushing is illustrated in Figure
2001a). More specifically, boundary conditions
11a. Conversely, a non-ductile failure mechanism
highlighted in Figure 10a correspond to conditions
involving connector shearing is illustrated in Figure
suitable to monotonic tests. Other design provisions
that deal with push-off tests for headed studs are the
lib. In these conditions all shear deformations
concentrate at the stud base, where the collar carries a
Japanese ones (JSSC, 1996). One may observe that the
large proportion of the shear force. Large slips are
test set-up shown in Figure 10a induces additional
achieved only with rigid studs, for instance 25 mm
compressive forces in connectors that can increase the
diameter studs in Figure lib, as significant stud
dowel strength up to 20 per cent compared to the
movements are impeded by the surrounding high
dowel strength obtained from composite beam tests
performance concrete.
(Oehler and Bradford, 1995, p. 56).
The testing equipment adopted to exert the
displacement Jona specimen under seismic loading is
illustrated in Fig. 10b, schematically (Bursi and
^':::·ίρ·;·.^: 4vv*";V;L·
Gramola, 1999). Due to the alternating nature of
loading, it is necessary to adopt post-tensioning bars in
order to transfer the reaction force from the concrete
slabs to the counterbeam. In detail, the fastening bars
are tensioned to exert a slight compression of about 0.5
N/mim. That stress level represents the case in which
the dowels are located near the inflection points of a
Figure 11. Potential failure of a headed stud connector
composite beam. Boundary conditions depicted in
embedded in a solid slab: a) Local concrete crushing; b) stud
Figure 10b impede any relative movement of concrete shearing
slabs and are conceived to simulate the passive
resistance offered by the continuation of concrete slabs
25
2.3 Behaviour under cyclic loading
In order to characterize the behaviour of connectors
under reversed loading predefined representative
loading histories are applied usually to connectors.
Especially under low-cycle fatigue the problem of
displacement pattern arises as unique relationships in
low-cycle fatigue are strictly valid only for constant-
amplitude loading. Real structures, however, seldom
conform to this ideal. Indeed, between the extremes of
constant-amplitude cyclic-displacement histories and
variable-amplitude random-sequence displacement
histories, there is a multitude of displacement pattern
of varying degrees of complexity.
As far as high-cycle fatigue is concerned, the
aforementioned problem was faced recently by Bode
and co-workers, in view of bridge applications (2000).
The influence ofthe stress-range history on the fatigue
of headed studs was investigated by means of stress
distributions generated through stochastic load models.
In the low-cycle fatigue regime, more specifically
seismic loading, the influence of cycles inducing a
high degree of inelasticity on subsequent cycles with
lower amplitude has not been clarified in a definitive
way. In the past, the problem was faced by Hwang and
Scribner (1984) who subjected concrete beams to
various cyclic displacement time histories. They
concluded that strength and stiffness degradation was
closely related to the maximum amplitude in each
cycle, but relatively independent of the sequence in
which large and small displacements were applied.
In fact, tests conducted by Bursi and Gramola
(1999) and Zandonini and Bursi (2000) on push-pull
specimens with headed studs endowed with 16 and 22
mm diameter (see Figure 12) showed that the influence
of headed stud performances under different histories
is remarkable.
a»
,
w w w each fixed amplitude and then increasing the
r τ' B....M. /s
,
1
*J
t» τ regarding the damage accumulation and strength
ft a K4W : *
X
Figure 12. Geometrical characteristics of specimens with a) 16
mm and b) 22 mm diameter stud connectors
In order to define conveniently displacement
patterns, a conventional elastic limit state
characterized by the displacement ey+ and the
corresponding force Fy + can be defined on the first part
of each response curve obtained from classical
monotonie tests as depicted in Figure 13,
schematically. These values can be defined by means
of bi- and tri-linear approximations of each curve,
determined on the basis of best-fitting and dissipated
energy-equivalence criteria (Zandonini and Bursi,
2000).
****%
r. 'W}
i f/K''
Ψ
WW)
- ßäineur uppror
e
Figure 13. Bi- and tri-linear fits of a shear force-slip envelope
In order to acquire comprehensive sets of
information from the specimens, the so-called
Complete Testing Procedure proposed by the
Technical Committee 1 of the European Convention
for Constructional Steelwork (TC 1, 1986) was used.
This controlled displacement procedure SPDP-1,
which is shown in Figure 14a, is characterized by sets
of equi-amplitude displacements (2 + 2k) ey , (k =
Ο,.,.,η). It is adopted to acquire data on the specimen
capacity such as the maximum shear strength, ultimate
displacement ductility, maximum absorbed energy,
etc. Moreover, it provides a convenient benchmark
against which to compare specimen performances
subjected to constant and random amplitude testing. It
is by far, the most common procedure used in Europe
to test components in a cyclic fashion. A similar
procedure (multiple-step test) is specified by the
Applied Technology Council (ATC, 1992), viz. (1 + k)
ey+, (k = Ο,.,.,η), in which two cycles have to be
applied with k > 3. The same procedure is also
suggested by the Appendix S ofthe Seismic Provisions
« ,ai
for Structural Steel buildings (AISC, 1997), though it
is referred to the qualification of beam-to-column and
link-to-column connections. However, the
aforementioned procedures embody some limitations.
More specifically, applying two or three cycles for
amplitude, does not provide direct information
1*1
0 o i degradation relevant to a defined ductility level, fn
fact, increasing amplitudes simulate the isotropic
strain-hardening behaviour that is opposite to the
strength degradation owing to low-cycle fatigue.
Recently, a revised loading history was suggested
by the Federal Emergency Management Agency
(FEMA, 1999). This loading history, which was
derived from a series of non-linear time history
analyses of the 3-storey Los Angeles model building,
adopts the interstorey drift angle as selected control
parameter. It comprises three sets with six cycles at
constant amplitude, one set with four cycles at
constant amplitude, and, finally, sets of two cycles
with constant amplitude. Demands provided by the
application of this loading history vary slightly from
26
 40 20
30 smp-i
15
l^ö/»-^
r&cauw>
20 '10

10 5
t o -f o
-IO -5
-20 -10
-30 -15
-40 -20
NUMidft

Hí
S - {%ti>P-i<\ ill Ül 6 . \$WP-i$
6 4
4
2

ίο r»
E3Ô
-2
-4 -4
-6
-S
-10
*c) III I -8
'à)
CYCLE NUMSER
CICLE NUMBER
Figure 14. Geometry Controlled displacement test protocol: a) Variable amplitudes; b) Constant amplitudes; c) Large amplitudes
superimposed upon constant amplitudes; d) Random amplitudes

those obtained in tests using the ATC 24 protocol
(1992).
The procedure SPDP-2 is also illustrated in Figure
14b and is characterized by a set of equi-amplitude
constant displacements at 10 ey+ in agreement with the
Cumulative Damage Testing Program (ATC, 1992).
This procedure provides the basis for developing
fatigue-life relationships in which strength and
stiffness degradation can be quantified. Moreover,
additional procedures were conceived to analyse the
slip sequence effects on cumulative damage. In detail,
benchmark against which to compare random
amplitude testing.
The shear force-slip responses provided by the
timber-concrete connection shown in Fig. 9 and
characterized by different Lt values are depicted in
Figure 15. The connection endowed with the largest Lt
exhibits a favourable strength and a ductile behaviour
(Ballerini et al., 2002). In addition, the shear force-slip
response of two specimens characterized with the
connection
4Se

the SPDP-10 test protocol illustrated in Figure 14c, is 4M

characterized by multiple large displacement cycles SSŰ ri
superimposed upon constant amplitude displacement m

fluctuations. In addition, the SPDP-13 test protocol,

plotted in Figure 14d, was tracked from pseudo-
dynamic tests on composite substructures exposed to
the N69W component of the 1952 Taft earthquake 15» Connection type "e"
(Bursi and Gramola, 2000). m imnnttmie test
tm mattetest
SÛ

9
'Χ
Average slip (mm)
Figure 16. Monotonie and cyclic shear force vs average slip of
specimens with "e" type (Figure 3) shear connection

shown in Fig. 3e, are illustrated in Fig. 16; the strength

and ductility of the connection appears not to be
influenced by the loading history. The behaviour of
three floors endowed with connections shown in Fig
3a, 3b and with an improved version of connection 3a
Λ4 ϋ.Χ 1,2
Average slip (mm) (with a timber groove with a larger diameter and a
Figure 15. Monotonie shear force vs average slip of specimens smaller height) are shown in Figure 17, respectively.
with concrete dowel-type connectors The favourable behaviour ofthe connections 3b and 3a
improved is evident. Moreover, the geometrical details
This type of sequence provides a convenient ofthe connection 3a penalize strongly its performance.
27
 36

2S
γ»*·*.
ί£20
/*
^" 1
ί

1
1

/
1

Ì
- connection type "a " ^..

-connection type "b"

/.< "\ \
τ
connection type "a" (improved)
τ-
V
40 60 70 Í.S xs S.S
Mitl- deflections (mm) A verace slips at beams ends (mm)
Figure 17. Mid-span deflections and average slips at beam ends of floors with concrete dowel-type connectors
specimens are illustrated in Figure 19. Reversed
The deflection at mid span and average slips at displacement cycles induce into the specimens a
beam ends of floors endowed with optimised concrete limited reduction of the shear strength owing to the
dowels (type "a" in Figure 3) and concrete dowels with high concrete strength whilst the ultimate slip
rebars (type b in Figure 3) are depicted in Figure 17. reduction is evident (Zandonini and Bursi, 2000).
The performance of the optimised connection al is In order to evaluate the headed stud performances
evident. from a seismic standpoint a conventional ultimate slip
The shear force-slip responses relevant to the RPM- eu+ is determined, to define the range in which each
01 and RPC-01 specimen, respectively, viz. specimens connector is able to provide a shear strength F+. Fy+,
with connectors embodying a 22 mm diameter, are as illustrated in Figure 13, schematically.
plotted in Figure 18. With regard to the cyclic These values can be defined by means of bi- and
response, the RPC-01 specimen exhibits both stiffness trilinear approximations of each backbone force-slip
and strength reduction owing to stud shearing and envelope determined on the basis of best-fitting and
fatigue. Due to the relatively high value ofthe concrete dissipated energy-equivalence criteria (Zandonini and
strength, collapse is governed by shearing at the shank Bursi, 2000). As a result, the ultimate slip ductility
base as depicted in Figure 1 lb. factor eu+/ey+ in the pushing regime can be estimated.
15. 160 With regard to the cyclic slip capacity, an
ÍEE20 PUSH attentive reader can observe from Figure 18 and 19
«0 that stud failure collapse is governed by low-cycle
fatigue stud shearing. This phenomenon leads to a
I 40
0
remarkable hysteretic energy dissipation owing to
stud yielding but it limits strongly the ultimate slip
-40
ductility (eu+/ey+). Indeed, one may observe that
-80
ttPM-91 shear connectors subjected to the ECCS procedure
2 -120
PULL RPC-m (TC 1, 1986) approach a maximum ductility of about
^ -160 15. Nonetheless, eu+ is greater than 5 mm for the
-16 -12 8-404 8 12 16
RPC-01 specimen, confirming the ability of 22 mm
SUP (mm)
headed stud connectors to sustain large slips even
Figure 18. Monotonie shear force vs controlled slip and
hysteresis loops of specimens with 22 mm connectors with the mechanism illustrated in Figure lib
160
(Zandonini and Bursi, 2000).

3.MODELLING
A connection between timber and concrete was
realized by grooves with a depth of 2 cm and a length
of 20 cm (see Figure 20, Gerold et al., 2000). For the
tensile force in the connection caused by the resulting
eccentric bending moment, screws with a diameter of
12 mm at a distance of 25 cm were installed.
Analysing the stiffness ofthe connection by the design
-2 0 2
method of Eurocode 5 (EC5) Annex B (CEN, 1993),
SUP (mm)
based on the measured mid-span deflections of decks,
Figure 19. Monotonie shear force vs controlled slip and
a stiffness value of a single groove of 860 MN/m2
envelope curves of specimens with 22 mm connectors
including the shear deformations of the deck was
Both monotonic and skeleton curves of all determined. A mechanical approach to the problem
according to the following Equations,
28

Figure 20. A concrete dowel connection coupled with
prestressed bolts and results of F.E. modelling
stiffness of a single dowel
-o V<V4»
6./.(ΰβ.Λ+<νΛ)
effective stiffness due to shear deformations:
5-G 'dJ{ -Ga>dM-dx
3*/fs 'G» >de + 3« /ία «?Λ ¿.
resulting stiffness:
i t .
C C c (3)
leads to a stiffness of 2235 MN/m2, including the shear
deformation of the deck, and to 8050MN/m2 without
the shear deformation (Kuhlmann et al. 2002).
A determination of the stiffness of the concrete
dowel by means of a FE model, (see Figure 20) results
in a stiffness value of 9307 MN/m2. Although there is a
wide range of determined values, there is only little
influence by these various values on the overall
stiffness.
While at the serviceability limit state, viz. the linear
regime, the load-deformation behaviour of headed
studs can be represented with a constant, the elastic
stiffness, at the ultimate limit state, viz. into the non
linear regime, the simulation of the load-deformation
behaviour requires more advanced tools. As far as the
static loading is concerned, several authors adopted
mathematical models or curvefitting techniques in
order to reproduce the shearforce slip response of
headed studs (Olgaard et al, 1971 and Oehler and
Coughlan, 1986). The most popular relationship was
proposed by Olgaard et al. (1971) and reads
P**/»* -(l-e-**)'
More specifically, Pmax+ is the maximum resistance
of a single shear stud, s is the slip whilst β and α
are constants which define the slip amount and the
initial slope of the curve, respectively. The
abovementioned equation was used recently by Rex
and Easterling, (2000) who modelled reinforced
composite slabs in tension as well as by Manfredi et al.
(1999) which modelled continuous composite beams
by means of discrete connectors, solving the resulting
system of differential equations in finite difference
form. On one hand, the above-mentioned mathematical
models can represent with extreme accuracy any shear
force-slip relationship but on the other hand, these
models are limited to the connector type used for
calibration. In addition, they are not able to show the
contribution of each physical component to the overall
shear connector response.
Based on its continuous progress, the finite element
technique or micromodelling is the other widely used
tool adopted to reproduce the connector response.
Matus and Jullien (1996) conducted parametric
analyses on push-off specimens employing solid and
interface elements with standard material models. A
similar approach was adopted by Kalfas and Pavlidis
(1997), which exploited beam elements to simulate the
behaviour both of headed studs and of steel beam
(i) stubs. It is apparent that proper material models and
refined meshes of these elemental systems grant
reliable finite element models and accurate analyses.
Within a study concerning the service load
behaviour of steel-concrete composite beams subjected
(2)
to repeated loading, the shear force-slip behaviour of
headed studs was idealized as a damaged region of slip
with zero stiffness followed by a constant stiffness
region, where the stiffness was taken to be the same as
for the linear region in a monotonie response (Taplin
and Grundy, 2000). This model was incorporated into
a non-linear interface element used in a finite element
mesh of a composite beam.
Regarding to the hysteretic modelling of headed
studs subjected to seismic excitation, a non-linear
constitutive shear force-slip law accounting for the
tangent component of the force was developed by
Salari (1999). The model is capable of accounting for
the deterioration of bond strength and stiffness during
load cycles. In order to calibrate the model, the results
of few pull-push tests carried out by Bursi and
Gramola (1999) were employed. Along the same line,
beam finite elements were developed by Ayoub and
Filippou (2000), among others, in which distributed
shear forces were adopted at the steel beam-concrete
deck interface. In detail, the shear force-slip
relationship was described by means of a model
similar to the one proposed by Eligehausen et al.
(1983) for the bond stress-slip response of anchored
reinforcing bars.
Bursi and Zandonini (2000) adopted one-
dimensional finite element beam models based on
(4) fibre concepts and on the Navier-Bernoulli beam
29
 ι ' ' . ' 1. 1 .
¡II', Ι. III llltm'll tim»l»llLlllLl!Égt
." .' ». Ί
Its performance on a 16 mm diameter shear connector
-V '· -T> :-»-. ^ response is illustrated in Figure 23. In the same figure,
tiJ^fri'^"-^. »mm
octj»ctch
OOUroMWr the performance ofthe standard connector Element 4,
l t tXM.
available in the DRAIN-3D library (Campbell et al.,
SSS^DB ynSBtttZTSi COr^ltCrQK
1994) is illustrated too.

4. STANDARDS
Design rules for mechanical shear connectors
subjected to monotonie loading are provided by
several codes, AIJ (1987), AISC LRFD (1993), CEN
Figure 21. One-dimensional finite element model of a (2001a) and CSA (1994), among others, which deal
composite beam part mainly with the design strength of dowels. In fact, the
strength of a shear connection is proportional to the
theory for composite beam modelling. Indeed, these compressive strength of concrete fi, to the tensile
models provide the best compromise between accuracy strength ofthe steel dowel fi and to the cross-sectional
and computational effort. Analyses were performed area Ad (Oehlers and Bradford, 1995, p. 58). The
with the DRAIN-3DX code (Campbell et al., 1994) design resistance of a headed stud embedded in a
and part of the relevant model is illustrated in Figure concrete solid slab varies from standard to standard
21. Forty-eight inelastic fibre beam-column elements with small modifications (AISC LRFD, 1993, CEN,
with distributed plasticity and with five fibre each are 2001a, CSA, 1994). However, several floor systems in
used to model the steel beam and the concrete deck. composite frame constructions make use of steel decks
Reinforcing bars are modelled by means of a where the troughs are transverse or parallel to the
distributed representation whilst shear and uplift are beam span direction. For transverse sheeting, the
simulated by means of joint elements. As the actual behaviour of headed studs is very sensitive to the stud
stud connector cyclic behaviour exhibits stiffness layout, the trough geometry and the stud geometry.
degradation and strength deterioration, for instance see Thereby, research is still ongoing in order to evaluate
the design resistance of studs under static loading. See
the recent modifications to design strength predictions
60 NPC-4>5 specimen
proposed by Johnson (2000), Rex and Easterling
Taj 40 (2000) and Wu and Hosain (2000), among others. In
fact, it must be emphasized that the majority of
0
composite steel floor decks used today have a stiffener
in the middle of each deck flange. Because of the
ξ -20 stiffener, studs must be welded off-centre in the deck
trough. Therefore, recent studies have shown that
ti -m Evperim. headed studs behave differently depending upon their
location within the trough. The so-called weak
-5 -4 -3 -2 -t 0 3 4 5
SLIP (mm)
1
(unfavourable) and strong (favourable) locations have
been identified (Easterling et al. 1993).
Figure 22. Hysteresis loops of a specimen embodying 16 mm
diameter headed stud shear connectors
In what follows the common case of stud shear
connectors automatically welded and with the troughs
< 60 NPC-OS specimen transverse to the beam span direction is considered. In
U3 40 1- this situation, EC4 (CEN, 2001a) specifies that the
design shear resistance of headed studs be evaluated
§ 20 using the following equation
t °
2-20
R
O .40 '°-«·Λγ 4 'r
Elemem 04 V
ä^o Element M >Rd »min«

-3 -2-10
SUP (mm)
Figure 23. Numerical representation of the NPC-05 specimen
hysteresis loops
Figure 22, constitutive laws that catch up with
experimental data need to be used. To this end, an
Element 06 able to reproduce the aforementioned
phenomena has been developed by Deng et al. (2000).
I
kr'O.il'a*
'&'ck' cm ?v
(5)
where Pud is the resistance of a single shear stud; fi is
the shear stud ultimate tensile strength of the stud
material but not greater than 500 N/mm ; d is the
nominal shear stud shank diameter; y v is the partial
safety factor which should be taken as 1.25 for the
ultimate limit state; fik is the characteristic cylinder
strength of the concrete at the age considered; Eon is
30
the mean value ofthe short-term elastic modulus ofthe
concrete; a = 0.2 f(h / d) + 1)] for 3 . h/d . 4 and
a =1.0 for h/d > 4; h is the overall height of the
stud; h = (0. 7/ .Nr)(bo / hP)[(h / hP)-l] . 1. 0; Nr is the
number of studs per trough . 2; bo is the width ofthe
deck trough; hP is the height of the deck trough.
Equation (5) should be verified by tests before being
used for studs of diameter greater than 25 mm.
Johnson (2000) warns out that Equation (5) can
provide both low (< 0.8 PexP) and high (> 1.5 Pex¡>)
predictions of the experimental strength PexP, because
Equation (5) takes no account of stud layout in the
trough or the sheeting thickness. A similar equation is
also specified by the current Canadian Standard (CSA,
1994) but with a value of 0.5 instead of 0.37.
It is worthwhile also to summarize the current
AISC LRFD specifications (1993) with the
modifications suggested recently by Rex and
Easterling (2000), when the troughs are transverse to
the beam span direction. The modified equation
follows
»Ν
><' π 'd'
Ö »* «0.5-
4 4 f -E áO.8 L loading, connectors need to carry heavy inertial forces,
(6)
where Qn is the strength for a single shear stud; f'c is
the characteristic cylinder strength of the concrete; Ec
is the mean value of the secant modulus of the
concrete; k = (0.85/ ^Nr)(bo / hP)[(h / hP)-l] < 0.75
for strong position stud and < 0.5 for weak position
stud. However, note that a reduction factor φ = 0.85
has to be applied to the design flexural strength of
composite beams. Moreover, Rex and Easterling
(2000) noted that even if the concrete is in tension,
there is little detrimental effect when the studs are
located in a composite steel deck with the trough
transverse to the beam span direction. Conversely,
Johnson et al (1969) observed a reduction of about 80
per cent for the headed stud strength compared to the
strength of headed studs embedded in solid slabs
subjected to compression.
For metal decks with troughs parallel to the beam
span direction, the behaviour in a push tests is not so
sensitive to the stud layout, provided that the EC4
(CEN, 2001a) rules for minimum spacing are
followed.
As far as bridge applications are concerned,
Equation (5) applies also to headed studs with a 25
mm diameter (Bergmann, 1999). However, for high
performance concrete, e.g. fc about 90 N/nrrm, the
maximum strength developed by headed studs is
always greater than the one predicted by EC4 (CEN,
2001a). More specifically, ratios of about 40 per cent
for studs with a diameter of 19 mm embedded in a
solid slab were obtained (Hegger and Döinghaus,
2000).
Welded stud shear connectors are essential
components of most composite bridges of steel and
concrete. Thereby, their behaviour in the high-cycle
fatigue regime has to be checked too. In this situation,
EC4-2 (CEN, 1997) indicates the following
relationship
Iog/V« 22.12-8· logâ% (7)
where N is the number of stress range cycles and Δ τ r
is the fatigue resistance of one stud defined as Δ τ r =
4 . Pr/ ( π d2). Nonetheless, considering that the design
strength should be based on regression of log N versus
Δ τ r, on the basis of the re-evaluation of 155 pull-
push test results, Johnson (2000) suggested recently
the following endurance relationship
log /V = 1 6.4-5.5· Iogurt (8)
Along the same line, Dorka and Bode (2000)
employed a Manson-Coffin relationship in order to
cover high-cycle and low-cycle fatigue ranges.
At present, design codes do not predict the shear
resistance of headed studs subjected to seismic
loading, though headed studs have an important role in
the global response of a composite steel-concrete
structure. In fact, besides their capacity to carry out
shear stresses and uplift forces owing to vertical
which have to be transferred from the floors directly to
the rigid cores. In this situation, the number of cycles
experienced by connectors during a strong motion is
rather small with very high shear stresses.
In order to verify the capabilities of design
provisions calibrated for monotonie loading, stud
strength calculations were performed according to EC4
(CEN, 2001a) and AISC LRFD (1993) by Zandonini
and Bursi (2000). These values have been evaluated
with and without partial safety factors y or reduction
factors φ, respectively, using measured rather than
nominal material properties. Predictions and
experimental strength values Pmax+ for individual
connectors of 22 mm diameter are indicated in Figure
24. It can be observed that codes predict correctly the
failure mode, viz. stud shearing as depicted in Fig.
lib. However, owing to reversed slip effects predicted
resistance values are unsafe. More specifically, test
and predicted resistance values draw the conclusion
that the design resistance of headed stud shear
connectors in dissipative zones can be obtained from
the design resistance provided by EC4 (CEN, 2001a)
by applying a steel penalty factor SPF equal to 0.75.
This value agrees with the reduction factor proposed in
the draft ofthe EC8 (CEN, 2001b) regarding specific
rules for steel-concrete composite buildings. More
specifically, it allows a partial safety factor Pmax,RPc-oi+
/ SPF Prd+ = 1.4 to be achieved, where Prd+ has been
evaluated with Eq. (5), SPF = 0.75, α = 1.0 and ki =
1.0.
Conversely, a SPF' equal to 0.55 applied to AISC
LRFD specifications (1993) allows a similar safety
factor, viz. SPF Prd+ / SPF' φ Qn+ = 1.03 to be
achieved, where Qn has been evaluated with Equation
(6), φ =0.85 and ib =1.0.
31
^ 180 with:
^_ LRPl>($**i) ar.o' = ar,o - Cv ds,Bü / 2 . 50 mm
ϊς tõô 22. drm. 25mm; 110. a. 250mm
IJtFDty-MS)
ds,L. 12mm;ds,Bü. 10 mm; a/ 2. s. a
< 140
P.C4(í^l)
S* Combined longitudinal and vertical shear
»5
,20 EC4fa-US) resistance:
,1 lôo O Cyclic
Δ Monotonie
logA? = 22.12-8-logArÄ (H)
*5 80
20 23 30 35 40 45 5Ö 55
f 'c . Á* (Μ Pa) Besides the given design requirements the following
Figure 24. Connector maximum strengths and code predictions conditions should be satisfied.
of specimens embodying 22 mm diameter stud connectors - The stirrups should be dimensioned for the
splitting force as follows.
As far as the American seismic provisions (AISC,
1997) are concerned, it must be observed that no Td^-F^L-U-doft/a,*) (Π)
penalty factors are provided to take into account the - To prevent the pull-out of studs at the edge ofthe
effects of reversed cyclic loading on the strength of slab the studs have to be sufficiently anchored behind
shear studs for composite beams in special moment the stirrups.
frames. However, the use of additional shear studs in - uncracked concrete: j3 < 30°
dissipative regions beyond those required in AISC or ν > max {1 10 mm; 1,7 . ar'; 1,7 . s / 2}
LRFD Specifications (LRFD, 1993) is suggested. - cracked concrete: j3 . 23°
Conversely, the static shear strength of headed or ν > max {160 mm; 2,4 . a/; 2,4 . s / 2}
studs connecting concrete shear walls to unencased - The carrying capacity for vertical studs in EC4
structural steel elements is reduced by 25 per cent (2001a) should not be exceeded.
(AISC, 1997). The equations including vertical shear have been
On the basis of Kuhlmann and Breuninger (2000) developed for an edge position and give values on
and Kuhlmann and Kürschner (2001a, 2001b) the the safe side for the middle position. Similar to the
following design equations have been derived for the design rules of Eurocode 4-2 (1997) for vertical
horizontally lying shear stud depicted in Figure 25. studs, but with the edge distance as an additional
Middle Posilion Edge Posilion parameter, a fatigue strength curve has been derived.

5. SEISMIC DAMAGE
Few theories of plasticity can reproduce the
behaviour of material undergoing cyclic loading in the
inelastic regime. However, these models do not
typically include the notion of failure, viz. a specific
level of damage that corresponds to an arbitrarily
defined residual capacity to resist further load. Yet a
material undergoing seismic loading will often fail
owing to low-cycle fatigue, i.e. an accumulation of
damage that results from a limited number of
Section A-A excursion, typically less than 100 well into the
Figure 25. Design details of horizontally lying shear studs inelastic range. The actual mechanism involves the
initiation and propagation of microcracks that
Longitudinal shear resistance: eventually lead to failure at the macroscopic level.
,0,4 A3
Pral » .42 - (fa - don · *')"* - (ο&Γ*
1 - A / γν (9) Since the processus very difficult to model (Lemaitre,
1992), the majority of investigations into the fatigue
process has dwelt upon phenomenological aspects of
with:
fatigue failure. In detail, the endurance of members
= a· Cv- ds,Bü 12. 50 mm
at'
under specific loading conditions is estimated without
19. doo. 25mm;ds,Bü. 8mm making a neat distinction between crack initiation and
110. a. 440mm;a/2. s. a;s/ar'. 3
propagation phases. Thereby, a considerable amount of
effort has been expended to define suitable damage
Vertical shear resistance:
variables in order to obtain the best correlation
IV 0 = 0,385 · (fkf\0,S' . {ar,oT'
,0.7
- W / ?v (10) between a damage variable and the life to failure. For
instance, damage parameters such as displacements or

32
plastic work or some functions of these have been of the maximum deformation capacity; β * is an
used. As a result, latest structural programs such empirical factor evaluated on experimental basis.
IDARC (Kunnath et al., 1992) and DRAJN-3DX The second damage model examined was proposed
(Campbell et al., 1994) embody fatigue failure criteria by Bernuzzi et al. (1997). Wöhler's approach (1871) is
for members and thereby, reliability analyses of the basis of the above-mentioned model in which a
existing structures and/or predictions for novel types of global parameter like the total displacement range is
structures can be performed. employed. For a typical displacement history made of
In order to validate damage models it is deemed n sets of cycles, Bernuzzi et al. (1997) demonstrate
necessary to define failure, i.e. to estimate the number that
of cycle to failure Nf. Due to the large uncertainty and
subjectivity in the failure determination, two different ΣΜ*
M
criteria are summarized here. The first one is energy d.
based whilst the second one is a design-based Nf á< (14)
approach. in which Aeeq is an equivalent displacement amplitude
For a safe estimate of the fatigue life, Calado and such that variable and constant amplitudes exhibit the
Castiglioni (1996) propose the ratio η f / η o < a same damage at failure. Only one test is required to
where α is equal to 0.5. In detail, η f represents the calibrate the fatigue life relationship being its slope
ratio between the absorbed energy at the last cycle equal to 1/m, where m was adopted equal to 3 based
before failure and the energy that might be absorbed in on steel component tests (Bernuzzi et al., 1997).
the same cycle if the component would exhibit an Therefore, its use is straightforward in design. As the
elastic-perfectly-plastic behaviour; 77 o is the approach relies on the Miner's rule for cumulative
aforementioned ratio but with reference to the first damage (1945), the interrelation between displacement
cycle in the inelastic regime. The second criterion and force (path dependency) is overlooked in this
considered here is the one suggested by Chai et al. model. Thereby, it is deemed to be necessary the
(1995). It relies on the design assumption that a application of a cycle counting method like the
component failure happens when the degraded strength rainflow or the reservoir technique (Ellyin, 1997).
F+ reaches the specimen yield strength Fy+, viz. F+ = The application ofthe model expressed by Eq. (13)
Fy+ = Fu+ = 1 according to Figure 13. Once failure is with Di= 1 entails the damage limit relationship
identified, it is appropriate to associate the behaviour
of a member subjected to cyclic loading to an E, .+{
equivalent monotonie response. More specifically, it is P.ß (15)
convenient to define a damage indicator, viz. a state
variable that enables a one-to-one correspondence in which Py = 467.8 kN, eum = 10.2 mm s = 1.13,
between a damaged state and a unique point of a i = 2.24 with a determination coefficient R = 0.57 for
specimen monotonie response. In these conditions, the specimens with 22 mm diameter headed stud
safety assessment is straightforward being the distance connectors (Zandonini and Bursi, 2000). The relevant
from failure univocal. Thereby, a damage index D is damage domain is illustrated in Figure 26.
introduced, i.e. a damage indicator normalized to the 2.5
failure level, for which zero corresponds to the virgin O MoHønmk
RPC-0.
state whilst one entails the achievement of failure. RPC-H3 OÇMc
2.0 O ORPC41
Different choices relevant to damage indicators are *
RPC-07**- XPOm
available. As matter of fact, two significant models are RPC4HtåRPC-06
summarized hereinafter. *.*Ί.5 RPC-09O RPM-OJ
"Ν
The first damage model was proposed by Chai et al. o -i.m O
hf
(1995) and derives from the widely used model of Park i.Ø RPC-OS
D-I.Û O
and Ang (1985). More specifically, this model takes RPM-02
R -0.57
into account the energy absorbed by the component 05
during a monotonie loading process, named Ehm, and Û.G 03 0.4 0,0 0.8 LO 3.2

only the surplus of cumulative energy (Eh Ehm) is

considered significant to damage, being Eh the Figure 26. Design details of horizontally lying shear studs
absorbed total energy. In these conditions, the damage
index reads The adoption of the damage model proposed by
Bernuzzi et al. (1997) and expressed by Eq. (14) for
D = 1 reads
' ν
'M« Ρν
* «* rMIM (13)
Log Åety-(Um)Log Ay + / (16)
where eM defines the maximum displacement reached
where m = 2.41, i = 1.1 16 with R2 = 0.45. Being the m
by the component so far; eum is the maximum
values not very close to the theoretical value 3 adopted
displacement under monotonie loading, i.e. a measure
for steel components (Bernuzzi et al. 1997) and as the
33
R values vary remarkably, several tests are necessary
to calibrate fatigue life relationships. The above-
mentioned trend happens indeed to Equation (15) as
well. The large scatter found depends mainly on three
factors: i) material and geometrical variability of
specimens; ii) loading variability; iii) exploitation of a
macroscopic approach which makes no distinction
among flexural, shear and friction mechanisms of
dowels which are treated altogether. Moreover, in
order to determine the m values in Equation (14), an
iterative procedure is needed. Hence, the damage
model of Bernuzzi et al. (1997) applied to headed
studs under low-cycle fatigue results very expensive
from a computational standpoint. Indeed, iterative
procedures related to the m determination as well as
the use of a cycle counting method are needed.
6 CONCLUSIONS AND RECOMMEND ATIONS
The contents reported in this paper have addressed
the understanding ofthe performance characteristics of
the shear connection for a reliable analysis and design
of timber-concrete and steel-concrete composite
members in which it is used. The following
conclusions have been drawn.
A large amount of work has been made to date in
the field of the shear transfer mechanism between
timber and concrete members and steel and concrete
members. The performance characteristics of different
shear connector types under monotonie, low- and high-
cycle fatigue loading appear to have been explored,
albeit not to the same depth.
The available experimental procedures for the
determination of the shear force-slip behaviour of
shear connectors under monotonie loading and
highcycle loading are well established. Nonetheless,
the calibration of test protocols for seismic loading,
which involves random loading and cycling in the
inelastic regime deserves further study.
Mathematical and finite element models of
connections are standard tools for the analysis of
composite beams, which can reproduce with extreme
accuracy any shear force-slip relationship. However,
mathematical models are limited to the connector type
used for calibration. Moreover, they are not able to
distinguish the contribution of each physical
component to the overall connector response.
A large amount of research has lead to better design
provisions for connectors mainly devoted to non-
seismic loading.
Procedures for damage assessment of headed stud
connectors under high-cycle fatigue are well
established. Nonetheless, the use of damage indices for
headed studs under seismic loading is uncommon,
owing to the relatively limited experimental database.
The following recommendations are made for
further study ofthe behaviour of shear connectors.
Additional tests are needed to accurately predict
strength, ultimate slip capacity and ductility of dowels
embedded in composite slabs under low-cycle fatigue
loading. Different geometrical characteristics of the
steel decking, especially with the troughs transverse to
the beam span direction and different reinforcements
need to be analysed.
Experimental tests with composite beam specimens
subjected to low-cycle fatigue loading embodying 22
mm or higher diameter headed studs in order to
estimate the corresponding slip demands are needed.
Constitutive models of the shear force-slip
behaviour of headed studs were calibrated with a
relatively limited experimental database. Additional
tests with different cyclic histories for an improved
description of shear stud constitutive models are
required.
Most ofthe available damage indices are based on a
macroscopic approach without making a neat
distinction between crack initiation and propagation
phases. Moreover, flexural, shear and friction
mechanisms of steel dowels are treated altogether,
explaining to a certain extent the large scatter found.
Future research should be directed towards developing
indices with distinct terms describing damage owing to
each above-mentioned mechanism.
All damage indices available determine clearly
failure, for which the damage index D is equal to
unity. However, much less work has been done with
regard to defining intermediate damage stages, like the
thresholds of damage and irreparable damage.
ACKNOWLEDGEMENTS
Results are sponsored by national grants or
industries for which the authors are grateful.
However, opinions expressed in this paper are those
of the writers, and do not necessarily reflect the
views ofthe sponsoring agency.
REFERENCES
American Institute for Steel Construction, 1993. Load and
Resistance Factor Design - Specifications for Structural Steel
Building. 1. AISC, Chicago, IL
American Institute of Steel Construction, 1997. Seismic
provisions for structural steel buildings. AISC, Chicago, IL
Applied Technology Council, 1992. Guidelines for cyclic
testing of components of steel structures. 24.
Architectural Institute of Japan, 1987. Standard for Structural
Calculations of Steel Reinforced Concrete Structures.
Aribert, J.M. and Lâchai, A. 1999. Comportement de
connecteurs acier-beton sous chargement cyclique repete en
vue du dimensionnement parasismique des connejdons.
Proc. Of Fifth National Colloquium AFPS, Cachan,
October, 19-21, 479-488.
Astaneh-Asl, Α., McMullin, K.M., Fenves, G.L. and Fukuzava,
E. 1993. Innovative semi-rigid steel frames for control of
the seismic response of buildings, Report UCB/EERC-
93/03.
Astaneh-Asl, A. 1995. Seismic design of composite structures
in the United States. Proc. of the Int. Workshop on
34
 Behaviour of Steel Structures in Seismic Areas STESSA '94,
Timisoara, Romania, 26 June - 1 July, 448-457.
Ayoub, A. and Filippou, CF. 2000. Mixed formulation of
nonlinear steel-concrete composite beam element. J. Struct.
Engrg, ASCE, 126(3), 371-381.
Ballerini, M. Piazza, 2000, EExperimental and numerical results
on timber-concrete composite floors with different
connection systems. WCTE2000 International Conference,
paper P50, Whistler, British Columbia, Canada.
Ballerini, M. Crocetti, R.Piazza, 2002, An experimental
investigation of notched connections for timber-concrete
composite structures. WCTE2002 7th World Conference on
Timber Engineering, (submitted), Malaysia.
Bergmann, R. 1999. Steel concrete composite beams and
columns. Proc. of the Int. Conf. On Steel and Composite
Structures, Delft, February, 24-25, 3.1-3.9.
Bernuzzi, C, Calado, L., and Castiglioni C. A. 1997. Ductility
and load carrying capacity prediction of steel beam-
tocolumn connection under cyclic reversal loading. J. of
Earthquake Engineering, Vol. 1 (2), 401-432.
Bode, H., Leffer, A. and Mensinger, M. 2000. The influence of
the stress range history on the fatigue behaviour of headed
studs. Composite Construction in Steel and Concrete TV -
Proc. of Engrg. Found. Conference, Banff, May 28 - June
2, Banff, Alberta, Canada.
Brunner M. and Gerber C. 2000 Composite declcs of concrete
glued to timber, World Conference on Timber Engineering,
P46, Whistler Resort British Columbia, Canada.
Bursi, O. S., and Gramola, G. 1999. Behaviour of headed shear
connectors under low-cycle high amplitude displacements.
Materials and Structures, RILEM, 32, 290-297.
Bursi, O. S., and Gramola, G. 2000. Behaviour of composite
substructures with full and partial shear connection under
quasi-static cyclic and pseudo-dynamic displacements.
Materials and Structures, RILEM, 33, 154-163.
Bursi, O.S. and Zandonini, R. 2000. Non-linear analysis of full
and partial shear connection composite substructures,
Composite Construction in Steel and Concrete IV - Proc.
Of Engrg. Found. Conference, Banff, May 28 - June 2,
Banff, Alberta, Canada. 249-261
Canadian Standards Association, 1994. Limit State Design of
Steel Structures. Standard CAN/CSA - S 16.1 - M94. 1.
CSA, Rexdale, Ontario
Calado, L. and Castiglioni, C.A. 1996. Steel beam-to-column
connections under low cycle fatigue: experimental and
numerical research. I lth WCCE, Acapulco, Paper No.
1227.
Campbell, S., Powell, G. H. and Prakash, V. 1994. DRAIN-
3DX Base program description and user guide - Version
1.10, Report No. UCB/SEMM-94/07.
CEN 1993. ENV 1995-1-1 : 1993, Eurocode 5: Design of
timber structures; Part 1-1: General rules and rules for
buildings"
CEN 1997. ENV 1994-2 Eurocode 4 - Design of composite
steel and concrete structures - Part 2: Bridges.
CEN 2001a. prEN 1994-1-1, Eurocode 4 - Design of composite
steel and concrete structures - Part 1-1: General rules and
rules for buildings.
CEN 2001b. prEN 1998-1-1, Eurocode 8 - Design of structures
for earthquake resistance- Part 1-1: General rules, seismic
actions and rules for buildings.
Chai, Y. H., Romstad, K. M., and Bird, S. M. 1995. Energy-
based linear damage model for high-intensity seismic
loading. J. Struct. Engrg, ASCE, 121(5), 857-864.
Civjan, S.A. Engelhardt, M.D. and Gross, J.L. 2000. Retrofit of
pre-Northridge moment-resisting connections. J. Struct.
Engrg, ASCE, 126(4), 445-452.
35
Cosenza, E. and Zandonini, R. 1997. Composite construction,
in Handbook of Structural Engineering, W.F. Chen ed. CRC
Press, Boca Raton, New York.
Deng, C-G., Bursi, O.S. and Zandonini, R. 2000. A hysteretic
connection element and its applications. Computers &
Structures, 78(1-3), 93-110.
Dorka, U.E. and Bode, H. 2000. Manson-Coffin relationship
for shear connectors." Composite Construction in Steel and
Concrete TV - Proc. of Engrg. Found. Conference, Banff,
May 28 - June 2, Banff, Alberta, Canada.
Easterling, W. S., Gibbings, D. R., and Murray, T. M, 1993.
Strength of studs in steel deck on composite beams and
joists. Engineering Journal, AISC, Second Quarter, 44-55.
Eligehausen, R., Popov, E. P., and Bertero, V. V. 1983. Local
bond stress-strain relationships of deformed bars under
generalized excitations. Report N. UCB/EERC-83/23.
Ellyin F., 1997. Fatigue damage, crack growth and life
prediction. Chapman & Hall. London.
FEMA, 1999. Seismic design criteria for new moment-resisting
steel frame construction. Federal Emergency Manegement
Agency, draft, November.
Fournely E., Racher P., Rivera N., 2001a. Comportement
dynamique de plancher bois-béton du système LIGNADAL
Intermediate report GEN-MAT Lermes Laboratory.
Fournely E., Racher P., Rivera N. 2001b, Comportement
dynamique de connexions multiples bois-béton du système
Sylvabat Report of contract 2001-5, GEN-MAT' Lermes
Laboratory.
Fournely E., Lecrompt Ch.,Racher P., Sauvat N. 2001c.
Comportement dynamique de plancher bois-béton du
système Sylvabat. Report of contract 2001-6, GEN-MAT
Lermes Laboratory.
Gattesco, N. and Giuliani, E. 1996. Experimental study on stud
shear connectors subjected to cyclic loading. J. Construct.
Steel Research,3S, 1, 1-21.
Gerold, M.; Kuhlmann, U.; Risio, T. D.; Sulzberger, L.;
Schänzlin, J. 2000. Verformungsmessungen von
Brettstapel-Beton-Verbunddecken. Kurzbericht
Forschungsvorhaben Deutsche Gesellschaft für
Holzforschung (DGfH), München, vgl. auch Institute of
Structural Design , University of Stuttgart (Mitteilung 2000-
27X)
Hegger, J. and Döinghaus, P. 2000. High performance steel and
high performance concrete in composite structures.
Composite Construction in Steel and Concrete TV Proc.
Of Engrg. Found. Conference, Banff, May 28 June 2,
Banff, Alberta, Canada.
Hwang, T-H and Scribner, CF. 1984. R/C cyclic response
during various loading. J. of Structural. Eng., 110, 3, 477-
489.
Hawkins, N.M. and Mitchell D. 1984. "Seismic response of
composite shear connections. /. of Structural. Eng., 110, 9,
2120-2136.
Japanese Society of Steel Construction, 1996. JSSC Standards
on Push-out Test for Headed Stud. Draft, N. 35
Johnson, R.P., Greenwood, R.D., and Van Dalen, K. 1969.
Stud shear-connectors in hogging moment regions of
composite beams. The Structural Engineer, 47(9), 345-350.
Johnson, R.P. 2000. Shear connection - Three recent studies.
Composite Construction in Steeland Concrete TV Proc. of
Engrg. Found. Conference, Banff, May 28 - June 2, Banff,
Alberta, Canada.
Kalfas, C. and Pavlidis, P. 1997. Load-slip curve of shear
connectors evaluated by FEM analysis, in Composite
Construction - Conventional and Innovative, Proc. Int.
Conf. Innsbruck, 151-156, IABSE, Zurich.
Kuhlmann, U., Breuninger, U. 2000. Behaviour of Lying Studs
with Longitudinal Shear Force. Engineering Foundation
Conference. Composite Construction. Banff.
Kuhlmann, U., Kürschner, K. 2001a. Liegende of Civil, Environmental and Architectural Engineering,
Kopibolzendübel unter Quer- und Längsschub im University of Colorado at Boulder.
Brückenbau. Institute of Structural Design. University of Schultz A. E., Hajjar, J.F., Saari, W.K. and Shield, CK 2000.
Stuttgart. No. 2001-1. Research Report financed by the Design formulas for the strength of headed stud connections
BASt. in composite infilled frames. Composite Construction in
Kuhlmann, U., Kürschner, K. 2001b. Behavior of Lying Shear Steel and Concrete TV - Proc. of Engrg. Found.
Studs in Reinforced Concrete Slabs. Symposium on Conference, Banff, May 28 - June 2, Banff, Alberta,
Connections between Steel and Concrete. 55u»Rilem Annual Canada.
Week. Stuttgart. 1076-1085. Taplin, G. and Grundy, P. 1997. Incremental slip of stud shear
Kuhlmann, U.; Gerold, M.; Schänzlin, J. 2002. Trag- und connector under repeated loading, in Composite
Verformungsverhalten von Brettstapel-Beton- Construction - Conventional and Innovative, Proc. Int.
Verbunddecken, Bauingenieur 77, Hl, S. 22-34. Conf. Innsbruck, September, 145-150.
Kunnath, S. K, Reinhorn, A. M., and Lobo, R. F. 1992. Taplin, G. and Grundy, P. 2000. Steel-concrete composite
IDARC Version 3.0: A Program for the Inelastic Damage beams under repeated load. Composite Construction in
Analysis of Reinforced Concrete Structures. Technical Steel and Concrete TV Proc. of Engrg. Found.
Report NCEER-92-0022, State University of New York at Conference, Banff, May 28 - June 2, Banff, Alberta,
Buffalo. Canada.
Lemaitre, J., 1992. A course on damage mechanics. Springer- Technical Committee 1, TWG 1.3 1986. Recommended testing
Verlag. Berlin. Leon, R. T., 1998. Analysis and design procedures for assessing the behaviour of structural steel
problems for PR composite frames subjected to seismic elements under cyclic loads. ECCS, No. 45.
loads. Engineering Structures, 20(4-6), 364-371. Uang, C-M., Yu, Q-S., Noel. S. and Gross, J., D. 2000. Cyclic
Machacek, j., Studnicka, J., Kipata, A. and S vitakova, M., testing of steel moment connections rehabilitated with RBS
2000. Perforated shear connector for composite steel and or welded haunch. /. Struct. Engrg., ASCE, 126(1), 57-68.
concrete beams, Composite and Hybrid Structures - Proc. Viest, I.M., Colaço, J.P., Furlong, R.W., Griffis, L.G., Leon,
of 6th ASCCS Conference, Los Angeles, March 22-24, 297- R.T. and Wyllie L.A. 1997. Composite construction design
304, ASCCS-6, Los Angeles, CA. for buildings, ASCE, McGraw-Hill. New York.
Manfredi, G., Fabbrocino, G. and Cosenza, E. 1999. Modeling Wöhler Α. 1871. Tests to determine the forces acting on
of steel-concrete composite beams under negative bending. railway carriage axles and the capacity of resistance of the
J. of Engrg. Mech. ASCE, 125(6), 654-662. axles." Engineering, 1 1 .
Matus, R.A. and Jullien, J.F., 1996. An experimental and Wu, H., and Hosain, M.U. 2000. Composite beams with
analytical study of a new shear connector for composite parallel wide ribbed metal deck: push-out and full size beam
steel-concrete beams. Proc. of Int. Conf. On Advances in tests. Composite Construction in Steel and Concrete TV -
Steel Structures, 1 1-14 December, Hong Kong. Proc. of Engrg. Found. Conference, Banff, May 28 - June
Matus, R.A. and Jullien, J.F. 2000. A high seismic performance 2, Banff, Alberta, Canada.
shear connector for composite steel-concrete structures Zandonini, R. and Bursi, O.S. 2000. Cyclic behavior of headed
subjected to strong earthquakes, Composite Construction in stud shear connectors. Composite Construction in Steel and
Steel and Concrete TV - Proc. of Engrg. Found. Concrete IV - Proc. of Engrg. Found. Conference, Banff,
Conference, Banff, May 28 - June 2, Banff, Alberta, May 28 - June 2, Banff, Alberta, Canada. 470-482.
Canada.
Miner, M. A. 1945. Cumulative damage in fatigue. Trans.
ASME, J. of Applied Mechanics, 67, A159-A164.
Oehler, D.J. and Coughlan, CG. 1986. The shear stiffhess of
stud shear connections in composite beams. J. of Construct.
Steel Research. 6(4), 273-284.
Oehler, D.J. 1991. Deterioration in strength of stud connectors
in composite bridge beams. J. Struct. Engrg., 116, 12,
3417-3431.
Oehler, D.J. and Bradford, M.A. 1995. Composite Steel and
Concrete Structural Members Fundamental Behaviour.
lstedn Elsevier Science.
Oguejiofor, E.C and Hosain, M.U. 1994. A parametric study of
perforbond rib shear connectors. Canadian J. of Civil. Eng.,
614-625.
Ollgaard, J.G., Slutter, R.G. and Fisher, J.W. 1971. Shear
strength of stud connectors in lightweight and normal
weight concrete. AISC Engrg. J., 8(2), 55-64.
Park, Y. J, and Ang, A. H.-S. 1985. Mechanistic seismic
damage model for reinforced concrete. J. Struct. Engrg.,
ASCE, 111(4), 722-739.
Rex, CO. and Easterling, S. 2000. Behavior and modeling of
reinforced composite slab in tension. J. Struct. Engrg., 126,
7, 764-771.
Rovnak, M., Duricova, A. and Ivanco, V. 2000. Nontraditional
shear connections in steel-concrete composite structures.
Composite and Hybrid Structures - Proc. of 6th ASCCS
Conference, Los Angeles, March 22-24, 1075- 1082,
ASCCS-6, Los Angeles, CA.
Salari, M.R. 1999 Modeling of bond-slip in steel-concrete
composite beams and reinforcing bars. Ph.D. Thesis, School

36
Repairing and strengthening for new requirements: use of mixed technologies
A. MANDARA
Department of Civil Engineering, Second University of Naples, Italy.

MUZEAU
J.P.
Department of Civil Engineering, CUST, Blaise Pascal University, Clermont-Ferrand, France

P. PERDLKARIS
Department of Civil Engineering, University of Thessaly, Volos, Greece

M. PIAZZA
Department of Mechanical and Structural Engineering, University of Trento, Italy.

C. SCHAUR
A-MBT Application Centre Mixed Building Technology ltd., University of Innsbruck, Austria

ABSTRACT: Basic principles and main application techniques underlying the use of mixed technologies in
the field of structural rehabilitation are shown and discussed in this paper. In particular, the current state-of-
the-art on the combination of possible intervention techniques is pointed out, with special emphasis on the use
of advanced materials such as special metals, fibre reinforced plastics and innovative connection systems. Re¬
inforcement of existing timber members is also addressed. Some practical examples of rehabilitation of exist¬
ing works are also presented, in order to highlight the potentials of mixed technologies in solving problems
occurring in structural restoration.

1 INTRODUCTION the use of modern and advanced solutions for

In the past, rehabilitation techniques for existing
buildings have been applied in such a way to restore
the original features of the work as is was at the end
of the construction stage. Such a practice embodies
still today the main policy for interventions carried
out on the building heritage and in particular on
monumental constructions, where saving as much as
possible of existing historical and architectural val¬
ues is the main purpose of the restoration project.
This has led, as a consequence, to a generalised use
of so-called "traditional technologies", mainly based
on the use of materials - such as masonry, stones
and mortars and techniques similar to those
adopted in the construction period of the building.
Nevertheless, when the upgrading of either structural
or functional features is explicitly required, then the
use of more advanced techniques, relying on the use
of modern materials and technologies is going to be
more and more common in the field of rehabilita¬
tion. The reasons for this new trend are manifold:
they include, in first place, the necessity to fulfil
more severe safety and/or living standards. Second,
with regard to the structural aspect only, which cov¬
ers the subject of this paper, the main motivations to
37
strengthening can be summarised as follows:
- Existing buildings, sometimes, do not possess
structural properties adequate to comply with
current structural codification;
- Existing buildings, as a rule, do nor offer ade¬
quate resistance to those actions not considered at
the design stage, e.g. a strong increase of live
load due to change of use;
- Existing buildings, most of times, are conceived
in such a way to be scarcely prone to modifica¬
tions ofthe structural layout, unless special provi¬
sions are adopted;
- Both structural scheme and materials adopted in
existing buildings, in most cases, do not allow to
resist significant seismic actions, except in those
cases designed with a specific account for anti-
seismic provisions (this normally occurs in case
of very recent constructions, only);
- Existing constructions, sometimes, suffer damage
and/or degradation of both materials and struc¬
tural elements, resulting in a global performance
even lower than that exhibited at the construction
stage.
Such considerations let easily understand the need
for solutions other than "traditional". The new ser-
vice conditions, as well as the new structural re¬
quirements, in fact, in most cases cannot be met by
means of a simple "conservative restoration", but
ask for more advanced solutions able to satisfy in¬
creased operational standards. The common re¬
quirements usually involve not only structural per¬
formance, reliability and durability of intervention
techniques, but also ease of inspection, maintenance
and monitoring. Also, in some cases, restoration op¬
erations are requested to be easily removed and sub¬
stituted, in order to avoid inappropriate interven¬
tions, or simply to proceed to a future transformation
of the building for a new use. Similarly, the eco¬
nomic necessity is sometimes felt to recycle materi¬
als for different purposes.
Considering the predominant constructional features
of structural elements in existing buildings nor¬
mally falling within the categories of masonry, r.c,
timber and steel the application of advanced sys¬
tems, materials and technologies usually results in
the use of "mixed" technologies, where the term
"mixed" is related to the inherent "composite" char¬
acter of the final assemblage. In this spirit, new
members, endowed with their own properties and
peculiarities, are added in order to supplement or
complement the existing ones with particular fea¬
tures. Similarly, integrative structures conceived
with up-to-dated criteria, or even advanced energy
control systems based of the use of special devices,
can be used for improving the performance of build¬
ings with regard to particular structural require¬
ments, such as the seismic actions. In this frame¬
work, two possible options in the use of mixed tech¬
nologies in structural rehabilitation are considered
herein. The first one is based on the use of innova¬
tive materials, including those with very special fea¬
tures (namely special metals, fibre reinforced plas¬
tics, composite materials, ecc.), in order to fulfil
quite particular strengthening requirements with
some special properties, typical of each material,
which can be tailored to the specific problem under
consideration. To this purpose, materials to be used
in mixed technologies should possess high strength,
ease of transportation and application, lightness,
easy market availability and, possibly, ductility and
reversibility. The second option is concerned with
the use of advanced solutions for obtaining a higher
performance level from existing structures. Such so¬
lutions include, as an example, special connections
systems, which have been developed in recent years.
This paper focuses the main aspects related to the
application of mixed technologies in the field of
structural rehabilitation. In particular, it addresses
some aspects of the research activity carried out by
authors, currently presented and discussed within the
meetings of WG1 and WG2 of COST C12 action
38
Table 1. Possible combinations between materials
Applied strengthening system
Base material
Concrete Metals Composite materials
Masonry ·
Concrete ·
Timber · ·
Steel ·
2 BASIC ASPECTS OF MLXED TECHNOLOGIES
From the structural point of view, the key point in
the application of mixed technologies to structural
problems is the mutual collaboration between exist¬
ing and added components. As far as rehabilitation
operations are concerned, two basic principles can
be followed, namely passive strengthening and ac¬
tive strengthening (Mazzolani, 2002, Mandara, 2002).
Passive strengthening aims at improving load carry¬
ing capacity of existing members simply by increas¬
ing either cross sectional areas or material mechani¬
cal properties. The full strength of the mixed assem¬
blage is exploited only through additional displace¬
ments, so that the effectiveness of such interventions
may involve a given amount of structural damage.
Typical practices are the insertion of coupling ele¬
ments alongside existing bearing elements in order
to enhance axial or flexural strength.
Unlike passive systems, active strengthening is
based on the introduction of an internal co-action
into the mixed system, so that a multi-axial state of
stress can take place. For this reason this practice is
able to develop its action without additional dis¬
placements, which is why it can contrast or even re¬
duce the damaging causes by means of suitable en¬
forcing procedures. Active systems can be imple¬
mented, for example, with steel tie-beams placed
alongside bearing walls or at the base of thrusting
structures, with prestressed steel bars into masonry
elements subjected to axial load, or by confining
compressed load-bearing masonry or r.c. elements.
All the above practices can be applied by combining
different materials and technologies according to
specific design requirements. Table 1 shows the
most significant combinations which can be consid¬
ered in practice when a mixed system is to be cre¬
ated. Such combinations have been selected among
those which maximise the mutual effect between
materials involved, providing in this way a good rea¬
son to adopt mixed technologies. A number of pos¬
sible combinations, nevertheless, has been not in¬
cluded in such framework, namely those based on
the used of materials and/or technologies character¬
ised by similar features (e.g. masonry or concrete
with special concrete mortar products, steel with
special metals, etc.). Even though, strictly speaking,
they belong to mixed technologies as well, their ef¬
fect is basically that to enhance the behaviour of the
base material, rather than creating a new system fit¬
ted with additional, improved features.
3 METALS FOR STRENGTHENING MASONRY
AND R.C. STRUCTURES
3.1 Strengthening with steel coupling elements
Steel elements can be used for both passive and ac¬
tive interventions. They offer ease of transportation
and erection, lightness and immediate availability of
load bearing capacity (Mazzolani, 1990). Coupling
techniques, based on the parallel arrangement of
new and old elements, are the most typical passive
technique. Coupling can be adopted for improving
the resistance against both vertical and horizontal ac¬
tions, including those cases when a portion or the
whole of the wall is removed due to new functional
requirements, in which case suitable provisions must
be adopted in order to substitute for the eliminated
parts. As the distribution of acting forces between
existing and added resisting elements depends upon
their mutual stiffness ratio, the introduction of new
elements can deeply affect the stiffness distribution
within the construction, so that a global change of
the structural layout can be required, including
foundation structures (Mandara & Mazzolani, 1991).
The insertion of braces for increasing the resistance
against horizontal loads is to be regarded as a further
case of coupling between additional and existing
structures. This is a typical retrofitting intervention
adopted in seismic areas to upgrade the safety level
of buildings in case of earthquake. In such applica¬
tion steel elements provide great effectiveness in
terms of stiffness with a negligible increase of
weight, especially when the concentric St. Andrew's
cross configuration is used (Figure 1). In some cases
it can be sometimes convenient to adopt less rigid
systems, in order to fit the global brace stiffness to
the building features. In particular, eccentric bracing
systems offer stiffness properties which can be eas¬
ily adjusted in order to match the actual horizontal
building stiffness. Also, they provide enhanced ulti¬
mate ductility as respect to the classical concentric
St Andrew's cross brace configuration.
3.2 Strengthening with metal tying elements
Iron ties can be considered as the oldest and most
common system of active consolidation, whose ef¬
fectiveness is worldwide recognised on the basis of a
millenary experience. Tie-beams can be profitably
used for many purposes, depending on the specified
design requirements. In particular, they can be in-
39
tended for linking together the walls of a masonry
building in order to enhance its box-like behaviour,
for improving the connection between floors and
vertical walls, for contrasting the thrust of arches
and vaults, or even for increasing the compressive
stress in the walls, in order to improve their strength
against shear actions. They can also be arranged ver¬
tically, so as to provide bearing walls with higher re¬
sistance to horizontal actions. Furthermore, in the
view of active interventions, tie-beams can be also
adopted for recovering both local and global dam¬
age, such as local spalling of masonry walls, arch or
vault displacements, wall out-of-plumbs and so on.
Steel
braces
Figure 1. Concentric steel braces for strengthening a masonry
building against horizontal actions (Mandara, 2002).
In general, the application of ties involves a signifi¬
cant increase of the global strength level of the con¬
struction. In addition, it also represents a fully re¬
versible intervention. Last, but not least, tying sys¬
tems are very cheap, whatever the technology
adopted, in both material and construction cost.
Confinement by tie-rods of load-bearing masonry
elements working in compression, such as columns,
males and walls, is very effective in terms of both
load bearing capacity and ultimate ductility. This is
due to the tri-axial state of stress occurring in the
element, which prevents masonry leaves from insta¬
bility phenomena and strongly reduces the onset of
cracks parallel to the direction of applied load. Con¬
finement can be applied in several ways, depending
on practical situations and performance requirements
(Mandara, 1992). The problem of transverse con¬
finement of masonry walls with internal steel tying
elements and end plates has been recently faced by
Mandara and Mazzolani, 1998. A theoretical model,
calibrated on the basis of both experimental and
F.E.M, numerical simulation, has been defined for Section C- C
predicting the behaviour up to failure of laterally
confined masonry walls under compressive load.
With a suitable 'choice of relevant parameters (a, a
and b), the model is able to interpret either experi¬
mental or numerical results with a satisfying degree Wooded spacer
of accuracy. The comparison is shown in Figure 2, Floor beam Cold formed profile
where curves referring to strength enhancement fac¬
tor due to confinement k, confinement stress σ« and
Poisson modulus ν are also plotted for a 800 x 800
mm masonry wall with thickness t = 500mm.

Ν
X
Cold formed
4 profile

Steel plet·

Woeoeii
refier

0 0.004 0.008 0.012 0.016 0.020 0.024

Figure 2. The model for confined masonry investigated by
Mandara & Mazzolani (1998) (top), and the comparison with
FEM results (bottom).

3.3 Strengthening offloors

Metal-based technologies are liable to be effectively
used in the strengthening of floors, beams and, in
general, of horizontal structures. The concept under¬ ¿at*- UK- .t^ufi

lying mixed applications can be in this case fully ex¬

ploited to make for some inherent lacks of these
structures such as, for example, inadequate load
bearing capacity or poor in-plane stiffness. Timber
floors are often typical of very old masonry build¬
ings and, for this reason, they can frequently exhibit
important damage conditions, which are strictly re¬
lated to their inherent features: high creep deforma¬
tions of both principal and complementary structure,
high susceptibility to biological degradation, easy
degradation of end restraints due to vibrations and
wood rotting, deterioration ofthe internal connection
between floor structure and filler layer, settling of
bearing walls and consequent floor damage.
40
J"
Figure 3. Strengthening of timber floors with steel elements
(Mazzolani & Mandara, 1991, 1992, Mandara, 2002).
Some possible solutions, based on the use of addi¬
tional steel profiles, are shown in Figure 3. Steel
profiles can be also applied upward, and connected
to timber beams by means of suitable connectors,
made of ties or nails. If such connectors are properly
designed so as to absorb shear slip, then a real com¬
posite system is obtained, which greatly improves
both strength and out-of-plane stiffness of the floor
structure (see also Section 4). The floor in-plane
stiffness can be improved by ad-hoc operations, such
as an additional timber boarding or a concrete up¬
ward cast, steel sheeting, or diagonal steel ties.
3.4 Strengthening of r.c. structures
Concerning damage conditions, it should be said that
r.c. structures seldom suffer severe structural prob¬
lems, apart from those conditions where highly ag¬
gressive environments can produce material degra¬
dation by chemical attack, which rarely occur into
buildings. In most cases a need for an upgrading of
such structures raises when higher loads must be re¬
sisted. This happens, for example, when a more se¬
vere use is planned for the building or when a seis¬
mic retrofit is necessary.
A|
M20 tl·»
¿deceive bonded
steel sheeting
Adheilve bonding
Section C - C
Κ.
'Ή /«deceive bended iteci etere
Figure 4. Additional reinforcement for r.c. beams with external
steel flats or cold formed profiles (Mazzolani & Mandara, 1991).
In the view of mixed technologies, steel elements
can be profitably used to supplement existing cross
sections in either tension and/or compression (Figure
4). Steel do have high strength, high stiffness, light¬
ness, immediate availability of load bearing capac¬
ity, chemical stability with concrete and negligible
creep. When used on both top and bottom of beams,
such elements can be connected to each other by
means of bolted ties working as common stirrups in
r.c. members. In order to avoid slip, additional steel
elements can be joined to the r.c. beam by means of
expansion bolts or special fasteners. As an alterna¬
tive, innovative products can be also used, such as
for example epoxy resins or adhesive bonding mor¬
tar. In such a case resins advantageously substitute
for mechanical fasteners (e.g. expansion bolts), that
could involve high stress concentration into con¬
crete. The cross section of steel profiles can be either
simply flat or variously shaped, depending on design
requirements. Cold formed profiles can be adopted,
in order to encircle the whole beam downside. In
some cases, when an increase of load bearing capac¬
ity in bending is only required, simple tie rods can
be allocated downside the beams with endpoints fas¬
tened to bolted contrasting elements (Figure 4). Such
application has the advantage to be easily applied
without interrupting the use ofthe structure. In addi¬
tion, it can be adopted for a local reinforcement and
is also completely reversible. In this application it is
recommended that the whole amount of shear in the
floor is born by existing concrete.
In case of r.c. columns, goals ofthe intervention can
be either restoring the original load-bearing capacity
of the column (in case of damage) or increasing the
load bearing capacity of the column, when a higher
structural performance is required. Damage in r.c.
elements can be purely external, when it is limited to
Adbestv« bending
the outer perimeter without involving the inner core
of the member, but also internal, when deep degra¬
CM fermd - dation of the element body is present. When damage
f·*1«
is limited to the external concrete cover and no in¬
u crease of acting load is planned, the cross section is
usually kept as it is and intervention is concerned
with the local repairing of damaged parts with ex¬
pansive concrete and/or resins. In such cases, con¬
finement with steel flat elements, according to the
^ technique of so-called "beton plaqué" can be used
for improving the load bearing capacity (Figure 5).
3.5 Interventions based on the use of special metals
In spite ofthe great effectiveness, steel elements are,
nevertheless, rather sensitive to corrosion, in particu¬
-si
lar when they are bonded outside members. They
also require a careful preparation of surfaces in order
to achieve the best performance. In addition, steel
elastic modulus and strength are sometimes too high
compared with masonry and concrete, and this can
cause problems in the stress transmission at the in¬
terface between materials. These considerations
have led in recent years to the use of special metals
for structural purposes in rehabilitation. Materials
most liable to this task are stainless steels, alumin¬
ium alloys, titanium alloys and shape memory al¬
loys. Despite their high cost, such materials meet in¬
creasing application in the field of structural reha¬
bilitation due to their peculiar features, which gener¬
ally offer several benefits when the whole lifetime of
the project is considered. For example, such metal
elements can be easily melted down and re-used for
different purposes. This contributes to increase their
long term economic sustainability compared with
traditional non-reversible technologies.
41

Welded
plates
Bottens
I
i&&'
-"C-profiles
Welded « formed
hoops
Bolts
Cold formed
profile
Angle profile
.
. Botten
Flat stiffener
?^: U^M'-*
AJÍ
Angle profile
Figure 5. Steel profiles for the strengthening of r.c. column and
beam-to-column nodes (Mazzolani & Mandara, 1991, 1992).
The typical features of all metals, namely great cor¬
rosion resistance, high strength-to-weight ratio, good
ductility, ease of fabrication and erection, aesthetic
appearance, reversibility, product availability, are
significantly enhanced with special metals. In addi¬
tion, all these materials possess architectural value,
thanks to their attractive look and nice harmonisation
with old materials (Mazzolani & Mandara, 2002).
Also, some of these metals, e.g. titanium alloys, al¬
low to reduce the effect of differential thermal
changes which can occur between the masonry and
the reinforcing elements due to the different thermal
expansion ratios. They offer, in fact, a very low lin¬
ear thermal expansion coefficient (ό+δχΙΟ^Ο0"1),
which is very similar to that of volcanic or meta-
morphic rocks, such as granite and marble. This al¬
lows titanium elements to be used in redundant or
prestressed systems with no risk to impair the effec¬
tiveness of intervention due to thermal changes. This
feature has been highly useful in the restoration of
the Parthenon in Athens and the Colonna Antonina
in Rome, where titanium alloy stirrups have been in¬
serted and hidden into existing stone blocks (Giuffrè
8c Martines, 1989). They proved to be far more ef¬
fective than conventional steel elements used before,
which had involved many cracks due to corrosion
and excess of thermal dilatation.
It is worth noting that the material choice should be
made not only on the basis of the range of mechani-
42
Flot
cal properties like strength, stiffness and ductility,
which is quite large within any family of metal ma¬
terials, but rather by considering other technological
factors, such as chemical-physical compatibility
with in-situ materials, corrosion resistance, as well
as the possibility to obtain particular surface appear¬
ances and shapes.
In the very last years, special devices based on the
Concrete
use of Shape Memory Alloys have been used in
bed combination with steel tying elements for the seis¬
mic protection of existing constructions (Indirli,
i 2000). Shape Memory Alloys (SMA), mostly Ni-Ti
or Cu-Al-Zn alloys, may be regarded as "smart" ma¬
Angle profile terials, as both their yield stress and modulus of elas¬
ticity strongly increase as long as temperature in¬
Batten creases within the so-called transformation tempera¬
ture range, due to a solid martensite-austenite phase
change. Such range is limited by M{ and Af, that is
' ¡»fes
by the temperatures where only full martensitic or
full austenitic structures can exist, respectively. The
above transformation can be induced by either me¬
Angle profile
chanical stress or temperature change, resulting in
the capability to recover, spontaneously or by heat¬
ing, large initial strains due to load (superelastic be¬
haviour, Figure 6a) with a corresponding amount of
dissipated energy due to different loading and
unloading paths. This allows the construction of
seismic protection devices (Croci et al., 2000, Dolce
et al., 2000, Pegon et al., 2000). A complete strain
recovery may also occur by heating material above
Af (memory effect, Figure 6b). This effect can be ex¬
ploited for impressing a state of co-active stress to
structural elements, by arranging SMA elements at a
temperature lower than Mf, where pure martensite
exists and material can be easily modelled up to the
required strain, and then heating it above Af, in order
to apply the required degree of co-action.
T<Af
unoading unloading
a% heating to
iiecumy iw Af
«f t ^
Figure 6. Basic behavioural principles of Shape Memory Al¬
loys (Mazzolani & Mandara, 2002).
4. STRENGTHENING OF EXISTING TIMBER
STRUCTURES WITH "NEW" MATERIALS
4.1 Introduction
Among the various techniques for the strengthening
of existing timber structures, the execution of "com-
posite" or "mixed" structures is currently proposed
as an effective restoration method to make up for
various static deficiencies, such as material decay
(damage or biological attack) or poor resistance and
stiffness of elements, or to adapt the existing struc¬
tural characteristics to new functional needs. Such
interventions are also in optimal accordance with the
current state of the art of the research and with the
normative requirements for the design and the exe¬
cution of works on ancient load bearing timber
structures (Piazza & del Senno, 2001).
When the "combination" of existing timber elements
and "new" materials is needed, the connection sys¬
tems play a leading role. In these cases, and when a
stiff connection is needed, traditional connectors are
inadequate and new kinds of connection are required
and must be used. The traditional connection sys¬
tems, both for ancient and modern wooden struc¬
tures, can be represented by carpentry joints (in
which the forces are transmitted directly from an
element to another, without special devices), or by
mechanical joints (in which special devices, usually
metal connectors, are used). These traditional con¬
nections certainly exhibit some advantages for the
designer and for the builder (the assembling easi¬
ness, a consequent building quickness, a sufficiently
large tolerances allowed when the joint is com¬
pletely assembled on site). On the other side, the
main disadvantages can be found in a limited
strength and stiffness of connections with metal de¬
vices acting in shear. Moreover, in the event of fire,
in which the mechanical characteristics of the por¬
tion of wood not charred remain unaltered, the pres¬
ence of not thermally insulated metal devices di¬
rectly exposed to the outside environment can seri¬
ously compromise the overall stability of the struc¬
ture or reduce its load carrying capacity.
Some of the most interesting new types of connec¬
tions, that have been proposed for possible future
development, are those utilizing glued-in steel ele¬
ments (bars or plates), completely hidden inside the
timber. There are now many examples of these con¬
nections in the field of modern constructions in glu-
lam (Buchanan & Fairweather, 1992, Natterer,
Herzog & Volz, 1994, Riberholt, 1986, 1988). Also
the interesting applications in the field of structural
repair and rehabilitation should not be forgotten (Pi¬
azza & Turrini, 1986, Piazza, 1990).
Although these new types of joints are usually more
expensive than the traditional ones, they can offer
some unquestionable advantages, when adopted in
existing structures, which can justify their use and
may lead to overall economical benefits: 1) high
connection stiffness values without significant set¬
tling; 2) the possibility of a ductile design of joints
through the yielding of steel elements (despite the
fact that the timber-glue-steel connection has an
elastic-brittle behaviour); 3) the potentials of these
43
kinds of joints against chemicals and fire, since the
glue and the steel elements are protected by the tim¬
ber; 4) the improved architectural aspect ofthe over¬
all timber structure since all the steel parts are hid¬
den inside the wooden elements.
On the other hand, it is necessary to mention two of
the main problems that can afflict this type of con¬
nections: the stresses induced by the dimensional
changes of wood due to moisture variations, and the
difficulties of assembling and gluing on site.
4.2 The basic connections with glued-in steel ele¬
ments
These connections have been investigated, experi¬
mentally and numerically, in previous researches
with different glue formulations (all two components
epoxy-based resins) (Cavalli et al., 2001). The basic
connections investigated can be divided (according
to the direction of the force they transfer) into axial
and shear connections. With regard to the former,
the force is transferred parallel to the steel bar axis
and parallel to the grain; with regard to the latter, the
force is at right angles to the axis of the steel bar.
These investigations have made possible, among
other things, to: 1) define the resistant mechanism of
the basic single joint, and both its strength and stiff¬
ness characteristics; 2) select the adhesive formula¬
tion with the best characteristics for the use from a
family of epoxy-type glues already identified for
other specific (not mechanical) properties; 3) verify
the technological problems related to gluing (by
pouring or injection) and ascertain the possible
variations of the results in terms of mechanical per¬
formance.
Concerning glued-in parallel-to-the-grain steel bars
subjected to tensile forces, as outlined by a number
of researches done on these kinds of connections
(Riberholt, 1986 & 1988, Buchanan et al., 1990,
Ballerini et al., 1994), the bearing capacity for ten¬
sile forces is essentially based on the adherence be¬
tween the three materials (wood-glue and glue-steel)
and, hence, on the development of shear stresses τ in
the layer of glue, which show - unfortunately - a non
uniform distribution. The bearing capacity ofthe ba¬
sic connection can be limited by the following fail¬
ure modes: 1) the pull out ofthe steel element and of
parts of the glue layer due to a partial failure of the
steel-glue adherence mechanism; 2) the pull out of
the steel element, the glue layer and some wood fi¬
bres due to the attainment ofthe shear strength in the
wood; 3) the pull out of the steel element, the glue
layer and some wood fibres due to the attainment of
the tensile strength perpendicular to the grain of the
wood which causes splitting failure in timber.
The crisis mentioned in (1) is usually excluded if the
metallic elements used have a very rough surface
and/or are fairly clean: this is the case of threaded or
ribbed bars. But a good glue-steel adhesion is far
more difficult to achieve when the steel surface is
large (like for plates), and when cleaning by sand¬
blasting may be insufficient if not followed immedi¬
ately by gluing. Then the first type of failure is, ba¬
sically, a sign of faulty gluing.
The types of failures spoken of in (2) and (3) are not
always physically distinguishable since the pull out
of the steel element nearly always takes place simul¬
taneously with the transversal splitting ofthe timber
element. In point (3), it is the actual splitting phe¬
nomenon that causes the subsequent pull out of the
steel element, while in (2) the splitting is the conse¬
quence of the opening mechanism which arises from
the pull out of the steel bar with the glue and some
wooden fibres, as shown in Figure 7. Failure mode
(3) is usually characterized by lower carrying capac¬
ity with respect to failure mode (2).
Figure 7. Tests on the basic connection under a tensile force,
before gluing (a) and after test (b) (Ballerini et al., 1994).
F(x)/F
0.8
\^ «^^ ·
A Experimental Test 1
Experimental Test 2
Experimental Test 3
0.6 *"^-
χ/φ
Figure 8. Tensile force along the glued-in bar: comparison be¬
tween numerical and physical results (Ballerini et al., 1999).
The typical numerical trend ofthe axial force F(x) in
the steel bar over the glued length, is reported in
Figure 8. The numerical results have been obtained
through a 3-D linear-elastic finite element model by
means of the ABAQUS code (Piazza et al., 2001).
Due to the great difference of the elastic Young's
modulus for different directions with respect to the
grain, the analyses have been performed assuming
for the wood an orthotropic behaviour with a cylin¬
drical symmetry while the glue has been considered
to have an isotropic behaviour.
When glued-in ρεφ6^ΐουΐ3Γ-Ιο-τ,1ιε^3Ϊη steel bars
subjected to shear forces are faced (Figure 9), as it is
easy to expect, the numerical model of this connec¬
tion is considerably complex both for the necessity
44
of inserting materials with different constitutive rela¬
tions and for the problems related to the interface
elements between various materials, but also for the
problems related to the numerical stability. Concern¬
ing the finite element formulation, some 3-D and 2-
D models have been set up using ABAQUS.
*t *»
,
.A
»*
4 ! '-·*
/» ·
i* *< *>
'- 'i.
»
Figure 9. Section ofthe specimen after the test, compared with
the results of a F.E. model at different load levels (Ballerini et
al., 1999).
In these models, two different behaviours are fore¬
seen for the glue layer: in the axial direction of the
steel pin and in the direction orthogonal to it. In the
first direction a mono-dimensional elastic element
has been assumed since the glue is in any case able
to prevent the slipping ofthe pin, also in presence of
permanent gaps near the surface of separation of the
two timber elements due to cyclic loads. In the sec¬
ond direction some mono-dimensional interface
elements, able to take into account the elastic behav¬
iour of the glue and the elasto-plastic unilateral be¬
haviour of the wood material, are used. This has
been done combining rigid elements, gap elements
(with an unilateral behaviour) and spring elements
with elasto-plastic stress-strain relations, that can re¬
produce permanent deformations in the wood
(Ballerini et al., 1999).
The connection with glued-in bars in shear can solve
also the problem of connection efficiency when used
in the well known timber-based composite elements
for the floor structures. The by now classic solutions
of "dry" connections (i.e. nails, bolts, dowels) have
in general shown not so satisfactory behaviour for
the problem dealt with.
4.3 Strengthening of timber members
A number of researches have been devoted to the at¬
tempt of improving the stiffness and the strength of
the single timber element, in order to fit more severe
structural requirements or for solving problems
caused by material "decay" (for biological attacks)
or by "damage" (fissures, fibre discontinuity, etc.).
A possible answer to many different prerequisites
(fire protection, aesthetical requirements, etc.) can F (KN)
160
be given by reinforcements with different materials jS^ > -p
completely embedded into the wood. 140

In Figure 10 the cross section of an old timber beam 120

(about 4.5 m long) is presented, with embedded steel 100

reinforcements glued along the whole element. In 80

Figure 1 1 the cross section of another old timber 60

beam (about 5 m long) is shown, with embedded 40

CFRP reinforcements lamellae, glued along the 20
whole element. In both cases, the epoxy adhesive 0 10 20 30 40 50 60 70 80 90 100 110
<5(mm)
have been utilized after the preliminary application
F(kN)
of an epoxy primer. The diagrams in Figure 12 (ap¬ 110
plied load F versus midspan deflection δ) show the 100

mechanical behaviour up to failure of both strength¬

ened timber beams, under the so called four point
bending test (EN 408). Obviously it is not possible
to make a direct comparison between the two tests,
but it is evident the very different behaviour of the
element with the embedded steel bars, which pre¬
sents also a marked difference from the well known
elastic-brittle behaviour of timber in bending.
0 5 10 15 20 25 30 35 40 45 50 55
<5{mm)
Figure 12. Mechanical behaviour of the timber beam strength¬
Surface treatment
ened with embedded steel bars (top) and with CFRP rein¬
\ with epoxy resin
forcements (bottom).

If the decay of timber is limited, the refurbishment

of existing floor structures is currently carried-out by
means of a concrete slab joined to the timber beams
by an adequate connection system (Figure 13). Usu¬
ally, the main needs which drive to the consolidation
i»me.n = 307 mm
of the floor structures are associated to lack of stiff¬
Figure 10. Cross section of a timber beam with embedded steel ness and strength with respect to the current stan¬
reinforcements. dards. Furthermore, the use of the concrete slab is
Surface treatment
,:i. '...«- !-,»
particularly effective in refurbishment, since it al¬
with epoxy resin
lows to link the masonry walls, to improve the
acoustic insulation of the floor obtained (also by
so
n-n means of its mass), to guarantee the incombustible
^Μ^ι^Μτ*!^ ι4 ja
element between different storeys (Piazza et al.,
£===¡3
CFRP lamella 1 .4 χ 50 mm ¡
1997). From the end of the seventies a number of
glued with epoxy adhesive ! theoretical and experimental investigations have
ι Timber element cross-section:
been carried-out to characterize different techniques
bmcm= 200 mm; íWm»= 2 10 mm
of restoration. Among the others, it must be men¬
Figure 11. Cross section of a timber beam with embedded tioned the mixed timber-concrete solution with steel
CFRP reinforcements. ribbed bar connectors epoxy-glued into the wood
(Piazza & Turrini, 1986), and the mixed timber-
4.4 Strengthening of members by composite systems timber solution, with different wooden slabs (ply¬
The use of a composite system is one ofthe most ef¬ wood, boards, etc.) and various connectors (Piazza,
fective method to reinforce floor timber beams in 1994). The main benefits achievable from the mixed
bending. It is evident that the correct restoration pro¬ structure, regarding its static performance, are: 1) a
cedure must obtain the desired results maintaining considerable improving of the strength and stiffness
for the existing timber elements their own, essential of the floor; 2) a greater sharing capacity of the slab
static function: thus composite structures, through with respect to the applied vertical loads; 3) a sig¬
the combination of existing timber beams with slabs nificant capacity, for the slab acting as an horizontal
made with other materials, are particularly interest¬ diaphragm, of linking the main vertical bearing ele¬
ing. ments ofthe building.
45
 properties is very low. If fracture of a fibre occurs in
the strip the progression of the crack will not spread
(E out like in other solid bodies, because neighbouring fi¬
bres stay intact and take over the stresses. The impec¬
Welded steel mesh
cable embedding of fibres in the matrix leads to the
Concrete slab
fact that the fibre can take over loading again very
close to the point of failure. The resistance against
chemical attacks is also very high. Therefore CFRP
CB
strips can be seen as durable towards pollutants
Metallic connectors which arise under normal conditions in buildings.

-^5*. 't·"'

CE

Figure 13. Commonly used glued-in connectors in a timber -

concrete composite structure (Piazza, 1994).

5 USE OF COMPOSITES
5.1 Generals

Since the middle of the nineties composites have

found their position as strengthening materials. In
Figure 14. REM picture of a CFRP strip.
the last decade thousands of structures have been
strengthened by means of externally bonded FRP
(Fibre Reinforced Plastics) reinforcement. The main
advantages of these materials are their very low
weight, a minimum of structural height, no corro¬
sion, easy application and simple use on the con¬
struction site. FRP systems are used in the form of
prefabricated elements (straight strips, shaped shells,
jackets or angels, wet lay-up systems) fibre sheet, /
fabric and fibre tows, and some special systems like
prestressed strips, etc. The suitability of each system
depends on the type of structure that shall be
strengthened. Practical execution and application
conditions, like cleanness and temperature, are very
important in achieving a good bond. Figure 15. Roving and CFRP strip.
The commonly used procedure for manufacturing
CFRP (Carbon FRP) strips is called pultrusion. Dur¬ Strips are manufactured with a width of 50mm up to
ing this continuous process rovings are saturated 150mm and a thickness of 1,2 mm and 1,4mm. They
with resin and pulled through a heated die, where the are laterally smoothed and tied on reels with a length
part is formed and cured. Traded strips consist of, of 250m. For example the total weight of strips with
depending on the width, 1 up to 3 millions of carbon a width of 50mm, a thickness of 1,2mm and a length
fibres with a diameter of 0,006 mm. The resin or the of 250m amounts only to 25kg per reel.
so called matrix binds together the fibres and is re¬ Likewise mechanical properties, thermal and electri¬
sponsible for the shape ofthe strips ( Figure 14). The cal properties show an orthotropic behaviour. In the
created material shows excellent unidirectional longitudinal direction the fibres conduct heat and
properties, which are determined by the content of electricity, in transverse direction they act like an in¬
fibres, which is normally about 70 %. CFRP strips sulating layer. This property is used for controlled
possess a nearly ideal elastic-plastic behaviour with¬ fast curing by means of special heating devices.
out any significant yielding or plastic deformations, Through the fibres the adhesive is heated up and its
connected with the tendency of brittle failure modes. chemical reaction is accelerated considerably.
As a result of the great number of fibres in the same Thanks to this effect externally bonded FRP can be
direction (Figure 15), the scatter of mechanical prop- applied within a few hours instead of days, which is
46
very important for bridges, industrial buildings or in Other strips usually require abrading and wiping
general where service interruptions have to be kept clean before use. In general they have to be verified
as short as possible. for possible damage resulting from transportation or
For more than 80 % of all applications, a standard handling. The adhesive is applied as a thin layer
CFRP strip with lowest modulus of elasticity is used. immediately after mixing. Thereby the roughened
Only if the verification of deflections or of crack surface is totally moistened. The adhesive is also ap¬
widths is the design ruling criterion, strips with plied to the strips in a dome shape, having slightly
higher properties are brought into play. more thickness along the centre line of the plate.
5.2 Installation problems This reduces the risk of forming voids when the
strips are applied. The strips are offered to the sur¬
Strengthening of existing buildings have to be car¬ face, applying pressure by means of rubber roller to
ried out in most cases under cramped conditions. In ensure good contact (Figure 16). The extra adhesive
many cases the workmanship is more difficult on is squeezed out along the sides and has to be re¬
account of existing installations or service lines. Due moved. In many cases strips have to be applied
to the flexibility of CFRP strips, it is possible to ap¬ crosswise. These crossings can be easily carried out
ply also elements of large dimensions without any by changing the thickness of the adhesive gradually.
special joint configurations. Special attention has to The second layer can be glued on the related strip at
be given independently of whether FRP products are the crossings. The occurring strains of the strips un¬
applied on reinforced concrete, masonry, timber or der design loads are too low to cause longitudinal
steel on the preparation ofthe surface. The condition cracks on account of transverse strains.
to provide an adequate bond with the adhesive is a
bearing substrate. It is important to make out and to
remove existing weak points.
The method of surface treatment depends on the ex¬
isting substrate. Concrete and masonry are prepared
preferably by means of high pressure blasting (water
or sand) or grinding. These methods produce an op¬
timal roughness of about 1-2 mm. The prepared sur¬
face should be dust free and normally dry. In those
cases where serious imperfections or unevenness oc¬
cur the related areas have to be prepared by applica¬
tion of special primers. Such products are offered by
the same suppliers as the strengthening composites.
A primer should be used according to the specifica¬
tions given by the manufacturer.
The available tensile strength on the surface is
measured by means of pull-off testing especially at
the anchoring points ofthe strips. The minimum ob¬
tained strength should be l,5N/mm 2 for the applica¬
tion of strips on concrete.
To avoid loss ofthe adhesive tensile strength, the as¬
sessment of the surface humidity is very important.
To prevent sweating on the strips the thawing tem¬
perature has to be determined. The influencing pa¬
rameters are the temperature and the humidity of the Figure 16. Application ofthe adhesive on substrates (top) and
air as well as the temperature of the surface. It is of CFRP strips (bottom).
only allowed to apply strips at a temperature which
is 3° C higher than the thawing temperature. These For some structures, profit can be taken bonding the
conditions have to be guaranteed during the whole strips on the surface in a prestressed state. The ser¬
time of the adhesives hardening, otherwise the sub¬ viceability of the strengthened elements is improved
strate has to be warmed up. by the prestressing force. Analogous to a conven¬
As a rule, one surface is already roughened by the tional precompression, the result is a convenient
manufacturer. The strips should be free from any change of the inherent stress distribution. By means
contamination. Strips can also be provided with an of this, the interior reinforcement is relieved, defor¬
in-built peel ply, to ensure a clean surface. The ply mations are reduced and existing cracks might be
should be removed immediately before application. closed again. The load bearing capacity increases
47
because the mutual vertical displacement of the
crack edge zones is hardened. Through this, the
peeling off caused at shear cracks can be reduced
and further advantages can by taken from the high
mechanical properties of the strips. The design and
construction of the end zones require for prestressed
strips special attention (Figure 17).
Many structures show not only a low flexural resis¬
tance but also the need for shear strengthening. For
beams pre-manufactured angles are used (Figure
18). CFRP angles differ from the straight strips be¬
side the shaping also from the another process of
manufacturing, but they are also made with a matrix
of epoxy. The surfaces of the angles are provided
with an in-built peel ply, which is removed immedi¬
ately before application. Proper anchorage in the
compression zone of a beam is strongly recom¬
mended. Where it is possible, the anchoring should
use the whole height of the compression zone, oth¬
erwise the usable height has to be reduced, so that
the member has a fictitious lower ultimate bending
resistance.
**>
I has not to take over such high anchoring strengths.
with a radius of minimum 25 mm. The dry anchor¬
ing areas are filled with the adhesive and the end of
the angles is introduced. These ends were also pro¬
vided before with the same adhesive after removal
of the in-built peel ply. Angles are applied in pairs,
overlapping in the tension zone of the structure and
also pressed with a rubber roller. Thanks to this ar¬
rangement it is possible to compensate variable
thickness of beam webs (Figure 19).
Material properties of CFRP fabrics and sheets are
determined by the way of weaving. The used fibres
or threads are crossing continuously during the proc¬
ess, where the threads are conducted alternating be¬
low and above (Figure 20). Hence fibres are not to¬
tally stretched and a little reduction of the stiffness
has to be put up. Where fibres run predominantly in
one direction they are named unidirectional. Where
fibres run in at least two directions multidirectional
fabric is produced.
Unidirectional fabrics are primary used for shear
strengthening of beams and confinement of columns.
In principle also flexural strengthening is possible,
'?.'
like with CFRP strips. In such cases the substrate

Figure 17. Anchoring of prestressed strips.

Figure 19. Use of prefabricated angles

- i - >-··« t. - '

'A.ÌA .' , Al*1 - _* -

fj.'-'z.--· t'.'i^V-VvVv
1*7 -a
I .
*<

>
- ·

·, ι·*>-;·* S~]
-** -J',
\H
ih
'-'"'ÍV
*f " >4a » i»* A *j

Figure 18. Prefabricated angels for shear strengthening.

The practical operation for strengthening against

shear by means of prefabricated angles is similar to
the application of straight strips with some addi¬
tional steps. The surfaces of the substrate are pre¬
pared in the same way. Furthermore all structural
members have to be provided with corners rounded
48
Figure 20. CFRP fabrics.
A great practical importance have fabrics with glass
fibres. They are mainly conceived for applications
on substrates with low tensile strength like masonry
or natural stones. Also, hybrid fabrics of different
types of fibres are offered in the markets.
Sheets show in contrast to fabrics no redirections of
fibres caused by crossings. They are manufactured
with one side on a backup layer, which keeps the fi¬
bres in their positions. In the most cases sheets are
used where fibres run predominantly in one direc¬
tion. But also sheets with multidirectional fibres are
manufactured, where differently orientated sheets
are knitted together. The preparation of concrete or
masonry surfaces happens in the same way as de¬
scribed for the application of CFRP strips. Attention
has to be paid that the roughness after blasting or
grinding is not so high compared to the application
of strips, which own a higher inherent stiffness. In
this way they are less sensitive to unevenness. Fab¬
rics and sheets are very flexible and therefore will
follow any unevenness. As a result, implications of
the unevenness are more important. To achieve the
required evenness often a primer is applied to the
surface.
The prepared surfaces are moistened with a certain
quantity of mixed resin with a low viscosity. Crucial
for the resin is its thixotropic behaviour, especially
for works at vertical areas or above someone's head.
It is very important that the resin does not run off or
drop down. Fabrics or sheets are tailored to the
wanted size and the fibres are orientated in the de¬
signed later loading direction. Then the parts are ap¬
plied by pressing them manually onto the adhesive
in such a way that it is stretched without introduction
of voids. Impregnation and further pressing of the
sheets is performed by applying adhesive on top of
the fabric or sheet with a roller brush (Figure 21).
Figure 21. Overcoating of sheets.
One of the advantages of wet lay-up systems is the
possibility to follow also irregular geometrical
shapes or curves during application. This operation
could be necessary in old masonry or for arches and
vaults (Figure 22). The use of materials resisting in
tension only, like FRP, can be useful, in fact, to pre¬
vent the onset of plastic hinges in the structure and
however, to increase both load bearing capacity at
collapse and ultimate ductility. This may be very
important, for example, in case of unsymmetrical
load distribution, which, under certain circum¬
stances, can result in the stress resultant to go out the
thickness of the arch or vault. The tensioned side of
the hinge is effectively constrained by the bonded
fibre strip, which can rely upon a large adhesion area
over the arch or vault surface. Considering the pos¬
sible arising collapse mechanisms of these struc¬
tures, it would convenient to place the composites on
both sides ofthe member (Valluzzi et al., 2001).
Confinement of columns or in general of members
in compression by means of CFRP sheets, directly
applied (Figure 23), significantly enhances the per¬
formance under axial load, bending and shear, be¬
cause of the increase in concrete compressive
strength, the increase in ductility and the higher re¬
sistance against buckling of the steel reinforcement
in compression. Column strengthening design meth¬
ods are mainly based on the effect that CFRP sheets
cause a triaxial state of stress in the concrete, similar
to the use of a helical reinforcement. By means of
this, a relevant improvement of load bearing capac¬
ity concerning normal forces can be reached.
Figure 22. Strengthening of a historic vault.
«es·"-'! : *
*>.
' ' **"
Figure 23. Strengthening of columns with CFRP sheets.
6 ADVANCED STRUCTURAL CONNECTIONS tions. More, because their installation is controlled
by the failure of a breaking groove acting as a fuse,
6.1 General
and because of their shape, it is possible to check the
Repairing or strengthening of structures often re¬ quality ofthe swaging action only by visual verifica¬
quires: tion (Muzeau et al., 1996).
- to reinforce or to improve existing connections (to On the other hand, because of the swaging opera¬
comply with higher loading conditions or to rein¬ tion, the installed fastener is comparable to a screw-
force a joint after more or less severe degradations nut system without gap. So, in case of vibrations, the
for instance), swaged bolt behaviour is much more efficient than
- to create new supports (in the case of modifications the one of classical high-strength bolts, the rotation
in the use of a structure for example). ofthe collar being quite impossible.
Numerous solutions exist to solve these problems.
As they depend on the type of members or elements
that must be connected together and on the materials
concerned, each problem is specific and it is difficult
to extract general rules. So, to illustrate some techni¬
cal possibilities, some examples of connections are
given thereafter, chosen in the field of mixed tech¬
nology, with a particular concern to innovative sys¬
tems (Muzeau, 2002). In all cases, when connections
are modified, the structure must be carefully locally
and globally checked to insure that it is able to resist
to these modifications: the increasing in the resis¬ Insertion of the pin and Positioning ofthe
tance of a connection acts on its stiffness and on the positioning ofthe collar. installation tool.
load distribution which may be strongly affected. In
other words, the improvement of a connection may
affect the mechanical behaviour of the whole struc¬
ture. Concerning the reinforcement of existing con¬
nections, most of the solutions are based on upgrad¬
ing the fasteners or reinforcing critical zones.
Innovative fastening systems may present a real effi¬
ciency due to their mechanical properties and/or
their installation process. In particular, they may of¬
fã ' ,_»_, f *2s-ksL
fer real advantages in case of rehabilitation (prob¬
lems of accessibility around the connection or needs
for ductility for example). Some of them may also Swaging operation. Separation of the pintail.
be attractive regarding the improvement of the me¬ Figure 24. Classical swaged bolts installation process.
chanical behaviour in seismic conditions.
6.2 Classical swaged bolts
The installation process of classical swaged bolts ν,ααΠίίΐΠ'^
consists in swaging a collar on a pin with a special
hydraulic tool. The internal diameter of the nose as¬
sembly being smaller than the external diameter of Figure 25. Huck-Fit bolt before installation.
the collar, it forces the collar material to flow into
the pin grooves (Figure 24). If most of the classical
swaged bolts present mechanical characteristics
similar to ordinary bolts or high-strength bolts grade
8.8, the Huck-Fit bolts (Figure 25) present mechani¬
cal properties similar to prestressed high strength
bolts grade 10.9 (Czarnomska et al., 1993, Muzeau
et al, 1998). Figure 26 shows this type of bolt after
installation.
Swaged bolts may be an efficient alternative in res¬
toration because their automatic and not noisy instal¬
lation allows them to be used in restrictive condi- Figure 26. Installed Huch-Fit bolt.
50
Finally, the ductility of such bolts must be empha¬ 6.3 Blind swaged bolts
sised: in a bolt loaded in tension, the ultimate resis¬
Different kinds of blind swaged bolts are available:
tance is the shear stripping resistance of the collar
the blind oversized mechanically bolt (BOM) and
grooves and not the tension resistance of the
the high strength blind bolt (HSBB) (Korol et al,
threaded cross-section of the pin (Figure 27). There¬
1993, Muzeau, 2000). They are presented respec¬
fore, this behaviour leads to a ductile response and
tively in Figures 29 and 30, respectively.
not to the brittle one of classical high-strength bolts
in tension. An example of such behaviour is show in
Figure 28.
Pintail

Sleeve

Figure 29. BOM fastener.

Figure 27. Swaged bolt after failure in tension.

k Internal loac in the bolt (kN)

450
- Maximum load Washer
sleeve
400 I

Pin
"~ Γ ^ j
i
350
- Remaining resistane e
ί Pintail
300 sleeve
collar
ì
ί 1

| , \
250 Figure 30. HSBB bolt.
!
200 As before, if the final steps of their installation con¬
cern the swaging of a collar on the pin, the first one
150 !

! !
consists in creating a head in the blind zone. Fig¬
100 ure 31 shows the installation of a BOM and fig¬
! ;

f ì
ure 32 represents a HSBB after installation.
50 Z.\/j ^ ì

'» ί

i ί ί ι ·>·

υ 5 10 15 20 25 30
Displacement ofthe connected plates (mm)
Figure 28. Mechanical behaviour of swaged bolts in tension.

It may be observed that a residual resistance remains

after reaching the maximum internal load in the bolt.
This is due to the fact that this maximum load does
not correspond to the pin failure but to a progressive
slip of the collar on the grooved zone. This behav¬
iour may be advantageous in seismic conditions as
the failure is progressive and not sudden and a large
displacement is required to separate completely the
collar from the pin.
This specific behaviour and the smooth shape of the
locking grooves explain the results obtained under
fatigue tests in tension (Missoum et al., 1997, Bap
tista et al, 1997). Huck-Fit bolts provide fatigue re¬
sistance higher than the one of similar high-strength
bolts (from 2 to 8 times higher, depending on the di¬
ameter and on the stress range). Figure 3 1 . Installation of a BOM bolt.
51
 Before installation

J _ ilïBffi ffliîMtrZJl
.ÍjZZJ !. I ¡¡al1' J¡.
m
miu ι

After installation

IM
BlftpwiW

Figure 32. Installation of a HSBB.

The tension capacity of a BOM bolt is close to 80%

of the one of a high-strength bolt grade 8.8, the fail¬
ure occurring by shear off of the bulb created in the
sleeve. Due to a heat treatment of the pin, its shear
resistance is much higher (about twice that of a HS
bolt grade 8.8). As a consequence, this bolt is poten¬
tially well suited for shear connections.
A HSBB has been designed to provide a clamp load
action close to the one of a HS bolt, grade 8.8. This Figure 34. Ultra-Twist blind bolt.
bolt provides an efficient resistance in tension but
that its shear capacity is reduced due to a smaller pin
diameter. Resin

6.4 The Ultra-Twist blind bolt

As an improvement to the HSBB, a new fastener has
been developed. Its components are shown in Fig¬
ure 33.

sleeve
Core
Nut

Figure 35. Injection bolt.

- It allows reliable and relatively cheap non-slip

Bulb sleeve Shear washer
Figure 33. Ultra-Twist blind bolt.
The installation of such a bolt (Figure 34) is based
on the use ofthe standard tool for TC bolts (an elec¬
tric torsion wrench).
Compared to the BOM system, this fastener repre¬
sents an improvement both in tension and in shear
resistance. Being based on the use of the standard
tool used to install TC bolts, it does not require the
special hydraulic swaging equipment.
shear connections to be created.
- It does not require specific torque control system
to be used.
- In the case of large holes (relatively common in
refurbishment due to the difficulty to respect se¬
vere tolerances), the gaps are fulfilled.
- In the case of low slip factors, injection may help
to create high strength friction grip bolted con¬
nections. This is also a common case in refur¬
bishment when the connected plated are not re¬
moved to allow a surface treatment to be done.
- Internal corrosion is prevented.

6.5 Injection bolts

Injection bolts (Jaspart, 1994) present interest when
the connection needs to be slip resistant: the clear¬
ance between the bolt and the wall of the hole is
completely filled with a two component resin (Fig¬
ure 35).
This type of bolt, which may be preloaded or not, of¬
fers the following advantages:
7 CONCLUSIVE REMARKS
This paper, which summarises a broad range re¬
search activity that the authors are developing in
these years, tries to outline the relevant state-of-the-
art in the field of "mixed technologies", with a spe¬
cial regard to building strengthening applications.
The use of systems based on mixed technologies, in
fact, sets out new frontiers in the field of restoration
of existing constructions. Advanced materials and
technical solutions ensure higher quality and longer
life to the intervention compared with traditional
techniques, leading to an improved structural per¬
formance from both engineering and economic point
of view. Also, materials and techniques illustrated in
this paper are characterised by really peculiar fea¬
tures, which can be considered somewhat comple¬
mentary to each other. This allows them to satisfy
special design requirements. For example, metal-
based techniques are fully reversible and recyclable
and, hence, environmentally friendly, because at the
end of their service life they can be re-used other¬
wise. On the contrary FRP elements, even though
hardly removable, have high strength and negligible
weight and, for this reason, they can be profitably
used in those cases when the addition of new
strengthening members is not possible. At the same
way, advanced connection systems make possible
the implementation of technical solutions not afford¬
able with traditional means.
In this framework, the authors wish to underline the
importance of the activity carried out within the
COST C12 Action, which gave them the possibility
to meet, to discuss and to compare their own experi¬
ences, together with researches coming from all over
Europe. This has led to gather valuable contributions
to the development ofthe state-of-the-art in the field
of structural rehabilitation, where experiences are
usually deeply different from Country to Country
and a strong need for homogenisation is felt. This
paper is a first result of this joined activity at Euro¬
pean level and, hopefully, can represent a good ex¬
ample of fruitful co-operation among researchers
working in different areas.
REFERENCES
Ballerini, M. Piazza, M. andZanon, P. (1994). Static
and cyclic behaviour of glued joints in timber.
Proc. of the Pacific Timber Engineering Confer¬
ence PTEC 94, Gold Coast (Australia).
Ballerini, M., Ferrari, A. and Piazza, M. (1999).
Plane grid glulam structures: development of
moment resistant connections with glued-in steel
rods, in "New Technologies and Structures in
Civil Engineering - Case studies on Remarkable
Constructions", (Dubina, D., Vagas, I. and Un¬
gureanu, V. Eds), Edit. Orizonturi Univ. Timisoa¬
ra (Romania).
Bailio, G., and Calvi, G.M. (1993). Strengthening of
masonry structures by lateral confinement. Proc.
of IABSE Symp. on Structural Preservation ofthe
Architectural Heritage, Rome.
53
Baptista, A.M., Missoum, Α., Muzeau, J.P. and
Ryan, I. (1997). Testing of swaged bolts for steel
connections, Int. Conf. New Technologies in
Structural Eng., LABSE-FLP, Lisbon, Portugal.
Bergmeister, Κ. (1999). Verstärken mit CFK Lamel¬
len, Heft 40/1999 Schriftenreihe des Österreichi¬
schen Betonvereins.
Bergmeister, K. (2001). Kohlenstoffaserelemente in
der Konstruktionspraxis, Fachtagung TU Inns¬
bruck.
Buchanan, Α. Η. and Fairweather, R. H. (1992). Ep-
oxied moment resisting connections for timber
buildings. Proc. of the Inter. Workshop on Wood
Connectors, Las Vegas, Nevada (U.S.A.).
Buchanan, A. H., Moss, P. J. and Townsend, P. K.
(1990). Reinforcing bars epoxy bonded in glue
laminated timber. Proc. of the Inter. Timber En¬
gineering Conference, Tokyo (Japan).
Cavalli, R., Piazza, M., Redigolo, N. and Urso, T.
(2001). Analisi funzionale dell'impiego di resine
epossidiche nei collegamenti di strutture di legno
lamellare. Monti e Boschi, vol. Lu, η. 2, 2001.
Croci G., Bonci A. and Viskovic A. (2000), The Use
of Shape Memory Alloys Devices in the Basilica
of St Francis in Assisi. Proc. Final Workshop of
ISTECH Project, Ispra (Italy).
Czarnomska, M., Muzeau, J.P. and Racher, P.
(1993). Etude expérimentale comparative du
comportement d'assemblages par boulons sertis
huck-fit ou par boulons hr. Constr. Métallique, n. 2.
Dolce, M., Cardone, D. and Nigro, D. (2000). Ex¬
perimental tests on seismic devices based on
shape memory alloys. Proc. of the 12th World
Conference on Earthquake Eng., Auckland.
EN 408 (1995). Timber structures Structural tim¬
ber and glued larninated timber. Determination of
some physical and mechanical properties. CEN
European Committee for Standardization.
FIB (2001) Technical report on externally bonded
FRP reinforcement for RC structures.
Giuffrè, Α., and Martines, G. (1989). Impiego del ti¬
tanio nel consolidamento del capitello della Co¬
lonna Antonina. Proceedings ofthe Convegno e
Mostra A.N J.A.SPE.R. Rome.
Indirli, M. (2000). The demo-intervention of the IS¬
TECH Project: the Bell Tower of S. Giorgio in
Trignano (Italy). Proceedings of the Final Work¬
shop of ISTECH Project, Ispra (Italy).
Jaspart, J.P. (1994). Perspectives d'avenir, Journée
d'études "Le boulon dans la Construction Métal¬
lique", CRIF, ULG-MSM, ProfilARBED, Namur,
Belgium.
Korol, R.M., Ghobarah, Α., and Mourad, S. (1993).
Blind bolting W-shape beams to HSS columns. J.
of Struct. Engrg., vol. 119, n.12.
Maier, M., Andra H-P, König G. (2001). Einsatz
vorgespannter Kohlefaserlamellen als Oberflä-
chenspannglieder, Beton-und Stahlbetonbau,
Ernst & Sohn Heft 12.
Mandara, Α. (1992). L'uso dell'acciaio nel consolida¬
mento di elementi murari verticali. L'Edilizia, 11.
Mandara, A. (2002). Strengthening Techniques for
Buildings, in Refurbishment of Buildings and
Bridges (F.M. Mazzolani & M. Ivanyi Eds).
CISM publications, Chapter 4.
Mandara, Α., and Mazzolani, F.M. (1998). Confin¬
ing of masonry walls with steel elements. Proc. of
the International LABSE Conference on Savings
Buildings in Central and Eastern Europe, Berlin.
Mander, J.B., Priestley M.J.N, and Park, R. (1998),
Theoretical stress-strain model for confined con¬
crete. Journal of Structural Division, ASCE, vol
107, No. STI 1.
Mazzolani, F.M., (1990). Refurbishment. Brochure
of the series "Information and Technical Advice",
ARBED, Luxembourg.
Mazzolani, F.M., (2002). Principles of rehabilitation
and design criteria. In Refurbishment of Build¬
ings and Bridges (F.M. Mazzolani & M. Ivanyi
Eds). CISM publications, Chapter 1.
Mazzolani, F.M., and Mandara, A. (1991). L'acciaio
nel consolidamento. ASSA.
Mazzolani, F.M., and Mandara, A. (1992). L'acciaio
nel restauro. ASSA.
Mazzolani, F.M., and Mandara, A. (2002). Modern
trends in the use of special metals for the im¬
provement of historical and monumental struc¬
tures. Elsevier Journal of Eng. Structures, n.24.
Meier, H. ( 2001). Grundlagen und Systeme für den
Einsatz von Faserverbundwerkstoffen im Bauwe¬
sen, Fachtagung TU Innsbruck.
Meier, H., Czaderski, C. (1998). Nachträgliche
Schubverstärkungselemente mit CFK-Winkel,
Schweizer Ingenieur und Architekt. Nr. 43.
Meier, U. (1994) Verstärkung von Bauwerken mit
Faserverbundwerkstoffen, VDI Ber. Nr. 1080.
Missoum, Α., Bouchaïr, Α., Muzeau, J.P. and Ryan,
I. (1997). Behaviour of swaged fasteners under
fatigue in tension. 18th Czecho-Slovak Int. Conf.
Steel Struct, and Bridges '97, Brno, Czech Repub¬
lic.
Muzeau, J.P. (2002). The use of special techniques
in refurbishment, in "Refurbishment of Buildings
and Bridges, (F.M. Mazzolani & M. Ivanyi Eds).
CISM publication, chapter 5.
Muzeau, J.P., Piraprez, E. and Kirby, P. (1998).
Special fasteners. COST CI Conf. Control of the
Semi-Rigid Behav. of Civil Engineering Struct.
Connections, Liège, Belgium, EUR 18854 EN.
Muzeau, J.P., Ryan, I, and Massena, P. (1996). Fixa¬
tions mécaniques: les boulons sertis Huck-Fit.
Mise en euvre et résistances. Assemblages, n.l.
Natterer, J. Herzog, T. and Volz, M. (1994). Cons¬
truire en bois 2. Lausanne (Suisse): Presses poly¬
techniques et universitaires romandes.
Neubauer, U. (2000) Verbundtragverhalten gekleb¬
ter Lamellen aus Kohlenstoffasern lBMB TU
Braunschweig.
54
Niedermeier, R. (1999), Verbundtragfähigkeit auf¬
geklebter Bewehrung, Mühchen Massivbau-Sem.
Pegon, P., Armelle, Α., Pinto, and Α., Renda, V.
(2000), The ELSA Laboratory and the protection
of cultural heritage. Proc. ofthe 5th Int. Congress
on Restoration of Arch. Heritage, Florence.
Piazza, M. & del Senno, M. (2001). Proposals and
criteria for the preliminary evaluation, the design
and the execution of works on ancient load bear¬
ing timber structures. In Wooden Handwork/
Wooden Carpentry: European Restoration Sites,
Elsevier, Paris.
Piazza, M. (1990). Mixed Systems in Reinforcement
of Old Structures. Proc. of LABSE Symp. on
"Mixed Structures incl. New Materials", Brussels.
Piazza, M. (1994). Restoration of timber floors via a
composite timber-timber solution. Proc. of the
Tech. Workshop RILEM "Timber: a Structural
Material from the Past to the Future", Trento.
Piazza, M. and Turrini, G. (1986). Advances in
technology of joints for laminated timber. Analy¬
ses of the structural behaviour. Proc. of 19°
C.I.B.-W18 Meeting & JUFRO S 5.02, Florence.
Piazza, M. and Turrini, G. (1986). The Influence of
Connector Flexibility on the Behaviour of Com¬
posite Beams. Costruzioni Metalliche, n.6.
Piazza, M., Cont, S. & Zanon, P. (1997). Some Re¬
marks on the Design of Timber Composite Struc¬
tures Exposed to Fire. Int. Conf. 'Comp. Consti. -
Conventional and Innovative', Innsbruck, Austria.
Piazza, M., del Senno, M. and Tornasi, R. (2001).
Behaviour analysis of axially loaded glued-in
timber-steel joints. LABSE Conf. "Innovative
Wooden Structures and Bridges", Lahti, Finland.
prEN 1995-1-1, Final draft 2001. Eurocode 5 - De¬
sign of Timber Structure - Part 1-1: General
Rules and Rules for Buildings. CEN European
Committee for Standardization, TC 250/SC5.
Riberholt, H. (1986/88). Glued bolts in glulam - Part
I & Π". Dep. of Struct. Engineering, Tech. Univ.
of Denmark, Series R, n. 210 & n. 228.
Spoelstra, M., and Monti, G. (1999). FRP-confined
concrete model. Journal of Composite for. Con¬
structions, ASCE 3(3), August 1999.
Triantafillou, T.C. (1998). Strengthening of masonry
structures using epoxy-bonded FRP laminates.
Jour, of Composite for Consti. ASCE 2(2).
Valluzzi, M.R., Valdemarca, M., and Modena, C.
(2001). Experimental analysis and modeling of
brick masonry vaults strengthened by FRP lami¬
nates. Journal of Composite for Constructions,
ASCE, August 2001.
 Service life and life cycle of building structures
Development of new solutions in view of economic and life cycle assessments

R. Blok, F. v. Herwijnen
University of Eindhoven, Netherlands
A. Kozlowski , S. Wolinski
Rzeszow University of Technology, Poland

Helena Gervásio
GIPAMB Consulting Engineers Ltd, Portugal

L. Simões da Silva
Univ. of Coimbra, Portugal

ABSTRACT: An overview of aspects which influence the lifetime and the life cycle of building structures is
given. The requirements of building structures are likely to change during their functional working life.
Therefore designers of building structures should strive for the best possible match between design working
life, technical service life and functional working life. A best practise design-example of a building structure
which aims at a technical service life and functional working life of twenty years is presented. The design
process should concern not only the lifetime, but the total life cycle of structures and their components in
order to avoid and minimise negative environmental impacts. Life cycle assessment methods are currently
under development. To minimise energy consumption the embodied energy in building materials is a
significant design factor. To minimise waste production the focus should be on renewable materials and
materials with a high recycled content. With regard to economy and life cycle assessment, optimisation ofthe
total expected life-cycle cost subject to minimum reliability requirements, optimisation of the whole life
building management, and evaluation of the remaining life-time of existing structures is discussed. As an
example of improving on material-efficiency, recommendations of semi-rigid joints in composite frames are
presented. Industrial, Flexible and Demountable (IFD) building is presented as a concept of building which
deals with aspects of sustainable building.

1. INTRODUCTION Quality of the individual materials and

The fact that building structures won't last
forever is not always in line with our expectancy. In
spite of this, it is beyond doubt that the time-aspects
involved in the planning, design, evaluation,
management, maintenance, demolishing, re-use and
recycling of building structures cannot be ignored.
Society changes fast and with it its demands on
buildings and their structures. Service life and life
cycle of building structures is a very broad subject or
rather wide range of subjects. The following chapters
treat some aspects of this very broad scope.
components
Standard of construction
Actions on the structures during life
Environmental effects on the structure
The use and changes in use ofthe building during
its service life
The decision processes regarding maintenance,
repair, adaptation and finally demolishing of the
structure.
For a good understanding, the following terms
regarding the lifetime of a structure are defined:

Design working life

2. SERVICE LIFE AND STRUCTURAL The design working life is the assumed period for
REQUIREMENTS which a structure is to be used for its intended
purpose (with anticipated maintenance but without
major repair).
2.1 Service life
The service life of a given building structure is Technical service life
affected by: The technical service life is the period for which a
Design of the building, its structure, joints and structure can actually perform according to the
components
55
structural requirements based on its intended
purpose (possibly with necessary maintenance but
without major repair).
Functional working life
The functional working life is the period for which a
structure can still meet the demands of its (possibly
changing) users (may be with repairs and or
adaptations).
Based on this we can define:
Technical durability:
The ability / capacity of a structure in its
environment to remain structurally fit for its
intended use during its technical service life (with
anticipated maintenance but without major repair).
Functional durability:
The ability / capacity of a structure to fit the
requirements for use despite changes in use.
2.2 Structural requirements
A structure in use will have to meet with the
following criteria during its lifetime.
For Technical durability:
During its Technical service life:
Safety (including loss of equilibrium):
Structural Resistance R > Solicitation S
Serviceability:
Structural Serviceability > Min. Serviceability
For Functional durability:
During its Functional working life:
Structural Functionality:
Structural Functionality > Minimum User
Defined Functionality
Structural Functionality can be defined as the
capability of the structure to fit the requirements of
users in terms of loading capacity, relocation of non¬
structural elements (e.g. ducts, staircases) and the
structures adaptability to accommodate changes in
these requirement (See also paragraph 2.4).
2.3 Technical service life
During its technical service life a building structure
has to meet with minimum required levels of
structural safety and, at the same time, has to meet
with minimum levels of serviceability.
a. Structural safety during the technical service life
of a building structure
The structural safety of a structure depends on the
resistance and the solicitation of the structure. Both
can be subject to seen or unforeseen changes in time.
The resistance of a building structure is likely to
change in time. Structural materials can be affected
by deterioration, for instance depending on the
moisture content of the materials. Sometimes the
expected life of a structure can be predicted more or
less accurate, for instance in the case of the
estimation of fatigue life. Failure can be predicted in
relation to the number of loading cycles. Γη the case
of time dependent (strength) properties of materials,
different models to predict the life of a structural
element with a minimum level of safety are
available.
Performance
Performance In time
Time
Figure 1: Deterioration of building components (Hermans,
1999)
Sometimes however the life of a building structure is
unforeseen short. The resistance, depending on the
properties of the materials, can show a time
depended behaviour different from our expectancy.
For instance, in the case of unforeseen chloride
induced corrosion of steel reinforcement in concrete,
or ASR, alkali silica reactions in concrete. Also the
solicitation of structures can be different from our
expectancy: Rare or unforeseen actions on buildings
can cause structures to fail. We sadly had to
experience an example of this on the 1 1th Sept. 2001.
Structural Integrity under exceptional loadings is an
important subject dealt with in WG 2.
b. Serviceability during the technical service life of a
building structure.
Apart from structural safety the serviceability of the
building and its structure plays an important role in
the life of a building structure. Both the level and the
requirements regarding the serviceability can change
in time. Deflections and cracking may increase in
time, even to such an extend, that serviceability
criteria are no longer met. Specification of
appropriate serviceability limits regarding
deflections, vibrations or cracking etc. may change
as well (partly depending on how they are perceived
by the users).
2.4 Functional working life
The functional working life depends on the
minimum user defined functionality on the one hand
56
and the functionality of the structure on the other
hand, combined with the ability ofthe structure to be
adapted in such a way that the user criteria are met.
With changing societies the demands of building
occupants become more and more dynamic. An
investigation carried out under companies in the
Netherlands showed the following figures:
15 -
22 % ofthe companies want to move
10 -15 % of the companies want to alter / rebuild
(Source: NVB 1998)
(Profit and Non-profit organisations in Holland
move with an average of once every five years).
If the functionality of a given structure meets the
(new) users demands, no adaptations to the structure
will be necessary. The functionality of a structure
depends largely on aspects determined at the design
stage. For instance:
The location ofthe structure (and building)
Positions of bearing walls
Column free area of floors
Bearing capacity of the structural elements to
accommodate functions with higher imposed loads
Height between floors to accommodate e.g.
service ducts
Ii" the users functional demands are not met by the
functional quality of the structure the functional
working life depends on the capability of the
structure to be adapted or else the functional
working life of the structure will end.
Adaptability: With regard to the adaptability of the
structure we can distinguish the following influential
aspects:
Bearing capacity of structural elements to
accommodate adaptations elsewhere.
Characteristics of the existing 'structure with
regards to possible strengthening or making new
voids for ducts or stairways.
The possibility to exchange structural elements
(connections etc.)
Other building aspects which influence the
possibilities for strengthening or exchange.
Possibility to disassemble and relocate the
structure
The relation between the functional working life and
the technical service life is illustrated in Figure 2.
Structures for which the functional working life is
less than the technical service life need to be taken
out of the market. (Their functional life can be
extended by finding new users with lower demands
on structural functionality or by improving the level
of functionality: adapting the structure). Structures
for which the functional life time (in terms of years)
is longer than the technical service life should be
57
repaired, upgraded or also dismantled because they
don't meet the safety or serviceability requirements.
Potential
Technical service life
(Years)
400
e>
20
Disassemble
recycle j £&&
W
ligio]
ι
"Renoiräte "
Restia urate
Adapt
1 20 400
Potential Functional working life (Years)
Figure 2: Potential Functional Working Life versus Potential
Technical Service Life (van Herwijnen, 1999)
Because of the large expenses often involved in
adapting building structures it can be advantageous
to strive for a functional working life equal to the
technical service life: the diagonal in the diagram
above.
A Best practice design-example of a project that
strives for "the diagonal" is the XX-office Building
in Delft, Netherlands. The design ofthe structure for
this building is discussed in chapter 3 "The load
bearing structure ofthe XX-office building Delft, the
Netherlands".
2.5 Economy
Several aspects are important in connection to the
decision process to maintain, repair, adapt or
demolish a building structure. Perhaps the most
important aspect in achieving an optimal
management strategy for new and existing buildings
is economy.
Optimisation of total expected life cycle cost subject
to minimum reliability requirements, optimisation of
the whole life building management, and also,
evaluation of the remaining lifetime of existing
structures is discussed in Chapter 5 "Economy and
life-cycle assessment".
2. 6 Building diagnostics
In order to obtain the right information to support
management decision processes, for instance
whether and when maintenance and / or repair are
needed, it becomes necessary to evaluate the
performance of the structure. Several techniques to
monitor, test and evaluate existing structures are
available, and are still being improved and further
developed. Building diagnostics play an important
role in predicting the life of existing structures. The
numbers of buildings built prior to the application of
limit state design outnumber the ones that are built A sustainable production process, costs- per unit
to latest views. Therefore building diagnostics play of resistance for wood are relatively low, a wide
an important role in the decision process to maintain, experience with wood construction, detachable
repair, adapt or demolish the structure. joints, possibility of wood recycling. LVL largely
reduces the disadvantage of large dimensions.
2. 7 The life-cycle of structures
3.2 Structural design
Before, the Design working life, the Technical
The spatial plan of the building shows two floors
service life and Functional working life of structures
with three naves with dimensions of 5 m in lateral
have been discussed. However it is necessary to look
direction and 6 m naves in longitudinal direction.
beyond the life of existing and future building
(Wooden) columns can be positioned on the
structures. The effects that building construction has
intersections of the main axes. The structural task
on the environment plays an important role in our
was to span bays of 5 χ 6 m as floor (first floor) and
well-being. The effects of the extraction of raw
roof structure.
materials, the production of materials with possible
waste or toxic emissions but also the use of the
products itself (with possible negative impacts) and
the possible re-use should be considered. In short:
the whole life cycle of the building structure.
Chapter 4 "Life-cycle of structures and
sustainability" deals with this topic.

3.THE LOAD BEARING STRUCTURE OF THE

XX-OFFICE BUILDING LN DELFT, THE
NETHERLANDS

3. 1Introduction
Traditionally the structure is the most durable part
of a building. A twenty years write-off time, was
aimed at with XX-office. This demand was
translated in the requirement for the structure (along
with the rest of the building) to be demountable,
sequentially re-usable or to be composted after this
twenty-year period. From the point of view of "Open
Building" a skeleton framework construction was
adopted. For the columns various materials were
considered. Cardboard columns filled with loam
were rejected because of their large dimensions and
susceptibility to changes in moisture content. Lime-
sandstone columns out of stacked drums were
rejected because of similar large dimensions and a
long mould production-time.
The final choice was made for wooden columns
out of Swedlam LVL (Laminated Veneer Lumber).
Decisive arguments for a wooden structure were:
;ESp:
Figure 3: XX office building Delft (van Herwijnen, 1999a)
Figure 4: XX-office: Wooden structure under construction
3.3 Floor structure
The floor structure consists of timber joists. Because
ofthe high loads (4 kN/m2) and the mass needed for
sound insulation, combined with the necessity to
limit the creep deformation, the span of the joists
was limited to 2 m. This was achieved by spanning
primary beams (A) from column to column in the
longitudinal direction, and secondary beams (B),
center-to-center 2 m, in the transverse direction. The
secondary beams are subjected to bending strains by
the joists; A Swedlam beam section 184,5 χ 300 mm
was sufficient for the 5 m span.
The span ofthe primary beams receives two point
loads out of the secondary beams. From this arose
the idea to construct these primary beams as a
subtended compound girder with tie rods and
vertical hinged compression members at the supports
of the secondary beams. Thus the whole of the
timber section is used to absorb the compressive
loading. The secondary beams provide sideway
support and prevent buckling in this direction. By
using these subtended compound girders (A) the
same section dimensions could be used for the
beams (A) and (B). Under and upper side of the

58
beams are thus in one plane. The joints are
constructed with pin fasteners and hidden steel
plates.
Figure 5: XX-ojfice: Interior overview
» n-
Figure 6: XX-office: Column head out of four wooden chocks
For the tie rods Rodan-anchors M52 were used. The
tension load at the head of the beam is kept in
equilibrium with the compressive loading in the
beam via a steel endplate. To support the primary
beams (A) the square wooden columns
(300x300mm) are equipped with a column head out
of four wooden chocks (linking sections); four bend
angle-cleats connect the structural elements together.
In a remaining encasing the more slender column
(only supporting the roof) on the first floor is placed.
The square columns are chamfered to achieve a more
slender elevation.
3.4 Roofstructure
For the roof structure the choice was made for
reinforced concreted wood fibreboard with a span
limited to 2 m to limit creep deformation. The
boards have a high thermal capacity and contribute
to the required sound absorption at the underside.
The layout of the roof structure is in fact identical to
the floor structure, except that, because of the
difference in live loads (roof: 1 kN/m2; floor: 4
kN/m2), the section dimensions of the beams,
columns and tie rods are smaller. Also for the cross-
section of the compressed upper-member of the
under spanned roof beams a cross-shaped section
was chosen. This shape of section could not be used
for the floor beams for reasons of fire-resistance.
3.5 Stability, Fire-resistance
Horizontal stability ofthe structure is achieved by
allowing the wooden plywood boards in the floors to
transmit the shear forces due to lateral loadings, and
by using bracings over the roof boards and vertical
wind bracings in longitudinal and transversal
directions.
The first floor, with the columns and wind
bracings at ground floor level are designed for a 30-
minute fire-resistance. The wooden beams and joists
were designed using a reduction in section
dimensions due to fire; for the steel elements a
reduction in yield strength was used. Because the
beams are supported by chocks, the load transfer is
achieved by contact-pressure. The steel joints at the
supports of the main beams loose their load
transferring ability after twenty minutes, thus
progressive collapse is prevented.
Ground floor and foundation
The original design was based on a foundation
out of steel-tube piles with cathodic protection
combined with steel integrated-section beams
between which the concrete prestressed hollow-core
slab spanned. This structure would be totally
demountable and suitable for re-use. Out of cost
considerations a traditional non-demountable
concrete foundation with concrete piles and beams
was used instead. After the 20-year period the
concrete at least can be re-used as crushed -concrete
granulate.
3.6 First floor construction
The first floor construction consists of timber
floor joists (59 χ 146 mm, 417 mm centre to centre)
with a fire-resistant board "Antivlam" (thickness 22
mm) glued to the underside of the joists, to obtain a
composite panel. The space between the joists is
filled with sand on a foil. The sand provides for an
important contribution to the acoustic insulation.
The space also accommodates service ducts. On top
the cement fibre-reinforced panels (24 mm
thickness) are supported on the timber joists using an
elastic intermediate layer. To the underside
concreted wood boards were fitted on a railing
system to provide for the sound absorption in the
ground floor areas. The total weight of this
floorsysteem amounts to 280 kg/m2. Thus a so-
called acoustic insulation index for airborne sound
insulation and impact sound of 0 dB is achieved
(laboratory-values), the minimum value for floors
which separate dwellings or offices with different
59
occupants (this was a predefined requirement for this
project). The acoustic insulation-level is compatible
with a 100 mm solid concrete floor with a 30 mm
thick cement screed.
. >\
Figure 7: XX-ofiîce: cross-section ofthe first floor structure
3.7 Advantages
Advantages of the designed construction are:
préfabrication ofthe floor panels; Moment reduction
in the continuous spanning timber joists fire-resistant
"Antivlam"-boards and the cementated boards
prevent exposure of the timber joists to fire
(minimum requirement: 20 min fire-resistance); the
cementated panels are regularly supported by the
timber joists; the total height of the floor
construction, underside of ceiling to top of the floor
is a mere 260 mm; The wooden floor system is fully
demountable; no contribution to the waste volume;
the "antivlam"-boards can be fastened directly to the
beams (A) and (B) to secure the transmitting of shear
due to lateral loadings through the floor to the
vertical bracings.
The wooden structure and the steel components
are not coated. This means that pure material can be
re-used or recycled, e.g. LVL-components can be
recycled to paper.
4. LIFE-CYCLE OF STRUCTURES AND
SUSTATNABILITY
4.1 Introduction
Buildings are among the largest artefacts
produced by men. As such they consume large
quantities of materials, land, energy, money and
human effort; resources which we used to consider
to be inexhaustible or infinitive, but have now been
found to be quite limited. We are starting to realise
that these resources will run out, in some cases
sooner than later.
Some fundamental questions about our economic,
social and environmental beliefs and actions are
arising and people are realising that our present way
of life, including how and what we build, may not be
sustainable into the future.
Buildings are often conceived as during forever, but
in fact, the end of the life of the buildings is
inevitable.
So it is important to think about the full life cycle of
a building from design and construction, operation,
maintenance, refurbishment and final demolition. It
is necessary to consider the consumption of material,
60
energy and other resources at each of these stages
and the interrelationships between them.
4.2 Sustainability
Sustainable Development implies that aspects
concerning the environment, the vulnerability of the
earth, exhaustible resources, and the quality of life
are all to be taken into account. Sustainable
Development is viewed as a major requirement for
future technology. It is often referred to as consisting
of three main points:
1. Integrated cycle management: the closing of the
material cycle from raw material to waste material
and emissions.
2. Reducing energy-consumption and the use of
sustainable natural energy sources.
3. Quality improvement: Raising the quality of the
products, the fabrication processes, minimising
waste production and negative impacts on
environment all aiming for a longer use of our
available natural resources.
4.3 Life cycle assessment
Life cycle assessment (LCA) methods have been
developed for several groups of products. ISO
standards series 14040-14049 deals with
environmental management and life cycle
assessment in general. Because of the complexity
and interaction between numerous influential factors
the strive for one overall figure indicating the
environmental impact of a product proofs not to be
effective. The most promising methods take the
description and evaluation of different product-
functions as a starting point.
4.4 Reducing energy consumption: Embodied energy
Embodied energy is the amount of energy
consumed by all of the processes associated with the
production of a building: from the acquisition of
materials resources to product delivery. This
includes the mining and manufacturing of materials
and equipment, the transport ofthe materials and the
administrative functions. Embodied energy is a
significant component of the life cycle impact of a
building.
Lightweight building construction such as a
timber frame is usually lower in embodied energy
than heavyweight construction. This is not
necessarily the case if large amounts of light but
high-energy materials such as steel or aluminium are
used.
There are many situations where a lightweight
building is the most appropriate and may result in
the lowest life cycle energy use.
In climates with greater heating and cooling
requirements and significant day / night temperature
variations, embodied energy in a high level of well
insulated thermal mass can significantly offset the
energy used for heating and cooling.
Each design should select the best combination
for this application based on climate, transport
distances, availability of materials and budget,
balanced against known embodied energy content.
4.5 Integrated cycle management: Building waste,
renewable materials and materials with high
recycled content
Waste as a result of building construction makes a
large contribution to the total production of waste,
therefore is an important issue to be dealt with. The
best way to "deal with waste" is prevention, or
recycling with highest possible value. Less wanted is
incineration or dumping. The prevention of building
waste is directly influenced by the possibilities of
disassembling, re-using, or recycling of structures,
components and materials together with minimizing
the use of materials and energy.
There are several ways of recycling a building. A
building can be modified in order to change its
functionality, or components of a building such as
steel or timber beams, bricks, etc, can be reused
(product-recycling). The reprocessing of materials
from the building demolition is another way of
recycling them (material-recycling).
Re-use of building materials can save up to 95% of
embodied energy that would otherwise be wasted.
Most building materials can be recycled. Some of
them are illustrated in the following list:
α Steel - conventional blast furnaces can
incorporate up to 30% steel scrap and this
percentage can go up to 100% in the most
modern ones,
α Aluminium - is 100% recyclable. Recycling
aluminium reduces embodied energy by
95%.
α Gypsum plasterboard - there are already
some companies recycling it. Plasterboard
disposed ofin landfill is poisonous,
α Timber can be re-processed into
horticultural mulch.
Concrete - un-set concrete can be "washed"
out at the plant to remove cement. Sand and
stone can be re-used. Set concrete can be
mixed and recycled as aggregate for new
concrete or road base and fill. (Mitsubishi
Materials Corporation have recently
developed a method to recycle also cement
from concrete rubble by means of thermo
shock, followed by a process of crushing and
sieving)
α Glass - most glass can be recycled. Glass
may be cut and re-used or recycled as
aggregate for concrete. Recycling glass
reduces embodied energy by 20%.
α Bricks and tiles - can be re-used when
appropriate or crushed a site for backfill,
aggregate and growl with portable crushing
plants.
Plastics - many plastics can be granulated
and re-used to make new plastic products.
The problems associated with the environmentally
safe and efficient disposal of waste continue to grow.
One answer to these problems lies in the ability of
society to develop beneficial uses for these waste
products. The construction industry can use some of
these products thus contributing to the
minimalization of these problems. Another answer
consists of optimising the combined use of different
materials in buildings to achieve improved
sustainability. This is precisely the aim of Mixed
Building Technology!
5.ECONOMY AND LIFE-CYCLE ASSESSMENT
Three main problems are usually considered in
connection with the economy and life cycle
assessment of building structures:
- Optimisation ofthe total expected life cycle cost
subject minimum reliability requirements,
- Optimisation of the whole life building
management,
- Evaluation of the remaining lifetime of existing
structures.
The optimum design of building structures requires
the consideration of structural reliability and the total
expected life-cycle cost as the acceptance criteria. A
concept for balancing structural reliability and the
total expected life-cycle cost offers the real potential
for integrating economic and reliability issues in the
design and the whole life structure management.
The formulation ofthe reliability-based optimisation
for minimum expected cost of a designed building
structure has been proposed by Murzewski (1989) as
follows:
min Ctot = min [Cc + (1 - ψ) Cds Ps + ψ Cdc (1 - Pf)]
(1)
where Ctot is the total expected life-time cost, Cc is
the basic cost (initial and other time-invariant), Cds is
the cost due to disturbances in service, Cdc is the cost
of decommissioning, ψ is the quality control
efficiency, Ps is the probability of disturbances in
service, and Pf is the probability of
decommissioning.
More general formulation of this problem has been
suggested by R. E. Melchers (2000):
61
min E [N(t)] = min { E [J] B¡] - E [ CB + Cc + CM
i structures depends on the safety assessment of these
+ PsCF]} (2)
where E [ ] is the expectation operator, B¡ is the i-th
benefit stream, Ps is the structural safety probability,
Cb, Cc, Cm, Cf are the basic cost, the design and
construction control cost, the maintenance cost, and
the failure cost, respectively.
In general, two possibilities for optimisation exist.
The first is direct optimisation or "design
optimisation" in order to determine the optimal set
of random variables or their design values. The
second form of optimisation is to determine the
optimal probabilities of failure, decommissioning
or/and disturbances. This is usually called the
"criterion optimisation".
A concept for balancing structural reliability and
total expected life-cycle cost offers the real potential
for integrating economic and reliability issues in the
design and the whole life structure management. The
mathematical formulation of the problem is that of
minimizing the total expected cost of structure over
its lifetime subject to minimum reliability
requirements (Frangopol, 1998):
mfnCtot = min(Q + Cm,i + CT + Cu + Cf) (3)
Pf(tif) < Pw(tif) (4)
where C,0t is the total expected life-cycle cost, C¡
is the initial cost, Cmji is the preventive inspection
and maintenance costs, Cr is the repair cost, Cu is
the user cost, and Cf is the failure cost, Pf and Pfd
are failure probability and target failure probability
over the life time of a structure tif. The initial costs
and costs of inspection, maintenance and repair of
building structures are relatively easy to determine.
The failure cost, especially the costs of injuries and
loss of human life are very difficult to assess. It is
also not easy to evaluate probabilities of disturbance,
decommissioning and failure.
In general, the time variant reliability method should
be used to perform a life-time cost analysis for
building structures. A possible solution is to find the
failure probability of a structure over the time
interval [0, %]:
Pf(t) = P[R(t)<S(t)] for te [0, tif]
(5)
where the resistance R and the sollicitation S are
functions of time.
Evaluation of the remaining life-time of existing
structures which in turn depends on each engineer's
experience, intuition and judgment. The capacity of
any structure to give a required performance during
the intended and remaining life-period is affected
both by direct actions applied to a structure and by
degradation factors connected with the
environmental actions such as chemical and physical
attack.
Two approaches to assessment of deteriorating
structures can be distinguished: the approach based
on implicit rules and semi-probabilistic methods
(these dominate in the present practice), and the
approach based on the time-variant probabilistic
methods. Different damage functions are available
for civil engineering structures (Yao, 1985).
Generally, three types of definitions for structural
damage can be distinguished: numerical, in terms of
repair or replacement costs, and verbal. Usually,
numerical values are also assigned to verbal
assessments. An example of the cumulative damage
function Dk can be given as follows:
Dk= Σ i
Wikdik/ Σ/
Wik '
dik = dij + [ (Zik - ya) / (cik- ya) ] (1 - dg) (6)
where wa is the cumulative importance factor for
the i-th element and events k, dik is the local damage
index for i-th element and events k, dy is the local
damage index for i-th elements and events j < k, Ζ is
the demand and c is the capacity for the demand, y is
the threshold for limit state considered.
Design, construction, inspection and evaluation,
repair and life-cycle cost may be integrated into an
optimal management strategy for new and existing
building structures. With reliable input data, an
optimal management strategy offers the valuable
potential for integrating safety and economic issues
in civil engineering practice.
6 EFFICIENCY OF USING SEMI-RIGID JOINTS
ΓΝ COMPOSITE FRAMES
In the light of sustainable building and from a
material-efficiency point of view, recommendations
of semi-rigid joints are presented. Analysis of six
frame configurations: two and four bays, two, six
and eleven storeys (Figure 8), each in two versions:
unbraced and with X-shape bracing, was presented
in (Koslowsky, 2001). Results of analysis, in the
form of drawings with relation between moments in
beams, masses of steel and composite frames, lateral
drift, section of elements and saving of steel, in
relation to secant joints stiffness are presented in
62
Figure 2 - 13 of paper (Koslowsky, 2001). On the Table 1: Optimal composite semi-rigid joints recommended in
braced frames
Saving of steel
Frame Recommended types in comparison
of semi-rigid joints to frame with
B/H = 1,54 pinned joints
A top and seat angels with web 14,6 %
angles (3G)
seat and web site plate (4G)
Β B/H = 3,08 Β Top and seat angels with web 19,8 %
angles (3G)
C top and seat angels with web 13,3 %
angles 3(G) seat and web site
plate (4G)
seat angels with web angles (2S
D 2G) top and seat angels with 10,7 %
web angels (3D - 3S - 3G)
seat and web site plate (4 S - 4G)
B/H = 0,54 unstiffened flush end plate (5D)
seat angels with web angles (2S
E 2G) top and seat angles with 5,3 %
web angels (3D - 3S - 3G)
seat and web site plate (4S - 4G)
unstiffened flush end plate (5D)
F top and seat angles with web 9,5 %
D B/H =1,08 angels (3G)
seat and web site plate (4G)

For unbraced frames, recommended joints are

presented in Table 2.

Table 2: Optimal semi-rigid composite joints recommended in

unbraced frames
Frame Recommended types Saving òf steel
of semi-rigid joints in comparison
ΪΈ B/H = 0,3 to frame with
rigid joints
top and seat angels with web
A angels (3G) 11,4%
seat and web site plate (4G)
unstiffened flush end plate (5G -
5D)
top and seat angels with web
Β angels (3S - 3G) 12,6 %
seat and web site plate (4S 4G)
unstiffened flush end plate (5S -
F B/H = 0,6
5G)
Figure 8: Analysed frame configurations
unstiffened extended end plate
(6D)
base of this analysis, it is possible to point out the C unstiffened extended end plate 0%
optimal classes of composite joints, which give the (6G)
best economical solution (the lowest steel D unstiffened flush end plate (5G)
consumption) in each of the analysed frames. For unstiffened extended end plate 5,9 %
braced frames, types of the most optimal composite (6D-6S-6G)
joints were collected in Table 1, together with the E unstiffened extended end plate 0%
percentage of steel saving in comparison to frame (6G)
with pinned joints. F unstiffened extended end plate 8,9 %
Í6S)

Recommendation of semi-rigid joints presented in

this paper can be used as guides for designers to
achieve material economy in structural design.

63
7 DESIGN, IFD BUILDING
Decisions made during the design ofthe building
structure and its elements play perhaps the most
important role in trying to minimise waste
production and extending the life of building
structures.
From the important notion to strive for
sustainable building rose the concept of IFD,
Industrial, Flexible and Demountable Building.
Industrialised and flexible building in itself is not
new. However, the combination with de-mountable
building is. The three elements of IFD building are
discussed and defined below.
7.1 Industrial
By industrial building in this context we mean
industrially making of building products.
In building the word 'prefabricated' is often used as
a synonym for 'industrial'. In this context
préfabrication means to perform in the factory what
was originally done on the building site, manufacture
here is project specific.
An example of this is prefabricated concrete for
columns, facade elements and the like.
Besides there are industries that have of old only
been able to realise their production in the factory,
because the material, the process and the necessary
machines require such. Examples are steel
construction workshops, but also industries that
produce prestressed floor slabs.
7.2 Flexibility
Flexibility is the quality of a building or building
component which allows adjustments according to
the demands and wishes of the users Millard and
Robinson, 1989).
Flexibility may relate to two stages:
- the design stage: variability in the composition
and the use of material;
- the user stage: flexibility to adjust the composition
and the applied building components to the changing
demands and/or wishes of the same or varying users
while in use.
On the structural level, flexibility in the user stage
may be translated into possible adaptability of the
floors to higher working loads (the so called extreme
live load) and the realisation of recesses for
stairwells, lift shafts or pipes and ducts. The placing
of extra floors or mezzanines and the construction of
extensions on ground level must also be counted
among these.
7.3 Demountable building
Demountable building is the construction of a
building in such a way that a building component
may be removed and possibly re-used or recycled,
soiled as little as possible by other materials, and
64
without damaging the surrounding building
components (Millard and Robinson, 1989). In
material recycling we do not use the complete
product, but only its raw or pure material.
Demountable building is also a means for the
realisation of flexibility, because building
components may be easily detached and replaced by
other (industrial) building components.
REFERENCES
CEN, Eurocode 1, Basis of design and actions on structures,
part 1: Basis of design
Damen Consultants, Market potential of IFD building for Dutch
Building industry (Dutch), Damen consultants Rotterdam,
1997
de Bekker, P.J.G.M. and Nijman, A.J.N., Environment and
Building production (In Dutch), E.I.B. Economisch Instituut
voor de bouwnijverheid, Amsterdam.
Frangopol D.M., A framework for balancing reliability and life-
cycle cost in bridge management. Proceedings of the VTT!
ΓΗΡ WG 7.5 Worldng Conference on Reliability and
Optimization of Structural Systems, 1998. The University
of Michigan, Ann Arbor, Michigan, USA, pp. 3-13.
Hermans, M., Dynamics in policy and society. Changing
requirements in plan development and building design (in
Dutch), KPMG N.V. 1999
Kilmister, M.B., Tall Buildings - Designing for economy,
serviceability and long life, Collected papers Tall Buildings
2000 and Beyond, L. S. Beedle 1991
Kozlowski Α.: Economic shaping of building structures by the
use of mixed technology and semi-rigid joints. COST C12
Doc. C12/WG1/01-04.
Leino, T., The service life design strategies, proceedings of the
conference Eurosteel '99, 1999
Melchers R.E., Optimization of Structural Reliability.
Proceedings of the LX IFIP WG 7.5 Working Conference
on: Reliability and Optimization of Structural Systems,
2000. The University of Michigan, Ann Arbor, Michigan,
USA, pp. 157-165.
Millard, W.R. and M.J. Robinson, M.J., The life of structures,
proceedings of The life of structures, G.S.T. Armer,
J.L.Clarke, F.K. Caras, Butterworths 1989
Murzewski J., Reliability of civil structures, (in Polish),
Arlcady, Warsaw, 1989.
van Herwijnen, F., Twintig duurt het langst, The structure of
XX-office (in Dutch) Het Houtblad, March 1999
van Herwijnen, F., Working life = Technical Service life,
Masterclass Heerlen TU/e e.o.1999
van Herwijnen, F., Development of a new adaptable and
demountable structural system for utility buildings, TU/e
research papers 2000, pp 55 67
Yao J.T.P., Safety and reliability of existing structures. Pitman
Publishing Inc., Boston, 1985.
WG2 - Structural Integrity Under Excepcional Actions
 Structural integrity under exceptional actions:
basic definitions and field of activity
F.M. Mazzolani
University of Naples 'Federico II", Naples, Italy
1 INTRODUCTION
The COST C 12 action "Improving Buildings'
Quality by New Technologies" covers a wide range
of problems, going from the mixed building
technologies to the urban design, passing through
the structural integrity under exceptional actions.
The last subject is dealt with by the WG2, which
started its activity in Prague in April 2001. After
about one year, WG2 presented its initial activity
during the Seminar held in Lisbon on 19-20 April
2002.
It was decided that this first presentation should
be devoted to the items of common interest among
the experts nominated by the European Countries
participating to this COST action. Accordingly, the
following subjects have been identified:
"Structural integrity of buildings under
exceptional earthquakes"
"Structural integrity of buildings under
exceptional fire"
"Evaluation of structural robustness of members
and connections"
"General methodologies for evaluating the
structural performance under exceptional
loadings"
In collaboration with WG1 (Mixed buildings
technologies), an additional subject has been
decided, dealing with:
Repairing and strengthening for new
requirements.
These subjects are covered by the expertise of the
participants of COST C12-WG2, operating in
various European Institutions.
It was agreed that the presentations on the above
subjects should represent a kind of state-of-the-art
report, where the activities of the concerned
Institutions are summarized, by trying to harmonize
them together and framing them into the
international literature. This first effort represents
the common bases to build the future activity, being
also a very suitable dissemination mean, according
to the COST scope.
In order to integrate this WG2 activity, two
additional very topical subjects have been selected
for the Lisbon Seminar, to be covered by external
experts:
- Recent development in ductility design of steel
structures;
- Twin Towers: qualitative collapse behaviour.
We have to recognize that the range of WG2 is so
wide that it is difficult, or rather impossible, to cover
all the concerned aspects. From now, it seems that
only two main exceptional actions, i.e. earthquake
and fire, can be dealt with within the group,
corresponding to the field of interest and the
knowledge of participants. Experts of other
exceptional actions are not available until now in the
group and, therefore, its activity is limited to the
existing expertise.
Nevertheless, it seems very important to provide a
general scenario of all possible exceptional actions
which can undermine the structural integrity of
buildings.
/ After these preliminary remarks, the main scope
of this introduction to the activity of WG2 is also to
provide some general definitions, which are
necessary to frame the present and to guide the
future work.
67

2 WHAT DOES "EXCEPTIONAL ACTION"
MEAN: AN ATTEMPT OF CLASSIFICATION
2.1 According to the Codes
The different national and international Codes define
the design actions, i.e. the loading conditions which
must be considered during the design process.
Referring just to the not gravity-
dependent actions, a clear distinction is made in the
codes between variable and accidental actions.
Eurocode 1 "Actions on Structures" specifically
devotes Part 1.1.7 to Accidental Actions,but this text
is not yet officially available, being still in progress.
According to Eurocode "Basis Structural Design"
(prEN 1990:2001), an accidental action (A) is
(1.5.3.5): "action, usually of short duration but of
significant magnitude, that is unlikely to occur on a
given structure during the design working life", with
the following clarifications:
Note An accidental action can be expected in
1.
many cases to cause severe consequences
unless appropriate measures are taken.
Note 2. Impact, snow, wind and seismic actions may
be variable or accidental actions, depending
on the available information on statistical
distributions.
In parallel, it gives for variable actions (Q) the
following definition: "action for which the variation
in magnitude with time is neither negligible nor
monotonie".
In addition, Section 4 "Basic variables" Eurocode
(prEN 1990:2001) points out that "certain actions,
such as seismic actions and snow loads, may be
considered as either accidental and/or variable
actions depending on the site location".
Therefore, following the Eurocode definitions, the
variable actions (type Q) are basically: wind, snow,
earthquake (for constructions in seismic zones) and
the accidental actions (type A) may be impact and
explosion, but also snow and seismic actions in
locations where they are not expected.
Different -factors are given by the Codes for the
two type of actions. Normally γ=1.5 for type Q
actions and γ=1 for type A actions.
In general, it is clearly stated that "accidental"
does not necessarily mean "exceptional", if the
action is regularly considered in the design process
according to the Code provisions. What about other
kinds of actions, which cause extreme conditions in
structures? Storms, floods, landslides, avalanches,
volcanic eruptions, and so on, are not explicitly
considered in the Codes. Are they really exceptional
68
actions? How to consider them in the reliability
assessment? This is an open question, which merits
an appropriate answer.
The situation is different for fire. In fact, special
attention is given to the fire actions in the Eurocode
project, being carefully considered in parallel for the
main constructional materials in the Part 1.2 ofthe
material oriented Eurocodes.
2.2 According to the physical model
Eurocode "Basis Structural Design" (prEN
1990:2001) indicates that (Section 4): "an action
should be described by a model, its magnitude being
represented in the most common cases by one scalar
which may have several representative values". And
also: "for accidental actions the design value should
be specified for individual projects".
So, the concept of modeling is expressed by the
Eurocode, but without giving a quantitative
provision.
The different nature of the so-called exceptional
actions makes it difficult to give a general model to
interpret their effects on the structural behaviour.
Except to snow, which is a static load and fire,
which modifies the mechanical properties of
material due to high temperature, we can recognize
that all the other actions are dynamic. But their
dynamic features and, therefore, the consequences
on the structural behaviour are different.
Wind and earthquake basically produce horizontal
forces on the structures, but with different frequency
and intensity.
Wind can become a storm or a typhoon as far as the
intensity of its action increases.
Also in case of earthquake, there are different types
of seismic actions (horizontal, vertical, impulsive,
cyclic), according to the position of the earthquake
epicenter respect to the urban settlement (near-field
and far-field earthquakes correspond to a recent
definition, in order to identify very different
consequences).
Other actions, like avalanches, floods, landslides,
volcanic eruptions, are characterized by a flow of an
amount of moving mass, which can be made of
snow, water, mud, lava, respectively. They are
differentiated by the mass speed, which is very slow
for the lava flow, but can vary in a quite large range
in the other cases. Hydrodynamic methods can be
used for the physical interpretation of these
phenomena.
Impacts and explosions are also dynamic actions,
which are characterized by a very big amount of
energy which is released in a very short period of
time. Models of interpretation of impacts and
explosions can be found in military engineering (see
Bulson, 1997)
2. 3 According to the design process
It can be observed that even a normal action can
become exceptional in some special cases, i.e. when
it produces some damage in the structural and/or in
the non-structural elements.
In order to cover all possible effects on the structural
integrity of buildings, a general definition of
exceptional action has to consider different cases. In
particular:
"The action is exceptional when:
- the action is higher than the design one;
- the action is not considered at all in the design
process;
- the action is just formally considered in the
design process, but in a wrong way;
- the action, even if correctly considered during the
design, after some time can act on a degraded
structure, due to the occurrence of an existing
damage, which can be produced by many causes,
i.e. age of the construction, ravage of time,
permanent effect of previous earthquakes, etc;
- the design action has been correctly considered,
but the execution of the structure has been bad,
being the constructional imperfections largely
greater than tolerance limits or the quality of
materials not conform to the design pre-requisites
for strength".
These definitions of "exceptional actions" are
proposed for identifying the WG2 activity.
3 EXCEPTIONAL SITUATIONS
3.1 General definitions
The Eurocode "Basis Structural Design" (prEN
1990:2001) gives the following definitions for
accidental situation (1.5.2.5): "design situation
involving exceptional conditions of the structure or
its exposure, including fire, explosion, impact or
local failure".
In addition seismic design situation (1.5.2.7) is:
"design situation involving exceptional condition of
the structure when subjected to a seismic event".
These clauses mean that an accidental situation
produces exceptional conditions, as well as a seismic
situation, independently whether they are included in
the design process or not. This is partially
contradictory with the clause of Section 4,
previously mentioned in 2.1. It seems that for the
69
Eurocode the concept of exceptional is related just to
the low probability of occurrence of a given
situation during the life period of a construction,
rather than to the level of produced damage.
Following to the previous definitions of
exceptional actions (see Section 2.3), it is possible to
define when an "exceptional situation" arises. It
corresponds to a combination of actions and
structural conditions which do not fulfill any code
provision, giving rise to different levels of damage.
They can be evaluated according to the performance
based design approach, which is very familiar to
earthquake design engineers, but the same concept
can be applied to any type of action.
The performance levels of a structure are
normally classified as follows:
- fully operational
- operational
- life safe
- near collapse.
The occurrence of each damage level is related to
a given level of design action, which becomes
exceptional when it produces an exceptional
situation according to the previous definition.
Reaching the last damage level (near collapse), the
action produces a disaster, which is the extreme
exceptional situation.
The actions producing disasters can be classified
according to their origin:
- natural (mainly earthquake when not considered
in the design; in addition there are: windstorms,
avalanches, volcanic eruptions, floods, landslides)
- artificial (i.e. actions produced by the intentional
or unintentional activity of man; they are: fire,
explosions, impacts).
3.2 Natural disasters
In Fig. 1 the worst natural disasters in the World are
listed since mid- 14th century. It can be observed that
the causes of the main disasters in terms of number
of deaths belong to medicine (bubonic plague,
influenza) and to wrong combinations of policy
versus economy producing famine (see China and
India).
But the other causes, like earthquake, flood,
typhoon, tsunami mainly produce fatalities due to
the unreliable behaviour of man works and,
therefore, they have to be carefully considered in the
engineering activities devoted to the disaster
protection.
 Deaths Date Cause Location
About 75,000,000 Mid-14* centurv Bubonic Plague (Black Death) Europe and Asia
About 22,000,000
About 20,000,000
1918
1969-71
Influenza
Famine
Worldwide
Northern China
Fig. 2 shows the percentage of deaths due to
1,500,000
1,000,000
1943-4 Famine Benna], India natural disasters in the twentieth century. It can be
1939 Flood Henan, China
900,000 1887 Rood Henan, China noticed that earthquake is the worst exceptional
830.000 1556
500,000 1970
Earthquake
Typhoon /Flood
Shanxi. China
Bangladesh action responsible of producing the largest number
300,000
300,000
1737
1881
Earthquake
Typhoon
Calcutta, India
Haiphong, Vietnam
of victims (58.13%), followed by floods (28.10%).
300,000 1642 Flood Huanxhe River, China The wind effects under form of hurricanes,
250,000 526 BC Earthquake AnbEoch. Syria
242,000 1976 Earthquake Tianjin, China tornadoes and typhoons represent the 10.67%. The
215.000 1876 Tsunami (Tidal Wave) Bay of Bensa). India
200,000 1703 Earthquake Tokyo. Japan last position is for the volcanic eruptions (2.80%)
180.000 1920 Earthquake
137,000 1730 Earthquake
Gansu, China
Japan
which are very often connected to seismic activities,
IO0.OO0
100.000
1228
1290
Flood
Earthquake
Friesland. Netherlands
Hebei. China
while all the other exceptional actions (avalanches,
100.000
100,000
1731 Earthquake BeiiinR. China landslides, etc.) have a very small occurrence
1911 Flood Chianfrjiang River. China
99,000 1923 Earthquake Tokyo, Japan (0.30%).
60,000 1693 Earthquake Catania, Italy
60,000 1755 Earthquake Lisbon, Fortuita! Looking to the great natural catastrophes in the
World during 2001 (Fig. 3), we have the
Figurei. World Worst Natural Disasters (www.geocities.com) confirmation that earthquake is the main killer
(14.000 deaths in India), but the storm is the main
cause of economical losses.
Volcanic eruptions Others This opposite trend clearly appears from the
2,80% o,30%
Hurricanes, tornadoes percentage distribution worldwide (Fig. 4). On the
typhoons, etc.
10,67%
basis of 25.000 fatalities, 64% are produced by
earthquake and only 8% by windstorm. Contrary, the
Floods
55% of economical losses are related to windstorm,
28,10% 58,13% while 24% to earthquake.
Fig. 5 gives the tragic balance of the natural
catastrophes in the period 1985-1999, in terms of
Figure 2. Percentage of deaths in the twentieth century due to
natural disasters (Gioncu &Mazzolani,2002)
both costs and deaths. The geographical distribution
shows that Asia pays in absolute the greatest
contribution with 45% of costs and 77% of deaths.

WORID MAP OF NATURAL CATASTROPHES 2001

*> -*<

Graat natural catastropha«

No. Oat* Εν«κ Fatatole* EMnornto lotaae TtMutadlrMa··
*^°" (USI ml (US* m)

Figure 3. Great natural catastrophes in 2001 4 13.1 El Salvador Earthquake, land-slid·! MS

14.000
1,500
4.500
300
100
3 ».1 India Earthquake·
(www.munichre.com). IS β-Μ.4 USA Sevra storm, hag
Tropical Storm AUleofl
1 5,500
«,000
1,900
3, KM
M 0-17.0 USA

70
 Others
9%
Others
10%
Earthquake,
volcanic eruption 3.3 Artificial disasters

There are not statistics about the number, the type

and the feature of artificial disasters, also because
Earthquake,
volcanic eruption they are a quite modern phenomenon strictly related
64%
to the development of terrorism.
Opposite to the damage produced by natural
a. b. disasters, they appear more and more in the most
Figure 4. Percentage distribution worldwide of fatalities (a) civilized Countries.
and economic losses (b) due to natural catastrophes in 2001 The objectives of terrorism are usually the most
(www.munichre.com). challenging constructions, like skyscrapers and long
span bridges, which collapse creates a spectacular
The number of deaths decreases as far as the image, shocking the public opinion. A catastrophic
civilization degree of the World region increases. cocktail of impact, explosion and fire produces
The limit example on the opposite side is given by tragic effects.
North America, where the percentage of death is the
minimum (1%), but the percentage of costs is quite
high (33%), second after Asia. It means that,
obviously, the society there spends much more
money to protect the human life than in less rich
Countries (South America), where the economy
cannot allow it. The same trend, but less marked, is
recognizable in Europe with 12% of costs versus
1.5% of deaths. · · -r. ·:·..
From this general overview it clearly appears that
the progress of civilization can reduce the amount of
fatalities side by side to the engineering activity,
whose task is to produce safer urban habitats by
providing more reliable design rules, more realistic
structural models and improved code provisions.

Figure 6. The WTC tragedy in New York

AFRICA SOUTH AMERICA
1.0% 2.0% (www.newyorker.com).
PACIFIC REGION CARIBBEAN
2,1)14 3.0%
UTIN AMERICA
4.0% The collapse of the Twin Towers represented the
collapse of a mythological symbol for all structural
engineers, who never believed that such a
catastrophe could arise (Fig. 6).
Despite the fact that these two buildings were
designed during the sixties considering the impact of
a.
PACIFIC REGION
an aircraft and the structure was fire protected, so
°·«** AFRICA
SOUTH AMERICA
these actions were considered in the design process,
EUROPE *sn
1.S%
10.0% CARIBBEAN
2.0%
the exceptional situation raised from the unforeseen
E--'" UTINAMERICA
4.0%
intensity ofthe effect of such combination of impact,
NORTH AMERICA
1.0%
explosion and fire.
The event of 11th September 2001 provided a bad
example, which has been recently followed in a
smaller scale. In April 2002 the "Pirellone" building
ASIA
77.0% in Milan, a high-rise r.c. building of 30 stories, has
b. been hit by a small airplane with a suicidal pilot,
who perfectly centered the 25th . floor (Fig. 7),
Figure 5. Geographical distribution of costs (a) and deaths (b) producing two victims.
for natural catastrophes in 1985-1999 (www.repubblica.it).

71
 Fortunately this impact was not followed by the Based on the observation since 1900, the frequency
building collapse and the damage level was limited, of occurrence of earthquakes is given in table of Fig.
being the situation not so extreme. Taking profit 8, where the magnitude (in Richter scale) is related
from this event, the Concrete Industry, not honestly, to the evaluation scale (from very minor to great).
predicted that r.c. is more safe than steel!
How to protect the human kind and his constructions 90000
80060
from the exception situations produced by the 80000
human madness is now one ofthe most pressing new
70000
problems for structural engineering. -5676-4"
60000
50000 44421

40000
30000
'15072"
20000
.9.049.
10000 75 1642 207 14
0
0.1 to 1.0 to 2.0 to 3.0 to 4.0 to 5.0 to 6.0 to 7.0 to 8.0 to
^ 0.9 1.9 2.9 3.9 4.9 5.9 6.9 7.9 9.9

Figure 9. Statistical distribution of earthquake magnitude since

1990 (www.usgs.gov).

Even earthquakes, which are classified as strong,

Figure7. The"Pirellone" impact in Milan (www.repubblica.it).
4 MAIN FEATURES OF EXCEPTIONAL
ACTIONS IN EXCEPTIONAL SITUATIONS
4.1 Earthquake
As it is the main responsible ofthe natural disasters,
the earthquake is the most worldwide studied action.
The US Geological Survey (USGS) estimates that
several million earthquakes occur in the World each
year. Many go undetected because they hit remote
areas or have very small magnitudes. The National
Earthquake Information Center (NEIC) now locates
about 50 earthquakes each day, or about 20.000 a
year.
major or great, do not necessarily create exceptional
situations, because they were considered in the
design process. Contrary, light or moderate
earthquakes acting on old and degraded buildings
can produce exceptional situations. The statistical
distribution of magnitude during the last twelve
years (Fig. 9) shows that the average value is located
between 4 and 5.
In the same period the correlation between the total
earthquakes per year and the corresponding
estimated deaths (Fig. 10) demonstrates that there is
not a correspondence between the numbers of
earthquakes for year, varying from 16.000 and
23.000, and the number of victims, which goes from
few hundreds to 52.000 (Iran, June 1990). Therefore,
we can recognize that not all earthquakes are
exceptional, but the number of deaths is
symptomatic of an exceptional situation.

Degree Magnitude Average Annually

Great higher
8 and 1

Major 7-7.9 18
Strong 6-6.9 120
Moderate 5-5.9 800
Light 4-4.9 6,200 (estimated)
Minor 3-3.9 49,000 (estimated)
Magnitude 2-3: about 1,000 per day
Very Minor <3.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Magnitude 1-2: about 8,000 per day
Figure 8. The earthquake frequency according to magnitude
since 1900 (www.usgs.gov). Figure 10. Earthquake versus estimated deaths since 1990
(www.usgs.gov).

72
The 8 to 8.3 magnitude earthquake occurred in 1556 high) is still very acting, producing a quite
in Shaanxi Province (China) and produced around continuous emanation of gas and lava (Fig. 12).
830.000 people died, it being a very exceptional
situation.

Earthquake Magnitude Fatalities \

January 1970 - Yunnan v< J' -
Province, China
7.5 C. 10.000
May 1970 -Peru 7.8 c. 66.000 ¡ X
February 1976 - Guatemala 7.5 C. 23.000

July 1976 - China (Tangshan) c. 225.000

8.0
(up to c. 655.000 unofficial estimates)
September 1978 - Iran 7.8 c 15.000
C. 9.500
September 1985 - Mexico 8.1
(up to 30.000 unofficial estimates)
December 1988-
7.0 c. 25.000
Turitey/USSR border
"' 41-
June 1990 - Iran 7.7 Up to 50.000
>

' August 1999 - Turkey 7.4 Minimum 17.000

13.805
January 2001 - India 7.9 (a further 14 fatalities were recorded In
Pakistan)
^Λ\-

Figure 11. Earthquake producing more than 10.000 deaths in

the last 30 years (www.btinternet.com).
In more recent times, the number of fatalities is
going to be progressively reduced thanks to the
progress in engineering, but it is still very high.
Table of Fig. 1 1 lists the earthquakes occurred during
the last 30 years with a number of fatalities greater
than 10.000. The balance from this scenario is
unacceptable, showing that a comprehensive activity
in seismic protection of buildings must be done
urgently, despite the unquestionable on-going
progress in this discipline.
Figure 12. Some pictures of the last Etna eruption in 2001
(www.educeth.ch).
The Vesuvius near the Gulf of Naples (Fig. 13)is
smaller than Etna(only 1000 m). Its last eruption was
in 1944 during the Second World War. Since that
time there was no activity, meaning that a new
eruption has to be expected, due to the large amount
of energy cumulated inside in a so long period.

4.2 Volcanic eruptions

The products of a volcanic eruption are basically

three: piroclastic ash, poisoning gas and lava.
The former can cover the top of
buildings giving rise to an increase of vertical loads,
which can produce the collapse of roof structures,
being an exceptional action, because it was not '",. t

considered in the design.

Against the second, there are no engineering
provisions, being the gas-mask the only immediate
protection for the individual man.
The lava represents a slow flow of incandescent
magma, which proceeds with a given speed and can
run over the buildings laying on its way. It can
destroy entire villages, but population has enough
time to evacuate. Engineering works can deviate the
original lava flow, by means of "ad hoc" explosions,
which create lateral way of outlet.
In Europe the main volcanoes are located in the
South of Italy: Naples (Vesuvius), Eolian Islands
(Stromboli) and Sicily, where the Etna (3000 m
Figure 13. Some pictures of the Vesuvius beside the Gulf of
Naples.
In the period 1600-1982 the total number of
deaths for volcanic eruptions was 238.867 and the
corresponding geographical distribution is given in
Fig. 14. The eruptions producing more than 500
known human fatalities are listed in the table of
Fig. 15, since the Vesuvius one in 79 B.C. which
destroyed the town of Pompei.
From the major causes of death, we have the
confirmation that lava is very rare and the most
frequent are ash and mud flows. Other eruptions
have been as big or bigger than these, but no one

73
lived nearby to be threatened (for example the
Valley of 10,000 Smokes eruption in Alaska in
1912). The Mt. St. Helens eruption in 1980 in
Washington state was a far less dangerous eruption
than these, only 61 humans died, although thousands
of deer and other animals perished.
The ancient eruption at Santorini Island in the
Mediterranean Sea in about 1650 certainly killed
thousands of people (and was the source of the
Atlantis legend), but there are no real estimates of
the number of deaths.
4.3 Landslide
This phenomenon can be produced by the variation
of the hydro-geologic regime in the territory, due to
natural or sometimes artificial causes. The flow of
mud or ground comes from the hillsides and can
invade the built-up areas. The energy of this semi¬
solid mass can produce an exceptional action against
the constructions, which are not designed to resist
such an action and, therefore, they can be seriously
damaged or even destroyed.

PHILIPPINES
1,6*
1.7S PAPUA MEW GUINEA
CENTRAI 1,5%
2,3% ALL OTHERS
0,9%

3.9%

JAPAH
8,1%

CARIBBEAN
12,9%

Figure 14. Percentage of deaths in each volcanic region from

1600 to 1982 (Blong,RJ.1984;www.vocano.und.
nodak.edu).

Deaths Volcano When Major Canse of Death

92,000 Tambora, Indonesia 1815 Starvation
36,417 Krakatau, Indonesia 1883 Tsunami
29,025 Mt Pelee, Martinique 1902 Ash flows
25,000 Ruiz, Colombia 1985 Mudflows
14,300 Unzen, Japan 1792 Volcano collapse, tsunami
9,350 Laki, Iceland 1783 Starvation
5,110 Kelut, Indonesia 1919 Mudflows
4,011 Galunggung, Indonesia 1882 Mudflows
3,500 Vesuvius, Italy 1631 Mudflows, lava flows
3,360 Vesuvius, Italy 79 Ash flows and falls
2,957 Papandayan, Indonesia 1772 Ash flows
2,942 Lamington, Papua N.G. 1951 Ash flows
2,000 El Chichón, Mexico 1982 Ash flows
1,680 Soufrière, St Vincent 1902 Ash flows
1,475 Oshima, Japan 1741 Tsunami
1,377 Asama, Japan 1783 Ash flows, mudflows
1,335 Taal, Philippines 1911 Ash flows
1,200 Mayon, Philippines 1814 Mudflows
1,184 Agung, Indonesia 1963 Ash flows
1,000 Cotopaxi, Ecuador 1877 Mudflows
800 Pinatubo, Philippines 1991 Roof collapses and disease
700 Komagatake, Japan 1640 Tsunami
700 Ruiz, Colombia 1845 Mudflows
Hibok-Hibok,
500 1951 Ash flows
Philippines

Figure 15. List of all eruptions with more than 500 known
human fatalities (Blong, R.J., 1984; www.volcano. Figure 16. Some pictures ofthe Samo landslide
und.nodak.edu). (courtesy Gianfredi Mazzolani).

74
 A wide landslide produced many damages in the
town of Sarno (Naples) and in the surrounding areas
(Bracigliano, Quindici, Siano) in May 1998. There
were hundred and sixty victims; fifty per cent of
buildings were damaged (20% collapsed and 80%
evacuated); large economical losses resulted there.
Some pictures of this disaster (Fig. 16) show the
mudflows progressing along the streets and the
consequent damage. The steel structure of an
electrical transmission tower totally collapsed;
facades of buildings were cut away; reinforced
concrete columns were folded and just the steel
reinforcements still remain, being the concrete
completely disaggregated; masonry houses cannot
resist this exceptional pressure.
This kind of action really creates exceptional
situations. It should be useful to interpret the effects
of this action on the structure by setting-up
mechanical models. They can be based on the
observation of damage and the subsequent
evaluation of the energy spent by the mudflow for
producing such a kind of damage. This energy can
be transformed into a force system acting on the
structure, as the action (exceptional) necessary to
cause the examined damage.
Collapse mechanisms have been proposed by C.
Faella & E. Nigro of the University of Salerno
(Italy), based on the damage produced on buildings
during the Sarno landslide. The comparison between
the ultimate load and the hydrostatic plus
hydrodynamic pressure due to the mud flow allowed
to evaluate the impact velocity, by assuming a given
flow depth under appropriate simplified hypothesis.
Many collapse mechanisms, both for r.c. and for
masonry structures, have been identified and the
formulations of the impact velocity as a function of
the ultimate characteristic of each mechanism have
been derived accordingly in a closed form.
Based on these results, special recommendations for
the mitigation of the landslide risk for new
constructions to be built in the damaged area have
been issued. The design parameters of the flow are
given to check the structure resistance, by assuming
different values of the flow velocity varying from 5
to 10 ms"1.
In 2001 the landslide events produced 1131 deaths
in the World, hitting particularly poor Countries of
Latin America (El Salvador, Ecuador) and Asia
(Indonesia, Nepal, India, Thailand).
According to statistics, the annual average number
of people killed is about 800-1000, for a total of 14
events in the World.
75
4.4 Fire
The effect of fire on structures is one of the most
studied subjects. Fire engineering is now a well
consolidated discipline and its codification is in a
very advanced stage.
Parts 1.2 in all Eurocodes are in parallel devoted
to Fire Design, according to an imperative directive
of the European Commission. Therefore, fire is not
belonging to the category exceptional actions, at
least for new constructions, where an appropriate
protection is provided. But there are some special
cases in which fire creates exceptional situations.
The most common case is the one of old masonry
buildings which were not designed against fire.
Their situation can be worsted, because the floor
structures are very often made of unprotected
timber.
After a strong earthquake, it frequently happen
that, due to gas pipeline broken, a big fire arises
(Fig. 17). It acts on structures, which are already
degraded because damaged by the earthquake. This
combination produces an exceptional situation, even
if the two actions were considered in the design
separately, but not acting together.
New studies are requested to evaluate the effects
of this combination of fire and earthquake. The
structure, damaged by the earthquake at a give level
of damage but not yet collapsed, still posses a
residual amount of safety. In case of fire, how to
evaluate the residual time of life of the structure
before collapse?
From this combination, an exceptional situation
can result in a catastrophe. This was the reason of
many collapsed and burnt buildings during the Kobe
earthquake (January 1995).
'V8*
i¿
SX*
Figure 17. Fire after earthquake in Kobe (1995)
 modification of the load carrying capacity of the
soil.

DATE PLACE COLLAPSED BUILDING CAUSE DEATH!

16/09/59 Barletta (Ba) Apartment building Unlegal superelevation 60
30/11/72 Rome Apartment building n.a. 15
13/1 1/79 Parma Hospital (3 stories) Gas explosion 20
12/10/81 Montecchio (Re) Apartment building Gas explosion 7
27/10/81 Pisa Historical building Gas explosion 9
18/12/82 Rome House Gas explosion 3
07/02/85 Castellaneta (Ta) Building Water infiltration 34
Λ 23/07/86 Rome Building Water infiltration 2
18/12/87 Lecco Historical building Gas explosion 7
04/03/89 Rome House Gas explosion 0
19/02/91 Rome Building n.a. 1

26/02/91 Pozzuoli (Na) Building under rehabilitation Explotion 8

24/01/92 Rome Fire brigade school n.a. 3
16/02/92 Rome Building Gas explosion 1
16/12/92 Naples Building Gas explosion 15
14/07/94 Motta Visconti (Pv) House for the old n.a. 28
04/10/96 Rome Building Gas explosion 1

15/12/98 Rome Apartment building n.a. 38

1 1/03/99 Palermo House n.a. 3
11/11/99 Foggia House (6 stories) n.a. 67
Figure 18. Fire due to terrorist attack in the Law Court of Figure 19. List of 20 collapsed buildings in Italy from 1959 to
Naples (Courtesy Pasquale Malangone). 1999 (www.repubblica.it).

Another dramatic combination comes when fire is
originated by an explosion. It was the case of the
Law Court Palace in Naples, the biggest in Europe
(Fig. 18). The tallest tower, some days before the
opening, was objective of a "camorra" terrorist
attack, by an explosion with subsequent fire. The
structure is made of a r.c. curved wall, which
provides to resist horizontal loads, integrated by a
pinned steel skeleton designed for vertical loads
only. The fire produced an exceptional situation,
which so seriously damaged the steel structures that
it was necessary to remove and replace them.
4.5 Explosions
The gas explosion is one of the main causes of
building collapse in the urban habitats. Fig. 19
shows a list of 20 collapsed buildings in Italy in the
period 1959-1999. In forty years 20 buildings
collapsed, producing all together 332 victims, an
average of about 9 for year. The buildings were
made of r.c. and masonry. The geographical
distribution around the Country corresponds to the
following locations: 5 in the North, 6 in the South
and 9 in the city of Rome.
The majority of buildings (18) were apartment
buildings or houses, some of them having a
historical feature.
Half of the collapse causes is due to
gas loss and consequent gas explosion. In many
cases, it was impossible to detect the actual reason
causing the structural failure, but it is reasonable to
suspect that some changes of the structural scheme
were illegally made, like elimination of structural
members and super elevations, giving rise to a
dramatic weakening ofthe overall carrying capacity.
There were only two cases of water
infiltration under the foundation, producing a
76
In general, these accidents has been caused by the
occurrence of an "exceptional situation", which
undermined the structural integrity of the buildings
until an extreme level of damage producing collapse.
One more general remark is that these buildings
were originally constructed in a bad way with poor
technology and later on degraded for a systematic
lack of maintenance. Their feature exactly
corresponds to the fact that even normal actions in
serviceability conditions can become "exceptional
actions", according to definitions of Section 2.3.
These situations are very often not foreseeable at the
light of a simple inspection of buildings, whose
robustness is accepted "a priori" as a dogma. It is
necessary to have a comprehensive overview of the
historical background in order to identify -
especially for old buildings - the modifications, the
changes in use and the maintenance, which the
building structure has undergone.
After the last failures of the apartment buildings in
Rome (1998, 38 deaths) and in Foggia (1999, 67
deaths), the Italian Government decided to rule by
law the assessment of buildings in urban habitats, by
creating the so-called "booklet of the building",
which has to collect all information about the
original features and the modifications of the
building structure ,related to the safety assessment.
Anyway, cases of unforeseen failure due to
accidental explosions are quite common worldwide.
It is well known the so-called Ronan Point incident
in U.K. in 1968. An apartment building, 23 stories
made of precasted concrete panels, collapsed due to
a gas explosion at the 18th storey. Due to the failure
of one panel, a progressive collapse was produced
from top to the bottom for a 25% of the building
plane.
As a consequence, the U.K. Government set-up masonry itself, without proper chaining by means of
design rules against this exceptional situation, steel ties, is unable to resist strong horizontal forces.
compulsory applicable for more than five stories
buildings. The effectiveness of these rules have been
also tested later on for the explosions produced by
the IRA terrorist attacks during the 80s.

5 STRUCTURAL INTEGRITY VERSUS

ROBUSTNESS
Figure 21. Masonry houses collapsed during the earthquake of
Campania-Basilicata (Italy) in 1980.
5. 1 According to the experience

The scenario of each disaster represents of natural

laboratory, where the structures are tested in full
scale, not only in elastic range under static loadings,
but also in plastic range under non static, dynamic
and/or impulsive loads until collapse. These tests
show the actual robustness of structural elements,
which very often is not enough to guarantee
sufficient structural integrity with a limited degree of
damage.
The most frequent lessons to be learned come
from the analysis ofthe post-earthquake scenario.
During an earthquake each structural material is
recalled to its responsibility.
The old constructions made of timber or masonry
are the first to collapse. It is interesting to observe
the collapse mode of a timber house during the
earthquake of 1999 in Turkey (Fig. 20). A classical
storey mechanism arises at the first floor, which
behaves as a "soft floor", due to lack of bracing.
a. b.
Figure 22. R.C. buildings collapsed during the earthquakes of
Campania-Basilicata (1980) (a) and Mexico City (1986) (b).
But what is absolutely unacceptable is the fact that
a modern r.c. structure, designed and executed in the
formal accordance with the code provisions, can
miserably collapse under a foreseen earthquake (Fig.
22).
Bad execution, poor materials, lack of control,
they are the main responsible of many catastrophic
events.
It happens not only for poor constructions, but
also for very modern and well designed seismic
resistant structures, when the loading condition
becomes exceptional, because it exceeds in a large
portion the design assumptions. It was the case of
the Kobe earthquake (1995) for many buildings (Fig.
23a,b) and bridges (Fig. 23c,d).

'BSjSi.

a. b.

V
Figure 20. Collapse mechanism of a woden structure after the
earthquake of 1999 in Turkey.

Many masonry houses collapsed during the c. d.

earthquake of Campania-Basilicata (Italy) in 1980 Figure 23. Failure of buildings (a,b) and bridges (c,d), made
(Fig. 21). It cannot be justified, but it can be of r.c. and steel, respectively, in Kobe (Architectural Institute
of Japan, 1995).
explained, due to the age of these poorly made
constructions and to the unquestionable fact that
77
 By generalizing these remarks and analyzing them
at the light of an engineered interpretation, it can be Annex Β of EN 1991-1-7 in parallel is devoted to a
pointed out that natural and artificial disasters, by general strategy for Risk Analysis and includes key
creating exceptional situations, often produce partial components that should be considered when
or overall collapse of buildings. In the more lucky designing building' structures. Rules for the
cases, they are strongly and irreversibility damaged. "foreseeable" risks are based on data and experience,
It happens because obviously they were not designed which has been collected for several years on
against this kind of actions and, therefore, according accidental damage caused by impact from ships,
to the above definitions, these actions create lorries and helicopter crashes. In the case of non-
"exceptional situations". In a more technical foreseeable risk analysis, rules are only provided to
language, we can say that the structural integrity of give threshold levels of robustness, typically level of
the construction has been partially or totally resistance against abuse.
violated. It means consequently that their structural During the last Coordination Group meeting
elements exhausted their own robustness. (Nicosia, 15-16 March 2002), it was noted that after
the Ronan Point incident in U.K. (see Section 4.5),
they undertook a review of national procedures
5.2 According to the Codes relating to risk analysis rules, recognizing that these
rules are subjective, because they can derive by
Both terms "structural integrity" and "robustness" emotive consideration.
are normally used in the common technical From this provisional overview, it clearly appears
language, but their meaning is often mixed-up. In that the Eurocodes are developing a very important
fact, in Eurocode "Basis Structural Design" (prEN activity in the field of actions, during the conversion
1990:2001) Section 2.2 "Reliability management", phase now in progress.
the levels of reliability relating to structural
resistance and serviceability are defined to be
achieved by suitable combinations of different 5.3 According to design
measures. Among them, mention is made to "the
degree of robustness (structural integrity)". So, here Beside the codification activity and waiting for its
both terms are used with the same meaning. final out-put, it seems now useful to propose "ad
Under form of provisional document, EN 1991-1-7 hoc" definitions for both structural integrity and
"Accidental actions" proposed in its draft Annex A: robustness from the design point of view.
"Robustness of Buildings: design for consequences In fact, when a structure is designed,
of localized failure from an undefined cause" a special distinction is made between the "requested
strategy that will ensure that a building is capacity", deriving from the loading conditions
sufficiently robust to sustain a limited extent of which the structural is asked to resist, and the
damage or failure, depending on the consequence "available capacity", which the structure has to
class, without collapse. provide to fulfill the above requirements.
The consequence classes are defined in the The assessment "requirement versus availability" is
informative Annex Β of EN 1990. They are shown the basis of structural design. Consequently, by
inTableofFig.24. generalizing this concept, the following definitions
can be given.
Structural integrity can quantify the capacity
Consequences
Description
Examples of buildings and which is required to the structure for resisting a
Class civil engineering works
given situation of loading with different levels of
High consequence for loss Grandstands and public
of human life, or economic, buildings, where damage. In case the situation is considered as
CC3
social or environmental consequences of failure are "exceptional", the required capacity of the structure
consequences very great high (e.g. a concert hall)
Mediam consequence for Residential and office
should correspond to the highest damage level,
loss of human life, buildings, public buildings which, according to the performance based design,
CC2 economic, social or .where consequences of
environmental consequences failure are medium (e.g. an
corresponds to "near collapse".
considerable office building) Robustness is the overall load bearing capacity,
Low consequence for loss
Agricultural buildings, according to strength, stiffness and ductility, which a
of human life, and
CCI economic, social or
where people do not structure or a structural element is able to provide.
normally enter (e.g. storage
environmental consequences
buildings), greenhouses This kind of capacity is clearly the available one,
small or negligible
having the feature of a basic pre-requisite
Figure 24. Definition of consequences classes (EN 1990).
independently ofthe specific cases, but which can be
suitably spent against exceptional situations.
78
 In case of exceptional situations, where the
materials are overloaded largely beyond the elastic
limit, the structural integrity of a construction can be
guaranteed provided that the structural elements
posses enough robustness.
From this point of view, robustness is, therefore,
related also to the capability of a structure to survive
under unforeseen circumstances.
The triad "strength-stiffness-ductility" is a basic Figure 26. Timber structures collapsed during the Kobe
earthquake (Architectural Institute of Japan, 1995).
requirement for assessing the robustness of
structures. The evaluation of the optimum relation¬ In any case, despite these recognized gaps, the
ship among these three basic properties is a modern earthquake engineering has reached a quite good
subject of research in different engineering fields level for the design of new constructions. The same
both material and structure oriented. It is well known is not completely true for existing constructions,
that the study of seismic resistant structures is a very which need repairing and up-grading. Consistent
advanced research subject and many improvements progress are expected in this direction, especially in
have been reached, also thanks to the lessons learned Countries, like Italy, where the percentage of old
from the earthquakes. It is well known that the lack buildings is very high in the historical centers, many
of ductility in the r.c. structures, is mainly originated of them belonging to the cultural heritage.
from the poor behaviour of nodes, very often due to
the complete absence of stirrups (Fig. 25).

Figure 27. Brittle behaviour of steel during the Kobe

earthquake (Architectural Institute of Japan, 1995).
Figure 25. Damaged nodes in r.c.
From a first glance, it seems that the situation in
the field of other "disasters" is less encouraging, in
Also in case of timber structures, the lack of
spite of the enormous cost paid in terms of both
ductility of connections is the origin of many
economical losses and human life.
failures (Fig. 26).
Therefore, beside "earthquake engineering"
Unexpected and unbelievable was the brittle
dealing with "seismic-resistant structures", or the
behaviour of steel during the Kobe earthquake
existing "wind engineering" and "fire engineering",
(Fig.27). In fact, it is generally accepted that steel is
an excellent material for seismic-resistant structures,
it is expected that new disciplines could cover with
the same scientific approach the other less
thanks to its strength and ductility.
investigated exceptional actions, giving rise to a
Contrary, in some cases during the Kobe
generalized way how to design "disaster-resistant
earthquake, steel behaved as glass, showing that this
structures".
kind of unexpected failure was produced by an
"exceptional situation" due to the combination of
many unfavorable effects and actions, which
6 CONCLUDING REMARKS
normally were disregarded before by both the
codification and the design practice. The main
At the beginning of the activity, this general
causes were:
overview has been finalized to identify the scope of
- presence of strong vertical quakes, due to the
special type ofthe earthquake (near-field); COST C12-WG2 "Structural integrity under
- effect of very high strain-rate; exceptional actions". It has been clearly recognized
- low temperature reducing toughness. that the field of interest is extremely wide, because
in principle it deals with very different phenomena
It was clearly underlined the need of a deep producing actions on the common types of buildings
investigation on these unexpected phenomena in structures made of various constructional materials.
order to transform an exceptional situation into a It is impossible to believe that the capacity of WG2
normal situation for the future earthquakes. members can carry-on such an ambitious program in

79
few years, also because many aspects of the whole
scenario are not covered by their own experience.
This first report is just an attempt to clarify the ideas,
giving some general definitions and pointing-out the
correct way to follow in order to face all the
structural aspects ofthe general problem by using an
unified philosophy and a homogeneous approach.
Virdi, K.S., Matthews, R.S., Clarke, J.L., Garas, F.K. 2000.
Abnormal Loading on Structures: experimental and
numerical modeling. E & FN Spon, London.
The pictures of the damaged structures after the
Kobe earthquake are taken from:
Architectural Institute of Japan (Japan Society of Civil
Engineers), 1995. "Records of the 1995 Great Haushin-
Awaji earthquake disaster", Maruzen.

REFERENCES All the information on other types of exceptional

actions producing disasters are taken from Internet.
The structural problems related to seismic A sincere acknowledgment to Simeone Panico for
engineering are dealt with in a wide technical his effective collaboration in the Internet research.
literature. In particular, from the Author's side, the
following volumes on seismic resistant steel
structures can be mentioned:

Gioncu, V., Mazzolani, F.M. 2002. Ductility of Seismic

Resistant Steel Structures. E & FN Spon, London.
Lungu, D., Mazzolani, F.M., Savidis, S. 1997. Design of
Structures in Seismic Zones: Eurocode 8 and worked
examples. Tempus Phare Project 01 198.
Mazzolarli, F.M. 2000. Moment Resistant Connections of Steel
Frames in Seismic Areas: Design and Reliability. E «fe FN
Spon, London.
Mazzolani, F.M. & Akiyama, H. 1997. Behaviour of Steel
Structures in Seismic Areas: Proceedings of STESSA '97 in
Kyoto. Edizioni 10/17, Salerno.
Mazzolani, F.M. & Gioncu, V. 1995. Behaviour of Steel
Structures in Seismic Areas: Proceedings of STESSA '94 in
Timisoara. Chapman & Hall, London.
Mazzolani, F.M. & Gioncu, V. 2000. Seismic Resistant Steel
Structures. CISM Course. Springer-Verlag, Wien-New
York.
Mazzolani, F.M. & Piluso, V. 1996. Theory and Design of
Seismic Resistant Steel Frames. Chapman & Hall, London.
Mazzolani, F.M. & Tremblay, R. 2000. Behaviour of Steel
Structures in Seismic Areas: Proceedings of STESSA 2000
in Montreal. Balkema, Rotterdam.

A general overview on Fire Engineering for steel

structure is given in:
Schleich, J.B. et al. 1993. International Fire Engineering Design
for Steel Structures: State ofthe Art. I.I.S.I., Brussels.

Information on volcanic eruptions can be found in:

Blong, R.J. 1984. Volcanic Hazard: A Sourcebook on the
effects of Eruptions. Academic Press. Orland, Florida.

Information on explosive loading are given in:

Bulson, P.S. 1997. Explosive Loading of Engineering
Structures: a history of research and a review of recent
developments. E & FN Spon, London.

Papers on abnormal loading due to avalanches,

snow, blast, fire, impact and earthquake are selected
in:

80
Twin Towers: Qualitative Analysis of Collapse Behavior
G. Olivete & A. Marinetti
Department of Civil and Environmental Engineering, University of Catania, Italy

ABSTRACT: This paper describes the qualitative collapse behavior ofthe Twin Towers ofthe World Trade
Center destroyed by a terrorist attack on 11 September 2001. After a brief description of the history of
construction and ofthe structural concept follows an illustration ofthe attack and ofthe damage caused by the
impact. An explanation is provided of why the impact alone could not cause the collapse, either total or
partial, ofthe towers. It is then shown how the heat developed by fuel burning must have been equivalent to
one thousand impacts ofthe aircraft type used by the terrorists. The damage caused by impact and the strength
reduction caused by the temperature rise are used to explain the different collapse mechanisms ofthe top parts
ofthe towers. For the north tower this leads to an estimate ofthe impact velocity. Finally elastic and elastic-
plastic wave propagation models are used to explain the collapse mechanism ofthe lower part ofthe towers.
In particular it is explained why the towers imploded on themselves and did not undergo local or global
buckling collapse.

concept behind their design and construction is

1 INTRODUCTION
On September 11th 2001 there was the terrorist
attack on the Twin Towers of the World Trade
Center in New York. This lead to the collapse ofthe
towers with the loss of more than 3,000 human lives.
Subsequently many structural engineers have
tried to explain why the towers collapsed and why
they collapsed in the way they did. Several theories
have been presented in the literature and in the
media, Bazant & Zhou (2002) and just recently a
report from the committee established by the Federal
Emergency Management Agency (FEMA) has
become available (May 2002).
The first author was invited on March 5th 2002 to
comment on a presentation by a member of the
association of engineers in Catania. This occasion
provided the stimulus to present a simple model for
the interpretation ofthe qualitative collapse behavior
of the towers. Subsequently this model was further
elaborated for a presentation within the seminar held
in Lisbon on April 19th 2002. Meanwhile the
publication of the FEMA report and the analysis of
additional material has provided the means for
additional interpretations.
This paper is organized as follows. Firstly a brief
history of the construction of the Twin Towers is
presented. Secondly the innovative structural
explained and the typical structural units and
materials are commented upon. Thirdly the attack is
briefly described together with the immediate
damage produced by the impact and penetration of
the aircraft. Fourthly the energy developed by the
deflagration of the aircraft fuel is evaluated and
compared with that from the impact. Fifthly an
explanation is provided for the collapse of the top
part of the towers. Finally a simple mechanical
model is presented to explain the collapse of the
lower parts ofthe towers.
Although this model cannot provide a detailed
explanation of the collapse, it provides some insight
in the actual qualitative collapse behavior of the
towers.
2 BRIEF HISTORY OF CONSTRUCTION
The architectural design of the World Trade
Center Twin Towers was conceived in the early
sixties by the American Japanese architect Minora
Yamasaki. He used his own firm Minora Yamasaki
& Associates assisted by the architectural
partnership Emery Roth & Sons.
Three engineering teams participated in the
design: Skilling, Helle, Christiansen and Robertson
for structural engineering, Jaros Baum & Bolles for
mechanical engineering and Joseph R. Loring &

81
Associates for electrical engineering. Finally the Port
Authority of New York and New Jersey provided
design services for site utilities, foundations,
basement retaining walls and paving. Ground
breaking for construction started on the 5 of August
1966. The steel construction of the actual towers
began in August 1968. The towers were partially
used before they were completed starting from
December 1970. The construction was completed in
January 1972 while the ribbon cutting ceremony
occurred on April 4th 1972.
/*·
Τ 'φ^
Figure 1. One stage in the construction ofthe WTC. The north
tower is in a more advanced state.
· .
» «
Hi
.*-_.
*-&.' ?S
·**·-·>
*··*» '«avi«.
Ï*
Figure 2. A more advanced stage in the construction of the
WTC. The towers are already occupied by the tenants while
work still goes on.
Two stages of construction derived from the web
(Great Buildings on Line- World Trade Center) are
shown in Figure 1 and in Figure 2. In Figure 1 the
north tower is more advanced while the south tower
82
is at an early stage of erection. In both towers the
perimeter walls and the central core are clearly
visible. The central cores are differently oriented in
the two towers.
Figure 2 shows a more advanced stage in the
construction. While the north tower is almost
complete, the south tower is only about two thirds
complete. The illumination shows that the towers are
already being used by the tenants while work still
goes on.
3 STRUCTURAL CONCEPT
The WTC design presented several innovations
for the time. These are described by Robertson
(2002), at the time in the team of structural
engineers. Some challenging aspects and innovations
used in the construction of the foundations are
described by Tamaro (2002), then working for the
Port Authority.
3.1 General innovation
Among the most significant innovations
considered in the design and construction are the
wind tunnel studies lead by Alan G. Davenport
(Robertson 2002), the use of viscoelastic dampers to
control vibration caused by wind gusts, extensive
préfabrication and novel architectural and structural
concepts. Here only the new structural concepts will
briefly be discussed.
3.2 Structural innovation
The framed tube concept with an inner core was
used as a structural system. Figure 3, taken from the
same website quoted above, shows a typical floor
plan. A more detailed representation may be found
in chapter 2 ofthe FEMA report (May 2002).
»-«.-·- m- tl·.-«.-«' »·--» ~»r-«~»~»~ »-»»-
/*" <m
I
I
Figure 3. Typical floor plan.
 behavior will be considered. A schematic view ofthe
There were 59 box sections columns on each of floor supporting system, derived from Clifton (2001)
the flat faces of the building. Only above the 84th is reported in Figure 5. The perimeter columns are
floor rolled or welded built-up I-shaped sections connected at floor levels by deep thick plates. These
were used. confer to the exterior walls a great in-plane stiffness.
A typical cross-section ofthe used box columns is The floor supporting structure is of the grid type
represented in Figure 4. These closely spaced with principal and secondary trusses disposed as
perimeter columns, 3 feet and 4 inches apart, center shown in Figure 6.
to center, provide most of the bending stiffness and
strength required to resist wind loading. Twelve
grades of steel, with yield strength varying from 290
N/mm2 to 690 N/mm , were used in the construction
of these box shaped columns. The thickness also
varied along the height from ΙΛ of inch in the upper
floors to 4 inches at the base ofthe building.
The core was also built of rectangular box
sections fabricated from A36 steel plates 36 inches χ
14-16 inches and thickness from 3Λ to 4 inches.
Above the 84th floor, rolled or welded built-up I-
shaped sections were used.

Figure 5. Schematic view ofthe floor supporting system and of

the perimeter column assembly.

Figure 4. Typical perimeter column box-section.

36 - steel plate; 38,39 - fire resistant plaster; 40 aluminum
façade; 42 - window glass; 43 - window frame.

The core resists mainly gravity loads with a

negligible contribution to wind loading.
This particular structural arrangement allowed for
office space free of columns. Moreover the
concentration of structural material along the
perimeter in the shape of closed spaced box columns
determined a greater stiffness and load carrying
capacity than structural concepts previously used in
tall buildings.
3.3 Floor supporting system
The floor supporting system is described in more
detail in the FEMA report. Here only the basic
elements necessary to understand the collapse
Figure 6. Schematic plan of the floor supporting system. 13 -
perimeter frame; 14 - principal trusses 900mm deep; 15 -
secondary trusses; 16 - horizontal floor bracing; 17 - core box
columns.
The principal trasses span from the exterior walls
to the inner core. The secondary trasses are
perpendicular to the principal ones and ran parallel
to wide side of the inner core. There are also
principal trasses spanning from the exterior walls to
the narrow side of the inner core. This time the
secondary trasses run parallel to the narrow side of
the core.

83
 On top of the secondary trasses a metal deck runs
parallel to principal trasses and supports a 4 inches
thick lightweight concrete slab.
Of considerable importance for subsequent
consideration are the support conditions for trasses
and deck. The former are shown in detail in Figure
2.6 of
the FEMA report. Here it suffices to say that the
principal trasses are simply supported at the two
ends. At both ends the connection to the columns
occurs at the level of the top chords. At the external
end the lower chord is connected to the column
through a viscoelastic damper.
Support to the principal trasses is also provided
for out-of-plane action at the level of the upper
chord.
In conclusion the floor supporting system, in the
whole, can be considered as grillage with simple
support conditions for out of plane action. These
conditions extend along the perimeter walls and
along the core perimeter.
This grillage or plate action is important to
understand the load redistribution after the aircraft
impact.
THE ATTACK
According to the FEMA report and other
available information the first plane, American
Airlines Flight 11, hit the northern face ofthe north
tower (WTC1) at 8.46 a.m. The second plane,
United Airlines Flight 175 crashed into the southern
face of the south tower (WTC2) at 9.03 a.m. Both
planes were Boeing 767-200ER with enough fuel for
a transcontinental flight from Boston to Los
Angeles.
East
West
Figure 7. WTC complex with directions of attack.
A schematic view of the directions of attack is
reported in Figure 7. Both planes have penetrated the
84
buildings and parts of their content have gone
through and ejected from the face opposite to that of
impact.
For instance the landing gear of AA Flight 1 1 has
been found several blocks south at the corner of
West Street and Rector Street, FEMA (2002). This
shows that not only the northern face of the north
tower has been damaged but, to a minor extent, also
the southern face.
Also UA Flight 175, after penetrating the building
from the south in a NE trajectory, ejects part of its
content through the northern face. For instance part
of the fuselage has been found on top of WTC 5,
Figure 8. Also the landing gear was found at 45 Park
Place and one engine at the corner of Murray Street
and Church Street. Both places are further north,
three blocks apart from WTC 5. More details and
pictures may be found in chapter 1 of the FEMA
(2002) report.
Examination of videos taken at and following the
impact seem to show that the aircraft may have
caused some damage also to the columns on the east
face, especially to those located near the north face.
The aircraft was able to penetrate most of the
building width and to propel some parts outside
through the northern façade because it invested an
area that was largely free of structural elements.
Figure 8. World Trade Center Complex. 1-North Tower; 2-
South Tower; 3-Mariott Hotel; 4-South Plaza Building; 5-North
Plaza Building; 6-U.S. Customs House; 7-WTC 7
The aircraft impacted the tower at a steep angle of
bank, Figure 9, so that several floors were affected,
from floor 78 to floor 84 according to the FEMA
(2002) report.
4.1 Damage from the impact
The two towers suffered different damage from
the impact. The north tower was hit at the level of
floors 94 to 98. Although the aircraft was not level at
the time of impact, the angle of bank was minimal.
Also the point of impact was nearly central to the
north face of the building, Figure 10. Another very
important aspect of the impact against the north
tower is that the central core is disposed with the
longer side parallel to the north and south faces and
therefore perpendicular to the direction of impact.
Therefore the aircraft, after having penetrated the
north face, arrests its run against the stiff central
core. The penetration of the peripheral wall being
easier because of the relatively low local stiffness.
This explains why only one landing gear was ejected
through the south face. From this analysis it appears
that only the external façade and the peripheral wall
ofthe inner core were damaged by the impact.

',-'' v'v.-l

Figure 10. North tower; detail of damage caused by impact.

Uiili.

Utf*

Figure 9. UA Flight 175 just before impacting the south tower.

The landing gear has probably not found

structural members on its way and has been ejected
through the south face due to its high momentum.
On the contrary the impact with the south façade
of the south tower has occurred at a high angle of
bank, Figure 9, and has consequently affected many
more floors as mentioned above. Furthermore the
impact has been eccentric and much closer to the
eastern wall. By considering that the internal core
was this time disposed with the longer side parallel
to the direction of impact, it becomes apparent that,
after penetrating the southern façade, the aircraft
found on its way mainly open office space. That is
why so many parts were ejected through the northern
façade near the northern eastern corner, part of
fuselage, landing gear and right engine.
85
Figure 11. South tower; detail of damage caused by impact.
4.2 The role of the floor system
The floor supporting system must have played an
important role in the damage mechanism following
impact. In spite of its light weight nature the floor
supporting system had a very large in-plane stiffness,
due to the grillage type of supporting beams, the
metal deck and the covering concrete. The shell type
structure of the aircraft must have behaved much as
an egg against the sharp stroke of a knife blade. The
aircrafts must therefore have been shredded in bits at
the impact with the floor systems.
5 IMPACT STRUCTURAL BEHAVIOR
Many structural engineers have tried to explain
why the towers did not collapse immediately after
the impact. The reasons that have been reported are
that the high redundancy of the structural system
together with the fact that the buildings had been
designed to withstand impact from an aircraft, albeit
much smaller, allowed a stress redistribution
satisfying global equilibrium. This explanation is
substantially correct.
5.1 Energy approach
The reason provided above, although satisfactory,
does not account for all the elements responsible for
the survival of the twin towers to the first impact.
These fall in two directions: on one side the actual
resistance of the towers and on the opposite side the
actual applied load.
To look at the problem in a better perspective it is
worthwhile to consider an energy balance. On one
hand the energy transmitted by the aircraft to the
tower, on the other the energy absorption and
dissipation capacity of the tower. The energy of the
aircraft before the impact is the sum of the potential
energy due to its weight and the kinetic energy
related to its mass and velocity. The potential energy
finally translates into weight and the weight of the
aircraft is a tiny amount as compared to the weight
of the structure above the impact level. The kinetic
energy is directly proportional to the mass of the
aircraft and to the square of its velocity.
5. /. 1 Weight ofthe aircrafts
The mass of the aircraft may be estimated from
the maximum take-off weight. Both aircrafts were of
the type shown in Figure 12.
Figure 12. American Airlines Boeing 767 - 200ER Length
54.9m, height 15.9m, wingspan 47.6m; take-off speed 310
Km/h, cruising speed 850 Km/h, landing speed 287 Km/h;
maximum take-off weight 179 tons; empty weight 86 tons,
maximum fuel capacity 73 tons, maximum payload 32 tons.
The total weight must have been somewhat
smaller that the maximum take-off weight because
86
neither plane was fully loaded and both had used a
minimal amount of the loaded fuel. Besides the
range of the aircraft was such that it could fly from
Boston to Los Angeles and back without refueling;
therefore it did not need to embark fuel to full
capacity. By considering full fuel capacity and the
actual number of passengers (81+11 for AA flight
1 1) the total weight of the aircraft may be estimated
at not more than 168 tons. If, instead, the aircraft had
embarked only the amount of fuel needed for the
'flight to Los Angeles, the total weight ofthe aircraft
may be estimated at not more than 131 tons.
In what follows a weight of 150 tons will be
considered for both aircrafts in the conviction that
this may be a safe upper bound. Therefore the mass
of the aircraft may be safely estimated at about
l50KN-s2/m.
5. 1.2Speed ofthe aircrafts
The maximum cruise speed ofthe aircraft is given
by the manufacturer in 490 mph («900 Km/h) at
35,000 feet. It is obvious that this speed cannot be
attained under low level visual flight conditions by
an inexperienced crew in the pursuit of a target.
Furthermore the flight operation manual for this type
of aircraft warns not to exceed the speed of 250 mph
(«460 Km/h) at or below 10,000 feet. Moreover
holding speeds between 200 and 220 mph are
recommended at 1,500 feet (Holger Ludwig,Eucom
Airlines & Terry Yingling).
From this information and from calculations
performed observing real time videos of the
approach of UA flight 175 to the south tower it may
be safely concluded that the speed of the aircrafts
must have been in the range 200-5-250 mph
(370-^460 Km/h)
5.1.3 The kinetic energy
The kinetic energy of the aircraft, as a function of
its mass and of its speed takes the expression
K = -mv2 (1)
2
For a mass of 150KN-S2 Im and a speed of
360Km/h it follows K = 750MN-m . It should be
observed that for increasing speeds the kinetic
energy increases as the square of the speed. For
instance, at a speed of 720Km/h the kinetic energy
is 3,000MV · m . Therefore the correct assessment of
the speed is an important parameter in the
characterization ofthe damage caused by the impact.
As explained before a speed of 360Km/h seems
much more realistic than the higher ones that have
been suggested in the literature, 590mph for United
Airlines Flight 175 and AlQmph for American
Airlines Flight 1 1, FEMA report, chapter 2.
The above considered speed is well above the
stall speed with flaps in zero position declared by the
aircraft operation manual in the range
224 + 302Km/h.
5.2 Global collapse could not occur
The kinetic energy ofthe aircraft must be used in
several ways at the impact with the tower. One part
is definitely used in plastic and rapture energy ofthe
aircraft structure. It is difficult to evaluate the
amount of kinetic energy which is lost in this way
without a complete non-linear structural model of
the aircraft and of the tower. Another part is lost in
vibration energy, or else in kinetic and elastic
deformation energy of the tower. In fact this part
goes in kinetic and/or strain energy according to the
value of the velocity. Finally another important part
goes into plastic and/or rapture energy ofthe tower.
It is easy to realize that most ofthe kinetic energy
in the aircraft is transferred either in plastic and
rapture energy in the aircraft or in plastic and rupture
energy in the tower, and that the vibration energy of
the tower is a very small amount ofthe total.
Let us evaluate the amount of strain energy
absorbed by the part of the structure designed to
withstand the wind action. A quick calculation may
be performed by assuming that the structural
strength can be varied to accommodate the variation
of the wind bending moment with a constant cross
section. This amounts to choosing the yield and the
allowable strength to be ofthe type ay{z) = k-z2 .
The origin of the reference frame is taken at the top
of the tower, so that the maximum yield strength is
ay{h) = k-h2 = 690MN Im2 .In this way the second
moment of inertia I of the cross-section resisting the
wind is given by the simple expression:
w-a 2 hJ. 2
= (2)
4·<τ,(Α)
where cra(h)=46QMN/m2 is the allowable stress,
w=2KPa the wind pressure, h = 415m the tower
height and a = 64m the length of the outer side of
the tower wall. For the sake of simplicity the part of
the cross-section resisting all the gravity actions is
neglected. With these values the strain energy
associated with the wind is 14MN · m . This is less
than 10% of the estimated kinetic energy of the
aircraft and would be from 6 to 9 times less
(4% -6%) if the cruise speed is considered as
reported in the literature. Therefore, it is clear that
some damage must occur to the structure and, in
principle, global collapse cannot be excluded.
However it may be easily seen that a reasonable
amount of plastic energy can easily account for the
kinetic energy of the aircraft. Still referring to that
part of the structure resisting the wind load, it may
be seen that this would account for all the kinetic
87
energy of the aircraft at the expense of a maximum
strain of only about 5 times the yield strain.
However, if the maximum cruise speed is
considered, a maximum strain of about 32 times the
yield strain would be required to account for all the
kinetic energy.
It is therefore evident that substantial damage is
to be expected from the impact, although this is not
large enough to cause the overall collapse of the
structure. Most ofthe damage is expected to be local
and, as has happened with the south tower and, to a
minor extent, with the north tower, some of the
kinetic energy might be retained by the parts that
after having perforated the entire building continue
their motion outside. It would be much like a bullet
that after having penetrated a body without
exploding continues until it exhausts its residual
kinetic energy.
5.3 Partial collapse
As far as equilibrium is maintained and the
stresses can be redistributed because of the
redundancy of the structural system, the partial
collapse of the upper part of the structure cannot
occur. For this to occur, a large part of the cross-
section should be severed, so that equilibrium could
no longer be satisfied by stress redistribution
because the remaining integral parts would be
yielding. Obviously, this did not occur for the north-
tower nor for the south tower. However the latter
was much closer to a state of partial collapse
because a larger part of the resisting cross-section
had been severed.
Actually the smaller time which elapsed between
the hit and the collapse ofthe south tower compared
to the north tower may be attributed to a condition
much closer to partial collapse. This was not so
much due to a larger speed, but to the angle of
impact and to the different orientation of the central
core.
6 CHEMICAL ENERGY
The authors have not found data on the amount of
fuel embarked by the two aircraft in Boston. If they
had embarked fuel to full capacity, then the
remaining fuel at the time of impact would have
been the difference between the full capacity and the
amount burned by the engines until the time of
impact. Because both planes flew for nearly one
hour before impact, 47 minutes AA Flight 1 1 and 49
minutes UA Flight 175, the amount of jet fuel
remaining at the time of impact would have been
about (90,770- 7,200) liters = 83,570 liters.
Another way of looking at the calculation would
be to consider that the distance between Boston and
Los Angeles is about 3,000 miles, which at a
cruising speed of 530 mph would take 5h and 40 The main consequence of this temperature rise,
minutes. Adding to these, 40 minutes for possible obviously not uniform, was a change in the material
emergency route diversion, would make 6h and 20 properties ofthe structural steel. Essentially the yield
minutes. The fuel loaded on the aircraft, at a strength and the elastic modulus are dramatically
consumption of 7,200 liters/hour, Jaros (2002), reduced by temperature increases. For instance the
would in this case have been about 45,000 liters. By yield strength versus temperature behavior of a
subtracting the amount burned before impact it typical structural steel is represented in Figure 13,
follows that the remaining fuel would be about taken from Outinen et al.(2001). It is important to
38,000 liters. notice the dramatic decrease of yield strength beyond
In conclusion the amount of fuel remaining in 400°C.
each plane cannot have been less than nearly 38,000 Similarly the dependence of Young's modulus on
liters, nor more than about 83,000 liters. The authors temperature is shown in Figure 14 for the same type
are inclined to believe that the first figure is by far of steel, Outinen et al.(2001). Here the decrease in
more likely because the excess of fuel carried by the Young's modulus starts at a considerable rate
aircraft would be at the expense ofthe payload. already at around 100°C.
The overall effect on the structural members and
on the structural skeleton is a considerable decrease
6. 1 Energy developed by fuel burning in strength and in stiffness.
The heating value of the jet fuel is 43,000 KJ/Kg
while its specific gravity is 0.81. Therefore, the 450
t
Itili
I 1 1 1
Steady state test results
amount of energy developed by burning 1 liter of jet 400 GL ; ; ; ?. 4..
O Transient state lest results
Φ Φ Λ Χ
fuel is 0.81Kg χ 43,000 KJ/Kg = 34,830 KJ. The 350
!
EC3: Part 1.2
extreme values of heat developed by burning the 1 300
! ! ^v i Îïït
I ^¿ t I t
residual fuel would therefore be: 2 1
1

t t\
1

liti
«>250
I \
Minimum 1,324GJ = l,324GiV · m
1 1 1 1 1 1
Γ

c¡»
1 1 1 \# 1 1 1

\ß)
Maximum 2,89107 = 2,891G/V · m fi
TS
t
t
ι
1
I
ι
1
1

'
>!
X.
1
1

'
1
1

'
t
I
'
2 190
By comparing these values with the estimated « IIII
till
\.
X
1
t
1 1
1
*"
1

100
kinetic energy of the aircraft at the time of impact, .ι I
ι
1
ι
1
ι
1
^Sjt
)1^*
1 *
ι
K = 750MN · m as a probable value and a maximum so
i
lililí
ιιιιιι
i i i i "^^Vg^X
^^^Ti-^gr.
γ-«»^
unlikely value K = 3,000MN m,it follows that the 300 400 S00 βΟΟ 7O0
heat developed by the fuel burning may be Temperatur« T. , (C]

compared, in energy terms, to the impact of a Figure 13. Yield strength as a function of temperature for
thousand aircraft. structural steel S355.

220000 I 1 p
1 " 1

6.2 Effects of heat on structural behavior 200000 ¡S^g. 4 J- Steady state test results "
<

® Transient state test results

The energy developed by the chemical reaction rr- iaoooo
ε EC3: part 1.2
due to fuel burning results in a huge amount of heat ε I6O0OO
till
till

released in the time interval required for the fuel 5

UJ
140000
& >< ¡ ! !
*
burning to complete. Besides this, additional heat is 120000
1 l\ 1 1 1

8
developed by the fires ignited by the fuel burning. φ
100000
1
1
1

1
\
\
1
1
1
1
1
1

The fire due to fuel burning and to other ~õ

3
aoooo
1
1
1
I
\l
\
1
1
1
1

3 00000
combustible materials was too big and too
3 40000
concentrated to be extinguished by the various lines ! î î Νι-^ρ !
20000
of fire defense available in the building. These are ' w ä -
ί
*
i
·
î 3-fi-í~!p Sí ?
described in some details in the FEMA report,
, 1

200 300 400 SOO SOO 700

appendix A.
The heat developed by the fire in part escaped
from the building through various openings that
became or were available, and in part contributed to
the raise in temperature in the building. The increase
in temperature may have reduced the fire protection
on the structural members, which must have been
seriously damaged in several members at impact,
and caused an increase in temperature in the
structural steel. This increase must have started in
the areas directly exposed to the fire, but migrated,
to some extent, to the entire steel skeleton according
to the Fourier law of heat flow.
Temperature Τ, , fC]
Figure 14. Young's modulus versus temperature for structural
steel S355.
7 COLLAPSE AT THE TOP
The collapse of both towers was initiated by the
fall of the part of the structure above the impact
zone. This was caused by the exhaustion of the load
carrying capacity of the structural members at the
level ofthe impact volume. As has been observed in

88
the videos the fall, or collapse, of the top parts has
occurred with different features in the two cases.
7. 1 Collapse ofthe south tower
In the case of the south tower, which collapsed
first although it was struck later, the collapse
occurred with a vertical translation accompanied by
two rotations.
Figure 15. Collapse ofthe top part ofthe south tower.
tjr
Figure 16. Twin Towers after attacks. Forefront: north façade
of north tower. Rear: north façade ofthe south tower.
As described by the FEMA report, chapter 2, and
shown in Figure 15 derived from the same report,
the rotation occurred first towards the east, around a
89
north-south axis, and then towards the south around
an east-west axis.
West East
West East
Figure 17. Schematic rotation of the south tower around a
north-south axis.
North South
Figure 18. Vertical translation ofthe south tower accompanied
by a rotation around an east-west axis.
This collapse mode can be explained quite
simply. The aircraft penetrated the tower through the
south façade and a substantial part of it exited
through the north façade. This implies that
substantial damage occurred in the north wall as well
as in the south one. This damage is clearly visible in
Figure 16, reproduced from the FEMA report. Apart
from undefined damage to the central core, the
substantial damage to the south wall and to a smaller
corresponding part of the north wall, close to the
north-east corner of the building, Figure 17, caused
the removal of the support for the floor supporting
systems of the upper stories along those walls. As a
consequence, these started to function more as a
beam like system than as a plate like one, therefore
 overloading the eastern wall as well as some
elements ofthe central core. South
When the temperature increase due to the fire
lowered the yield strength of the structural steel to "V"A A » A A A k"V~
the level of stress present in the structural members,
the eastern wall began to yield, causing an eastward Τ* Τ '' " " Τ " "Y*"
rotation along a north-south axis, Figure 17. This
motion also required plastic flow in the central core
and plastic rotations in the western wall.
This was immediately followed by additional
stress redistribution causing the failure of the
remaining part of the south wall and a drop of the
top part of the building along that wall to gain
support. This explains the southward rotation of the North
top part of the building along an east-west axis
clearly visible in Figure 15. It is worth noticing that
Figure 19. Nearly uniform load distribution in north tower after
while the eastward rotation, according to this impact.
interpretation, occurred smoothly, the southward
rotation was caused by a sudden drop, through
several damaged stories, along the southern wall and
resulted in a dynamic impact against the lower part
of the tower.
W '

7.2 Collapse ofthe north tower

The aircraft penetrated the building through the
north wall and probably arrested its motion against
νΤΎ~<
the central core that must have also been damaged to Ν syrn
some extent. Only one piece of landing gear went
trough the opposite wall, and given its limited size it
might have moved across openings without South North
producing significant structural damage.
Generally speaking, damage was restricted to a
limited volume, in this case, and the stress
redistribution resulted in a more uniform stress
distribution than in the case described previously. Figure 20a. North tower: schematic representation of symmetric
This situation is schematically depicted in Figure 19. buckling.
With the increase in temperature, the yield stress at
the level of impact gradually lowered almost
everywhere to the nearly uniform value of the stress
distribution. At this point there must still have been
equilibrium between internal stresses and external
load, essentially gravity load. However, if a positive
temperature gradient existed, this may have driven
the yield stress in the structural steel even lower,
¿TU
thus rendering static equilibrium impossible. Ν asym
Of course the possibility could exist, as well, that
the collapse ofthe upper part was due to buckling of South North
the structural members in the surrounding of the
damaged volume. However there is one big
objection against this hypothesis. The upper part of
the north tower was seen moving vertically in an
accelerated motion. No combined vertical-horizontal
motion was seen at the time of collapse, nor by Figure 20b. North tower: schematic representation of skew-
observing many times the available videos. This symmetric buckling.
would point to a kind of symmetrical buckling as However it is well known that an unsymmetrical
that shown in Figure 20 a. or sway buckling ofthe type depicted in Figure 20 b,
would occur under a much smaller load. Until this

90
circumstance is not clearly explained it can
reasonably be assumed that collapse was originated ù{t) = a0 (8)
by plastic yielding.

7.2.1 A mathematical model

u(í) = aoj (9)
The condition of dynamic equilibrium is
schematically shown in Figure 21 for the top part of After denoting with tc the collision time of the
the north tower. upper part with the lower part and with A the
distance to be covered (drop height) it follows that:
A mii{t) 1

6h_ >3
t = (10)
\aoJ
W

'6/Λ'
,WJM uc=~a0 (11)
Y u(t) \aoJ
These results lead to the following conclusions:
- collapse of the upper part occurs with an
accelerated motion;
- the acceleration gradient depends on the
temperature gradient.
Video pictures seem to confirm the first conclusion.
Figure 21. Equilibrium ofthe top part ofthe north tower.

In mathematical terms this may be expressed as

7.2.1.1Numerical results
The evaluation ofthe time of collision tc and the
follows:
velocity of impact uc requires the knowledge of the
ay(e(t))-A + m'u(t) = W = mg (3) height h and the parameter a0 . The latter depends in
turn on the temperature and on the stress gradients at
where θ(ί) is the temperature at time t. For the time of incipient collapse. These two parameters
are very difficult to quantify. Here a different
t = 0, θ(θ) = θ0 and σ,(θ)· A = W = mg strategy is adopted. On the basis of the time of
A series expansion of ay(ø(t)) and of 9{t) allows collision tc , which may be estimated from videos,
for the following to be written, after retaining only and for different but feasible heights h the velocity
terms up to the first order: of impact uc is evaluated.
A rough estimate ofthe time of collision provides
values in the range 2s<tc< 3s while the height
-t (4) h may be expected in the range 12m < h < 18m . As
Οθ Je=e
may be seen from Table 1, the corresponding range
Substitution into the equation of motion gives: of velocities of impact is 12m is < uc < 21ml s . This
result will be particularly useful when considering
(da. Λ
the collapse ofthe lower part ofthe tower.
·θ0 t + m-ü(t)=0 (5) The reliability of the results reported in Table 1
Κοθ; *««, may be checked against the times of collision and
velocities of impact for a free fall from the same
By setting
height. These are given by the well known
A_ 'δσ,Λ expressions:
θ=-α (6)
m οθ 'Ο-θ.
i = (12)
the equation of motion (5) may be written as:
ü(t) = a0-t (7) ùc=gtc=J2gh (13)
A double integration and application of The corresponding numerical values are shown in
homogeneous (rest) initial conditions yields:
Table 2.

91
Table 1. Velocity of impact ùc\ml s) in tenns of eight A and
time of collision tc .
h(m)
tis) 12 15
2.0 18 23
2.5 14 18
3.0 12 15
Table 2. Time of collision tc and velocity of impact uc in free
fall conditions.
him) 12 15
«.W 1.56 1.75 1.92
uc(mls) 15 17
As expected the times of collision for the given
heights are much smaller. However velocities of
impact larger than those calculated in free fall
conditions are not admissible. Therefore only the
results referring to the last row of Table 1 will be
accepted.
7.2.2 Another model
The collapse ofthe top part ofthe tower may also
be due to buckling or yielding of the central core in
the surroundings of the damaged volume.
Symmetrical buckling may have occurred because
any lateral motion would be prevented by the
exterior walls. The buckling stress is extremely
difficult to evaluate theoretically because of the
many parameters affecting such an evaluation. End
conditions, second moment of area, Young's
modulus are all little known, with the latter
depending also on temperature.
The actual buckling stress, however, would be
known to a good degree of approximation. In fact,
according to the FEMA report, Appendix A, the
central core was built of A3 6 structural steel with
very similar properties to those of S3 5 5 steel
previously described. Because the core provides only
a marginal contribution to the wind resistance of the
building, a working stress in the central core
columns of about two thirds of the yield stress may
be expected. The central core started to buckle or to
yield when the buckling or yield stresses was
lowered by the rise in temperature to the level of the
working stress. It is also likely that any buckling
must have been of the elastic-plastic type, with
buckling stress not very much different from the
yield stress.
This is only a working, albeit quite reasonable,
hypothesis in view of the impossibility of any more
reliable estimate ofthe buckling stress. Therefore the
yield stress may be considered in the range of
200MNIm2 <σ Λθ)< 250MNIm2
From the graph of Figure 13 the corresponding
temperature range is given by
500°C<0<600°C
18
27 The corresponding range for the Young's
22 modulus may be derived from Figure 14, providing
18 eOGNIm2 <E<120GNIm2
This model is physically justified by the
following:
- the central core is more stressed than the outer
18 walls in the absence of wind;
- the fire may have increased the temperature more
19
in the columns ofthe central core than in those of
the outer walls because the heat would flow
outside through openings. Moreover the external
air would cool the exterior columns;
- the stronger and stiffer exterior columns would
prevent any swaying motion ofthe core.
- As a consequence the central core would pull the
exterior columns inwardly as buckling or yielding
progressed.
The theoretical developments at section 7.2.1 still
apply in this case.
8 COLLAPSE OF THE LOWER PART
In this section an explanation will be provided of
why the towers imploded over themselves without
developing a potentially more lethal lateral collapse
mechanism.
The main point that must be understood is that the
collapse ofthe top part resulted in a kind of hammer
blow for the lower part. The wave propagation
theory used, for instance, to describe the stress
propagation in pile driving, Clough & Penzien
(1975), may be used to explain why at a given time
only one part of the tower is affected by the blow
while the remaining part, separated from the other by
the wave front, is in a condition of static
equilibrium. Therefore damage proceeds with the
velocity ofthe stress wave propagation.
8.1 Basic principles
The following basic principles will be used in
deriving the waveform and the velocity of wave
propagation that will provide the main results of this
analysis.
First Newton's second law of motion resulting in
the equation of motion of an axially vibrating beam
will be used.
Next D'Alembert solution for this equation will
be used to establish the waveform and the velocity of
wave propagation.
92
 Finally the principle of conservation of linear
momentum will be used to establish initial u(z,t) = f t (19)
conditions and stress intensity. ρJ

8.2 Equation of motion

8.4 Initial conditions
Direct equilibrium of a beam segment, like the The lower part of the tower is at rest before the
one shown in Figure 22, provides the following
upper part collapses. The displacement may be
equation.
written as follows:
u(z,t)=0 ;Vi<0;Vz>0 (20)

îΝ This implies
f(x) = 0;\/x<0 (21)

u(t) Τ î mu (O The initial condition in terms of velocity may be

written as
u(z,t) = 0 ;Vi<0;Vz>0 (22)
dN A
1 N + õz dz
ΛΓ

which implies
'(*)= 0 ; Vx<0 (23)
After the impact, which is here assumed to occur
Figure 22. Equilibrium of a beam segment. at time t = 0 , the velocity may be derived from the
conservation of the linear momentum. If the mass of
the top part of the structure is denoted by Μ , then it
- Ν - müdz + Ν + dz = 0 (14) may be written:
õz
Simplifying and using the constitutive law Mii(0,0+ )+ Jp4«(z,0+ )dz = Mùc

N = EA = EAu' (15)
Ôz or

yields the equation of motion: Λ

EAu" -mii = 0 (16) Mu(0,0+)+ ¡pAf' 0+- dz = Mu, (24)

PJ

From equation (21) it follows that

8.3 D 'Alembert solution
The solution to the equation of motion may be
conveniently given in the form provided by
f' 0+-Í- =0 (25)
PJ
D'Alembert, Achenbach (1973):
û(0,0+)=ù (26)
u{z,t) = f t +g t + - (17)
V ρJ ρ J
The two functions /(x)and g(x) represent two 8. 5 Waveforms and wave velocity
waves propagating with velocity vp respectively The initial condition
along the positive and negative direction of the ζ
axis. That is, f(x) is a forward propagating wave,
while g(x) represents a backward propagating wave. predicts a sudden jump in velocity at the tip of the
For the problem being considered no backward lower part of the tower that from a rest condition
propagating wave can exist, giving: takes the velocity of the upper part at the collision
time. The general expression for the velocity may
ζ
g t+ =0 (18) therefore be written as follows:
V vpj ( Λ f Λ
ζ ζ
u{z,t)=f' t- = ùrH Γ (27)
V vpJ Ι νρ)
93
where H(x) is the Heaviside step function. This f
shows that the velocity uc propagates with velocity N(z,t) = -^-ùcH t (32)
vp downwards along the tower. v, VPJ
From the expression for the velocity, the
corresponding expressions for the displacement and (
for the acceleration may be easily derived σ(ζ,ί) = ùcH t (33)
v
respectively by integration and differentiation, V VPJ
obtaining: The graphical representation of the waveforms is
f \ provided in Figures 23 and 24. Each physical
z quantity propagates downward with velocity v .
u(z,t)= f(z,t) = ùl t-- H (28)
This may be derived from the equation of motion by
p J K VPJ
setting
f λ f
z z ζ
ü{z,t) = fi" t = ÙS t (29) u(z,t)=f Γ (34)
v. vpJ \ "pj
Next the axial force and the axial stress may
y be obtaining
calculated by the following formulae:
¡EA
(35)
N{z,t) = EAu'(z,t) (30)
''"VT
a(z,t) = Eu'(z,t) (31)
which yield 8.6 Numerical results
Some numerical results in terms of velocity of
wave propagation and overstress due to impact will
w
be presented in the following. Due to lack of
knowledge of specific mechanical parameters, only
V orders of magnitude will be established. For instance

uS t
Ì the specific weight of the tower (weight per unit
length) and the corresponding areas ofthe structural
cross-section will be considered according to the
v V values reported in Table 3.

7ZZ Table 3. Specific weight and cross-section area.

Specific w<îight Cross-section areas
ü{z,t) ù(z,t) u(z,i) MN/m 2
m
"max "min
Figure 23. Waveforms: acceleration, velocity, displacement. 6 10.00 7.50
7 12.00 9.00
8 14.00 10.50
u « 12 21.00 15.75
-EA -E
The two columns concerning the cross-section
refer to the maximum and minimum cross-section
area which is considered compatible with the given
weight. *

Table 4. Wave propagation velocities vernis).

Specific E
Weight (GN/m2)
(MN/m) 60 120 200
v////////.
"max "min -"max -"min "-max "nun
N(z,t) σ(ζ,ή 6 990 858 1400 1213 1808 1566
Figure 24. Waveforms: axial force and axial stress. 7 1004 870 1420 1230 1834 1588
8 1015 879 1435 1243 1853 1604
12 1015 879 1435 1243 1853 1604

94
 Another parameter that must be considered in values, the expected range of collapse propagation
order to evaluate the wave propagation velocity and velocity would be 29ml s <vc < 36m I s . These
the corresponding overstress is Young's modulus. values are much smaller than the free fall velocity
Three values will be considered: E = 200GNIm2, from the top, that for h = 350m would be
E = 120GNIm2, E = 60GNIm2. The first value Vff = 83m/s. Videos clearly show that the debris
refers to the material unaffected by the temperature falls at a much higher velocity than the velocity of
rise, the next two values correspond to temperature collapse propagation.
values of 500°C and 600°C respectively. The velocities reported in Table 4 are of two
The values of the corresponding wave order of magnitude larger than the observed collapse
propagation velocities are given in Table 4. velocity. This is not contradictory because the elastic
The overstress depends on the impact velocities. theory cannot predict collapse.
Three feasible values, derived from Table 1, are However, the overstresses given in Tables 5, 6
considered, namely wc=12m/.s, uc =15mls, and 7 are much greater than the largest yield stress
ùc = 18m I s . The corresponding values of the for the structural steels used in the construction of
overstress are reported in the following tables. the tower. These overstresses cannot be developed
because the material yields.
The main result that can be gathered from this
Table 5. Overstress a0(MN/m2); u = 12m I s . analysis is that the tower could not withstand the
Specific E impact caused by the collision ofthe upper part with
Weight (GN/m2) the lower part. Also overall buckling could not occur
(MN/m) 60 120 200 because at any given time only a tiny part of the
-"max -"min "max "min "max ^min tower was subject to an extra axial force which was
6 727 839 1029 1187 1327 1533 sufficient to make it yield. Therefore the buckling
7 717 828 1014 1171 1309 1511 length was always negligibly small.
8 709 819 1003 1158 1295 1496
12 709 819 1003 1158 1295 1496
8. 7 Elastic-plastic analysis
To make the analysis consistent and to limit
Table 6. Overstress o0(MN/m2); uc = 15ml s .
complexities to a bare minimum, the material will be
Specific E considered elastic-perfectly plastic. The equation of
Weight (GN/m2) dynamic equilibrium
(MN/m) 60 120 200
A "min ON
-"max "min "max "min -"max
mii = 0 (36)
6 909 1049 1286 1484 1659 1916 õz
7 896 1034 1268 1463 1636 1889
8 887 1024 1254 1448 1619 1870 must be supplemented by the yield condition
12 887 1024 1254 1448 1619 1870
N = EAsy = ayA (37)

and by the compatibility relation

Table 7. Overstress σ(ΜΝ/ιη2); ùc=18m/s .
Specific E (38)
ε =
Weight (GN/m2) dz
(MN/m) 60 120 200
"max "min -"max "min "max A ·
"mm

6 1091 1259 1543 1781 1991 2299 a = (39)

7 1076 1241 1521 1756 1963 2267
8 1064 1229 1505 1738 1943 2244
12 1064 1229 1505 1738 1943 2244 By using the same waveforms as in the elastic
case, that is
The results reported in Table 4 should be
ζ
compared with the observed collapse velocity. u(z,t) = f t (40a)
Observation ofthe videos concerning the collapse V vpj
ofthe north tower provides two elements particularly
useful for the present analysis. First, collapse
propagates at a constant velocity as predicted by the u(z,t)=r t (40b)
wave propagation theory. Second, the total ρJ
propagation time may be estimated in about 10-12
seconds.
Also the height of the lower part may be
estimated at about 350-360 meters. From these
95
 f \
ζ õz
ü{z,t) = f" t (40c) (43)
V VPJ
Also
the wave propagation velocity becomes
du õu õz
a = £--V (44)
VP = (41) õt õz Õt "
1 +a
At collapse initial conditions may be set as
In the case of elastic behavior it is ε = 0 and
a = oo . For rigid-plastic behavior it is «ç =0 and u = u (45)
a = 0. y
It can immediately be recognized that for a = oo ε = ε. (46)
the elastic wave propagation velocity is regained,
while for a = 0 the wave propagation velocity drops It follows that the velocity of collapse
to zero. propagation may be calculated through the formula
For the case being considered, considering an
«
A3 6 steel which was used for the construction ofthe (47)
vp=-
central core and a Young's modulus
E = 200GNIm2 corresponding to ambient
temperature, it follows that <? sl.8-10"3; Also the velocity of collision may be calculated
s. = £r = 30% and a = 6 · 10"3 . Also by the equation

a
ν,=ετ·
c r
vρ (48)
= 0.077 (42)
cc +1 By considering the three feasible values of impact
velocities ùc=12ml s , ùc=15ml s , ùc=l%mls
For lower values of Young's modulus , that is at
and two extreme values for eri namely
higher temperatures, the ultimate or rapture strain
ε, = 24% and ε, = 36% , the values shown in Table
becomes much larger and the parameter a
are found for the collapse velocity.
comparatively smaller. For lack of information on
the actual experimental values, the upper bound
provided above (0.077) will be used also for Table 9. Expected values of collapse velocity ν ¿mis).
E = 12QGNIm2 and for E = 60GNIm2. In this
way, the wave propagation velocities shown in Table ε,(%) îic(m/j·)
4 take the values reported in Table 8. 12 15 18
24 50 62 75
30 40 50 60
Table 8. Wave propagation velocities vp(m/s). 36 33 42 50
Specific E
Weight (GN/m2) From the values shown in the table, the most
(MN/m) 60 120 200 probable one seems to be ν = 33m I s
"max A
"mm
.
-"max A
"min
.
"max "min corresponding to uc = 12m I s and^r =36%.
6 76 66.3 108 93.7 140 121
7 77.5 67.2 109.7 95 142 123
8 78.4 67.9 110.8 96 143 124 9 FINAL REMARKS
12 78.4 67.9 110.8 96 143 124
A qualitative analysis of the collapse behavior of
The wave propagation velocities are still higher the Twin Towers of the World Trade Center has
than the estimated collapse velocity, but the order of been presented.
magnitude is now correct. It should also be kept in In particular it has been shown that the buildings
mind that the values pertaining to E = 120GNIm2 could have not withstood the impact of the aircraft
and E = 60GNIm2 are only upper bounds to the without suffering substantial structural damage.
actual velocities. However, as shown by history, this damage was
mostly local, that is confined to the volume of
impact and its immediate neighbourhood.
8.8 A simpler model
The kinetic energy ofthe aircraft before impact is
During an infinitesimal time interval, dt, a a key factor determining the amount of damage that
disturbance propagating at a velocity, vp , describes a the building could suffer. However this energy
spatial interval, dz . Therefore cannot be determined on the basis of the cruising
speed of the aircraft; the velocity just before impact

96
must have been much lower for several reasons.
Furthermore this energy was not totally transformed
into damage of the building. A considerable part of
this energy went into fracture and plastic energy of
the aircraft. Another part, albeit smaller, was
transformed into vibration energy for the building
and the large number of viscous dampers must have
been of paramount importance in its dissipation.
The towers would probably not have collapsed
without the intervention of exceptional events such
as earthquakes or hurricanes if the aircraft fuel had
not caught fire. Even in these cases, probably only
the top part would have toppled with a different
collapse mechanism from the one exhibited on
September 1 lnth.
The energy developed by fuel burning was
enormous and probably equivalent to about one
thousand aircraft impacts. Most ofthe heat may have
escaped through openings, but a substantial part has
given a temperature rise that disabled locally the fire
proofing system and caused a substantial
temperature increase in the steel skeleton. This
action was sustained by the permanent fires fed by
combustible material present in the buildings.
The temperature raise in the steel skeleton caused
the lowering of its stiffness and strength properties.
The reduction in strength was substantially
responsible for the collapse of the top part of the
building. The collapse mechanism was different for
the two towers and caused by the type of damage
directly caused by impact. The two different
mechanisms have been explained in the text on the
basis ofthe sustained damage.
The collision of the top part of the structure with
the lower part acted much as a hammer blow which
started a stress wave propagating in the lower part.
The overstress generated by such wave was large
enough to cause material failure. The collapse
propagation velocity cannot be explained by an
elastic wave; the overstress would be much larger
than the yield stress of the material and wave
velocity would be too large. An elastic-plastic theory
allows at least for a good estimate of the order of
magnitude of the velocity of the wave propagation.
The overstress was, of course, of such a magnitude
as to satisfy yield conditions.
Finally a very simple model is derived which
relates in a simple way the velocity of collapse
propagation, the impact velocity and the failure
strain. The observed quantities seem to be very well
correlated with the model.
The most important aspect of the models
presented is that they clearly explain why the towers
imploded on themselves and did not show any kind
of buckling, either local or global. The wave
propagation mechanism shows how failure occurred
straightaway with the overstress propagation, in such
a way that, at a given instant, there was one part of
structure that was not overstresser! the one beyond
97
the wave front, and another part that was entirely
collapsed. In this way the buckling length is always
negligibly small and buckling cannot occur.
It is possible that in reality the collapse
mechanism may not have adhered completely to the
models described inasmuch as failure may have been
localized at weaker elements. However it is
important to note that this does not undermine the
substance of the analysis and even a solid beam
under impact would have shown the same type of
collapse.
REFERENCES
Achenbach, J. D. 1973. Wave Propagation in Elastic Solids.
North-Holland Publishing Company.
Bazant, Z.P. & Zhou, Y. 2002. Why Did the World Trade
Center Collapse? Simple Analysis. J. Eng. Mech. 128(1),
2-6.
Bazant, Z. P. & Zhou, Y. 2002. "Addendum to ' Why did the
World Trade Center collapse? Simple analysis'".". J.
Eng. Mech. 128(1), 369-370.
Clifton, G.C. 2001. Collapse of the World Trade Center
Towers. HERA, New Zealand.
http://www.tenlinks.corn/ABOUT/PEOPLE/AUTHORS/clif
ton.htm
Clough, R. W. & Penzien, J. 1975. Dynamics of Structures.
McGraw-Hill, Inc.
Federal Emergency Management Agency (FEMA). 2002.
World Trade Center Building Performance Study.
Great Buildings on Line. World Trade Center.
htfo://www,CTeatbuildmgs.corn/buildmgs/World Trade Ce
nter Images.html
Holger Ludwig, Eucom Airlines & Terry Yingling. Boeing
B767-200 Aircraft Operation Manual.
http://www.eucomairlines.de
Jaros, K. 2002. Boeing 767 Specifications.
http:/Æome.primusnetz.de/Tgaros/aviation/lib/767.hrm
Minora Yamasaki Associates, Architects. World Trade Center
Fact Sheet .
ht(p://vv^\w.m-vamasakixorri/prois/wtcfs.html
Outinen, J., Kaitila, O. & Mäkeläinen, P. 2001. High-
Temperature Testing of Structural Steel and Modelling of
Structures at Fire Temperatures. Research Report Helsinki
University of Technology, Laboratory of Steel Structures.
TKK-TER-23.
Robertson, L.E. 2002. Reflections on the World Trade Center.
The Bridge 32(1):5-10.
Tamaro, GJ. 2002. World Trade Center "Bathtube": From
Genesis to Armageddon. The Bridge 32(1): 1 1-17.
Recent development in ductility design of steel structures

V. Gioncu
Department of Architecture, University 'Politehnica ' Timisoara, Romania

ABSTRACT: The paper presents the development in design methodology of steel structures under strong
earthquakes, where the ductility design plays a leader role. The factors influencing this ductility are included
in design: effects of strain-rate for near-field earthquakes and accumulation of plastic rotations for far-field
earthquakes, ultimate limit state due to plastic deformations or to ductile or brittle fracture.

1 INTRODUCTION zones, brittle fractures were relevant. In other cases,

The myth of steel as perfect material for seismic
areas has seriously compromised by the seismic
decade 1985-1995. During the Michoacan (1985),
Northridge (1994) and Kobe (1995) earthquakes the
performances of some steel structures was very bad,
clearly showing that there are some lacks in our
current practice. Because in all cases the damage
occurred also when the design has been performed
in perfect accordance with code provisions, it results
that something was happened, which was not
foreseen in the design practice (Mazzolani, 1995).
A common approach in current seismic design
practice is to characterize the earthquake effects by
smoothed design spectra. Elastic design spectra are
obtained by amplifying peak ground motions with an
amplification factor. The inelastic design spectra are
obtained by reducing the elastic spectra with the
behaviour factor. This methodology provides good
results for the great majority of earthquakes, which
can be framed in the category of ordinary events
(moderate distance from the epicenter, normal soil
conditions, etc.). However, as more earthquake
records were obtained, especially in the last period,
became very clearly that the use of this simple
methodology is inadequate in some cases, the
ground motions having characteristics dramatically
different from the ones considered in codes.
Seismic-resistant steel structures are designed
relying on their ability to sustain high plastic
deformations. But in some special conditions as high
ground velocities, low temperatures, inadequate
welding, instead of plastic rotations of pre-destined
the number of important pulses is very high
producing ductile fracture due to accumulation of
plastic deformations. In these circumstances,
questions have been raised if the actual design
philosophy is effective in some special conditions as
produced by exceptional earthquakes.
Examining this situation and the current code
provisions some main conclusions result:
- provisions refer only to ordinary earthquakes,
without considering in a proper way the cases of
some exceptional earthquakes;
-ground motion modeling do not considers the
exceptional earthquakes as the ones in near-field
sites or produced by very bad soil conditions;
-design measures concerning the structure
protection against exceptional earthquakes are
insufficiently.
Therefore, it is acutely required to have practical
methods to design structures against exceptional
earthquakes. The paper present a comprehensive
methodology based on the required-available
formulation for ductility design, as a way to protect
structure against these exceptional earthquakes.
Only the ductility of structural members is
presented, considering that the joint over-strength is
sufficiently to impede the participation of
connections at energy dissipation.
2 SOURCE CHARACTERISTICS
Due to the many factors influencing the ground
motions each earthquake possesses unique
characteristics. Strong ground motion is

99
characterized by many physical factors associated
with magnitude, faulting process, source-site
geometry, wave type, wave propagation in
heterogeneous soil and geological structure, spatial
variation of waves due to local site conditions, etc.
Current ground models based on recorded strong
motion data predict ground motion parameters using
a simplified model in which the effects ofthe source
are represented by the magnitude. The effects of
wave propagation from source to the site are
specified by distance and the site conditions by site
soil category (Somerville, 2000). The analysis of
recorded data indicates that the ground motions
vary, in many cases, significantly from the predicted
values. This observation indicates that the empirical
ground models have a large degree of uncertainty,
because other conditions that are considered in
modeling are not treated as parameters of these
simple models. These conditions include source
effects in near-field sites, propagation paths and site
soil effects for far-field sites.
In recent years the Engineering Seismology has
scored remarkable achievements due to the heavy
implementation to record seismic data of high
earthquake risk areas and the improvement of
along the fault direction, and last for considerable
time. These earthquakes are characterized by large
magnitude, relative long duration, high peak values
of ground velocities and displacements and long
period of vibration. In near-field sites, these
earthquakes are very devastating due to high
velocities, impeding the formation of plastic hinges
and, therefore, producing brittle fracture of structural
members. In the near-field sites, the vertical
components are also very important. The ground
motions have long duration pulses. The Northridge
earthquake is an example of this earthquake type.
Alluvial deposit
Interplate earthquake Intraplate earthquake
due tu lhcliun due to bending,
.stresses tension-compression
or shearing stresses
(a)

geological prospecting methodologies. The results of Epicenter Afl'ecled areas

these research works give the possibility to improve / /

the design methodology with some important data %sTotk J70W N\
concerning the main factors which are not included ' Shallow crustal

in current design provisions: source mechanism, \ ^¿/ / Normal crustal

\ 300 kni
deep of source and soil conditions. So, three
/ Intermediate

Λτ/-
i
different earthquake types must be considered in
structural design (Fig. 1).
-Subduction intra-plate shallow crustal earthquakes. Profound
(b) Hypocenter
In this case the fault is situated in a surface tectonic
plate and the earthquake is produced by plate
Fig.l Earthquake types
bending and/or shearing. Generally, the maximum
credible magnitude produced by these faults is no
-Subduction inter-plate deep earthquakes. In this
very high, the frequency of occurrence is small, the
case the two plates involved in the ground motion
natural period of vibration is short and area of
are deep. The area involved is very large and the
influence is reduced. The rapture length is short and
influence of soil site condition is very important.
the ground motion have a pulse-like characteristic.
The number of pulses is large, the natural periods
But, in special soil conditions, these earthquakes can
are long and the spectra are based on effective peak
be very devastating if the affected area is an
of ground motions. The magnitude is not so large as
urbanized and economic developed zone. The great
in the case of shallow source, but, due to the long
danger of this earthquake type is its hidden
duration of ground motion, the destractiveness can
characteristic: they can occur in area where no
by very high for both near-field and far-field
previous or very rare events have been recorded and
earthquakes. The Mexico City earthquake is an
where the current knowledge would suggest the
example of like this earthquake type.
existence of a quiescent area. The Kobe earthquake
All these examples refer to earthquakes produced
is an example of this earthquake type.
out of the Europe and the crucial question is if the
-Subduction inter-plate shallow earthquakes. The
similar events can occur in Europe. Examining the
earthquake occurs by sudden movement of two
Mediterranean area, the most active zone in Europe,
adjacent tectonic shallow plates at their boundaries.
one can see that the majority of earthquakes belong
They produce slipping on very large area, especially
to the intra-plate types. There are only reduced zones
100
where inter-plate earthquakes occur, caused by
collision between African and Eurasian geologic
plates. The shallow inter-plate faults are situated in
Turkey, Greece, Adriatic coast and Gibraltar, while
deep faults are the ones from Romania, South Italy,
Greece and Gibraltar (Marceliini, 1998).
Unfortunately, we have no relevant information
about the behaviour of steel structures in these
areas, the number of erected structures being
insignificant. In these conditions, the European code
periods the number is 8 to 12, while for short periods
this number is 10 to 20 (Yang and Popov, 1997).
There also some other important characteristics as
pulse type of velocity and displacement, vertical
components, accidental torsions, etc.
1 Accel (a
Held of steel ,

provisions must be elaborated on the knowledge «tniniiir«*; !

Fault normal
learned from the USA and Japanese earthquakes, Average
.UBC Near-field
considering the peculiarities of European steel
structures. UBC Far-field

Kuult parallel

3 ASSESSMENT OF EXCEPTIONAL
EARTHQUAKES

During the last decades large sets of strong motion

recordings were obtained from numerous
earthquakes, with characteristics beyond the ones
considered in current codes. It is matter of
earthquakes in near-field sites or in sites with very
bad soil conditions. In case of these ground motions
even for moderate magnitude earthquakes, the
destractiveness can be very severe due to some 3 Time (sec)

characteristics not considered in code provisions. Number of yield reversal

25·
-Near-field earthquakes refer to the ground motions Árlela
situated in an area within distance of source depth. Newhall
Sylmar
This earthquake type is very significantly for Santa Monica
shallow sources. The main characteristics refer to
influence of'forward rupture directivity. The normal
fault components are more severe than the fault
parallel components (Fig. 2a). The amplitude of *·%...·· Period (sec)
ground accelerations containing forward directivity (c)
effects are 50% larger than those for average
directivity. New provisions of UBC 97 for near-
fault motions appear to provide an adequate value Fig. 2 Characteristics of near-field earthquakes
for periods smaller than 0.8 sec, but significantly
lower for longer periods (Somerville et al, 1996),
very important range for steel structures. The high -Far-field earthquakes refer to ground motions
velocity influence is also very important, values of situated far from epicenter. For this earthquakes the
100-200cm/sec were recorded at soil level, but the main characteristics is influenced by source-soil
velocity spectrum shows amplification until resonance, as the effect of soil conditions,
400cm/sec (Fig.2b). This high velocity induces high increasing of duration, and period (Fig. 3a) in
strain-rate in structural members, impeding the comparison with near-field earthquakes, and
development of plastic rotations, especially in case increasing of number of important pulses to 15 to 20
of low temperatures, when the structure is not cycles (Fig. 3b).
protected. The influence of yield reversal number is
presented in Fig. 2c, for some Northridge buildings,
in function ofthe structural periods. For long periods
3 to 6 yield reversals are produced, for medium

101
 Mazzolani, 2002). But there are significant barriers
Acceleration (g) KC 1940 El Centro to implementation of this new methodology in the
PA 1966Piitíield
MI 1973 Mh/nguken-Ola design practice (Hamburger, 1996). Design
CH 1985 Chile
|MX 1985Μβο«Π procedures conforming to current code requirements
LP I9H9 I.« Prieta
NO 1994 Northridge employ simple analytical method based on a single
KO 1995 Kobe
demand level. Even the procedure of EC 8 and UBC
97 for two level demands consider the serviceability
MX ormitene« ol soll
ooncbuons) limit state in a very primitive manner, by
diminishing the ultimate limit state response with a
reduction coefficient of ultimate limit state values.
In order to provide buildings capable to reaching
Acceleration (g)
these multi-level limit states it is necessary to
Far-field earthquake
evaluate the structure performances using a multiple
analyses. As a result, engineering offices are
unlikely to provide this methodology. So, a crucial
problem for structural research is to perform a
simplest possible methodology and to try to
0 1 2 4 5 6 7 « 9 10
introduce this one in the code specification. In this
Acceleration (g) Nwir-fleM eartt»qtiake context, the RSD (Rigidity, Strength, Ductility)
0.40

method is proposed by Gioncu and Mazzolani

0.20
0 ffhflatiArWr+rÊ·***-
-0.20
-0.40
Ü 12 3 4567
(b)
Fig.3 Characteristics of far-field earthquakes
4 COMPREHENSIVE DESIGN METHOD
The main goal of current seismic design is to
protect life and structure collapse for very rare
severe earthquakes. But the seismic events have
shown that the buildings are also capable to
withstanding more frequent less severe events with
damage asking the interruption of functionality of
many buildings, evacuation of people, losses in work
places for varying period of time, monetary losses,
etc. In some cases these situations give rise to
damage more important that the structure collapse.
So, the concept of damage control for multi-level
state limits must be adopted as a basic design
philosophy.
-Multi-level design approaches. The subject of
multi-level state limits was in the interest of many
reports, presented in international conferences and
papers published in journals. In function of taking
different limit states for structures and non-structural
elements some multi-level states have been
proposed: four, three or two levels (Gioncu and
(2002), based on the verification of seismic design
through required-available formulation for three
limit states:
-serviceability limit state (SLS), when the damage of
8 9 ~ 10
Time (sec)
non-structural elements for frequent earthquakes is
controlled through the verification of structure
rigidity:
R: required rigidity < available rigidity
-damageability limit state (DLS), when the damage
of structural elements for occasional earthquakes is
controlled through the verification of strength:
S: required strength < available strength
-ultimate limit state (ULS), when the collapse of
structure for rare earthquakes is controlled through
the verification of ductility:
D: required ductility < available ductility
The required demands are determined at the level
of overall structure, in function of earthquake
characteristics. The available demands are
determined at the local level in function of
mechanical and geometrical characteristics of
members. The rigidity and strength criteria can be
performed without difficulties, being well covered
by code provisions. Only for ductility criterion there
are no coherent methodology, the provisions
referring only for constructional rales.
-Ductility design criterion. The recent earthquakes
have shown that structures fail to perform their
intended under exceptional event exceeding the
design provisions and consequently, they may suffer
local damage, by loss of strength in a single
102
member, or in some portions of the structure. The -correction of obtained values considering the
designers are now fully aware of the importance of dynamic effects of seismic loads.
ductility, as the ability of a structure to undergo
plastic deformations without any significant
reduction of strength, represent the alone measure of
the suitable structure behaviour during the
5 REQUIRED DUCTILITY
exceptional earthquakes. Unfortunately, the design
specifications contain only very vague provisions
concerning the ductility checking."... When plastic Static required ductility. The base shear force for
global analysis is used the member shall be capable ultimate limit state results from:
of forming plastic hinges with sufficient rotation
capacity to enable the redistribution of bending
(1)
moment to develop" (Eurocode 3, Section 5.3.1). g
"...It shall be verified that both the structural
elements and the structure as a whole posses where au represents the acceleration corresponding
adequate ductility taking into account the expected to the strongest ground motion expected on the site.
exploitation of ductility, which depends on the If the base acceleration is the one corresponding to a
selected system and the behaviour factor (Eurocode return period of 475 years, the maximum values can
8, Section 4.5.2.3). The EC 3 provisions refer to the be evaluated for a return period of 970 years. The
static analysis, while the EC 8 considers the seismic acceleration corresponding to ultimate state limit can
analysis of structures. Both two examples show be determined from the relationship:
some very indefinite provisions in codes. For the
n 0.28... 0.40
structural engineer involved in the designing of a a.. 970
structure in a seismic area is essential to have a clear (2)
aA 475
definition what means sufficient rotation capacity
(EC 3) or adequate ductility (EC 8). There are two
ways to consider this aspect in the structure analysis, inferior limit is given by ATC, while the superior
one by MCT 97 provisions (Lee et al, 2000). It
admitted by EC 3. The first one requires the
estimation of rotation capacity by an analysis, but results:
without any indication concerning the method for do
it. The second one refers to constructional rales, the a= (1.22.. .1.33K (3)
cross-sections being classified in four classes in
function of slenderness of compression elements. The designer can choose the proper value in function
This classification contains many shortcomings on the local conditions.
(Gioncu and Mazzolani, 1995) and refers only for Due to the fact that ductility is determined for
static loads, the extension for seismic loads, as the highest value of seismic load a global mechanism
proposals from EC 8, is a very disputable rale. and the kinematic methodology can be used for
structure analysis corresponding to ultimate limit
In these conditions, it is acutely required to have
state. An idealized elasto-plastic behaviour may be
practical method to checking the structure ductility
considered instead of the actual curve (Fig. 4a),
at the same level as the verification for
using the methodology established by Mazzolani
displacements and strengths. For that a
and Piluso (1996). The equilibrium equation
methodology, considering the influence of main
corresponding to the plastic curve behaviour results
characteristics of near-field and far-field earthquakes
as (Gioncu and Mazzolani, 2000):
(strain-rate or accumulation of plastic deformations,
respectively) must be elaborated. But it is a very
a = cc, cc26r (4)
difficult task, due to the great number of factors
influencing the actual behaviour. Therefore, a
simplified methodology for practical purposes where ai is the collapse multiplier:
represents a pressing need. In order to set-up a
method to be used in design practice, two steps must 32X <1.0 (5a)
be developed: «1 =
2 FbuH
-estimation of required and available ductilities, with
using a static and monotonie approach;

103
 due to the increasing of yield stress, the ultimate
», »b strength remaining constant (Fig. 5a).In order to
ΣΜρ = ΣΜΡο,η
ι=1
+ΣΣ2Μρ»Μ
*=1 >1
W consider this effect, a modified Souroshian-Choi
equation is used (Souroshian and Choi, 1987):

and α2 , the characteristic of post-critical slope:

ysr 1.46 + 0.0925 logg
= CTCW (10)
α2 =
Py 1.15 + 0.0496 log¿
(6)
2aulg
where cT influence of low
considers the
6r being the plastic rotation of plastic hinge. The temperature, if structure is not protected (cr =1.1
α-θ curve obtained for an ideal elasto-plastic for T° =-20°C) and cw, the influence of welding
behaviour (Fig. 4b) considers that plastic hinges
simultaneously occur in all plastic zones and the (cw = 1 for rolled profiles and cw = 1.15 for welded
structure behaves elastically until fully plastic profiles). Due to the modification of yield ratio, an
mechanism is formed. But, in reality, the plastic increasing of plastic moment occurs:
hinges occur successively and the cusp formed by
the perfect elasto-plastic behaviour is smoothed, the
maximum load bearing capacity being reached for Mpsr=^Mp (11)
Py
ccma]iFbu . In order to obtain a simple procedure to

determine the required ultimate rotation, the actual resulting the increasing of multiplier a, (Fig. 5b):
continuous curve can be substituted by a three-linear
curve OABC, where the plateau AB is obtained for
amax = 0.9a, . So, one can obtain the static required
«Ur = Zz-M. (12)
ultimate rotation:

fe*L.=o.i V+^ a
(7a) Plaslis rncchunisni
behaviour
ν 2 J

and the required rotation capacity:

i-r__j__L (7b)
\ a2 eeJ

The ratio:

^a, =2
WH
(8)

is predominant in the required rotation. One can see

that the characteristics of seismic actions are directly
absent. But the dependence is inserted through the Fig.4 Simplified push-over method
values of plastic moments, determined in function of
these seismic actions and the behaviour factor q. The coefficient a2 remains unaffected by the strain-
Influence of strain-rate must be considered in case rate. So, the required rotation or required rotation
of near-field earthquakes, where the velocity ground capacity increases to:
motion can be very high. The effect of strain-rate ε
Pysr \0ru
produces the increasing of yield ratio: (13)
' req,sr Py
' y W
s.r" J req.st

Py =
L (9)
Influence of plastic rotation accumulation,
characteristic situation for far-field earthquakes.
fu

104
 history analysis including the effects of
accumulation of plastic rotations. One can see an
fw /fv important increasing of required rotation capacity
1 Wright
2 Rao due to multiple pulses. Due to the fact that the
3 Soroushian
4 Walluce
number of pulses producing plastic rotation
5 Kassar increases with the reducing of collapse multiplier
6 Kaneko
a, value, the amplification factor must consider this
aspect (Fig. 6b). So, the required plastic rotation or
1.0
Field of rotation capacity for the case of accumulation of
near-source
earthquakes plastic rotations results from the relationship:

2 log s (1/sec)

= r,op (14)
req,ap Uv req,st
/Strain-rate influence

where γ is a coefficient considering the effects of

Static behaviour plastic rotation accumulation, given in Table 1.

Table 1. Coefficient γαρ

Far-field Near-field
α, Beams Columns Beams Columns
Ou Ousr
0.2 3.5 4.0 1.5 2.0
Fig. 5 Influence of strain-rate on required ductility 0.4 2.5 3.0 1.3 1.6
0.6 1.5 2.0 1.0 1.2

0rx 10¿
Differences are considered for beams and columns,
D Maximum plastic rotation under static action (push-ovw method)
M Accumulated plastic rotation under cyclic action (lime-history method)
the influence of plastic accumulation for columns
being more important than for the beams. The
-WM- reduced number of pulses, in case of near-field
earthquakes, is also considered.
-W- 4x3.50
Influence of plastic rotation accumulation
produced with high-velocity. In intermediate field it
4.00
!7.5σ 7.50- i.w is possible to have a cumulative effects of strain-rate
and accumulation of plastic rotations. In this case the
required ductility can be determined by affecting the
ßarni Column (a)
<* 4 required ductility for strain-rate with the coefficient:
^hlgh otj value
(15)
^low aj value \tna
1 req,sr+ap req3sr

6 AVAILABLE DUCTILITY

The same methodology as for required ductility is

Fig.6 Influence of plastic rotation accumulation on
required ductility
Fig. 6a shows the effect of accumulation of plastic
rotation due to multiple pulses (De Matteis et al,
2000). The plastic rotation values have been
obtained using a static pushover method and time-
used, by determining the static available ductility
and correcting these values for seismic actions.
Standard beams. The available rotation capacity
must be determined at local level, taking into
account that the member belongs to a structure with
complex behaviour. But this is a very difficult task,
due to great number of factors influencing the
behaviour ofthe actual member. Thus it is important
to simplify the analysis by using, for the actual
member, a simple substitute member with a similar

105
behaviour. This member is the standard beam, where the coefficients α contain the geometrical
proposed by Gioncu and Petcu (1997). One can see and mechanical properties of members. A computer
(Fig. 7) that in a complex structure the inflection program DUCTROT M (DUCTility for ROTation of
points divide the frame members in two parts, with Members) are performed (Petcu and Gioncu, 2002),
positive and negative bending moments and each in order to obtain the ultimate rotation or rotation
actual member can be replaced by a combination of capacity, for different section types and geometrical
two standard beam types: and mechanical member properties. Fig. 9 shows
-SB1 type, central concentrated load beam for case some pages from DUCTROT M computer program
of members under linear moment gradient; for determining of available ultimate rotation and
-SB2 type, distributed load beam for case of Fig. 10, the comparison of computed values with
member under quasi-constant moment. experimental results.
Idealized behaviour*
uroskin

-r'

Iniltxlioo. L
poinl Χ φ)
tam * ,., Tj ÖkjOuO.9

, fairrríirrrrrj __ . Plastic
t») Idolized behaviour Y/mecha-

ismi natii -.iiii,,,... ,* t

Ì ΤΛ Γ, "
ip issa ^q
.
«
iî î
min
ι τ
ι n nm
χ
1 L 1 4 L· \
)Lpg( 'i-pQ}
Isiül EE2 Quasi-constant moment - SJ32
(O

(b) up ft,80.9

Fig.7 Definition of standard beams

/
Static available ductility. The behaviour under
static loads are presented in Fig. 8a,b. One can see
that the ultimate plastic rotation is limited by the
occurrence of local plastic buckling of compression
flange. For the profiles frequently used in design
practice, two types of local plastic buckling occur,
in-plane and out-of-plane buckling modes (Fig. 8c). (O

During the experimental tests one can observe that

some plastic mechanisms are formed in plastic Fig. 8 Behaviour of standard beams
zones. The large rotations are concentrated in these
local plastic mechanisms. The actual behaviour Brittle fracture. This aspect is very actual, because
curve can be substituted by three different curves, the development of steel production and application
for elastic, elasto-plastic and post-critical of steel in civil and industrial buildings have the
behaviours. Using the rigid-plastic analysis, the tendency to introduce high strength materials.
following post-critical curves are obtained: Recent advances in important structures require to
use thick plated with high strength, high fracture
-for in-plane mechanism: toughness and high corrosion resistance. All these
requirements give rise to a decreasing of steel
M ductility properties. The main factor controlling the
= «> +«2 -ÜT (16a)
plastic properties is the yield ratio (see relation (9)).
The importance of the yield ratio is increased in a
-for out-of-plane mechanism: structure due to moment gradient along the
geometrical axis, the member behaving in the strain-
M 1
= a,+a2 1/2 +«3
(16b) hardening range. Generally, it is considered that
M, θ θ 3/4 yield ratio with ρ < 0.75 assures a good ductility,

106
while yield ratio with ρ > 0.75 impedes the
development of a good ductility. The mild steels (Fe
360, Fe 430, Fe 510) used for structures frame in the
category of low yield ratio. The problem of yield
ratio is very important for recent steel production, in
which the new steels with high strength have high
yield ratio. In case of high yield ratio some yield
lines with high plastic rotations may be changed in
fracture lines and cracks, producing a brittle failure
of members (Fig. 11a). The fracture of plate,
considered when the ultimate strain is reached, can
be determined for the rotation (Gioncu and
Mazzolani, 2002):

1
α,=ξ -1 (17) O'cxp
\r,

ξ being numerical coefficient including the steel

a Fig. 10 Correlation between theoretical and
characteristics, L, the distance between the inflection experimental values for rotation capacity
points of local plastic mechanism, t, the plate
thickness, su, the ultimate strain. Fig. lib shows
the influence of European, USA, and Japanese steel
qualities on the fracture rotation of a plate. A very
important reduction of fracture rotation may be
noticed as far a the yield stress and yield ratio
increase, showing the importance of steel quality in
the protection against premature collapse.

ft^jrji^rj^ £*.*« »>*. a«·· &«

2 Muniti churawi.iTtti' ISccI»» tkariitcfhtlka | SLCTIO"/
wh tEaiEN Tw&vs ciw o CHARACTERISTICS
rtsjLi ,

aucun Γο 4·** M$n

4 »Vfì. .IN»*«
ITO!
tt 5 -It m«
WiCÎ t* 'Α. *<!*«

th*£4=te££lic! «cíe drforozjifJ τ*πι{

tbr f tumatitittfnlifí

A ' Ili,

1 -tf < «tan Λ» t ¡jvm btr

' g.wä»liFU!
iJhrcUing triad·« *A»tCtm*
fcpliM badJae, aurntf ij.iilUat I Ojl·«*;^^ it
*" Vt.\.jMk ,1

JSeh^inotiieWliJtTrtellfatrie»·«!»«»* Si1*« o»«c lUin (nancri*«lcteadlB| ί jjj_

^ίΓί-ΊΥΜίίίΙΡΙυίΓΐ "ECtS *«»tÎI*»>ICÎ'QÏEî'îtC'Î
or

4*£ **»*, «n pai 1 5ui

Fig. 1 1 Influence of steel quality on plate fracture

»*.-t_-_r

Fig. 9 Determination of ultimate rotation with

DUCTROT-M computer program

107
 fracture rotations are presented in Fig. 14b. At the
same time the ultimate plastic rotations are plotted.
The danger of compression flange for near-field
New Japanese Steel
Py=0.74
earthquakes exists for profiles with slenderness less
W200x4C than the limit proposed by Gioncu and Petcu
(1997):
203

'c^
\<fJ (19)
1.126 ( Ε χ

[{1 + 0.61 py){l + 0.192E I Eh)\l2{fy ^

Fig. 12 Influence of yield ratio on moment rotation where E and Eh are the moduli for elastic and strain
curve
hardening ranges;
At the level of member, the load-deflection curves of
Fig. 12, obtained by using FEM analysis with
ABACUS program (Galambos, 1999), shows a
tremendous difference on inelastic rotation capacity
of structural members, due to the yield ratio. The
fracture rotation of a buckled compression flange, P).=0.9Ü9

0f , can be determined by the relationship: l*M»/ /o.a.«

/Ο.ΧΠ

«W (l + )3 dt) "
where :

^=1.5^ (18b)

considering that the ultimate strain in a member is

greater than that determined on tension tests. Fig.
13a shows the influence of yield ratio for which the
fracture of buckled flange occurs. For py > 0.817
fracture can occurs before reaching the ultimate
plastic rotation. Fig. 13b shows the influence of
yield ration on the beam rotation capacity. For a
given yield ratio the plastic rotation capacity is
changed in a fracture collapse. One must note that
the thicker flanges are more sensitive to fracture
than the thinner ones.
Influence of strain-rate. There are three effect types
of strain-rate on the ductility:
-reducing the ultimate available ductility due to the
increasing of plastic moment given by relationship
(1 1) as result of yield ratio increasing (Fig. 14a);
-increasing of brittle fracture danger (Fig. 14b), the
fracture rotation due to strain-rate resulting by
changing in equation (18a) of py with pysr . For
some profiles with different flange slendernesses the
Fig. 13 Transformation of plastic collapse in fracture
collapse
-increasing of nil-ductility (ND) temperature, which
is the temperature for which a ductile behaviour is
transformed in a brittle fracture. For static loads, this
temperature is about -95 C for rolled profiles and -
65 C for welded profiles (Akiyama, 2000). For
seismic loads, having strain-rate range from
5 · 10"1 / sec to 9 · 10"1 / sec the temperature increases
with 40 ...60 C (Fig. 14c), being -55... -35 C for
rolled profiles and -25 ...-5 C for welded
profiles. For rolled profiles, the ND temperature is
out of practical situations, but for welded profiles, it

108
is very closed to possible temperature when the
structure works in open air.
Influence of repeated cyclic actions. For available
rotation the control parameter is selected to be the
elastic rotation θρ (in the spirit of ECCS
Recommendations) and the cyclic loading type,
constant increasing or decreasing amplitudes (Fig.
15a). Under cyclic actions, the plastic collapse
mechanism is formed by super-imposing, as in a
mirror, of two local plastic mechanisms (Fig. 15b).
The cyclic action effects are presented in Fig. 16a
for constant amplitudes. There are not differences
between monotonie and cyclic actions until the
buckling of compression flange at the cycle:
Θ.
+1 (20)
Θ.
where [ ] denotes the integer part of a real number. If
η is the number of strong pulses and nr is the
number of pulses produced after the local buckling,
being the reason of reducing of static ductility:

nr =n-nh (21)

Local plastic mechanism So, the ultimate rotation for cyclic actions results as:

(Θ
V
) .
russets =θ ru -ΔΘ u (22)
\""/

where:

n(nr 1)_ . ,. T
θ , increas. amplitude
2
A(9U = -j (nr- 1)θρ , const. amplitude (23)
(1 1/2" )θρ , decreas. amplitude

Plastic mechanism for M

Fig.l4 Influence of strain-rate ρ ysr Fig. 15 Influence of accumulation plastic rotation

109
 wnue ror rar-neia earmquaices, tne accumulation or
-/*¥,
/ Γλ
plastic deformations.
jíj j j
/il Ά liti
ii o+o

-- tiili Jfcl
/¿--Tjr
s
Mp REFERENCES
lJ^rir"
Mk"
(a)
Akiyama H. (2000): Evaluation of fracturai mode of failure in
steel structures following Kobe lessons. Journal of
^Constant amplitudo Constructional Steel Research, Vol. 55, 211-227
De Matteis G., Landolfo R., Dubina D., Stratan Α. (2000):
Influence of the structural topology on the seismic
performance of steel framed buildings. In Moment Resistant
Connections of Steel Frames in Seismic Areas. Design and
Reability (ed. F.M.Mazzolani), E&FN Spon, London, 5113-
538
Galambos T.V. (1999): Receent research aand design
development in steel and composite steel-concrete structures
in SUA. Journal of Constructional Steel Research, Vol. 55,
No. 1-3, 289-303
Gioncu V., Mazzolani F.M. (1995): Alternative methods for
assessing local ductility. In Behaviour of Steel Structures in
Fig. 16 Reduction of rotation capacity due to the
Seismic Areas, STESSA 94 (eds. F.M. Mazzolani and V.
accumulation of plastic rotation Gioncu), Timisoara, 26 June-1 July 1994, E&FN Spon,
London 182-190
Gioncu V., Mazzolani F.M. (2002): Ductility of Seismic
Resistant Steel Structures. Spon Press, London
Gioncu V., Petcu D.(1997): Available rotation capacity of
wide-flange beams and beam-columns. Journal of
150 150 Constructional Steel Research, Vol. 43,No.l-3, 161-217
>=*I4 Hamburger R.O. (1996): Implementation performance based
40

300 *- seismic design in structural engineering practice. In 11th

»1, = 9 cycles
World Conference on Earthquake Engineering, Acapulco, 23-
30
30 θρ =0.00771 red 28 June, CD-ROM 2124
^Constant rotation amplitude
Lee L.H., Lee H.H., Han S.W. (2000):Method for selecting
design earthquake ground motions for tall buildings. The
20 1«
IIII Increased rotation amplitude Structural Design of Tall Buildings, Vol. 9, 201-213
Marceliini Α., Daminelli R., Pagani M. Riva F. (1998): Seismic
hazard of the Mediterranian area. In 11* European
Conference on Earthquake Engineering, Paris, 6-11
September 1998, CD-ROM 269
0 0.65 0.
Py Mazzolani F.M. (1995): Some simple considerations arising
from Japanise presentation on the damage caused by Hanshin
Fig. 17 Cyclic number of ductile fracture in function earthqquake. In Stability of Steel Structures (ed. M. Ivanyi),
Budapest, 21-23 September 1995, Akadémiai Kiado,
of yield ratio Budaapest, Vol. 2., 1007-1010
Mazzolani F.M., Piluso V. (1997): A simple approach for
evaluating performance levels of moment-resistant steel
frames. In Seismic Design Methodologies for the Next

110
 Generation of Codes (eds. P.Fajfar and H. Krawinkler),
Bled, 24-27 June 1997, Balkema, Rotterdam, 241-252
Petcu D., Gioncu V. (2002): Computer program for available
ductility analysis of steel structures. Computers and
Structures (paper submitted for publication)
Priestly M.J.N. (1997): Displacement-based seismic
assessment of reinforced concrete buildings. Journal of
Earthquake Engineering, Vol. 1, No.l, 157-192
Somerville P.G. (2000): Reducing uncertainty in strong motion
predictions. In 12th World Conference on Earthquake
Engineering, Auckland, 30 January-4 February 2000, CD-
ROM 2688
Somerville P.G., Smith N.F., Abranamson N.A. (1996):
Account for near-fault rupture directivity effects in the
development of design ground motions. In 11th World
Conference on Earthquake Engineering, Acapulco, 23-28
June, 1996, CD-ROM 711
Souroshian P., Choi K.K. (1987): Steel mechanical at different
strain rate. Journal of Structural Engineering, Vol. 113, No.4,
863-872
Yang T.S., Popov E.P. (1997): Comparison of MRF
connection damage due to Northridge quake with theory. In
Behaviour of Steel Structures in Seismic Areas, STESSA 97,
(eds. F.M. Mazzolani and H. Akiyama), Kyoto, 3-8 August
1997, 10/17, Salerno, 897-904

111
General methodologies for evaluating the structural performance under
exceptional loadings

M. Fischinger, P. Fajfar, M. Dolsek, D. Zamfirescu* & A. Stratan*

University of Ljubljana, Slovenia, * visiting researcher from Technical University of Civil Engineering,
Bucharest, Romania, * visiting researcher from "Politehnica " University of Timisoara, Romania
T. Leino
VTT Building Technology, Finland
R. Vãcãreanu, T. Cornea & D. Lungu
Technical University of Civil Engineering, Bucharest, Romania
S. Wolinski
Rzeszow University of Technology, Poland
E. Álarcon
"Politehnica " University of Madrid, Spain

ABSTRACT: The traditional code procedures are typically based on experience. As such they may not work
well in the case of exceptional loadings and innovative structural systems. Moreover, the investors and
society may choose higher level of performance than embedded in traditional codes. The alternative is offered
by the Performance Based Engineering (PBE), which is defined as the integrated effort to produce the
engineered facilities for multiple performance objectives. The possibilities of having different target
performance levels increase the competition between producers and products. The decision about the
acceptable performance is inherently linked to the problem of risk and reliability, even more so in the case of
exceptional loading. General overview of the required performance levels as well as methods of risk and
reliability evaluation is therefore addressed first. The second - major part ofthe paper presents more specific
application ofthe PBE in earthquake engineering (Performance Based Seismic Engineering - PBSE), where a
major breakthrough of new performance based oriented methodologies has been observed after some
catastrophic events in the recent past. Two specific procedures - the Cornell's approach and the probabilistic
methodology for seismic evaluation of existing buildings implemented in HAZUS are explicitly addressed.
The Cornell's approach is illustrated by the case study of RC frames with masonry infills. The Monte-Carlo
simulation is used to calibrate the fragility function parameters given in HAZUS.

1 INTRODUCTION design procedures are often empirically based and

COST C 12 action is oriented towards "Improving
buildings' structural quality by new technologies".
The special objective of the WG 2 is to study the
behaviour of innovative (i.e. mixed systems studied
by the WG1) as well as existing structural systems
under exceptional loadings, being defined as actions
not considered in the design of the structure. Four
natural elements: fire, air, water, earth, as well as
human activity are sources of exceptional loadings
acting on structures due to: fires, wind-storms,
floods, earthquakes, explosions, impact of heavy
vehicles, etc. The performance that we expect from
the building products in such circumstances is very
much related to the characteristics of the extreme
loading and the structural integrity of the building,
but also to the physical properties of the assemblies
ofthe factory made building products.
It should be realised that better performance as
well as quality in general can and has to be achieved
by using the improved methodologies of structural
design and evaluation. Namely, the present code
113
they operate with vague terms and imprecise
definitions as "moderate action", "some damage" or
"less safe than". As such, they often cannot satisfy
(a) the owner's desire for sound judgment on the
cost and benefits of a particular structural solution,
(b) the society's needs for informed decision making
and (c) the designer's desire for understanding ofthe
procedure (Krawinkler 1997).
Since the traditional code procedures are
frequently based on experience, they usually yield
acceptable results in the case of the standard
structural solutions and everyday actions. However,
they could not work as well in the case of the
exceptional loadings and (in the absence of the
empirical evidence) they will certainly not work in
the case of the new technologies and innovative
structural solutions.
The alternative is offered by the Performance
Based Engineering (PBE), which is defined as the
integrated effort to design (Performance Based
Design - PBD), construct and maintain the
engineered facilities of predictable and quantified
performance for multiple performance objectives.
The issue of performance and associated
problems of risk and reliability are inherently and at
least implicitly present in every engineering activity.
Consequently this very broad topic is first addressed
in the paper by a general overview of the required
performance levels in the development of building
products as well as the whole range of methods of
risk and reliability evaluation. The second part ofthe
paper presents more specific application of the PBD
in earthquake engineering (Performance Based
Seismic Engineering - PBSE), where a major
breakthrough of new performance based oriented
methodologies has been observed after some
catastrophic events in the recent past.
2 PERFORMANCE /REQUIREMENT/ LEVELS
ΓΝ THE BUILDING PRODUCT
DEVELOPMENT
2.1 Product development
By definition the building products are factory made
parts (building components) providing the user some
kind of a performance, which is not merely space or
floor area for his business or the load bearing
capacity. The types of performances of the building
products discussed here are the protection of people
and goods against the effects of wind or snow
overloads, or the other exceptional strains like rain,
cold weather, humidity, ice, fire, pollution, blast or
the impact loads etc. The performance needed in
such circumstances is very much related to the
extreme loading and the structural integrity of the
building, but also to the physical properties of the
assemblies ofthe factory made building products.
It has been fully realised, however, that there are
possibilities of having different target performance
levels. This increases the competition between
producers and products, and lets the designers have
more degrees of freedom in the design of buildings
and structures. Also the society has more options to
define the level of performance related to public and
other similar new building projects. Allies between
producers of different products may increase the
amount of possible good solutions for the builders.
With the definition of different performance levels it
becomes obvious that all structures need not be built
to last forever. The suitable choice between the costs
of the new building compared to the costs of its
maintenance gives more possibilities of competition.
The general ideas regarding how to produce and
manufacture a competitive building product can be
found for example in texts by M. Porter. He has
studied the external factors having an influence on
the success of the product. They are called the "Five
Forces Affecting Competitive Strategy". These
forces or threats are:
114
- (1) The threat of entry by new competitors,
- (2) The intensity of rivalry among existing
competitors,
- (3) Pressure from substitute products,
- (4) The bargaining power of buyers,
- (5) The bargaining power of suppliers.
The most important internal factors having an
influence on the product development and the
success of the product on the market have to do
with:
- (6) The careful selection of requirements that the
product is supposed to meet,
-(7) The methods and machinery available in the
production process ofthe firm,
-(8) The decisions taken concerning the level of
performance ofthe product.
2.2 Requirements and performance levels
The principle of the Performance Based Design of
building products can be expressed with the
following chart, which describes the relations
between the properties ofthe building products, their
requirements and how the products are proved that
their properties meet the different requirements.
The figure describes the process of proving the
performance of the product. Having different quality
or performance levels depends on the loads or
product types as explained earlier. The process of
the development of building products must comprise
three defined parts, which are interrelated. They are:
- The study ofthe properties ofthe building products
and structures,
-Definition of the requirements or requirement
levels for different uses of products,
- The definition and development of testing methods
to prove the products properties.
The box defining the different possible levels of
performance has in many building codes been
divided into three or four levels. For example in the
new seismic codes these levels are frequently
defined according to the tolerable damage.
Regarding any ofthe other properties, the basic level
is always expressed in the national building design
codes. Unless there is a design error there is always
a reserve, i.e., the products and structures meet the
basic requirements very well (see Fig. 1), because of
the relatively discrete division of the dimensions of
the products.
However, in practice the structural resistance
properties (Ultimate and Serviceability limit states)
are the only ones where the basic requirements are
sufficient and no additional requirements need to be
made. Regarding most of the other properties there
is a tendency of defining them to meet a higher
performance level. Higher level is always more
expensive to acquire but there may be good enough
reasons to see that for example a manufacturing
process can continue no matter what kind of loads or decision-maker's preferences. The weights are
strains have an effect on the buildings. combined with

The Performance Based Design in

development of building products

PROPERTIES OF THE REQUIREMENT LEVELS MEETING THE

BUILDING PRODUCT"
The builders end end usei
define the quality levels,
standards and goals for tl
building project.
This includes definitions
the levels of safety,
perrformance and quality
different building produc
and their design and serv
life, etc.
4 -PRODUCTS REQUIREMENT
;ty, and the builder (and Testing methods
defined by society
s!) define requirements, i.e.,
indards, design codes, and and/or the builder,
ific customer needs are the and designed for
testing the product
ents for building project, or
tie level of quality of work.. properties.
st also define the methods of For the resistance
properties basic
rove the quality of products.
requirements are
; requirements, because of
sufficient.
asons, it is the society who
he basic minimum levels.

Quality improving, or
Red = Unacceptable Reserve jw better performance level
performance
-
f\ Basic requirements Basic performance Performance
from design codes level in reality requirements
for the project

Figure 1. Performance based design of building products.

3 METHODS OF RISK AND RELIABILITY

EVALUATION
The decision about the acceptable performance is
inherently linked to the problem of risk and
reliability, even more so in the case of exceptional
loading. In such a case the risk analysis of a
structure would provide information that makes
possible to select concepts for risk reduction and
minimization considering: risk to humans, risk to
environment and the cost of safety measures.
Three types of the risk and reliability evaluation
methods are generally accepted for usual structural
design and assessment: qualitative, semi-quantitative
and quantitative.
Qualitative methods are based on implicit rules
and on engineer's judgment. Rational
understanding of a long and successful building
tradition may lead to a reliable and economical
design, especially when more quantitative approach
usually fails from a lack of sufficient data.
Deterministic risk-based exceptional loading
analysis is usually based on the standard
Quantitative Risk Analysis (QRA), which is focused
on the combined effect of frequency and
consequences. A large number of scenario are
usually considered in form of an event tree.
Semi-quantitative methods based on the
hierarchical models consist of different levels and
decision weights, which are chosen according to the
115
parameters grades which can be used to calculate a
risk value.
Quantitative methods are focused on the
combined effect of loadings, using different levels
probabilistic or fuzzy-probabilistic assessment
methods: analytical, numerical and simulation-
based.
All components of the structural design and
evaluation processes involve various uncertainties.
These uncertainties can be put into two categories
with regard to causes: natural (e.g. unpredictability
of loads such as: wind-storm, earthquake, flood) and
human (including departure from an optimum or
safe design and operations such as: calculation
errors, lack of knowledge, overloading, etc.).
Different types of uncertainties can be observed in
the performance of structures: random and fuzzy.
A random variables, parameters and states are
usually considered as a function of three factors: the
physical variation factor which represents the
parameter's variation that is inherent in quantity
being considered, the statistical variation factor
which represents uncertainty from estimating
parameters due to a limited sample size and the
model variation which represents uncertainty due to
simplifying assumptions, unknown effects, etc.
Fuzzy variables, parameters and states of a
structure can't be precisely described either in
deterministic or statistic way, but they can be only
roughly estimated. For example, a load due to
explosion can be defined as a fuzzy number Q =
"about Q".
A notion of the "performance function" or the
"limit state function" or the "safety margin" is
associated with the generally accepted Limit States
Design Method, and can be defined as: Ζ (R, S) = Ζ
(Χι, X2,..., Xn) = R - S , where R is the resistance, S
is the load effect and X¡ are variables and parameters
representing resistance and load components,
dimensions, analysis factors, etc.
Various methods of the structural design and
reliability evaluation can be distinguished:
-Deterministic methods (level 0), where the F-
function is a vector of real numbers.
- Semi-probabilistic methods (level I), where the F-
function is a vector of fractiles of unknown
rank.
- Simplified probabilistic or first order reliability
methods (level II), where the F-function is the
normal n-dimensional random variable.
- Fully probabilistic or exact probabilistic methods
(level III), where the F-function is a random
variable with the resultant probability distribution.
- Fuzzy-probabilistic and fuzzy, unconventional
methods (level IV), where the Z-function is a
random n-dimensional variable with fuzzyfied
boundaries or elements, or Ζ is a fuzzy set.
Safety measures associated with these methods
are defined by:
- The global or component safety factors (level 0)
Sg= Rm/Sm > 1 ΟΓ Sg.^R.ni/S.n.i > 1 (1)
where Rm, Sm are mean values of the resistance
and the load effect,
- Partial safety factors γ \ (level I)
Ti = Xd,i/Xk,i ~~
where Xd,i and X^ are the design and
characteristic values of random variables X¡ ,
- The global or component reliability indices β
(level Π)
ß = E[Z(R,S)]/V[(R,S)]>ßf, or
ßR = E[R]/V[R]> ßfjR and
ßs = E[S]/V[R]> ßf,s
where E and V are operators of the expected value
and the mean standard deviation,
- The probability of failure Pf (level ITT)
Pf= J Z(Xi)dxi < Pf,d
FA
The probability of fuzzy event Pf (level TV)
Pf = j μζ(χί)Ζ(Χί)άχί <pfd
R oriented probabilistic approach for seismic
or representative values of fuzzy safety factor
(level TV)
ί8[μ(ί)]=Α/5 > sgA (6)
where Pf and P^ are the imprecise probability and
its target value, FA is the failure area, μ ( . ) is the
membership function of fuzzy variable, sz and ss¿
are the fuzzy global safety factor and its target
value.
Code design procedures, including seismic design
procedures have been traditionally based on the level
I approach using partial safety factors. However, in
seismic design there has been a recent breakthrough
of higherlevel methods as it will be demonstrated in
the subsequent chapters. Yet again there has been
also a tendency to simplify these procedures and to
relate them to the traditional approaches (see some
features ofthe Cornell's approach for example).
4 4. APPLICATION OF A PERFORMANCE
BASED METHODOLOGY TN SEISMIC
DESIGN AND EVALUATION
4.1 Introduction
Several recent earthquakes have caused many
building owners and design professionals to question
performance objectives embedded in the standard
seismic design codes. The codes have been
traditionally focused on ensuring a life safe
performance with the empirically based expectation
that the other performance objectives will be
automatically satisfied. First of all a substantial (but
not quantified) amount of damage was anticipated
and tolerated in the case of major earthquakes.
However, after the recent earthquakes the owners
and public policy makers have been not willing to
accept large economic losses. It has been recognised
(2)
that performance objectives with the ability to
control property and business interruption losses
need to be developed and implemented. This has led
to the formal development of Performance Based
Seismic Engineering guidelines (PBSE), which are
briefly presented in subsequent sections.
First the conceptual framework for PBSE is
summarised. Then, a brief overview of simplified
(3) non-linear analysis procedures is presented. A
breakthrough of this type of approaches occurred in
last years and they have been already implemented
in several modern codes and guidelines for
earthquake resistant design of new buildings as well
(4) as for evaluation of existing buildings. An inherent
feature of structural properties and ground motion
parameters is their large uncertainty and
randomness, which makes necessary to formulate
PBSE approaches in a reliability-based
(5) (probabilistic) format. Recently an engineering-
performance assessment has been developed at
Stanford University (Cornell & Krawinkler, 2000).
116
In this paper, it is very briefly summarised and
applied for analysis of a reinforced concrete frame
building with masonry infill. Finally, the paper
presents an alternative methodology for the
assessment of seismic behaviour and fragility of RC
buildings using HAZUS (1997). The Monte-Carlo
simulation is used to calibrate the fragility function
parameters given in HAZUS (1997).
4.2 Conceptual frameworkf or PBSE
In the introduction the Performance Based
Engineering (PBE) has been defined as the
integrated effort of design, construction and
maintenance needed to produce engineered facilities
of predictable performance for multiple performance
objectives. A related term is Performance Based
Design (PBD), which encompasses only aspects
related to design. Performance Based Seismic
Engineering (PBSE) and Performance Based
Seismic Design (PBSD) apply to aspects related to
seismic effects. The goal of PBSE is to prevent
casualties due to earthquakes and to minimize
earthquake-related costs to the building owner over
the life ofthe building.
The following definitions and explanations are
taken, in abbreviated form, from a paper by Bertero
(1997).
Design Performance Objectives. A design
performance objective is an expression of the
desired performance level for the building for each
earthquake design level. In the case of a building
design, design performance objectives should be
selected based on the building's occupancy, the
importance of functions occurring within the
building, economic considerations including costs
related to building damage repair and business
interruption, and considerations of the potential
importance of the building as a historic or cultural
resource. According to the above definition, design
performance objective couples expected or desired
performance levels with levels of possible seismic
hazards.
Earthquake Performance Level
Life Safe Near
>
Frequent
O O
(43 year)
& Occasional (for New
CO
(72 year)
o O
n
Rare
(475 year)
s
CS
Very Rare
UJ (970 year)
Figure 2. Seismic performance design objective matrix (Vision
2000).
117
This is illustrated by the Design Performance
Objective Matrix shown in Figure 2, which has been
proposed for buildings by the Vision 2000
Committee.
Definition of the Performance Level. A
performance level is an expression ofthe maximum
desired extent of damage to a facility given that a
specific earthquake design level affects it. In the
case of buildings, the condition of structural
elements, non-structural elements, and contents are
all considered in the performance levels, as are
availability of site utilities necessary to building
function. Performance levels are expressed in two
different but related ways:
- In qualitative terms meaningful to the lay public
including building owners, occupants, and public
policy makers.
- In technical terms useful for engineering design
and evaluation, including the extent of
deterioration and degradation permissible to
individual structural and non-structural elements,
as well as overall global behaviour ofthe structural
systems.
As indicated in Figure 2, the Vision 2000
Committee has selected and defined four individual
performance levels among the entire spectrum of
damage states that a building may experience when
subjected to earthquake ground motions of
increasing severity (damage potential). The
performance levels recommended by Vision 2000
(SEAOC 1995) and indicated in Figure 2 are as
follows: Fully Operational or Serviceable (facility
continues in operation with negligible damage);
Operational or Functional (facility continues in
operation with minor damage and minor disruption
in non-essential services); Life Safety (life safety is
substantially protected, damage is moderate to
extensive); and Near Collapse or Impending
Collapse (life safety is at risk, damage is severe, and
structural collapse is prevented).
Definition of Earthquake Design Levels. For
many sites, there is a continuous spectrum of
earthquakes, which may affect it, ranging from
distant, small magnitude events that produce
negligible hazards, to local, large magnitude events,
which produce hazards that have the potential of
being highly damaging.
PBE seeks to control the levels of damage
experienced by a facility over the full spectrum of
events, which may occur. In order to permit
practical application of this approach, it is necessary
to select a series of discrete earthquake events from
among the entire spectrum, which may occur. These
should represent the range of earthquake severity for
which a particular facility performance is desired.
These discrete earthquake events are termed
earthquake design levels. For the case of building,
the Vision 2000 Committee expressed the
earthquake design levels as a set of earthquake
ground motions and associated hazards with
specified probabilities of occurrences and/or in
terms of a mean recurrence interval. The definition
ofthe earthquake design levels will vary from site to
site depending on the seismicity of the region on
which the site is located as well as the socially and
economically acceptable levels of damage to the
user community.
Collapse
Prevention
' io "LS "CP
Lateral Displacement
Figure 3. Pushover curve from FEMA (Hamburger, 1997).
Minimum Design Performance Objectives. As
indicated in the matrix of Figure 2,
recommendations are provided for minimum design
performance objectives for building of three
different occupancies and uses: (1) Basic Objectives
are defined as the minimum acceptable performance
objectives for buildings of normal occupancy and
use; (2) Essential/Hazardous Objectives are the
minimum acceptable for Essential Facilities (which
are critical to post earthquake operations such as
hospitals, police stations, fire stations,
communication centres, emergency control centres,
and shelters for emergency response vehicles) and
Hazardous Facilities which are those that contain
large quantities of hazardous materials, but where
the release of those materials would be contained
within the boundaries of the facility and the impact
to the public minimal; and (3) Safety Critical
Objectives which are those that contain large
quantities of hazardous materials, the release of
which would result in unacceptable hazards to wide
segments of the public. The above three minimum
objectives are illustrated in the Performance
Objective Matrix of Figure 2 by the diagonal lines.
Enhanced Objectives. Individual owners and
tenants may desire on a case-by-case basis to design
their facilities for better performance or lower risk
than the minimum objectives recommended in the
design performance objective matrix of Figure 2.
These design performance objectives which result in
less damage at one or more earthquake design levels
than the minimum recommended are termed
"Enhanced Objectives."
Selection of Design Performance Objectives. This
selection should be made by the client in
consultation with the designer based
consideration of the client's expectations, the
seismic hazard exposure, economic analysis, and the
acceptable risks using the design performance
objective matrix (Figure 2).
An illustration of structural performance levels is
presented in the static pushover curve shown in
Figure 3, (FEMA-273). (Note the difference in
terminology compared to Figure 2, which is based
on Vision 2000.) Although, as indicated in Figure 3,
FEMA-273 describes the structural performance
levels based on the global behaviour ofthe structure,
actual evaluation is performed on a component
basis.
Analysis Procedures. A key step in the PBD
methodology is performance evaluation
(acceptability) analysis. The design procedure
typically endeavours to provide a strength,
deformation, or energy capacity at several
performance levels. The evaluation analysis is used
to check the adequacy of the design and to make
modifications as needed.
The current procedures used to design buildings
for seismic effects have been developed based on
evolutionary concepts and observed behaviour of
buildings and have been formulated to suit the
designer's needs. The seismic loads that are
prescribed can be easily combined with other
loading conditions. These techniques, by their very
nature, are not suitable to predict the actual
behaviour of the structure in a specific earthquake.
Such predictions require the use of improved
demand and capacity characterisations together with
more realistic evaluation techniques.
Several evaluation procedures with potential
applicability to PBE have been developed. They
include elastic, equivalent elastic and inelastic
analysis. When selecting the most appropriate
evaluation procedure, it should be taken into account
that great uncertainties are involved in the
description of seismic input and of strength and
deformation capacities of elements and structures.
Evaluation procedures need not be more accurate
than is merited by the uncertainties involved.
Inelastic analysis procedures are needed to
evaluate the structural behaviour at performance
levels at which significant inelastic deformations are
expected. The existing inelastic procedures need
refinements to capture the collapse safety. For code
purposes only simple methods of non-linear analysis
are suitable. One analysis option for performance
evaluation at multiple performance levels is the non¬
linear static (pushover) analysis.
Non-linear time history analysis with recorded or
simulated ground motion records provides the most
accurate means for predicting seismic demands - for
the specified ground motion and presuming that the
hysteretic characteristics of all important elements
can be modelled accurately. For a general
on
performance evaluation, a set of ground motion
records that represents the full range of frequency
118
and intensity characteristics of the hazard at the site
needs to be used. For the time being, non-linear time
history analysis is appropriate only for analysis very
important structures and for research.
Performance evaluation is based on a comparison
of computed demands for critical response
parameters with limiting values of those parameters
associated with the selected performance levels.
Consensus has been reached that the interstory drift
is a very useful performance indicator. The
important component(s) of drift will depend on the
evaluation objective, considering that interstory drift
may be caused by shear distortion within the story,
cumulative flexural rotations (e.g., in walls), and
rotation at the bottom of the structure due to
foundation flexibility.
4.3 Simplified nonlinear methods
For the time being, the most rational analysis and
performance evaluation methods for practical
applications seem to be simplified non-linear
procedures, which combine the non-linear static
(pushover) analysis of a relatively simple
mathematical model and the response spectrum
approach. In recent years, a breakthrough of these
procedures has been observed. They have been
implemented into the modern guidelines and codes.
Although different methods may yield in many cases
similar results, they differ in respect to simplicity,
transparency and clarity of theoretical background.
In this chapter, an overview of simplified methods
and their basic features are presented.
Freeman et al. (1975) developed a rapid
evaluation method, which can be considered as a
forerunner of the today's "Capacity spectrum
method". Saiidi & Sozen (1981) proposed to
perform non-linear dynamic analyses on an
equivalent SDOF system. Based on this idea, in mid-
1980s Fajfar & Fischinger (1987, 1989) developed
the first version of the N2 method (N stands for
Non-linear and 2 for two mathematical models - a
SDOF and a MDOF model). However, the
earthquake engineering community has not paid
much attention to simplified non-linear approaches
until mid-1990s, when a breakthrough of this
approaches occurred. Present examples of simplified
non-linear methods are the so-called Capacity
spectrum method, which is in different variants
applied in ATC 40 (1996), in Trisevices' manual
(see Freeman, 1998), and in Japanese Building
Standard Law (see Otani et al. 2000), the non-linear
static procedure, applied in FEMA 273 (1997) and
further developed in FEMA 356 (2000), the N2
method (Fajfar & Gaspersic 1996, Fajfar 2000),
which has been implemented in the draft Eurocode 8
(CEN 2001), and the Modal pushover analysis
(Chopra & Goel 2002). A method, similar to the
non-linear static procedure in FEMA 273, has been
119
included in the most recent NEHRP (FEMA 2001)
provisions. Several simplified non-linear methods
are implemented in the SEAOC (1999) Blue Book.
All methods combine the pushover analysis of a
multi-degree-of-freedom (MDOF) model with the
response spectrum analysis of an equivalent single-
degree-of-freedom (SDOF) system. Inelastic spectra
or elastic spectra with equivalent damping and
period are applied. As an alternative representation
of inelastic spectrum the Yield point spectrum has
been developed (Ascbheim & Black 2000).
With the exception of the methods implemented
in FEMA documents, these methods are formulated
in the acceleration - displacement (AD) format. In
this format, the capacity of a structure is directly
compared with the demands of earthquake ground
motion on the structure. The graphical presentation
(Figure 4) makes possible a visual interpretation of
the procedure and of the relations between the basic
quantities controlling the seismic response. The
capacity of the structure is represented by a force -
displacement curve, obtained by non-linear static
(pushover) analysis. The base shear forces and roof
displacements are converted to the spectral
accelerations and spectral displacements of an
equivalent single-degree-of-freedom (SDOF)
system, respectively. These spectral values define
the capacity diagram. The definition of seismic
demand spectrum represents the main difference
between different methods. In all cases, the
intersection of the capacity curve and the demand
spectrum provides an estimate of the inelastic
acceleration (strength) and displacement demand.
Dd* D/ S^Sfc sd
Sae acceleration demand of elastic system
acceleration capacity and demand of inelastic
ay system
Sa(j design acceleration
c _c displacement demand of elastic an inelastic
öd Ode .L
α °e system
Dy yield displacement
~ * displacement determined by elastic analysis
d with design seismic forces
μ ductility
T* period
Figure 4. Elastic and inelastic demand spectra versus capacity
curve for equivalent SDOF system.
All methods are basically limited to planar
structures. Recently, attempts have been made to
extend the applicability of the methods to
asymmetric structures, which require a 3D analysis.
(Ayala & Tavera 2002, Moghadam & Tso 2002,
Fajfar 2002).
A study performed on a simple planar structure
(Zamfirescu & Fajfar, 2001) indicates that the
procedures, based on pushover analysis, generally
yield results of adequate accuracy. The global
quantities (like top displacement) are generally more
accurate than the local ones (like rotations at
member ends). However, different approaches differ
in regard to simplicity, transparency, and the clarity
ofthe theoretical background.
The essential difference is related to the
determination of the displacement demand (target
displacement). In the "capacity spectrum method"
and in other methods based on equivalent
linearization, inelastic seismic demand is estimated
from a highly damped elastic spectrum. An
equivalent linear system is determined with lower
stiffness and higher damping than those of the actual
system. The equivalent (effective) elastic period of
the system corresponds to the secant stiffness at the
maximum displacement. Different procedures have
been proposed for determination of equivalent
viscous damping, which is intended to take into
account the hysteretic energy dissipation. The
equivalent damping is usually a function of the
ductility demand μ. If an equivalent elastic spectrum
is used, displacement demand depends on equivalent
stiffness and equivalent damping. On the other hand,
both quantities depend on the ductility demand,
which is related to displacement demand.
Consequently, all quantities are interrelated and, in
principle, iteration is needed. The iterative procedure
is supposed to converge to the final value of the
target displacement. However, it may not converge
as demonstrated by Chopra and Goel (1999). The
use of highly damped elastic spectra for the
determination of seismic demand is a controversial
part of the capacity spectrum method and other
methods based on equivalent linearization. The
quantitative values of equivalent damping, suggested
by different authors, differ considerably. It is
interesting to note that Freeman, the author of the
"Capacity spectrum method", derived the equivalent
damping (employed in Triservices' manual
procedure) by equating the peak deformation of the
equivalent linear system, determined from the elastic
spectrum, to the peak deformation of the inelastic
system, determined from the inelastic (Newmark-
Hall) spectrum. The question arises, why this
detour?
The inelastic design spectra can be used directly
and they do not require iteration. So, a more
straightforward approach for the determination of
seismic demand is based on the use of the inelastic
strength and displacement spectra, which can be
obtained directly by time-history analyses of
120
inelastic SDOF systems. A simpler alternative is
determination of inelastic demand spectra from
smooth elastic spectra by using reduction factors
obtained from statistical analyses. Following the
idea of Bertero (1995) and Reinhorn (1997), it has
been recently formulated in the AD format (Fajfar
1999, 2000). Inelastic spectra, which have been
applied also by Chopra & Goel (1999), can be used
not only for analysis and performance evaluation,
but also for direct displacement-based design as
indicated in (Fajfar 1999) and demonstrated by
Chopra & Goel (2001). The above arguments
suggest the superiority of the procedures based on
inelastic spectra to those based on equivalent elastic
spectra. They are based on a more solid theoretical
background and they do not require iteration.
Different procedures differ also in the assumed
lateral load pattern, used in pushover analysis, and in
the displacement shape, used for the transformation
from the MDOF to the SDOF system (and vice
versa). Bilinear idealization of the pushover curve is
required for the methods using inelastic spectra. The
procedures employing equivalent elastic spectra do
not need initial stiffness. However, the equivalent
damping is usually based on the ductility, therefore a
bilinear idealization is needed also for these
procedures (partial exception is the approach in the
Japanese code). If the bilinear idealization depends
on the displacement demand, than the computational
procedure becomes iterative even in the case when
inelastic spectrum is used. It is questionable if this
complication is warranted.
For practical applications and for educational
purposes a graphical representation of the procedure
is extremely important. A breakthrough of the
simplified methods was possible when the
acceleration displacement format was
implemented, which allows a visualization of
important demand and capacity parameters (even if
all results can be obtained numerically). It seems
that the use of this format is essential for the
appreciation ofthe procedure in practice.
To summarise: The simplified non-linear
methods, which are, like any approximate method,
subject to several limitations (see, for example,
Fajfar 2002), provide a tool for a rational yet
practical evaluation procedure for building structures
for multiple performance objectives. The
formulation of the methods in the acceleration -
displacement format enables the visual interpretation
of the procedure and of the relations between the
basic quantities controlling the seismic response.
This feature is attractive to designers. In general, the
results obtained using the simplified methods are
reasonably accurate, provided that the structure
oscillates predominantly in the first mode.
4. 4 Cornell 's approach for probabilistic seismic
performance assessment basic concept
At Stanford university the foundation of an approach
for probabilistic seismic performance assessment
was developed (Cornell & Krawinkler 2000). It
represents a generic structure for coordinating,
combining and assessing the many considerations
implicit in performance-based seismic assessment
and design. Recent research in Pacific Earthquake
Engineering Research (PEER) Center as well as in
some other institutions, including University of
Ljubljana, takes advantage of this foundation.
The proposed foundation in its generalized form
presumes that the basis for assessing adequacy ofthe
structure or its design will be a certain key Decision
Variable, DV, such as the annual earthquake loss
and/or the exceedance of one or more limit states
(e.g., collapse). These can only be predicted
probabilistically. Therefore the specific objectives of
engineering assessment analyses are quantities such
as Åicoii, the mean annual frequency ( MAF, which is
approximately equal to the annual probability for the
small values of interest here) of collapse, or a similar
quantity related to loss express in money terms.
Then the practical and natural analysis strategy
involves the expansion and/or "disaggregation" of
the MAF, λ(ϋν), in terms of a structural Damage
Measure, DM, and ground motion Intensity
Measure, IM, which we can write symbolically as:
MDV) = j$G(DV\ DM) dG{DM\ IM) d/\IM) (7)
Here G(DV|DM) is the probability that the
decision variable exceeds specified values given
(i.e., conditional on knowing) that the engineering
Damage Measure (e.g., the maximum interstory
drift) are equal to particular values. Further,
G(DM|TM) is the probability that the Damage
Measure exceeds these values given that the
Intensity Measure (such as spectral acceleration at
the fundamental mode frequency or peak ground
acceleration) equal particular values. Finally λ(ΓΜ)
is the MAF ofthe Intensity Measure.
Note that λ(ΙΜ) is in fact just a seismic hazard
curve, the determination of which is conventional
Probabilistic Seismic Hazard Analysis (PSHA).
Secondly, the estimation of G(DM|TM) is the
objective of linear and nonlinear dynamic analysis or
seismic "demand analysis" under accelerograms
with a given intensity level. Examples of G(DV|DM)
when the DV is a binary damage state indicator
variable are various "fragility curves". In the case of
the more recent HAZUS software it is the likelihood
of a given discrete loss state.
The generic equation (1) is an example of the
Total Probability Theorem of probability theory. It
"de-constructs" the assessment problem into the
three basic elements of hazard analysis, dynamic
121
demand prediction, and failure or loss estimation, by
introduction of the two "intermediate variables",
DM and ΓΜ. Then it re-couples the elements via
integration over all levels of the selected
intermediate variables.
The choice of appropriate intermediate variables
is part of the science and art of successful
engineering analysis. Ongoing research is concerned
with this problem.
Using the above foundation, a methodology that
focuses on specific performance levels (which may
range from continuous operation to collapse
prevention) and an acceptable annual probability of
exceeding these levels (the "performance objective"
approach advocated in present guidelines). This
approach can be based on an LRFD-like format
(capacities > demands). The open question is how
to quantify acceptable performance (or capacities) at
the various levels. Much work has been done in
quantifying drift capacities for collapse, but other
performance levels have not yet been addressed in a
quantitative manner.
4.5 Cornell's approach for probabilistic seismic
performance assessment a case study
The described seismic risk assessment methodology
is illustrated by a case study of a four storey three
bay RC frame with masonry infills. The structure
had been designed to reproduce the design practice
in European and Mediterranean countries about forty
to fifty years ago. However, it may also be typical of
buildings built more recently, but without the
application of capacity design principles (especially
the strong column - weak beam concept), and
without advanced detailing.
In our example peak ground acceleration was
adopted as intensity measure because the fact on
seismic hazard were available in terms of this
quantity (note that use of first mode spectral
acceleration usually results in less dispersion). For
the case study the probabilistic representation of the
seismic hazard at Krãko Nuclear Power Plant
(Slovenia) was adopted (Fajfar et al, 1994). The
hazard curve can be approximated by the
form Ή = k0· a ~k ,where ag is peak ground
acceleration (intensity measure), and k0 and k are
constants depending on the site (in our case
£0=1.1 1E-05 £=2.9). Maximum storey drift was
selected as damage measure. It is a good measure for
damage of structure caused by seismic actions,
especially for structures where storey mechanism is
created. For life-safety performance level fixed
value of limit storey drift C of 2% of storey height
was chosen. It simplifies the capacity model,
because capacity is now a deterministic variable and
its standard deviation ac is equal to zero. Attention
is focused on the demand model. Maximum storey
drifts were calculated with nonlinear dynamic
analyses for different accelerograms and different
values of peak ground acceleration. More about
mathematical model of the structure can be find in
Dolsek & Fajfar (2001). Fourteen accelerograms
were selected from the European Strong-Motion
Database. All acceleregrams were recorded on stiff
soil. Scaled acceleration spectra are presented in
Figure 5. It can be observed that average spectrum
of the selected group of accelerograms is similar to
the EC 8 spectrum.
EC 8, Β
Mean
measure can be obtained. In our example we first
calculated sample mean and unbiased sample
variance. Maximum median drift D\a ) and standard
deviation of natural logs <7D, were determined with
assumption of lognormal distribution of maximum
drift. Finally maximum median drift was
approximated with expression D = a-(ag)b(sQQ
Figure 6), where a and b are constants (for our case
α = 41.1, è = 3.2 and σ·,, =0.455).
Hazard curve, demand curve and capacity value
are coupled by the total probability theorem. Result
is mean annual frequency (MAF) of limit drift
violation P, which can be determined as (Cornell et
al. 2000)

Ρ = #(<£,) -exp \·^\<,+°1)

2 b
HtâfycÁS)
C/3 ,

From the above expression the MAF of the limit

0 0.5 lg), corresponding to median drift capacity C
0.0 0.5 1.0 1.5 2.0
Period (s)
Figure 5. The elastic acceleration spectra scaled to lg.
As a result of non-linear dynamic analyses a set
of curves which link maximum storey drift versus
peak ground acceleration can be constructed (see
Figure 6). Each curve represents several non-linear
dynamic analyses for particular accelerogram and
different values of peak ground acceleration. In our
example fourteen curves are plotted, one for each
accelerogram. After the curves are determined,
statistics of the drift demand given the intensity
drift violation can be obtained simply by substituting
peak ground acceleration capacity a^ (in our case
2.5 3.0
(in our case limit drift equal to 2% of storey height,
5.4cm) (see Figure 6) into the hazard curve, times a
correction factor Cf depending on the slope of
hazard and demand curves, standard deviation of
natural log for maximum median drift given peak
ground acceleration, and standard deviation of
natural log for median drift capacity. Evaluation of
above expression yields the MAF of limit drift
violation Ρ = 7.6 ·10"5 which is much less than the
commonly acceptable mean annual frequency value
Ρο=4·10~4, which corresponds to 2500 years
return period. Note that our prediction is based on
50% confidence level. With higher confidence level
MAF ofthe limit drift violation increases and it may

0.80

0.70
a
gj
0.60

0.50
a
< 0.40
cu
0.30
- -O· · - ·
t
Median Drift Demand according to Po
0.20 - - O - - - Median Drift Capacity

0.10
\ » Median Drift Demand
Fitted Median Drift Demand
· Median Drift Demand +-SD
0.00
If' = 0.79cm 4 C=5Acm 6
MaEximum Drift (cm)

Figure 6. Curves representing maximum drift versus peak ground acceleration for different accelerograms, median drift demand and
fitted median drift demand.

122
happen that for a high confidence level MAF of the
limit drift violation exceeds the allowable MAF. The
computational procedure for confidence levels
different from 50% is more complex and will not be
discussed here. An advantage of the Cornell's
approach is also that the method can be evaluated in
load resistance format design. In this case we
assume allowable MAF of limit drift violation P0 .
From the hazard curve the peak ground acceleration
corresponding toP0(a^/ = 0.29g) was calculated
and from ap°f the median drift corresponding
to P0 (Dp = 0.79cm) was determined (Figure 6).
Load resistance format checking is then expressed
with
<p-C>X-D j°'
f Ik Λ
φ = exp σν
2b
'\k_
A = exp (9)
2ba%
where φ is the capacity factor and λ the demand
factor. In our example φ - 1, because
ac = 0 (capacity was deterministically selected),
and A = 1.10. From expression (9) we obtain
1.0 · 5.4cm > 1.10 · 0.79 = 0.87cm . Demand (0.87cm)
is less than capacity (5.4cm) indicates that the MAF
of the limit drift violation is much less than the
allowable MAF and that the structure is safe.
Although the frame was poorly designed the
structure can be accepted as safe. Strength of the
masonry infills significantly contributes to low risk.
In addition to the infilled structure bare frame was
also studied. In this case risk Ρ = 1.66 · 10"3 exceeds
acceptable one.
4.6 Assessment of seismic behaviour andfragility of
buildings using HAZUS
The probabilistic methodology for seismic
evaluation of existing buildings implemented in
HAZUS (Earthquake Loss Estimation Methodology
- HAZUS - Technical Manual, 1997) assess the
seismic fragility using the "capacity spectrum
method" (Freeman, Nicoletti & Tyrell, 1975).
The step-by-step procedure to assess the seismic
fragility of existing buildings is highlighted in the
following, HAZUS (1997):
1. Create a computer model of the structure;
perform a pushover analysis; plot the roof
displacement - base shear (Aroof - Vì) curve;
2. Convert the capacity curve to the capacity
spectrum; any point Vu Aro0fon the capacity curve is
converted to the corresponding point Sa¡, S<u on the
capacity spectrum;
3. Obtain the elastics demand response spectrum in
ADRS format; the spectrum might be processed
123
from an actual seismic motion or might be a smooth
design spectrum;
4. Plot on the same graph both the demand and the
capacity spectrum;
5. Select a trial performance point on the capacity
spectrum; corresponding to this point one computes
the equivalent viscous damping, β<$(%):
ßeff=ßo + 5 (10)
where ßo is the hysteretic damping represented as
equivalent viscous damping and 5% is the viscous
damping inherent in the structure. The term ßo can
be calculated as, Chopra (1995):
a-'4π E£» so
(H)
where Ed is the energy dissipated by damping and
Eso is the maximum strain energy.
6. The spectral reduction factors, SRa and SRy are
derived. The equations for the spectral reduction
factors are given by (Newmark & Hall, 1982):
SRA=(3.21-0,681n(ßeff))/2. 12 (12)
SRv=(2.31-0.411n(ßeff))/1.65 (13)
7. Reduce the elastic acceleration spectrum
according to the computed equivalent viscous
damping using spectral reduction factors; plot on the
same graph the reduced demand spectrum and the
capacity spectrum.
8. Intersect the capacity spectrum with the reduced
demand spectrum. If the displacement at the
intersection of the demand spectrum and the
capacity spectrum is within 5% of the displacement
of the trial performance point, the selection made in
step 5 is correct and the trial performance point
becomes the actual performance point. If the
intersection of the demand spectrum and the
capacity spectrum is not within the acceptable
tolerance, then a new trial point is selected and the
process is repeated starting with step 5. The
performance point represents the maximum
(spectral) structural displacement, S¿¡, expected for
the demand earthquake.
9. Determine the building fragility functions. The
conditional probability of being in, or exceeding, a
particular damage state, ds, given the spectral
displacement, S<j, is defined by:
V
Ρ[4^] = Φ 1
In (14)
'* \Od,ds j
where:
Sd,ds is the median value of spectral displacement at
wnich the building reaches the threshold of the
damage state, ds,
ßds is the standard deviation of the natural logarithm
of spectral displacement for damage state ds, and
 Φ is the standard normal cumulative distribution 4.7 Monte-Carlo simulation for obtainingfragility
function. function parameters
10. For the maximum (spectral) structural
displacement, S¿, expected for the demand
HAZUS (1997) includes the fragility function
parameters, S¿,ds and ßds appropriate for each type of
earthquake ground motion, one determines the
structural damage state probabilities using fragility
building corresponding to USA practice of design
functions from step 9. and construction. In order to calibrate the fragility
function parameters appropriate for structural
As discussed in Chapter 4.3, the equivalent
systems, which are different from USA practice, the
linearisation approach, based on highly damped
elastic spectra, implemented in the "capacity
Monte-Carlo simulation technique can be used
(Rubinstein, 1981).
spectrum method", can be replaced by approach
based on inelastic spectra. The superiority of the
In damage analysis, the uncertainties associated
later approach is demonstrated in Figure 7, which with seismic demands and structural capacities need
to be modeled. The Monte-Carlo technique as ap¬
presents elastic displacement spectrum, inelastic
displacement spectrum for ductility factor μ = 4, and
plied for the development of fragility function
parameters involves the selection of values of the
highly damped elastic spectrum, determined
input capacity random variables required for
according to ATC-40 procedure employing Eqs. 12
and 13. pushover analyses, the performance of pushover
analyses and the simulation of damage to the
40 structure.
Elastic spectrum
μ=4 The direct Monte-Carlo technique requires a large
35
-- ~ - --"" - number of simulation cycles to achieve an
y
tO acceptable level of confidence in the estimated
NONSPEC
25
/
probabilities. The Latin hypercube technique might
^' be used to reduce the number of simulation cycles.
20 , ..--.. Using the Latin hypercube technique for selecting
·- values ofthe input variables, the estimators from the
15

/ - - · ATC-40
simulation are close to the real values of the
10
/ .-' Inelastic quantities being estimated. The Latin hypercube
5 spectra
technique uses stratified sampling of the input
0 ·--' variables, which usually results in a significant
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 decrease in the variance ofthe estimators.
T.s In brief, the simulation technique implies the
Figure 7. Elastic and inelastic (/=4) displacement spectra - following steps (Vacareanu, 2000):
INCERC (N-S) 1977 seismic motion (Cornea, Vacareanu & - simulation of structural parameters;
Lungu, 2000). -random permutations of structural random
variables;
In the alternative approach (Chopra & Goel, -performing push-over analyses using generated
1999) using inelastic spectra based on strength samples;
reduction factors, the expected spectral displacement - sample statistics of results of analyses.
Sd of an inelastic system with properties T, ßo, and^ The compressive strength of concrete and the
is determined by the following steps (which replace yield strength of steel are, as a minimum, the
steps 5-8 from original methodology): parameters that should be treated as the random
1. A constant-ductility design spectrum is variables. Following (Galambos et al. 1982), a
established by reducing the elastic design normal probability distribution for concrete strength
spectrum by appropriate ductility-dependent and a lognormal probability distribution for steel
factors that depend on T. strength might be used.
2. The capacity curve is plotted on the same graph. ¡3 Storit j

3. The yielding branch of the capacity diagram

__ ._

^'
intersects the demand diagram for several ^
μ values. One of these intersection points, which * .
-μ
remains to be determined, will provide the - μ-α
...

μ+σ

deformation demand. At the one relevant "'·-- .**"

intersection point, the ductility factor calculated 0.04 Λ'

1
from the capacity diagram should match the 0.02

ductility value associated with the intersecting

demand curve.

Figure 8. Capacity curve, Monte-Carlo simulation

124

The outcome of the pushover analyses is
of capacity curves, which can be described
or mean plus/minus one standard deviation capacity
curves, Figure 8.
For calibration of fragility function parameters
stipulated in HAZUS methodology it is necessary to
establish a correlation between Park&Ang (1985)
damage index and interstory drift at threshold of
damage state.
multiplying the interstory drift by the height of the
a family building and by the fraction ofthe building height at
as mean the location of push-over mode displacement.
Once the parameters of fragility function, Sd,ds
and ßds, are obtained one can compute and plot the
functions using equation 14, Figure 10.
0.7

The more recent slightly modified version of the SJO.5

Slight

Park&Ang index, in which the recoverable deforma¬ Moderate

Extenttve

tion is removed from the first term might be used: 03

Compttte
£xpectrd

D = Dm-Dy_ JdE_
(15)
Du-Dy FyDu 20 40

where Dm = maximum displacement; Du =

ultimate displacement; Dy = yielding displacement; Figure 10. Fragility functions, Monte-Carlo simulation
ße - strength deterioration parameter; Fy = yielding
force and E = dissipated hysteretic energy.
The correlation between Park&Ang damage ACKNOWLEDGEMENTS
index and damage state is given in table 1 (Park,
Ang & Wen, 1992):
D. Zamfirescu and A. Stratan at the University of
Table 1. Relations between damage index and Ljubljana was sponsored by European Commission
damage state
within the research training Network Safety
Assessment for Earthquake Risk Reduction
Range of damage index Damage state
None (N)
(SAFERR) Contract No HPRNCT 1999 0035.
D<0.1
0.1<D< 0.25 Minor (Mi)
0.25 < D < 0.40 Moderate (Mo)
0.40 <D< 1.00 Severe (S) REFERENCES
D > 1.00 Collapse (C)
Aschheim, M.A. & Black, E.F. 2000. Yield point spectra for
Using the definition of Park&Ang damage index seismic design and rehabilitation. Earthquake Spectra
and the capacity curve, one can determine the 16(2): 317-35.
ATC 1996. Seismic evaluation and retrofit of concrete
correlation between Park&Ang damage index and buildings: (1), ATC 40, Redwood City: Applied
interstory drift as mean and standard deviation Technology Council.
values, Figure 9. Ayala, A.G. & Tavera, E.A. 2002. A new approach for the
evaluation of the seismic performance of asymmetric
5.0 buildings. Proceedings ofthe fh National Conference on
4.5 Earthquake Engineering, Boston. Oakland: EERI.
4.0 Bertero, V.V. 1995. Tri-services manual methods. In Vison
M
3.5
--- μ-σ η 2000, Part 2, Appendix J. Sacramento: Structural Engineers
3.0
μ+σ Association of California.
ξ 2.5 Bertero, V.V. 1997. Performance-based seismic engineering: A
critical review of proposed guidelines. Seismic design
1.5 methodologies for the next generation of codes. Rotterdam:
1.0
AABalkema.1-31.
0.5
Building Research Institute, Ministry of Construction, Japan,
0.0
1978. Digitized data of strong-motion earthquake
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
accelerograms in Romania (March 4, 1977) by
Park&Ang Damage Index
Observational Committee of Strong Motion Earthquake,
Figure 9. Correlation between Park&Ang damage index and Kenchiku Kenkyu Shiro No.20, January.
interstory drift CEN 1994. Eurocode 8 - Design provisions for earthquake
resistance of structures, European prestandards ENV 1998.
Making vertical sections in Figure 9 for the Brussels: European Committee for Standardization.
threshold values of Park&Ang damage index given CEN 2001. Eurocode 8 - Design of structures for earthquake
resistance, Part 1, European standard prEN 1998-1, Draft
in Table 1 one can identify the mean and standard No.4. Brussels: European Committee for Standardization.
deviation values of interstory drift at threshold of Chopra, A. K. (1995). Dynamics of structures: theory and
damage state. The median value of spectral applications to earthquake engineering. Englewood Cliffs,
displacement at which the building reaches the N.J.: Prentice Hall.
threshold of the damage state, Sd,ds is obtained by
125
Chopra, A.K. & Goel, R.K. 1999. Capacity-demand-diagram
methods for estimating seismic deformation of inelastic
structures: SDF systems, Report PEER-1999/02. Berkeley:
Pacific Earthquake Engineering Research Center,
University of California.
Chopra, A.K. & Goel, R.K. 2001. Direct displacement-based
design: use of inelastic vs. elastic design spectra.
Earthquake Spectra 17(1): 47-64.
Chopra, A.K. & Goel, R.K. 2002. A modal pushover analysis
procedure for estimating seismic demands for buildings.
Earthquake Engineering and Structural Dynamics 31: 561-
82.
Cornea, T., Vacareanu, R., & Lungu, D. 2000. Technical
Report for Collaborative Research Center (CRC) 461 of
SFB, Germany; "Strong Earthquakes: A Challenge for
Geosciences and Civil Engineering", Karlsruhe University,
64 p.
Cornell, C.A. & Krawinkler, H. 2000. Progress and Challenges
in Seismic Performance Assessment. PEER Center News
(4)1:1-3.
Cornell, C.A. et al 2000. Seismic Reliability of Steel Frames.
Proc. Of 9* IFIP WG 7.5 Working conference on
Reliability and Optimization of Structural Systems.
September 2000, Ann Arbor, MI.
Dolsek, M. & Fajfar, P. 2001. Soft storey effects in uniformly
infilled reinforced concrete frames. Journal of Earthquake
Engineering (5)1:1-12. Imperial College Press. London.
Fajfar, P. & Fischinger, M 1987. Non-linear seismic analysis of
RC buildings: Implications of a case study. European
Earthquake Engineering 1(1): 31-43.
Fajfar, P. & Fischinger, M. 1989. N2 - A method for non¬
linear seismic analysis of regular buildings. Proceedings of
the 9* World Conference on Earthquake Engineering,
Tokyo, Kyoto; Tokyo: Maruzen, 5:111-16.
Fajfar, P. & GaäperSio, P. 1996. The N2 method for the seismic
damage analysis of RC buildings. Earthquake Engineering
and Structural Dynamics 25:23-67.
Fajfar, P. & Krawinkler, H. (Editors) 1997, Seismic design
methodologies for the next generation of codes. Rotterdam:
AABalkema, 1997.
Fajfar, P. 1999. Capacity spectrum method based on inelastic
demand spectra. Earthquake Engineering and Structural
Dynamics 28: 979-93.
Fajfar, P. 2000. A nonlinear analysis method for performance-
based seismic design. Earthquake Spectra 16(3): 573-92.
Fajfar, P. 2002. Structural analysis in earthquake engineering -
a breakthrough of simplified nonlinear methods. In:
Proceedings of the 12th European Conference on
Earthquake Engineering, London, Kynote paper.
Fajfar, P. et al 1994. Probabilistic assessment of seismic hazard
at Krsko Nuclear Power Plant. Revision 1 prepared for
Nuclear Power Plant Krsko.
FEMA 1997. NEHRP guidelines for the seismic rehabilitation
of buildings, FEMA 273, and NEHRP Commentary on the
guidelines for the seismic rehabilitation of buildings,
FEMA 274. Washington, D.C.: Federal Emergency
Management Agency.
FEMA 2000. Prestandard and Commentary for the Seismic
Rehabilitation of Buildings, FEMA 356. Washington, D.C.:
Federal Emergency Management Agency.
FEMA 2001. The NEHRP Recommended provisions for new
buildings and other structures, FEMA 368. Washington,
D.C.: Federal Emergency Management Agency.
Freeman, S.A. 1998. Development and use of capacity
spectrum method. Proceedings of the &h U.S. National
Conference on Earthquake Engineering, Seattle, CD-ROM,
Oakland: EERI.
Freeman, S.A., Nicoletti, J.P. & Tyrell, IV. 1975. Evaluations
of existing buildings for seismic risk A case study of
Puget Sound Naval Shipyard, Bremerton, Washington.
126
Proceedings of the Is' U.S. National Conference on
Earthquake Engineering, Berkeley: EERI.l 13-122.
Galambos, T.V., Ellingwood, B., MacGregor, J. G., & Cornell,
C. Α. 1982. Probability based load criteria: Assessment of
current design practice. Journal of Structural Division,
ASCE, vol. 108, No. ST5, 959-977.
Hamburger, R.O. 1997. Defining performance objectives.
Seismic design methodologies for the next generation of
codes. Rotterdam: AA Balkema. 33-42.
HAZUS - Technical Manual 1997. Earthquake Loss Estimation
Methodology, 3.
Krawinkler, H. 1997. Research issues in performance based
seismic engineering. Seismic design methodologies for the
next generation of codes. Rotterdam: AA Balkema. 47-58.
Moghadam, A.S. & Tso, W.K. 2000. 3-D pushover analysis for
damage assessment of buildings. Journal of Seismology and
Earthquake Engineering (Tehran) 2(3): 23-31.
Newmarlc, N. M., and Hall, W. J. (1982). Earthquake Spectra
and Design. Berkeley, Calif. Earthquake Engineering
Research Inst.
Otani, S., Hiraishi, H., Midorikawa, M. & Teshigawara, M.
2000. New seismic design provisions in Japan. Proceedings
2000 Fall ACI Convention, Toronto.
Park, Y.-J. & Ang, A.H.-S. 1985. Mechanistic seismic damage
model for reinforced concrete. Journal of Structural
Engineering. ASCE. 111(4): 722-739.
Porter, M.E. 1998. Competitive Strategy: Techniques for
analyzing industries and competitors. Simon & Shuster
Trade.
Reinhorn, A.M. 1997, Inelastic analysis techniques in seismic
evaluations. In: Fajfar P, Krawinkler H. Editors. Seismic
design methodologies for the next generation of codes.
Rotterdam: AA Balkema, 277-87.
Rubinstein, R.Y. 1981. Simulation and the Monte-Carlo
method. John Wiley & Sons, New York.
Saiidi. M. & Sozen, M.A. 1981. Simple nonlinear seismic
analysis of R/C structures. Journal of Structural
Engineering, ASCE: (107): 937-52.
SEAOC 1995. VISION 2000, Performance based seismic
engineering of buildings. Sacramento: Structural Engineers
Association of California.
SEAOC 1999. Recommended lateral force requirements and
commentary. Sacramento: Structural Engineers Association
of California.
Vacareanu, R. 2000. Monte-Carlo Simulation Technique for
Vulnerability Analysis of RC Frames - An Example.
Scientific Bulletin on Technical University of Civil
Engineering of Bucharest, 9-18.
Zamfirescu, D. & Fajfar, P. 2001. Comparison of simplified
procedures for nonlinear seismic analysis of structures. In:
Proceedings of the 3rd US-Japan Workshop on
performance-based earthquake engineering methodology
for reinforced concrete buildings, Seattle, Washington.
Structural integrity of buildings under exceptional earthquakes

E.S. Mistakidis
Dept of Civil Eng. University of Thessaly, 38334 Volos, GREECE
V. Bosiljkow
Slovenian National Building and Civil Engineering Institute, SLOVENIA
G. DeMatteis
of Structural Analysis and Design , University of Naples Federico II , 80125 Napoli - ITALY
Dept.
F. Dimi
Romanian Academy, Centre of Advanced and Fundamental Technical Sciences, Timisoara, 1900, ROMANIA
M. Fischinger
Dept. of Civil and geodetic Eng, University of Ljubljana, Jamova 2, SI- 1000 Ljubljana, Slovenia

ABSTRACT: The paper tries to identify the effect of exceptional earthquake loading on Civil Engineering
structures. Several reasons leading to exceptional action are identified and their effect on the structures is
discussed. The main problems and the strategies to address them are summarized. The paper summarized the
latest developments concerning the characteristic factors affecting the structural integrity and its evaluation.
Three different cases are considered depending on the specific structural material (concrete, steel, masonry).
Finally, conventional and non-conventional methods for the optimization ofthe structural response under the
exceptional earthquake action are summarized.

1 INTRODUCTION problems introduced in the first part of the paper are

An exceptional action is an action that has not been
considered in the initial design of the structure.
Although, in principle, this condition should be
avoided, it is a fact that in many cases the structures
are asked to undertake larger forces than the ones for
which they were designed. The reasons for that are
connected with the character of the action (snow,
wind, live load, seismic load) but also with the
structural system itself, as it will be explained later.
The paper tries to summarize the reasons leading to
exceptional seismic action on structures and to give
some material dependent solutions to the structural
deficiencies that might occur.
The first part of the paper contains material
introductory to the notion of exceptional seismic
action on structures. The reasons than can lead to
seismic loading not taken into account during the
initial design process are summarized. The results of
the exceptional action are identified and general
procedures for the evaluation of the structural
integrity are given. This part summarizes also the
general methodologies used for the optimization of
the structural response.
The next sections are related to specific materials
(concrete, steel, masonry). Material related reasons
leading to poor structural response are presented as
e.g. poor detailing in concrete structures, brittle
failure of concrete elements (shear), poor moment-
resisting connections in steel, brittle fracture of
welds in steel, combination of high stiffness and low
strength in masonry structures, etc.. Also, material
related procedures for the anticipation of the
127
presented, as e.g. specialized procedures for the
optimization of structural performance under
earthquake action.
2 REASONS LEADING TO EXCEPTIONAL
EARTHQUAKE ACTION ON STRUCTURES
The case of earthquake forces on structures is a
rather characteristic case where an action can be
exceptional. It is admitted that there exists a high
probability that the value of the seismic forces will
at some time exceed the value prescribed in the
design. This fact is related to the inherent
uncertainty nature of the seismic action but also to
incomplete or inadequate knowledge of the
structural behavior at the time of the design of the
structures. In the following, several reasons for
which a structure might be subjected to exceptional
seismic forces are identified.
A very large number of structures were designed
with a regulative framework, which today seems to
be insufficient. This is for example the case in
Greece, where the lateral forces prescribed by the
later seismic codes are significantly higher than the
ones prescribed by the seismic codes before the year
1985. Of course, this happens because, in the
meantime, significant scientific knowledge has been
embodied in the seismic codes. This knowledge has
to do with the scientific field of seismology but also
with the comprehension of the behavior of the
structural materials and their combinations. For
example, during the last 20 years, a huge number of
tests have been performed in concrete, steel,
composite, aluminum and wood specimens that have
allowed the development of rational procedures for
the calculation ofthe strength and the ductility ofthe
various structural elements.
Also, due to the developments in both hardware
and software that took place during the last 20 years,
the design engineers are capable to perform more
sophisticated seismic analyses, something that has
also been reflected in requirements of the seismic
codes.
From the viewpoint of the seismology, it is well
known that it is difficult to obtain exact values for
the seismic actions. But recently, there have been
recognized several reasons for which the design
forces should be increased in certain cases. For
example, during the past 20 years, a significant
number of recorded strong motion data has indicated
that the characteristics of the ground motion vary
significantly between recording stations. This
phenomenon is magnified for stations located near
the epicenter. As a result, two main regions with
different types of ground motions are considered
(Gioncu et al 2000).
The near-source region that is the region within
few kilometers of either the surface rapture or
the projection on the ground surface of the fault
rapture zone.
The far-source region situated at some hundred
kilometers from the source.
Unfortunately, the characteristics of the design
spectra and the design methods adopted by the
majority of the seismic codes have been based on
records obtained by far-source fields. Therefore,
they are incapable to describe the seismic intensity
in the near-source region. Lately, the so called "near
source factor" has been introduced in some
advanced seismic codes in order to take into account
the amplification of the earthquake forces in the
near-source region (ATC-40 1996, SEAOC 1999).
Moreover, the vertical component of the seismic
action in near-source field could be greater than the
horizontal ones. Notice that, until now, this vertical
component is only in special cases taken into
account in the design of buildings. Also, in near-
source areas, due to the very short periods of the
ground motion and the pulse characteristics of the
loads, the significance of higher vibration modes
increases. Due to the pulse characteristics of the
actions, developed with great velocity and especially
due to the lack of restoring forces, the ductility
demands could be very high.
All the above have been dramatically verified
after the Kobe earthquake. The observations after
this earthquake show, by both ground motion
recordings and observed damage to buildings, that
earthquake loading conditions in the near-source
128
region subject buildings to more severe conditions
than previously assumed.
Another aspect of the seismic design whose
significance has been recognized only during the last
decades is connected to the ground conditions. It is
now well known that the properties of the site soils
affect the intensity of shaking that can be expected at
the building site. Various parameters such as the
thickness of the soft and stiff soil layers, the shear
wave velocities of the rock and soil layers, the
soil/rock impedance ratio, the layering properties of
the soil layers etc. influence the amplification or
attenuation of the seismic action on the structures.
Moreover, landsliding, liquefaction and surface fault
rapture has led to catastrophic results on buildings
during the last decades and their significance has
been only recently introduced and quantified in
some seismic codes (ATC-40 1996).
Another reason leading to exceptional
accelerations on structures (i.e. accelerations greater
than the design ones) is connected with
magnification that sometimes occurs in the short
period range. This was for example the case in the
earthquake that hit Athens in the September of 1999.
This was a moderate intensity earthquake (5.9 on the
Richter scale) that resulted in several collapses and
deaths and a lot of damages in a large area. The
accelerations recorded in various neighboring sites
led to acceleration spectra, which presented a peak
in the short period range (see Figure 1). This peak
acceleration occurred in a period of 0.25 sec and was
about 2 to 3 times greater than the design one for the
specific site. As a result, a lot of rather stiff and
limited ductile structures (low-rise buildings, some
of them designed according to the latest seismic
code) were significantly damaged. On the other
hand, high-rise buildings with a dominant period
greater than 0.4 sec appeared no damages at all.
ε
<
tu
α.
P«r!od, Τ(«·<0
Figure 1. Acceleration response spectra (horizontal
components) of the September 7, 1999 earthquake in Athens
and elastic design spectra ofthe Greek Seismic Code.
The distinction of the seismic zones defined in the
seismic codes is based on historical earthquake data.
However, various difficulties arise concerning the
determination of the faults that gave significant
earthquakes during the historical ages. The most
important difficulty is connected with the existence
of extensive sea-covered areas in active seismic
areas. In these areas, the faults that generate
earthquakes cannot be explored with the usual
geological methods. Another one is based on the fact
that the biggest earthquakes in Europe are not very
strong (usually less than 7.0 in the Richter scale) and
for this reason, the surface indications of the faults
are relatively small and therefore hardly
recognizable. Therefore, some faults or the potential
of existing faults for the generation of strong
earthquakes, are identified only after giving new,
relatively strong earthquakes. This was for example
the case in Greece where a strong earthquake
(Kozani-Grevena, 1995, 6.6 in the Richter scale)
occurred in a region characterized in the lowest
earthquake hazard zone. After this event, this area
was classified in a higher seismic hazard zone,
therefore, all the structures build before 1995 in this
area may be subjected in an exceptional earthquake
as they were designed for a lower value of the
probabilistic acceleration.
It should be also realized that seismic design
philosophy has dramatically changed in the past
decade. Most traditional codes have been
concentrated on life save performance during large,
infrequent earthquakes, accepting significant
damage as long as the collapse is prevented.
However, there has been growing evidence (e.g.
Northridge and Kobe) that such, frequently
uncontrolled damage is not tolerated by the society.
Consequently, new performance objectives
demanding operational structures after fairly large
earthquakes impose totally new requirements that
have been seldom or never considered in the
traditional design.
Except the above, there are also reasons more
closely connected to the structural system, for which
a structure might be submitted to an exceptional
earthquake action. The most important is the fact
that after moderate earthquakes, some damage
occurs in the structure. In the case that some
elements are significantly damaged, a procedure for
repair is usually initiated. But, in the case that the
damage is hidden or underestimated, the structure
remains with some of its strength reduced; therefore,
the next seismic motion will find it in a condition
not considered in the design. Usually, this kind of
damage is mainly connected with the degradation of
stiffness. Studies that have been performed during
the last years on this topic (FEMA307 1998)
indicate that the main effect of prior damage to the
seismic response of a structure is the increase of the
displacement (and therefore the ductility) demands.
However, it was found that a relatively limited prior
damage does not play a dominant role on the overall
seismic response.
129
Finally, arbitrary change of the usage of the
structure should be mentioned as a case where the
structure is submitted to forces for which it has not
been designed. For example, a building designed for
usual importance and later used as a hospital, may
not meet the advance performance requirements for
medical buildings. Also, the conversion ofthe floors
of an ordinary building to storage compartments
may lead to exceptional seismic action on it.
3 EVALUATION OF THE STRUCTURAL
INTEGRITY
Cases as the ones described in the previous
paragraph can be identified in almost any
seismically active area and there is a strong need for
some measures against this unfavorable situation. In
this respect, a close examination of the situation is
needed to identify the reasons leading to the possible
exceptional seismic action on the structure. In
general, the treatment of such a case requires the
steps described in the following.
The first step is connected with the determination
of the seismic action. As it has already been
mentioned, modern seismic codes provide
information about the consideration of the various
unfavorable effects. For this reason, data on soil
surface and subsurface conditions at the site shall be
obtained from existing documentation, or from a
program of site-specific subsurface investigation. In
any case, the seismic hazard should be identified and
quantified using the latest scientific knowledge.
In the second step, information should be gathered
about the specific structure. A meaningful structural
analysis of the probable seismic behavior of the
building requires good understanding of the existing
components and their interconnection. Therefore, the
configuration of the structural system, as well as the
type, detailing, connectivity, material strength and
condition of the structural elements comprising the
building shall be determined. It must be noticed that
the strength and deformation capacity of existing
components should be computed based on derived
material properties and detailed component
knowledge. As the conclusion that an existing
structure does not meet some specified criteria can
have considerable consequences, it is important that
the evaluation is based on the best available
information ofthe properties ofthe materials and the
components, rather than on very conservative
assumptions. Existing component strengths must be
determined for two basic purposes: to allow
calculation of their ability to deliver load to other
elements and components, and to allow
determination of their capacity to resist forces and
deformations.
Finally, an analysis of the building should be
performed in order to determine the forces and the
deformations induced in the various structural
components by the seismic action. Here, the
following methods should be used.
Static or dynamic elastic analysis. These are the
most classical methods of analysis, however,
concerning the analysis of existing buildings,
various problems may occur. Although an elastic
analysis indicates where first yielding will occur,
it cannot predict failure mechanisms and account
for the redistribution of forces during
progressive yielding. For this reason, the results
of the linear procedures can be very inaccurate
when applied to buildings with highly irregular
structural systems, unless the behavior of the
building is nearly elastic.
Static nonlinear analysis. This is generally a
more reliable approach for the assessment of the
behavior of a structure than are linear
procedures. However, it is not exact, and cannot
accurately account for changes in dynamic
response as the structure degrades in stiffness or
account for higher mode effects.
Dynamic nonlinear analysis. This is the most
accurate analysis, however, it is at present
extremely time consuming. It involves time
history analysis of a three-dimensional
mathematical model using simultaneously
imposed consistent pairs of earthquake ground
motion records along each ofthe horizontal axes
ofthe building.
It should be emphasized that the inelastic analysis
procedures should be preferred as they permit the
comprehension ofthe behaviour ofthe buildings, the
identification of the failure modes and the
quantification of the potential for progressive
collapse. However, it must be noticed that they
should be applied only by engineers experienced in
nonlinear analysis. As a balance between simplicity
and applicability the static nonlinear analysis is
promoted by various codes (FEMA273 1997,
FEMA274 1997, ATC-40 1996, EC8).
Finally, the results of the analysis should be
compared with the calculated force and deformation
capacities of the structural members in order to
verify the structural integrity under the considered
seismic action. For this reason, the performance
level should be defined, i.e. the requirements of the
engineer should be quantified (see also the article
"General methodologies for evaluating the structural
performance under exceptional actions" in the same
Volume). Depending on the results of this
comparison, the following main deficiencies may be
identified.
Lack of stiffness
Lack of strength
Lack of ductility and dissipative capacity.
The following measures can be understood as a
general description of the options provided to the
130
design engineer for the optimization ofthe structural
behavior, according to the deficiencies identified by
the analysis procedure.
Strengthening and stiffening. These are the most
common methods for the enhancement of the
seismic behavior of structures. They are applied
mainly in cases where the lateral force resisting
system is inadequate. The strengthening results
to higher values of the lateral force that initiate
damage events in the structure. However, the
elastic characteristics of the structure do not
change significantly. Stiffening changes the
fundamental period of vibration. The modes of
vibration might change as well. System
stiffening and strengthening are in most cases
performed as concurrent strategies, since most
systems that will strengthen a structure also
simultaneously stiffen it. Similarly, most
stiffening techniques usually increase the
strength as well. Typical systems used for
strengthening and stiffening include the addition
of new elements such as shear walls, braced
frames, buttresses, etc. and also the
strengthening of existing elements, e.g. column
jacketing, strengthening of beams through
external reinforcing materials (steel or
composites), etc.
Improvement of the deformation capacity. This
method can be applied in order to improve the
seismic performance of systems that include
brittle elements. Methods of enhancing
deformation capacity include the addition of
confinement to existing elements (through shear
jackets or fiber reinforced polymeric materials),
strengthening of columns to avoid soft stories,
etc. This methodology is mainly applied for the
optimization of the behavior of concrete
buildings and is effective when the required
modifications involve a relatively small number
of elements.
Base isolation. This approach requires the
insertion of flexible bearings at a single level of
the vertical load carrying system of the building.
Typically, these bearings are placed near the
base and are designed to have relatively low
stiffness and extensive lateral deformation
capacity. In some cases the bearings may also
have superior energy dissipation characteristics.
The installation of an isolation system results in
a substantial increase of the building's
fundamental response period and subsequently to
a dramatic reduction of the seismic forces
applied on the building. As the isolation bearings
have much greater flexibility than the structure
itself, the seismic deformations tend to
concentrate on the positions in which they have
been installed. Together these effects result in
significantly reduced deformations on the part of
the building above the level of the isolation
 bearings. Base isolation has most commonly comparison with some traditional structural systems.
used in the past for the optimization of the Unfortunately, many past events (e.g. Figure 2) have
seismic performance of historical buildings, warned that this should not be taken as granted.
because it usually allows the substantial Heavy damage of RC structural systems occurred
reduction of the interventions in the since the demand was greater than expected and/or
superstructure. the structural capacity and integrity was not
There are several base isolation systems, which sufficient. Both factors are briefly analyzed below.
are based on: viscous fluid dampers, sliding
While all general conclusions regarding the
bearings, lead rubber bearings, sliding bearings
reasons leading to exceptional seismic action on
with viscous fluid dampers, elastomeric bearings
structures (Chapter 2) apply, some additional
with controllable fluid dampers, etc. In the paper
of Chang and Makris (2000), the efficiency of considerations relate to RC buildings.
these dissipative mechanisms has been Design seismic action and related seismic force
investigated. An alternative to these classical reduction (behavior) factors have been more or
solutions is the use of the shape memory alloys less empirically determined for the average and
(SMA). SMA' s are special alloys, which can be reasonable regular representatives of typical RC
made of various metals: Copper, Zinc, structural systems (like reasonably regular cast-
Aluminium, Nickel, Titanium, etc.). The new in-situ frames for example). However, it should
system based on this material is capable to be realized that for various reasons many
provide a wide range of performances (Dolce buildings do not possess the anticipated average
and Marnetto 2000). However, despite this dissipation capacity, which has been traditionally
advances offered by base isolation, in recent considered in the design. Therefore larger
years the actual efficacy of this technology has (exceptional) seismic forces should be used to
been also questioned by several seismologists, verify such buildings.
essentially because base-isolated buildings result
to be vulnerable to large pulse-like ground
motions generated at near-fault locations, due to
the large displacement demand occurring for
isolators. In order to reduce these displacements,
supplementary dampers have been suggested.
Nonetheless, existing results show that the use of V'
supplemental dampers in seismic isolation is a
misplaced effort and alternative strategies to
solve the problem should be implemented (Kelly
1999).
Addition of supplemental damping through
energy dissipation systems. These systems
increase the ability of the structure to dampen
the earthquake response through either viscous
or hysteretic damping. This approach requires
the installation of energy dissipation units within
the lateral force resisting system. These units
dissipate energy and therefore, reduce the
seismic displacements of the structure. The
installation of these units often requires also the
installation of vertical bracings to serve as a Figure 2. disintegration frame (Kocaeli
mounting platform for the units and therefore, earthquake, 1999)
typically results in a simultaneous increase in
system stiffness. Energy dissipation systems This is particularly true for specific or less
typically have a greater cost than conventional investigated existing systems (i.e. frames with
systems for stiffening and strengthening but have infills, prefabricated industrial buildings and
the potential to significantly enhance the even structural walls) as well as for systems not
behavior ofthe structures. designed for earthquake loads in the past. An
example of numerous collapses of limited ductile
infilled frames during the Kocaeli/Turkey
4 REINFORCED CONCRETE
earthquake is illustrative. Apparently regular
structures lost the infills in the lower stories at
Engineered reinforced concrete (RC) buildings are
the very beginning of the earthquake (Figure 3).
in general considered earthquake resistant in
131
 Consequently they behaved like soft-story
frames, which was never anticipated in the
design (Dolsek, 2001). Such increased local
ductility demand could not be withstood with
very poorly detailed columns.
Even more so, such considerations apply for
innovative systems, where no empirical evidence
exists. Special studies should be made for such
systems to evaluate their dissipative capacity and
related level ofthe design earthquake loading.
It should also not be forgotten, that typical
values for seismic force reduction factors
(behavior factors) of RC structures are based on
the anticipated global ductility in the range of
approximately 3 to 6. The related local ductility
demand may well exceed the number of ten.
Such large deformations of RC members are Figure 4. Combined shear axial compression due to
inevitably associated with large cracks and the lack of confinement

severe damage. There has been growing

evidence that such damage is not tolerated by the
society. This actually implies that larger
(exceptional) earthquake action than the one
considered in traditional design should be used
to verify if the damage is within the acceptable
limits.
> itV
1/
r. · disintegration in RC structures are shear failure, lack
Λ'-*··"1·::·
- ·.*
Figure 3. Soft story mechanism formed after the fall out ofthe
infills in the lower two stories (the situation never considered
in the design)
Fortunately, most well designed RC buildings
posses enough dissipative capacity and overstrength
that they maintain their structural integrity even in
the case of exceptional earthquake loading. For
example, most buildings in Kobe, built after 1981
survived well the earthquake, which was much
stronger than anticipated.
In short, structural integrity of RC structures
depends on two compatible yet very different
materials - ductile steel and brittle concrete, as well
as on the bond between them. In the case of
earthquake loading, structure's integrity can
therefore be ensured if (a) favorable global
mechanism develops, (b) ductile reinforcement
governs the response of elements, (c) bond between
the two materials is maintained even in the case of
repeated cyclic loading.
On the other hand, in addition to the bad
conceptual design (i.e. soft story), which is a
common cause of failure for all structural systems
and materials, the most typical causes of structural
of confinement (Figure 4), and bond failure in
anchorage zones (i.e. in beam-column joints; Figure
5). Other examples can be found in EASY -
Eartquake Engineering Slide Information System
(Fischinger, 1997).
The question arises what analysis is needed to
identify such risks in the case of exceptional
earthquake loading? It is clear that standard elastic
analysis based on the behavior factor as a single
empirically defined parameter to account for seismic
energy dissipation, is not capable to provide an
adequate answer.

132

Vv
1 !
-. / general micro models were somewhat more
// ''
} λf 1 *,
Figure 5. Disintegration of a poorly detailed beam-column joint
Still in the range of elastic analysis, the capacity
design procedure offers a viable solution, ensuring
that ductile flexural response is achieved by
inhibiting non-ductile deformation modes. Most
typical application in RC structures is to preclude
shear failure by suitable design and detailing of
ductile flexural elements. Since the shear demand (in
general any demand on the non-ductile mechanism
or element) is limited below shear capacity
regardless the level of the earthquake loading
applied, this method seems to be perfect in the case
of exceptional action. However, the problem is to
know, if the capacity of the protecting mechanism
(i.e. the rotational capacity of the RC beam) is large
enough to survive the exceptional action. Very few
countries possess a suitable data-base of their
specific structural systems to solve this problem.
Therefore the most recent solutions in the
advanced codes have moved towards new inelastic
design methodologies. However, although relatively
simple static (push-over) methods have been
introduced, the question arises if at the present the
modeling of inelastic behavior of RC structures is
good enough to serve the purpose. Even in the pre-
yielding stage the stiffness of the cracked section is
difficult to evaluate, not to mention the complex
behavior of confined concrete and bond-slip
mechanisms in a post critical stage near failure.
In principle, two major groups of models exist.
Micro models analyze inelastic stress-strain response
of (confined) concrete and steel and are sometimes
considered as "exact" models. Macro models consist
of a finite number of discrete springs following
prescribed force-displacement relationships. Which
group is to be chosen, depends on the specific
problem. Among plentiful analyses, it is worth to
mention the international benchmark study CAMUS
(Combescure, 2000). The blind prediction of the
seismic response of a RC cantilever wall was made
for four consecutive earthquakes, among which at
least two could be considered as exceptional. Eleven
participants of the benchmark study predicted the
»Uc1 response with varying success. A number of
predictions can be considered (surprisingly) good. In
successful (if the problems like local failure criteria
and bond slip were adequately addressed). However,
^ some predictions with macro models were also good
or at least acceptable (Fischinger, 2002), giving hope
that such models could serve the purpose in the case
V' ' of more complicated real buildings.
5 STEEL STRUCTURES
5. 1 Characteristic factors affecting the structural
integrity and its evaluation
5.1.1 General
Also in case of steel structures, seismic events
recently occurred worldwide produced unacceptable
damages and economical losses, destroying a
number of buildings and bridges (De Matteis et al.
2001). In particular, the 1985 magnitude 7.3
Michoacan (Mexico City) earthquake evidenced the
first collapse of an important high-rise steel
building. Also, in magnitude 6.7 Northridge (Los
Angeles) earthquake more than 100 low- and high-
rise steel buildings suffered unacceptable damages
and exhibited fracturai failure modes. Similarly, in
the magnitude 7.2 Hyogoken-Nanbu (Kobe)
earthquake severe damages were detected in a
number of modern steel constructions. Two main
remarks have to be pointed out: (1) Most of such
structures were built within the previous 10 years
and therefore were designed according to recent
design and constructional methods for seismic-prone
Countries; (2) Damages interested a number of
welded beam-to-column connections of moment
resisting framed buildings, emphasising poorness of
basic material and detailing.
In the whole, these events clearly shown that even
a high-ductile material like steel could suffer brittle
failure modes and something in customary design
assumptions and construction technique had to be
revised. Akiyama (2000) attributed the causes of the
above damages to a combination of the following
factors: poor quality of steel, poor welding, error in
estimate of inelastic deformation capacity and
133
excessive seismic input. In other words it could be
concluded that the natural hazard related to
earthquake was too high, high intensity earthquake
actually, representing an exceptional loading
condition for current steel building and bridges. This
is particularly true for near-source earthquakes due
to the effects of strain rate and superior vibration
modes, which are not correctly taken into account by
the present design codes (Gioncu and Mazzolani
2002).
As a matter of the fact, aiming at guarantying the
structural integrity of steel structures against
earthquake attacks, several laboratory and field
investigations as well as comprehensive research
programs have been undertaken worldwide,
promoting several detailing and design
improvements. In particular, it has to be mentioned
that a significant contribution to the improvement of
the knowledge on the seismic response of steel
structures has provided by the RECOS INCO-
Copernicus European Research Project (Mazzolani
2000). A number of aspects have been clarified, with
special emphasis given to (1) beam-to-column joint
behaviour, (2) detrimental effect due to high strain
rate and (3) detrimental effect due to welding, but a
number of other aspects remain not yet completely
solved, the present scientific research on this topic
being still now very strong-minded.
5.1.2 Ductility criteria
In case of high-intensity earthquakes, the main
concept which recent code provisions are based
upon is that the actual strength capacity of structures
is lower than the one that could be developed if the
structure itself behaved linearly elastic. Then,
structures resist earthquake due to structural ductility
and energy dissipation worked out by some crucial
structural elements undergoing large plastic
deformations. As a consequence, the performance of
the whole structure is strongly dependent on the
behaviour of these crucial elements and the
dominant criterion for assessing the collapse of the
systems is the ductility demand to ductility capacity
ratio of the energy-dissipation zones. Ductility
demand is strongly influenced by the structural
typology and relative results are highly sensitive to
adopted structural models. Ductility capacity
essentially depends on the structural component
itself, but is also related to the typology of the
applied loading. In fact, local ductility (defined at
level of members and joints), which is already
difficult to be evaluated under monotonie loading,
may be strongly affected by the type of seismic
ground motion, essentially because: (1) cyclic action
produces deterioration of mechanical properties due
to repeated plastic excursions, (2) dynamic effects
134
induced by high peak velocities affects the ductility
of the material and may induce brittle fracture
modes. A consistent procedure for ductility checking
of steel structures in seismic areas is presented in
Mazzolani and Gioncu (2001).
5.1.3 Beam-to-column connections
In case of steel moment resisting frames, the seismic
response of the whole structure is severely
influenced by the behaviour of beam-to-column
connections, where plastic hinges are likely to occur.
Hence, the assessment of the seismic performance is
essentially based on the ratio between rotation
demand imposed by earthquake and the rotation
capacity available in connections.
In order to assess the available rotation capacity of
connection details used in practice, accounting for
low-cycle fatigue, several tests have conducted in
recent years and fatigue (S-N) curves have been
proposed (Calado et al. 1998). Among many others,
excellent overviews on the relevant experimental
evidence are presented in (Engelhardt & Sabol 1997,
Tsai & Popov 1997, Mazzolani 2000, Roeder 2000).
In particular, as an output of these investigations it
has been arisen that there are several important
issues that have a strong effect on the ductility of
practised moment connection configurations under
cyclic dynamic actions, such as the quality and detail
of welding, stress concentration due to the adopted
connection detail, concrete slab, randomness of
actual mechanical properties, high strain rate.
In order to overcome failure of joints under
seismic loading, recent specifications are aligning at
addressing higher benchmark values of available
rotation capacity for connections. As a consequence,
procedures for improving connection behaviour so
to allow for this higher ductility have been proposed
and implemented in recent construction practice and
code provisions. They include improved welding
practice (higher toughness welding electrodes,
modification of size and shape of weld access holes)
and more accurate details, as well as different
structural configurations, the latter being essentially
based on two complementary strategies: (1)
weakening of the beam cross section at the joint
location; (2) straightening of the joint. The
feasibility of these new-proposed connection
typologies has been shown and discussed by several
Authors and for instance by Suita et al. (2000), while
a general study on their influence on the global
seismic performance of moment resisting steel
frames is presented in Anastasiadis et al. (1999).
Anyway, the correct prediction of moment
connection behaviour, both bolted and welded type,
still represents a critical issue within the scientific
debate, it being worthy of further research activity
and laboratory investigation.
5.1.4 Effect of welding
The poor behaviour of welding was one ofthe main
causes of brittle fracture of connections during the
last important seismic events (Northridge, 1994 and
Kobe, 1995). This has favoured the development of
a number of experimental and analytical
investigations on the performance of welded
components (Mazzolani & Gioncu 2001).
Among the possible causes of brittle fractures in
welded joints the following ones have been
identified:
workmanship (welding defects);
detailing (stress concentration at the root or the
toe of welds);
design practices (larger beam and column sizes
than those tested);
poor welding practice (low-toughness weld
metal, poor quality control);
unusually high seismic input (high strain rates),
Various experimental and numerical
investigations (Dexter et al (2000) Mao et al (2001),
Dubina et al (2001)), show that, from the several
factors affecting the welding behaviour, weld access
hole and welding technology have the highest
impact.
5.1.5 Effect of strain rate
As was previously presented in the introduction of
this chapter, one of the main causes for the
unsatisfactory behaviour of steel structures during
the last great earthquakes was the poor ductility of
basic material and detailing. It is well known that the
mechanical properties of steel, such as yield strength
and ultimate strength and thus the ductility vary with
strain rate. The loading-rate effect during an
earthquake was considered negligible, especially for
earthquakes occurred before Northridge and Kobe
events, where moderate velocities were recorded.
However, after these very important and special
earthquakes, when the recorded velocities have been
very high, many specialists consider that the
loading-rate may be a possible cause of the
unexpected bad behaviour of steel structures.
The first research work concerning the effect of
strain rate on the behaviour of metals performed
during the third decade of last century indicated a
very important increasing of the yield stress with an
increase of strain-rate, especially for strain-rate
greater than 10'Vsec. The increase of ultimate tensile
is moderate, the influence of strain rate being less
important than the yield stress. Consequently, the
yield ratio defined by the ratio between yield stress
135
and tensile strength, increases as far as the strain-rate
increases, with the tendency to reach the value 1.
More recent results (Wakabayashi et al 1994,
Dubina et al 2001, Mazzolani 2000) have confirmed
the previous results. More detailed research works
have shown that the modulus of elasticity is not
influenced by the strain-rate variation and the upper
yield stress is more strain-rate sensitive than the
lower stress. Furthermore, the increasing of strain
rate produces a rapid increase of yield plateau, but
contrary a slow increase ofthe ultimate strain.
Based on the theoretical and experimental results, it
was also concluded that: (1) the mechanical
properties such as yield strength and ultimate tensile
strength, increase with increasing strain rate; (2)
strain rate influence increases with decreasing the
yield strength ofthe material.
5.1.6 Hysteresis models for ductility demand
evaluation
Evaluation of seismic performance of steel
structures is usually carried out assuming the elastic-
perfectly-plastic model for plastic zones. This type
of model may be associated with a conventional
available ductility related to the maximum
deformation beyond which strength degradation is
likely to occur. This methodology leads to
conservative results and is adequate to obtain
reliable information on deformation demands only
when the above ductility limit is not exceeded.
However, when the evaluation of the actual
structural integrity under exceptional loading is of
concern, more realistic cyclic load-deformation
characteristics should be considered, taking
explicitly into account the actual features of
members and connections likely to undergo large
plastic deformations.
Existing relevant experimental tests show a
number of complex aspects that could have a major
impact on seismic demands and that therefore should
be accounted for in accurate structural system
modelling. In particular, strong non-linearity,
kinematic hardening of the monotonie restoring
force characteristic, cyclic hardening, damage of
mechanical properties due to repeated plastic
excursions, pinching ofthe hysteretic cycle are some
ofthe essential aspects arising from laboratory tests.
For instance, the importance of such effects has
been evidenced in Krawinkler and Seneviratna
(1997), where the evaluation of target displacement
to be used for estimating deformation and strength
demands to be compared with available capacities
determined via pushover analysis is focussed on. In
particular, the opportunity and feasibility of adopting
simple either period-dependent or constant
coefficient to amplify displacement demands of
ideal elastic-perfectly elastic systems gathering the
ones of actual systems, which are affected by
pinching and strength deterioration, is shown and
discussed. Further, a parametric study assessing the
effect of the above phenomenological aspects on
seismic demands via time history dynamic analyses,
based on a complex but reliable mathematical
model, has been carried out in Della Corte et al.
(2001). Firstly, according to several existing test
data, it has been shown that the proposed model is
able to represent quite well all the above
phenomenological aspects having a major impact on
the load-deformation characteristics of elements
(members and connections). Then a wide numerical
study dealing with moment resisting frames has been
performed, showing that, in some cases and
depending on the type of strong ground motion
under consideration, deformation demand evaluated
through non-conventional hysteresis model may be
remarkably different from the one evaluated by
using elastic-perfectly plastic model, especially at
collapse limit state ofthe building.
5.2 Non conventional methods for improving
structural integrity
5.2.1 General
Structural integrity of steel structures may be
improved by adopting several special design and
constructional techniques on both new and existing
structures. The majority of such methods aim at
limiting deformation demands to main structural
elements allowing for the satisfaction of higher
performance levels under design and exceptional
earthquake loading. In the following, some non
conventional design strategies that can be adopted
for both new and existing steel structures in order to
reduce the effects of earthquakes on the main
bearing structure are briefly presented and discussed.
5.2.2 Smart use of cladding panels as shear walls
Steel plate shear walls are being used more and more
to resist earthquake and wind forces in steel
structures. Major direct proceeds related to their
application are improved structural energy
dissipation capability and increase of building lateral
stiffness. Such a system provides also several other
advantages when compared to the other usual lateral
load resisting systems, namely steel savings, speed
of erection, reduced foundation cost, increased
usable space in buildings. In contrast to these issues,
in the field of steel structures the spreading out of
this structural typology has been quite limited, both
under the constructional point of view and with
regard to code provisions and development of
reliable design procedures.
Actually, there are several ways to profit of shear
wall effect to improve economy and structural
integrity of steel buildings. For instance, at European
level, there have been many attempts to consider the
136
diaphragm effect provided by common lightweight
cladding panels as either the primary or the
secondary lateral load resisting system of the
building, giving rise to promising new trends for the
design of steel structures. In such a case, due to the
poor mechanical characteristics of the adopted
system (both metal sheeting and connecting system),
generally the contributing effect of shear panels is
quite limited. Nevertheless, it has been already
proved that it can be profitably taken into account in
low- and medium-rise moment resisting frames for
increasing the lateral stiffness ofthe whole structure,
especially at the serviceability limit state (De
Matteisetal. 2001).
On the other hand, steel plates continuously and
rigidly connected to the external frame have been
successfully proposed and utilised worldwide. In this
case, shear walls are definitively conceived as the
primary system for absorbing external horizontal
actions, while frame members have the main role of
carrying out stationary loads. The main problem
concerning this type of system is related to the earl
occurrence of shear buckling phenomena, which
produce a poor dissipative behaviour of the shear
wall, resulting in a slip-type hysteretic response
(Driveretal. 1997).
Lastly, some applications of shear walls based on
stiffened plates made of special low yield strength
steel have been proposed in Japan in recent years
(Torri et al. 1996). In such a case, due to their
exceptional hysteretic behaviour as well as their
capability to early undergo plastic deformations,
shear walls are mainly conceived as hysteretic
dampers and therefore their employment is
essentially based on the concept of seismic structural
passive control. As a consequence, shear walls have
to be designed in such a way to avoid any buckling
phenomenon, checking that the yielding of the panel
under shear is of concern.
All the above cited research works demonstrate
that steel shear walls may remarkably improve the
economy and/or the performance of steel buildings,
especially providing both additional lateral stiffness
and additional energy dissipation sources, which
represent key aspects in the design of high-rise
buildings under earthquake loading at serviceability
and ultimate limit state, respectively. Nonetheless,
two main features inhibiting a more extensive use of
steel shear walls have to be recognised: lack of full
understanding the design procedure and not full
comprehension the actual seismic behaviour of the
system. For these reasons, additional research efforts
appear to be necessary.
5.2.3 Use of specially designed eccentrically
bracedfirames
Eccentrically Braced Frames (EBF) may be
classified according to the link length:
short link
 long link
Depending on their length, the links will behave
predominantly in shear or bending. EBF are
configured so that the shear/flexural plastic hinges
are conducted within the link. The beam outside the
link, connections, braces and columns must then be
proportioned to remain nominally elastic as they
withstand the deformations.
hing«
Detail A
LEJD
ί ΛΙ J. -L
Figure 6. Structure and link configurations
For EBF with short links, it is possible, as
suggested by Dubina et al (2002), to design
removable dissipative links, by using a lower
(compared to the rest of elements) yield steel (Figure
6). The inelastic frame behaviour is proportioned so
that the required plastic deformation of the frame is
accommodated through the development of shear
plastic hinges within the links. The other structural
elements - members and joints- behave elastically
until "Operational" and "Damage Controlled"
performance levels are attained. This is a repairable
state of the structure, when the damage is located in
links, only. In the "Life-safe" or "Damage State"
levels, limited plastic zones are accepted in the other
elements.
This solution permits, at the same time, an easy and
low-cost intervention for replacing the damaged
links.
The experimental tests carried out worldwide have
shown a very good behaviour of this innovative
system under static and cyclic loading, both in terms
of strength and ductility.
6 MASONRY
6.1 Introduction
In contrast with other construction materials,
masonry has much longer tradition as other materials
have. That is why masonry heritage as well as
contemporary masonry strongly depend on available
materials, climatic and functional requirements,
technical knowledge and traditional practices
specific to different regions. Classification of
masonry buildings can be done according to:
137
materials used for construction (adobe, stone, brick,
block), structural systems (unreinforced, confined,
reinforced, prestressed and as infill), place of
construction (rural, urban) and use of buildings.
Allthrough its history, masonry was usually seen
as a compression type of material and while
designers often used arches, vaults and domes to use
compression to span spaces, the role ofthe structural
walls was solely to support the floors and roofs.
Little or no attention was paid on the resistance of
the structural walls due to the seismic loading.
Although some of the monumental masonry
buildings were sometimes designed on the basis of
experiments and the simple theory of structures, all
through the history the rale of thumb was
predominantly used by designers and architects and
measures for improvement of structural integrity and
resistance of the buildings due to seismic loading
were more the exceptions than the rale.
6.2 Masonry structures under seismic loading
The previous discussion is leading us to some
general overwhelming opinion among designers that
masonry is a stiff, heavy and low resistance material
with almost no ductility and ability for energy
dissipation. It is true that some of the great number
of masonry buildings subjected to exceptional
earthquakes, many were severely damaged and
collapsed. On the other hand, there were cases where
some buildings survived the earthquake only slightly
damaged or even undamaged, although they have
been built at the same location as the damaged
buildings.
Figure 7. Typical damage of masonry building.
Learning from the past and awareness of the
necessity to preserve our historical heritage,
improved our knowledge ofthe mechanical behavior
of structural masonry. By analyzing the crack
patterns and damage of structural masonry we were
able to clearly identify the weak and good points of
different structural systems. Once the damage
analysis was clearly evaluated and a failure
mechanism was defined, the forces that were
developed under seismic actions were determined.
Depending on the region, different types of
masonry structures with different morphologies of
their structural elements can be found. Nevertheless,
single and multistory structures are representing the
majority of our masonry heritage and when exposed
to seismic loading some basic crack pattern (Figure
7) both of the structure and its elements can be
generalized and classified as follows (Binda et al.
1999):
cracks between walls and floors
sliding between r.c. tie beams (or plates) and the
masonry
cracks in presence of discontinuity in the
masonry (closed openings, chimney pipes)
cracks and failure mechanisms in masonry vaults
overturning of standing out or overhang
elements (balcony, eaves, chimney pot)
overturning ofthe gable wall
cracks at the corners and at wall intersections
out-of-plane collapse of outer leaves (external
panels)
cracks in spandrel beams (lintels) and/or
parapets
diagonal cracks in structural walls
partial disintegration or collapse of structural
walls and
partial or complete collapse ofthe building.
However, for better understanding of masonry
buildings under seismic loading, also additional
experimental investigations have to be carried out.
The results of these tests have consistently revealed
that masonry, when properly proportioned, detailed
and constructed with good workmanship, provides
adequate resistance against seismic forces. Today,
contemporary masonry buildings, designed and
constructed according to requirements of modern
seismic codes, behave adequately.
6.3 Architectural and structural design concept
The following basic principles should always be
considered when designing a seismically resistant
masonry structure:
simple and regular plans, with symmetry where
is possible
regularity in elevation (both of the geometry and
the variation of story stiffness)
integrated foundation system
rigid floor diaphragm and
robustness.
When subjected to seismic forces, the structural
walls in masonry buildings should be always
distributed in two orthogonal directions. They
should be firmly connected (either by steel ties or
r.c. ring beams) to a rigid floor diaphragm and their
number and strength should be sufficient to resist
seismic load. Only in the case of rigid floor action
the seismic forces will be evenly distributed to the
individual shear walls proportionally to their lateral
stiffness. The distance between structural walls
should be also limited according to the structural
system and the zone of seismicity.
138
In order to provide torsional stability due to
differences in ground motion along the length of the
building in the case of earthquakes, the
recommendation is that the length of masonry
buildings of all masonry structural systems or their
separated parts should be limited up to 40 m in the
zones of high (ag > 0.3 g), and up to 50 m in the
zones of moderate and low seismic intensity (ag <
0.3 g).
The distribution, size and position of wall
openings, such are windows and doors, have a
strong effect on the in-plane resistance of shear
walls. Furthermore, when subjected to lateral
loading, the concentration of the stresses first occurs
in the corner of the openings, thus provoking large
strains and consequently resulting in cracking of
masonry elements and degradation in stiffness
(Bosiljkov et al 1997).
Finally, an important factor in designing masonry
structures is also the design of details, connections
and non-structural elements such as: partition walls,
chimneys, masonry veneer, ornamentations, etc.
6.4 Seismic resistance verification
6.4. 1 Structural walls (Tomazevic 1999)
Masonry buildings are typical shear-wall structures.
Masonry shear- walls in two orthogonal directions of
the building, which are linked together with floors,
represent the basic resisting elements for both
vertical gravity loads and horizontal seismic loads.
Consequently, basic principles, hypothesis and
mathematical models used for seismic resistant
design of r.c. shear walls and shear-wall structures
can also be applied to masonry buildings. However,
mathematical models developed for seismic
resistance verification of r.c. structures should be
modified to take into account the specific
mechanical characteristics of masonry and
constituent materials, as well as specific structural
characteristics of various types of masonry
construction .
L ~_LJ
rcn
,tiT\
JT^fpy
Figure 8. Typical mechanisms of behavior of masonry shear
walls (after Tomazeviö 1999)
Concerning the structural configuration, shear
walls can be either solid or pierced by window and
door openings in each story. They represent the
basic structural elements of a masonry structure
which resist the seismic loads. According to their
configuration, type of construction and resulting
seismic behavior as well as failure mechanism,
masonry shear walls are classified into three main
categories: cantilever walls, coupled walls with pier
hinging and coupled walls with spandrel hinging
(Figure 8).
In the case of cantilever walls (Figure 8a), two or
more walls are connected together in the same plane
with floor slabs, which distribute the lateral loads
among the walls in proportion to their stiffness, but
do not transfer any moments resulting from the
bending of the walls. It represents a structural
system preferred for ductile seismic response. To
provide ductile behavior of the structure, strong
reinforcement and careful detailing at the lowermost
sections of the cantilevers is required due to large
bending moments developed in the walls.
Traditionally, masonry shear walls are pierced by
window and door openings. Above and below the
opening, spandrels connect the walls and transfer the
seismic forces. Depending on the proportion of
openings, either piers are relatively weaker than
spandrels (Figure 8b) or spandrels are relatively
weaker than piers (Figure 8c). In the first case,
which is most often the case of traditional
unreinforced masonry (URM) construction, the
damage will first occur to the piers, which may be
considered as fixed above and below them.
Depending on the geometry, boundary conditions
and quality of masonry materials, the piers will
either fail in shear due to diagonal compression or
rock until crashing of masonry will occur at the
compressed zones. Although the pier action of URM
walls is not so fatal for a masonry structure as is
hinging of columns in the case of a r.c. frame, the
non-ductile behavior of weak piers can be improved
by means of adequately distributed bed joint
reinforcement.
6.4.2 Analytical models
In general, the analysis of masonry structures can be
done either through usage of lumped parameter
models (LPM), structural element model (SEM) or
finite element models (FEM). Since for ordinary
masonry buildings, the first vibration mode shape is
the predominant one, there is usually no need for
sophisticated non-linear dynamic and the LPM
models are quite rare.
SEM approximates the actual structural geometry
more accurately by describing individual structural
elements such as piers and walls. In the case of
single or multistory buildings due to their regularity
and simplicity an equivalent static analysis in two
orthogonal directions by using SEM can provide
reliable information regarding the seismic safety
under expected seismic loads. Nevertheless, since
seismic loading can exercise the structural system to
and beyond its maximum resistance capacity, the
SEM models usually has to be used with static non¬
linear analysis. In that case a step-by-step procedure
is followed, using decreased stiffness values under
139
increasing lateral loads. Nonlinear element behavior
is prescribed in the form of nonlinear lateral
deformation-resistance relationships, depending on
the boundary conditions and failure mode of
masonry elements. Usually the bi-linear or tri-linear
behavior of SE is considered. The storey resistance
envelope is calculated by stepwise drifting of the
storey for small values. The SE's are deformed
equally (due to the rigidity of floor structure) and
internal forces are induced according to the assumed
shape of resistance envelope of each SE. In the case
of torsional effects (due to relatively displacement of
the mass centre to the centre of stiffness of the
storey) the displacements of individual SE are
modified.
Masonry is a composite, heterogeneous, non¬
linear structural material. As with other composite
materials, also with masonry the mechanical
properties are conditioned with the properties of
composite elements, their volume ratio and the
properties of bond between the units and the layers
of mortar. Moreover, the properties and behavior of
masonry is strongly affected by the orientation ofthe
main principle stresses towards the bed joints.
Following the aforementioned, the main strategies
for application of FEM can be adopted for the
masonry as follows:
simplified micro-modeling - usable for small
elements and shear wall with openings, where
expanded units are represented by continuum
elements whereas the behavior of the mortar
joints and unit-mortar interface is represented by
discontinuous elements;
homogenization - is aimed to solve the problem
of modeling of large masonry structures, by
treating masonry as a homogeneous material.
Mechanical properties of masonry are predicted
from the properties of its constituents, i.e. units
and mortars.
macro-modeling for modeling whole structures
where masonry is regarded as an anisotropic
composite material (constitutive models).
Each of those FEM strategies has advantages and
disadvantages. Nevertheless, when analyzing old
historical building even a simple elastic FEM
analysis should be very useful as an anticipation
for effective and less time consuming SEM
seismic verification.
6.5 Optimization of the seismic response
Improving the seismic response of the structure can
be done through repair, strengthening or seismic
upgrading of the existing or damaged structural
system of building. The main goals are always the
same - to increase the strength and ductility. The
measures that will be applied depend on numerous
factors, but basically the seismic resistance of the
building is the governing one.
6.5. 1 Methods of strengthening of masonry walls
Strengthening ofthe building elements in existing
masonry constructions should be always done
systematically. Some of the basic methods of
strengthening of masonry walls are:
repair of cracks
repointing the joints with stronger or the same
mortar as it has been already built in
application of reinforced cement coating
(jacketing) on one or both sides ofthe walls
application of FRP or CFRP stripes on the walls
grouting with cement, modified cement, or
epoxy grout
prestressing the walls in one or both directions
(vertical or horizontal) and
reconstruction of most damaged portion of the
wall.
Special attention should be paid in the case of
historical monuments, where the proposed methods
should fulfil the basic requirements of restoration
and conservation of cultural monuments. In that case
the application of advanced materials should fulfil
the requirement of compatibility as well as durability
with the original architectural and mechanical
elements ofthe existing structure.
6.5.2 Methods of improvement of structural
integrity
In order to improve the structural integrity of the
building system, beside the strengthening of
individual structural elements, the following
measures should be considered:
tying of walls with steel ties
interventions in floor structures and roofs
repair of corners and wall intersection zones
strengthening of walls by confinement and
seismic isolation.
In the practice, we usually have to combine
several of those measures together. Nevertheless,
our final goal is always the same: to achieve
monolithic response of the structure with increased
load bearing capacity and energy dissipation.
REFERENCES
Akiyama, H. 2000. Evaluation of fracturai mode of failure in
steel structures following Kobe lessons. Journal of
Constructional Steel Research, Vol.55 (1-3): 211-227.
Anastasiadis, Α., Mateescu, G., Gioncu, V. and Mazzolani,
F.M. 1999. Reliability of joint systems for improving the
ductility of MR frames, In Dubina and Ivanyi (eds.),
Stability and Ductility of Steel Structures (SDSS'99), 259-
268. Oxford-Elsevier.
ATC-40 1996. Seismic Evaluation and Retrofit of Concrete
Buildings, prepared by the Applied Technology Council
(Report No. ATC-40), Redwood City, California for the
California Seismic Safety Commission (Report No. SSC
96-01).
140
Binda, L., Gambarotta, L., Lagomarsino, S. & Modena, C.
1999. A multilevel approach to the damage assessment and
the seismic improvement of masonry buildings in Italy.
Seismic Damage to Masonry Buildings, A. Bernardini (ed):
179-194. A.A.Balkema
Bosiljkov, V., Kralj, B., ¿arnie, R., Pande, G.N.& Middleton,
J. 1997.Finite element analysis of model shear tests.
Computer methods in structural masonry 4. Pande, G.N.,
Middleton, J. & Kralj, B. (eds): 153-160. E&FN Spon
Calado, L., Castiglioni CA., Barbaglia, P., Bernuzzi, C. 1998.
Seismic design criteria based on damage concepts. ll'h
European Conference on earthquake Engineering, Paris, 6-
II September 1998.
Chang, S.-P., Makris, N. 2000. Effect of various energy
dissipation mechanism in suppressing structural response.
Proc. 12th World Conference on Earthquake Engineering,
Auckland, New Zealand, 30 January- 4 February 2000.
Combescure, D., Chaudat, T. (2000). ICONS European
program - seismic tests on R/C bearing walls _ CAMUS3
specimen. CEA, Rapport DMT/00-014/A.
De Matteis, G., Landolfo, R., and Mazzolani, F.M. 2001.
Contributing Effect of Cladding Panels in the Seismic
Design of MR Steel Frames. In Zingoni A. fed.), Structural
Engineering, Mechanics and Computation: 937-946.
Oxford: Elsevier.
De Matteis, G., Mazzolani, F.M., Landolfo, R. 2000. The
behaviour of connections in steel MR-frames under high-
intensity earthquake loading. In Virdi, Matthews, Clarke
and F.K. Garas (eds.), Abnormal Loading on Structures:
Experimental and Numerical Modelling: 58-72. London: E
& FN SPON.
Della Corte, G, De Matteis, G., Landolfo, R., and Mazzolani
F.M. 2001. Seismic Analysis of MR Steel Frames based on
Refined Hysteretic Models of Connections. In Zingoni
(ed.), Structural Engineering, Mechanics and Computation:
955:964. Oxford: Elsevier.
Dexter, R.J., Melendrez, M.I. 2000. Through-thickness
properties of column flanges in welded moment
connections. Journal of Structural Engineering, ASCE,
126(1): 24-31.
Dolce, M., Mamerto, A. 2000. Passive seismic devices based
on shape memory alloys. Proc. 12th World Conference on
Earthquake Engineering, Auckland, New Zealand, 30
January- 4 February 2000.
Dolsek, M., Fajfar, P. (2001). Soft story effects in uniformly
infilled RC frames, Journal of EE, Vol. 5, No. 1, pp. 1 - 12.
Driver, R. G., Kulak, G. L., Kennedy, J. L. and Elwi, A. E.
1998. Cyclic Tests of Four-Story Steel Plate Shear Wall.
Journal of Structural Engineering, ASCE, 124(2): 112-120.
Dubina, D., Stratan, Α., Moisa, T., Mihu, P. 2001. Behaviour
of welded connections of moment resisting frames beam-to-
column joints, ECCS TWG 10.2-01-012 Document.
Dubina, D., Stratan, Α., Chitina, Α., Fulop, L., Dinu, F. 2002.
Reliability of seismic resistant structures in Romania.
CNCSIS/World Bank Research Grant No. 16, Research
Report on 3rd year.
Engelhardt, M.D and Sabol T.A. 1997. Seismic-resistant steel
moment connections: developments since the 1994
Northridge earthquake. Progress in Structural Engineering
and Materials, Vol. 1, No. 1: 68-77.
EUROCODE 8 2001, Design provisions for earthquake
resistance of structures, Draft No. 4 of prEN 1998-1.
FEMA273 1997. NEHRP Guidelines for the Seismic
Rehabilitation of Buildings, prepared by the Building
Seismic Safety Council for the Federal Emergency
Management Agency, Washington, D.C.
FEMA274 1997. NEHRP Commentary on the Guidelines for
Seismic Rehabilitation of Buildings, prepared by the
Building Seismic Safety Council for the Federal Emergency
Management Agency, Washington, D.C.
FEMA307 1998., Evaluation of Earthquake-Damaged
Concrete and Masonry Wall Buildings Technical
Resources, prepared by the Applied Technology Council
(ATC-43 Project), for the Federal Emergency Management
Agency, Washington, D.C.
Fischinger, M., Cerovsek, T., Turk, Z. (1997). EASY, CD
ROM, University of Ljubljana, FGG, IKPIR;
http://www.ikpir.fgg.uni-lj .si/EASY
Fischinger, M., Isakovié, T., Kante, P. (2002). CAMUS 3
international benchmark - Report on numerical modeling,
blind prediction and post-experiment calibrations,
University of Ljubljana, IKPIR Report EE-1/02.
Gioncu, V, Mateescu, G, Tirca, L. and Anastasiadis (2000), A.
Influence of the type of seismic ground motions, in:
Moment resistant steel frames in seismic areas: design and
reliability, London: E & FN Spon.
Kelly, J. 1999. The role of damping in seismic isolation
Earthquake Engineering and Structural Dynamics, Vol. 28:
3-20.
Krawinkler, H., Seneviratna, G.D.P.K. 1998. Pros and cons of
a pushover analysis of seismic performance evaluation. In
Engineering Structures, Vol. 20, No. 4-6: 452-464.
Mao, C, Ricles, J., Lu, L.W., Fisher, J. 2001. Effect of local
details on ductility of welded moment connections. Journal
of Structural Engineering, ASCE, 127(9): 1036-1044.
Mazzolani, F.M. 2000. Moment resistant connections of steel
frames in seismic areas: design and reliability. London: E
& FN Spon.
Nakashima, M., Tateyama, E., Morisako, K., Suita, K. 1998.
Full-Scale test of beam-column subassemblages having
connection details of shop-welding type. Structural
Engineering Worldwide, Elsevier Science (CD-ROM),
Paper Ref. Tl 58-7.
Nakashima, M., Roeder, C.W., Maruoka, Y. 2000. Steel
moment frames for earthquakes in United States and Japan.
Journal of Structural Engineering, ASCE, 126(8): 861-868.
Roeder, C.W. 2000. SAC program to assure ductile connection
performance. In Mazzolani and Tremblay (eds.), Behaviour
of Steel Structures in Seismic Areas (STESSA 2000): 659-
666. Rotterdam: Balkema.
SEAOC, 1999. Recommended Lateral Force Requirements and
Commentary, Sixth Edition, Structural Engineers
Association of California, Sacramento, California.
Suita, K., Nakashima, M., Engelhardt, M.D. 2000. Comparison
of seismic capacity between post-Northridge and post-Kobe
beam-to-column connections. In Mazzolani and Tremblay
(eds.), Behaviour of Steel Structures in Seismic Areas
(STESSA 2000): 271-278. Rotterdam: Balkema.
TomazeviiS, M. 1999. Earthquake-resistant design of masonry
buildings. Imperial College Press. London
Torii, S., Teramoto, T., Kihara, H., and Kitamura, H. 1996. The
response control design of high-rise building with low yield
strength steel wall. Proc. 11th WCEE, Elsevier, CD-ROM:
Paper No. 97.
Tsai, K.C and Popov, E.P. 1997. Seismic Steel Beam-Column
Connections, Report No. SAC-95-09, FEMA Background
Report N. 288: IV/l-IV/39.
Wakabayashi, M., Nakamura, T., Iway, S. and Hayashi, Y.,
1994. Effects of strain rate on the behaviour of structural
members, Proc. 8th World Conference on Earthquake
Engineering, Vol.4: 491-498. San Francisco.

141
Structural integrity of buildings under exceptional fire

F. Wald
Czech Technical University in Prague, Czech Republic

V. Bosiljkov
Slovenian National Building and Civil Engineering Institute, Slovenia

G. De Matteis
University of Naples Federico II, Italy
P. Haller
TU Dresden, Germany
A. Santiago, L. Simões da Silva
University of Coimbra, Portugal
P. Vila Real
University of Aveiro, Portugal

ABSTRACT: The paper summarizes the European knowledge in the field of the behaviour of the structures
made of the main structural materials under fire action. The new findings are stressed to prepare background
for studying ofthe exceptional fire conditions. The exceptional fire is the commonly observed problem ofthe
fire acting after or together with other accidental loading as earthquakes, windstorms, heavy snow loading, gas
explosions, bomb blasts, impact of vehicles, rapidly changing loading situations (temperature) or ice loads.
The exceptional cases include also the situations not taken account during the design or in fire the
unexceptional conditions due to the different fire scenario and burning materials (different to expected
cellulose or hydrocarbon fires). The structural integrity is the major goal of the designers to ensure the
resistance of the structure under accidental actions and enlarge the safety under exceptional condition. It is
guaranteed by the robust of structural elements and its joints, which is described by its balanced stiffness,
strength and ductility between its members, connections and supports.

INTRODUCTION
1.
The existing European findings, models, and
regulations about the structural safety are generally,
with few exceptions, concerned with rules aimed to
ensure an adequate safety level of constructions
under, normal loading conditions. The overall
structural safety is investigated simply by assuming
an increase ofthe multipliers ofthe service loads up
to reach the collapse value. This approach has led to
a satisfying degree of accuracy in the prediction of
the safety margins under serviceability load
conditions. A greater accuracy in the evaluation of
structural safety is possible when a probabilistic or
semi-probabilistic approach is followed in the
determination of both actions and structural
resistance. In this way is possible to get a good
investigation of structures subjected to random
actions. These approaches form the basis of most
recent developments in the field of regulations and
are now part of any relevant structural code,
including of course the Eurocodes, where specific
allowance for accidental loading conditions is made.
Fire resistance is used to characterize the
performance of elements of structure in fire. The fire
resistance is the time for which elements performs
its functions under specified conditions. These
functions may include the ability: not to collapse, to
limit the spread of fire, to support other elements.
All materials progressively lose their ability to
support a load when they are heated. If components
of a structure are heated sufficiently, they may
collapse. The consequences of such a collapse may
vary, depending on how critical the component is in
controlling the overall behaviour of the structure. In
order to limit the threat that a fire poses to people in
a building and to reduce the amount of damage that a
fire may inflict, large buildings are divided up into
smaller fire compartments using fire resisting walls
and floors. Parts of a fire compartment may be
divided up by fire resisting construction to protect

143
particular hazard within them. The performance of
fire separating elements may rely heavily on the
ability ofthe structure that supports them to continue
to provide that support under fire conditions
(Lennon, 1997). The criticality is the degree to
which the collapse of an individual structural
element affects the performance of the structure as a
whole. All main components of a structure are
generally expected to exhibit fire resistance
proportionate to the nature ofthe perceived risk. The
nature of the risk is usually assessed on the basis of
the size and proposed use of the building in which
the structural element occurs, which is an important
part of a fire safety risk analysis.
The definition of the fire resistance is the ability
of construction or its element to satisfy for a stated
period of time load bearing capacity, integrity and
insulation (same or all of the criteria. As a
consequence of European harmonization, fire
resistance is increasingly being expressed in terms of
R, E and I. R means the resistance to collapse, i.e.
the ability to maintain load-bearing capacity. E is the
resistance to fire penetration, i.e. an ability to
maintain the integrity of the element against the
penetration of flames and hot gases. I is the
resistance to the transfer of excessive heat, i.e. the
ability to provide insulation from high temperatures.
The term the elements of structure is in fire
engineering applied to main structural elements such
as structural frames, floors and walls. Compartment
walls are treated as elements of structure although
they are not necessarily load-bearing. External walls
such as curtain walls or other forms of cladding that
transmit only self weight and wind loads, and that do
not transmit floor loads, are not regarded as load
bearing, although such walls may need fire
resistance to satisfy other requirements in connection
with a need to restrict fire spread between buildings.
Load bearing elements may or may not have a fire-
separating function. Fire-separating elements may or
may not be load bearing.
A fire safety engineering approach takes into
account the total fire safety provides a fundamental
and economical solution than the prescriptive
approaches to fire safety. The modelling of a
structure involves three stages. The first stage is to
model the fire scenario to determine the heat energy
released from the fire and the resulting atmospheric
temperatures within the building. The second stage
is to model the heat transfer between the atmosphere
and the structure. Heat transfer involves three
phenomena (conduction, convection and radiation)
all contribute to the rise in temperature of the
structural materials during the fire event. The third
144
stage is the determination of the response of the
structure basic simple checks, engineering
advanced models and sophisticated discrete models
based on all data available.
The requirements for the codes contain simple
checks, which provide an economic and accessible
method in the majority of buildings. For complex
problems considerable progress has been made in
recent years in understanding how structures behave
when heated in fires and in developing mathematical
techniques to model this behaviour. It is now
possible to predict the behaviour of certain types of
structure with a reasonable degree of accuracy. The
most common form of analysis is the finite element
method (Newman at al. 2000). It may predict
thermal and structural performance. In fire, the
behaviour of a structure is more complex than at
ambient temperatures. Changes in the material
properties and thermal movements cause the
structural behaviour to become non-linear and
inelastic.
In the fire resistance tests the gas temperature is
increased to follow a predefined time/temperature
curve. This heating regime is very different from that
occurring in real fires. The maximum temperature
attained in a real fire and the rate at which
temperatures increase depend on a number of factors
relating lo the fuel available, the geometric and
thermal properties of the compartment and the
availability of openings through which oxygen can
be supplied to the fire. Techniques have been
developed to mathematically describe a natural fire.
The analysis determines the rate at which heat is
released from the available fuel, see Schleich at al,
(1999). This is a function of the amount of
ventilation available and the density and distribution
ofthe fuel itself. Heal loss from the compartment via
convection and radiation from the openings, and
conduction through the other solid boundaries is
calculated before the resulting atmospheric
temperatures may be determined. There are the two
forms of fire used in standard fire resistance tests.
They are timber fires or hydrocarbon fire in
petrochemical industry.
The periods of fire resistance specified in
regulations attempt to relate the damaging effect of a
real fire to an equivalent period of exposure in a
standard fire resistance test. Safety factors are
introduced to account for building use and height.
The modelling of the fire depends upon several
factors such as the fire load density, the size and
shape of the ventilation openings, and the thermal
characteristics of the enclosing compartment. Time
equivalence is useful when comparing the
performance of an element in a natural fire with the
known performance of the same element in a fire
resistance test, see Schleich at al, (2001). It is useful
for researchers and fire investigators to know that
the natural fire was approximately equivalent to so
many minutes of standard test.
The fire load is the amount of combustibles
available to burn. It is expressed in terms of the
equivalent weight of timber (kg/m2) or in terms of
energy released (MJ/m2). The reference area is
normally the floor area only sometimes the total
internal surface area ofthe compartment is used.
The ventilation relates the size of openings, such
as windows and doors, to the size of a fire
compartment or room.
The heating of element depends on the incident
heat flux. This is the amount of heat, which an
element receives in a fire. The heat flux is made up
from radiation and convection. Tn fully developed
fires the radiation is the dominant flux. A parametric
fire is a mathematical idealisation of a natural fire in
a compartment. The gas increases to a maximum and
then declines, as it would in a real fire. The fire
temperature is a function of the ventilation factor,
the fire load and the thermal properties of the wall
linings. Advanced computer programs use a
technique of computational fluid dynamics.
The structural response and its modelling under
fire condition depend on the applied structural
materials. The difference may be found by modelling
of the whole structure, in the modelling of the
elements as well as joints (Allam at al 1999). The
resistance under fire condition does not guide the
difference. The aluminium and steel are transferring
the heat. The timber, masonry, concrete and
lightweight concrete has insulate itself. The added
insulation may be economical for the aluminium,
steel and timber structures. The design models are
based on previous check under ambient temperature,
if possible. The advanced model of global analysis
as well as element and connection check may be
applied taking into account large deformation, the
changes of structural properties and models during
heating simulating the real fire.
2. MASONRY STRUCTURES
2.1 Introduction
Following the Great Fire of London in 1666,
Charles Π insisted on brick and stone buildings
instead of the wooden ones that have been
previously used. Furthermore during the bombings
of the Second World War, brick structures suffered
less damage than steel or concrete buildings. This
actually proves the fact that the masonry is not only
the high fire resistance but also it has a large
capacity to resist accidental damage due to the some
forms of exceptional loading.
In general, brick and block masonry materials are
incombustible and cannot start or spread a fire.
Furthermore masonry is rarely seriously damaged by
fire in the sense like other materials can be - it does
not buckle like steel, spall like reinforced concrete or
burn like timber. On the other hand, the high
temperatures can influence the strength of the units
their deformations and can destroy the mortar-brick
junction and thus can provoke some further
reduction in residual compressive strength and sharp
drop in flexural and shear strength of the masonry
elements exposed to the fire.
2.2 Materials for use in masonry walls
The masonry is strongly composite material
consisting of unit, mortar, and concrete infill and
reinforcing steel. From that point of view, the fire
behaviour of a masonry wall depends on:
- masonry unit material - clay (C), calcium silicate
(CS), autoclaved aerated concrete (AAC) or
dense/lightweight aggregate concrete (LC);
- type of unit - solid or hollow (type and pattern of
holes, percentage of holes), shell and web
thickness;
- characteristics of the units - strength, density,
porosity;
- type of mortar - general purpose, thin layer or
lightweight mortar;
- type of the execution of the pretend (head) and
bed joints (fully filled, stripes, unfilled, tongue
and groove systems etc.);
- with or without renders (plasters) on one or both
sides;
- relationship of the applied load to the resistance
ofthe wall and
- slenderness ofthe wall.
2.3 General requirements related to masonry walls
From the point of view of fire protection, a
distinction is made between non-load-bearing walls
and load-bearing walls and between separating walls
(walls along escape ways, walls of stair wells, or
compartment walls) and non-separating walls
(within single fire compartment).
It is assumed that walls span or extend from one
floor to the next floor or to the roof, and that those
floors or the roof provide lateral support to the top
and bottom of the wall, unless its stability under
normal function is achieved by other means, for
example buttresses or special ties. Since masonry
walls can be either single or multiple leaf elements, a
general attention is also focused on the number of
leaf, their thickness and if they are load-bearing or
not. Also perforated masonry units should not be laid
145
so that the perforations are at right angles to the face all masonry material and the standard renderings.
of the wall, i.e. the wall should not be penetrated by Building code requirements for fire resistance
the perforations. typically vary from Vi hours up, depending on type of
Special attention has to be paid in designing the building, occupancy and criteria of failure (R - load
joints as well as recesses and chases for fixtures, bearing, E - failure due to cracks, holes, etc, and I-
pipes and cables in the walls. insulation failure). Having in mind all variety of
materials that can be used in masonry walls it is
2.4 Design Practice worth to mention that the values presented in Table
1 reflect only the small portion of those requirements
Masonry walls themselves do not contribute fuel to
stated in different codes.
the fire, are not subject to flame spread, and do not
produce smoke or toxic gases in the presence of fire. Table 1: Ultimate fire Resistance Periods for Load-bearing
They do provide solid non-combustible barriers to Clay and Shale Walls in the United States" (Drysdale 1994)
the spread of fire from the original fire area and can Ultimate Fire Resistance Period (h)
be equally useful in creating safe compartments or
No-Combustible Combustible
escape routes from the building. Therefore, masonry Members Framed Members
construction is ideally suited to minimizing the Into Wall Framed Into W.
potential for ignition of fires and the consequences Wall Wall No Plast. Plast. No Plaster on
of fires. Thickness Type pi. on one on two pi. exposed
The main fire resistance aspects (Drysdale et al. (mm) side sides* side*
1994) of the design of masonry walls relate to the 100 Solid VA 1% 2/2 .

following: 200 Solid 5 6 7 2 2Î4

- ability to maintain sufficient load-carrying 300 Solid*'· 10 10 12 8 9
capacity to support floors and roofs in load- 300 Solidt 12 13 15 -

230-250 Cavity 5 6 7 2 214

bearing construction
- ability of nonload-bearing firewalls to maintain
sufficient strength during and after specified
duration of fire to avoid collapse
- thermal characteristics ofthe wall so as to prevent
temperature rise that could possibly cause new
ignition
- impact of the failure of other structural elements
on the stability ofthe masonry
- impact of the use of other materials as part of the
wall system
2.5 Assessment by testing and tables
While most building codes for other materials relate
to both fire prevention and fire protection design
procedures, for the masonry this is not a case. Due to
the general belief that the masonry is incombustible
material little or no investigation were made on
development of analytical procedures for assessment
of fire resistance of the masonry structures or its
components in a real building fire.
All of the current building codes for masonry
(EC 6, BS 5628, DIN 4102, UBC, NBC etc.) relate
to the fire design ofthe masonry structures by:
- avoiding premature collapse of the structure
(load-bearing function) and
- limiting fire spread (flames, hot gases, excessive
heat) and temperature rise beyond designated
areas (separation failure).
The requirements for masonry materials for
achieving those goals are stipulated by means of
numerous tables, which are derived mainly from fire
resistance tests of masonry walls with standard
temperature-time curve. The tables are drawn up in a
standardized form to give consistent information for
* To achieve these ratings, each plastered wall face must have
at least 13 mm 1:3 gypsum-sand plasters.
** Based on load failure (for load-bearing walls)
t Based on temperature rise (for nonload-bearing walls)
As it concern Eurocodes for masonry structures
(ENV1996-1) it is worth to mention that for the
moment in the latest version there are no input
values in the tables for minimal thickness for fire
resistance for different types of masonry
(ENV 1996-1, 1995). In the previous version ofthe
EC 6 (ENV 1996-1, 1990) there were some
provisional values that were based largely on BS and
DIN standards, but with the comment that the
method of test according to BS and DIN were not
exactly the same. In general the values for given
periods of fire resistance are sometimes higher
according to the German results while there is some
British data that gives a lower thickness that has
often been used.
2.6 Assessment by calculation
It has been only recently (Hahn 2001) that first
attempts were made to assess the fire resistance of
masonry walls by taking into account the relevant
failure mode in fire exposure, the temperature
dependent material properties and effects of thermal
expansions and deformations.
In general the calculation methods may be:
- a simplified calculation model for specific types
of walls (analytical solution) and
- global structure analysis for simulating the
behaviour of structural masonry elements and the
entire structure (FEM solution).

146
 Although there are lack of data for other
numerous types of masonry materials, some good
results were obtained through simplified calculation
procedure both for pillar and wall elements, where in
comparison to the experimental results, the results of
simplified calculations were a little bit too
conservative. Most of the problems that are limiting
the usage of both analytical and FEM solution of fire
resistance of masonry structures are large variety of
materials that have been used in masonry and lack of
experimental data for verification of those models.
Moreover, it will be quite demanding to prepare a
general solution for all types of the masonry. As it
can be seen from Figure 1, different types of the
units have different behaviour due to thermal
actions. Furthermore some types of the units like
AAC and CS are even gaining on the strength at
some elevated temperatures, while LC and C units
are having constant declination of the strength with
increasing the temperatures.
14
Autoclaved areated concrete
12 Calcium silicate
Lightweight concrete
10
Cla
8
indicates the period of time it can maintain the load
bearing capacity and tightness. Both properties have
to be taken into account. Wood and steel are
diametrically opposed with respect to these two
definitions. Wood burns, however, since charcoal is
produced at a constant rate so that the time to failure
of wooden construction elements can be easily
predicted. Steel is incombustible, however, it has to
be protected against heat in order not to loose
strength and stiffness. The design of multi-story
wood buildings requires good fire resistance, which
is in general, achieved by cladding wood structures
with incombustible components or, by the use of
construction elements especially floors made of solid
wood. Floors with a long span have been designed as
timber concrete composite, which successfully
resisted significantly more than 90 minutes in case
of fire.
a
S

' 6
X
4

_ J---W !-__*

0 100 200 300 400 500 600 700 800 900 1000 1100 1200
Temperature [°C]

Figure 1: Shrinkage and expansion under changes of

temperature for different types ofthe units (Hahn 2001)

3. TIMBER STRUCTURES
3. 1Behaviour of wooden constructions
The protection of buildings against fire is an
important topic, which puts constructions made of
Figure 2:
combustible materials such as wood, and wood
based material at a disadvantage (Kollman and Coté
1968). Experiences with devastating fire in the
history of our cities led to regulations that forbade
the use of wood first as a roofing material and on the
ground floor. Afterwards when alternative building
materials and design were available wood was more
and more replaced especially in urban areas. Today,
there is again a tendency to construct multi-storey
buildings in wood where the risk of fire plays a key
role, see STEP (1995). These buildings require a
careful planning with respect to fire protection,
which focuses on an organizational, conceptual,
constructive and technical approach. Due to the
Figure 3: NLT covered (according table 2, Nr 2) after testing
complexity of this topic, the following concentrates
on constructive fire protection. Combustibility is an
important material property. Fire resistance,
however, characterizes a construction element and

147
3.2 Experimental results on floor elements made of
solid wood and wood-concrete composite
Combustibility and fire resistance determine the fire
behaviour of wood and wooden elements.
Instinctively most people are susceptible to attach a
high risk to the combustibility of a construction, see
Tab. 2. However, if one consider that fire resistance
of masonry and wooden building are equivalent, it
becomes obvious that the course of a fire depends
decisively on the resistance of the wall and floor.
Elements from solid wood have a good fire
behaviour since there is no heat conduction through
hollow spaces. The tightness of the element is
crucial and can be achieved by gluing or cladding
with other materials. In the past, the fire behaviour
of different solid construction elements and whole
buildings has been tested. The test results encourage
to a further use of wood in the multi-story sector.
The following summarizes experimental findings on
nailed laminated timber floors (NLT) that can
alternatively be provided with a thick layer of
concrete, which guarantees tightness and makes the
floor work as a structural composite where the wood
is subjected to tension and the concrete to
compression. Nailed laminated wood is to be
distinguished from glued laminated wood with its
tight bond line.
Table 2: Fire resistance of selected solid floor elements and
composites
No. Floor and layers Fire resistance
1 NLT (thickness: 110 mm) 23 min
uncovered
2 NLT (thiclcness: 110 mm) 35 min
+ insulation (mineral: 4 cm)
+ particle board (16 mm)
3 Wood concrete composite > 90 min
NLT (thickness: 140 mm)
+ concrete (thickness: 80 mm)
Figure 4: Wood concrete section on the basis of NLT
148
4. STEEL STRUCTURES
4. 1 Reached Stage
The design methods are used in practice to
increase safety of steel structures under fire
conditions. New studies have been made in the area
of material properties, complex as well simple
design models of elements and joints, and fire
modelling in last ten years. Very complex and
progressive was the research in fire modelling and
its application into the design. Among the work done
in this field there are two topics design that will be
presented here: the lateral torsional buckling of steel
I-beams at elevated temperatures and the component
model for the behaviour of steel joints in case of fire.
4.2 Lateral-Torsional Buckling of Steel I-Beams
The problem of lateral-torsional buckling of steel
I-beams at room temperature is well recognised by
practice. The same problem at elevated temperature
was studied by Bailey et al., 1996, who uses a three
dimensional computer model to investigate the
ultimate behaviour of uniformly heated unrestrained
beams.
The behaviour of steel I-beams at elevated
temperature has been analysed numerically (Vila
Real & Franssen, 1999, 2001) leading to a new
proposal for the evaluation of its lateral-torsional
buckling resistance. This proposal was based on the
numerical results from the SAFIR program, a
geometrical and materially non-linear code specially
established for the analysis of structures submitted to
fire (Franssen, 1995). The capability of this code to
model the lateral-torsional buckling of beams has
been shown (Vila Real & Franssen, 1999) at room
temperature by comparisons against the formulas of
ENV 1993-1-1, 1993.
It must be emphasized that the simple model
presented here (Vila Real et al. 1999, 2001), was
established on the base of numerical simulations
using characteristic values for initial out-of-
straightness (L/1000) and residual stresses
(0,3*235MPa), which are unlikely to be
simultaneously present in a test or in a real fire. In
the experimental work we have done, the
geometrical imperfections and the residual stresses
were measured as well as the nominal yield strength
of the material and the Young Modulus. These
measured values were used in the numerical
calculations.
A set of 120 full-scale tests based on a reaction
frame and on a hydraulic system has been carried out
for beams of the European series ΓΡΕ 100 with
lengths varying from 0.5 to 6.5 meters. Three tests
have been done for each beam length and for each
temperature level. The beams were electrically
heated by means of ceramic mat elements, heated by
a power unit of 70 kVA. To increase the thermal
efficiency a ceramic fibre mat has been used around
the beam and the heating elements. According to the
proposal the design buckling resistance moment of a
laterally unrestrained beam, with a class 1 or 2 cross
section type, can be calculated by

Ίμ,Γι
where%ir;î,isgivenby

%LT,fl - (2)
§LTß,com + V l$LTß,com ì ~ i^LTß.com ì
with a)

YLT fi.com ~ rs L + a^ LT fi, com + \^ LT, β, corn) J (3)

and
a = 0.65p35/fy (4)
where fi is the nominal yield strength ofthe material,
Wpi is the section modulus, ky,siCom the reduction
factor for yield strength, relative slenderness
^LT,e.com' and YMfi is the partial safety factor for steel.
The lateral-torsional buckling curve now depends on 0.4 0.6 0.8
the steel grade as it can be seen in Figure 5. Experimental

Mb,fi.l,M IMßfiJtd b)
1.2 Figure 6: Experimental behaviour, for elevated temperatures
iiii EC3, S235 or S355 (above 400 °C).
; ; -O- New Proposal - S355
1.2

ί- -j -i \ '- -4 ί -i

i 0.6

: i : I i

' '. ' '. !

0 0.2 0.4 0.6 0.8 1.2 1.4 1.6 1.8 2

SAFE

Figure 5: Presented design buckling curve compare to the 0.4 0.6 0.8 1 1.2

ENV 1993-1-2 predictions Numeric

a)
Both experimental and numerical results have been
compared with the simple formulas from ENV 1993-
1-2 and the new proposal. The results of these
comparisons are shown respectively in Figure 6 and cp
Figure 7. In these figure the regression line is much . to
more close to the ideal continuous line in the case of =
cu
0.6
t
numerical calculation than for the experimental Z 0.4
op

results.

SAFE

0 0.2 0.4 0.6 0.8 1 1.2

Numeric

b)
Figure 7: Numerical behaviour, for elevated temperatures
(above 400 °C).

149
 4.3 A component modelfor the behaviour of steel fire-load curve: anisothermal, isothermal or transient
joints at elevated temperature situation. An application to a cruciform flush end¬
Recent experimental evidence have highlighted the plate beam-to-column steel joint is presented and
need to evaluate the behaviour of steel joints at compared to the experimental results obtained under
elevated temperatures, since they exhibit a distinct various loading conditions (Al-Jabri et al. 1997).
change of its moment-rotation response under Comparison with experimental results has shown
increasing temperature, that affects the global good agreement with the proposed methodology, as
response ofthe structure. In terms of cold design, the represented in Figure 8.
'component method' constitutes today the widely The fire response of steel joints involves additional
accepted procedure for the evaluation of the various complexity to the corresponding cold analysis. This
design values (EN 1993-1-1/A2). analytical methodology greatly simplifies this
It was purposed an analytical procedure problem, allowing a direct solution from the
capable of predicting the moment-rotation response knowledge ofthe response at ambient temperature.
under fire conditions (Simões da Silva et al., 2001).
This procedure is based in the 'component method',
that consists of modelling a joint as a extensional
springs and rigid links, whereby the springs
represent a specific part of a joint make an identified
contribution to one or more of its structural
properties (component). Each component exhibits a
non-linear force deformation response (here taken as -FBll:4KNm
-FB12:81dNm
a bi-linear approximation), characterised by four -EB13:lZ8kNm
-FB14:17kNm
properties: elastic stiffness, post-limit stiffness, limit - EXPER1M. RESULTS
-AN AIJT. RESULTS
load, yield displacement and limit displacement. Component 4.1
Component 5.
The evaluation of the fire response of steel Component 4.2
Component 2
joints requires the continuous change of mechanical ι 1 ι 1 1 1 1 1 1 1 φ (rad)

properties of steel (yield stress and Young's 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100

modulus) as temperature increase. In the context of

Figure 8: Anisothermal temperature-rotation response
the component method this is implemented at the
component level:
Fί;θ ~ "y.
1
k θ LFi;20°C s (5)
5. ALUMINIUM STRUCTURES
&ί;θ ~ * Ε,θ *^¡;20°C (6)
>
5.1 General
The demand of aluminium structures has increased
s = κF'*
iß
;θ
,
ΚΕ;Θ
ui;20°C , (7) almost continuously in the last 100 years, the main
fields of application being aeronautic, shipping and
automotive industries. Attention to aluminium is
where F is force in component, Κ is stiffness of the also paid by building and construction markets,
component, kE;e is reduction factor relative to E, ky: <?
especially with regard to offshore structures, bridge,
is reductionfactor relative to^,, and <5is deformation reticular space structures, wall panels, scaffolding
of component. Introducing these equations in any and windows. Attraction is essentially due to
evaluation of moment-rotation response of steel lightness, versatility of product shapes (allowed by
joints at room temperature yields the required fire the extrusion fabrication process) corrosion high
response: resistance, ease maintenance, good appearance and
M j;0 *>;<? M¡;20°C'> (8) low-cycle costs, but also high structural efficiency
(large strength to specific weight ratio) (Mazzolani
1995).
j. Mi;9 ky;B A .
On the other hand, in the field of structural
Ψι:θ ~ ς, ~~ , Yi;20°C > (9)
ύί;θ ΚΕ;Θ engineering, full utilisation of special features of
aluminium has not been always possible, due to lack
_ Ε ζ2
CgΖ of technical data and difficulty of application of
^;θ ~ ι ~ ΚΕ:Θ )Jí;20',C ; (10) special calculation methods required by the peculiar
Στ^
k: ί;θ
features of the material. This is the case of structural
fire design, which is presently carried out by
adopting simplified material schématisation and
where φ is rotation of joint, M is bending moment
calculation procedures. In fact, while design rules for
moment, and Sj is the joint stuffiness. This procedure
steel structures subjected to fire are based on
may be applied to a fire event, characterised by any
comprehensive research activities and laboratory
150
tests as well as on long experience, in case of
aluminium structures it has to be revealed a poorness
of available data and a weakness of existing design
rules (Forsén 1995, Lundberg 1995).
Traditionally, the behaviour of aluminium alloy in
fire is thought to be a critical task for the structural
design. This is due to the rapid decrease of material
resistance at elevated temperatures, as respect to
other traditional materials like steel and reinforced
concrete. In particular, aluminium alloys melt
between 600 and 650 °C, the exact value being
dependent on the type of alloy under consideration,
but at 200-250 °C, most of the alloys have already
lost approximately 50 % of their original strength at
room temperature. For this reason, when for
aluminium structural components resistance to fire is
required, the adoption of reliable prediction models
would be compulsory, allowing accurate design
calculations and safety assessment to be carried out.
5.2 Thermal and mechanical properties of
aluminium
Features of aluminium alloys are changing under
exposure at high temperatures. Relevant values of
both physical and mechanical properties are given in
the European Prestandard ENV 1999-1-2 (1998).
The first ones allow the temperature distribution in
the cross-section to be determined by thermal
analysis, while the second ones allow the bearing
capacity of the structure at elevated temperature to
be evaluated by means of structural analysis. In
general, it has to be emphasised that almost all the
material properties are dependent on the type of
alloy, but some common features may be identified,
allowing some general trends to be recognised.
In particular, it has to remarked that: (1) thermal
elongation of aluminium is about 2,5 -10"5 0C_1,
therefore, more than twice that of steel; (2) specific
heat of aluminium alloys ranges from 0,9 U/kg °C
(at room temperature) to 1,1 kJ/kg °C (at 500 °C),
therefore about two times the one of steel
(nevertheless due to the ratio of specific weight
between steel and aluminium, for a given cross
section and fire exposure, the former has an higher
volumetric heat capacity than the latter, resulting in
more rapidly heat up of aluminium structures than
steel structures); (3) unlike steel, thermal
conductivity of aluminium is increasing with
increasing temperatures and ranges from 140-190
W/m°C at room temperature to 180-220 W/m°C at
400 °C, it being therefore more than three times
higher than for steel; (4) the emissivity of aluminium
surface is dependent on the surface finish, making it
difficult to be determined accurately (suggested
values are 0,3 and 0,7 for clean uncovered surfaces
and for painted and covered surfaces, respectively,
while for steel the suggested value is 0,625).
From the structural design point of view, the most
important difference between aluminium and carbon
151
steel is the shapeofthe stress-strain curve. In fact, in
case of aluminium, the stress-strain curve is non¬
linear and there is no well-defined yield stress due to
continuous hardening. Besides, it has to be
considered that as a function of both the chemical
composition of the alloy and of the possible type of
applied heat-treatment, there is a great variety of
aluminium alloys having mechanical features
substantially different from each other in terms of
limit strength, ductility and shape of the stress-strain
curve. On the other hand, the mechanical properties
of aluminium at transient high temperatures are
complex and cannot be easily generalised. A wide
preventive determination of experimental data and
accurate material models is therefore necessary for
an accurate and reliable determination of load-
bearing capacity of aluminium structures under fire
conditions.
The main parameters for characterising the
mechanical behaviour of aluminium alloys in the
elastic range are Young's modulus (E) and 0.2%
proof strength (f0.2), the latter being used as
conventional elastic limit ofthe material. The former
is practically independent from the adopted alloy,
while the latter is related to the type of alloy under
consideration. In Figure 1, values suggested by
Eurocode 9 for typical aluminium alloys and tempers
at elevated temperatures are depicted. It can be
observed the remarkable variation of the
conventional elastic limit with temperature (/ο.2,τ)
due to type of alloy and temper. For instance,
relative strength values at 200 °C are 90 and 40
percent for 5052-O and 7075-T6 alloy, respectively.
Even though there is any possibility to gather typical
common trend among different groups of alloy, it
can be noted that the beneficial effects of heat
treatment and work hardening, used to improve the
mechanical properties at normal temperatures,
diminishes at high temperatures.
On the other hand Eurocode 9 (ENV 1999-1-1) does
not supply any further information regarding the
material behaviour in the plastic range, not allowing
for taking into account at elevated temperatures the
rather important contribution provided by the
hardening features of aluminium alloys.
Several additional data gathered by test results,
including ultimate strength (fu) and ductility ($,) of
typical aluminium alloys are provided in ASM
(1994). From the analysis of relevant results, the
following remarks may be drawn: (1) for heat treated
alloys (T) and work hardened alloys (H) the variation
of hardening ratio (fjfio.2) with temperature is slight,
while it is considerable for alloys in the annealed
stage (O); (2) the inelastic behaviour of the alloy is
independent from the adopted treatment for
temperatures higher than 200 °C; (3) there is a
remarkable increase of uniform strain with
increasing temperature for all types of alloys; (4)
Inelastic features of aluminium alloys are strongly
 affected by the time of exposure at elevated
Initial modulus of elasticity in % of normal temperature, showing higher ductility and lower
value at ambient temperature ultimate strength as far as the exposure time
100% - increases, see Conserva et al. (1992). Characteristic
temperature dependence curves of the above
80% - parameters for some typical alloys are shown in
Figure 10, permitting some ofthe above conclusions
iS 60% to be stressed.
ω 40%

20% -
5.3 Ongoing research projects in Europe

η0/, -
Many issues related to the structural behaviour of
aluminium exposed to fire have to be still solved in
() 100 200 300 400 500 600
order to allow the designer to apply accurate and
Temperature (CC)
rational procedure for the safety assessment of
0.2% proof strength in % of normal value at
members and structures. Such aspects are concerned
ambient temperature with both more realistic material modelling to be
1.00 , 5052-Ο used for global structural analysis and the actual
5052-H34
* 5083-O response of aluminium members at elevated
Κ 5083-H113
0.80 β 5454-0 temperatures.
5454-H32
6061 -T6 The behaviour of aluminium in fire is presently
50.60 θ 6063-T6
r-
'- -) 6082-T6 under investigation at University of Naples Federico
-ώ 3003-O
τ>Γ
3003-H14 Π. The general scope of the research activity is the
*%40 \ -Β
-Β
5086-O
5086-H112 definition of reliable rational methods for structural
-· 7075-T6 fire design. Therefore, mechanical models able to
0.20
account for the material behaviour at elevated
Temperature (°C) temperature in both elastic and inelastic ranges have
0.00
been proposed. Such models are based on the well
0 100 200 300 400 500 600
known Ramberg-Osgood law (Mazzolani 1995),
Figure 9: Main mechanical properties of aluminium at elevated whose parameters are given in a closed-form as a
temperatures as suggested by EC9.
function of the corresponding values at room
temperatures and specific temperature-dependent
relationships, which have been calibrated on the
basis of existing test results (Ponticelli, 1999). Τη
cooperation with University of Liege, the above
material model has been introduced into the well
know computer program SAFIR (Franssen et al.
1995), allowing the behaviour of complex
aluminium structures exposed to standard fires to be
predicted up to plastic collapse by means of an
accurate but direct and quite simple approach.
Wide research activities are also carried out at the
NTNU of Thondheim. In particular the behaviour of
100 200 300 400
rectangular hollow cross-section columns made of
150.0 6082 aluminium alloy (both temper T4 and T6)
5052-O
5052-H34 subjected to fire has been recently investigated by
125.0 -9 6063-T6 Langhelle and Amdahl (1998) by means of both
- -O- - 6063-T1 experimental tests and complex numerical analyses.
100.0
Several interesting conclusions are drawn in the
¿ 75.0 above study, emphasizing the necessity to undertake
further research activities to obtain more
50.0 comprehensive database and a better understanding
of some fundamental phenomenological aspects
25.0
Temperature (°C) which have been evidenced by tested aluminium
0.0 alloys contrarily to the simplified schématisation
100 200 300 400 suggested by the present design code. In particular,
some of the important issues arisen are: (1) creep
Figure 10: Typical inelastic behaviour of aluminium alloys at
elevated temperatures (exposure time at testing temperature up effects are irrelevant when constant heating rate is
to 10000 h). applied, while they are important in case of tests

152
carried out under constant temperature; (2) EC9
yields reasonable capacities for tested columns,
except for those tested with constant heating rates,
due to the fact that creep effects are not correctly
taken into account in such a condition; the difference
between experimental and design capacities depends
on both the type of temper and the fabrication
process, it being larger in case of welded columns
and temper T6; (4) reduction of yield strength of AA
6082 at elevated temperatures is different for temper
T4 and T6, while design codes give similar reduction
values for both tempers (in particular T4 exhibited a
less pronounced loss of capacity due to the artificial
ageing caused by the applied heating); (5) The
strength of the alloy is invariant with respect to
temper at 285 °C, and the detrimental effect of
welding completely vanishes at this temperature.
6 CONCLUSIONS
The European knowledge of fire design brings good
engineering tools for modelling of the structural
behaviour under fire conditions. Three steps of
procedure may be separated: the model of fire
scenario in the compartment, the heat transfer to the
structure, and the response of the structure. For all
steps are ready for use the design models. The
simplest design models are supported by design
charts and design rules. The engineering models are
based on natural fires scenario, refined transfer of
the heat between the atmosphere and the structure
and non-linear global analyses. The complex
models, based on the FE modelling of fire scenario
and 3D non-linear behaviour of structure, are ready
to be applied for prediction of the structural
behaviour under exceptional fire loading. All design
and knowledge steps are under development by the
experimental, numerical and analytical modelling to
address and to reach the tomorrow asked level of
safety.
REFERENCES
Allam, A. M., Fahad, M. K., Liu, T.C.H, Burgess, I. W., Plank,
R. J. & Davies, J.M.(1999): Effects of Restraint on the
Behaviour of Steel Frames in Fire. In: "Proceedings ofthe
Conference Eurosteel '99, Praha, pp. 279-282.
ASM (1994). Aluminium and Aluminium Alloys. ASM
Speciality Handbook, J.R. Davis Ed., The material
Information Society.
Bailey, C. G., Burgess, I. W. & Plank, R. J. (1996) Computer
Simulation of a Full-scale Structural Fire Test, The
Structural Engineer, 74 (6), pp. 93-100.
Conserva, M., Donzelli, G., Trippodo, R., (1992). Aluminium
and its applications. EDIMET, Brescia.
Drysdale, D. D. (1999) An Introduction to Fire Dynamics, 2nd
ed, West Sussex, England, John Wiley & Sons.
Drysdale, R.G., Hamid, A.A. & Baker, L.R. (1994) Masonry
structures: behavior and design. Prentice-Hall. Inc. New
Jersey.
153
ENV 1991-2-2 (1995) Eurocode 1, Basis of design and actions
on structures - Part 2-2: Actions on structures - Actions on
structures exposed to fire, Brussels.
ENV 1993-1-1 (1992) Eurocode 3, Design of Steel Structures -
Part 1-1: General rules and rules for buildings, Brussels.
ENV 1993-1-2 (1995) Eurocode 3 - Design of steel structures
- Part 1-2: General Rules - Structural fire design, Brussels.
ENV 1996-2 (1990) Design of Masonry Structures. Part 10:
Structural Fire Design, Brussels.
ENV 1996-1-2 (1995) Design of masonry structures - Part 1-
2: General rules - Structuralfire design, Brussels.
ENV 1999-1-2 (1998) Eurocode 9: Design of Aluminium
Structures - Part 1-2: General rules -Structural Fire
Design. CEN, Brussels.
Forsén N.E. (1995) Fire resistance, in F. M. Mazzolani
"Alurninium Alloy Structures", 2nd Edition. E&FN Spon,
London.
Franssen, J.-M.; Schleich, J. B. & Cajot, L.-G., (1995) A simple
model for fire resistance of axially-loaded members
according to Eurocode 3; J. Construe. Steel Research, Vol.
35; pp. 49-69.
Hahn C. (2001) Alternative Wege zur Ermittlung der
Feuerwiderstandsdauer von Mauerwerk. Mauerwerk
Kalendar 2001.: pp.353-377. Verlag Ernst & Sohn, Berlin.
Kirby B.R. (1995) The Behaviour of High-strength Grade 8.8
Bolts in Fire. J. Construct. Steel Res., 33, 1995, 3-38.
Kollman F.F.I, Coté W.A. (1968) Principles of wood science
and technology, Springer Berlin , p. 592, ISBN 03-
87042970.
Langhelle N.K. and Amdahl, J. (2001) Experimental and
Numerical Analysis of Aluminium Columns Subjected to
Fire. In Proc. Eleventh International Offshore and Polar
Engineering Conference, Stavanger, Norway, 17-22 June,
406-413.
Lennon T. (1997) Cardington Fire Tests: Survey of Damage to
the Eight Storey Building, BRE internal Report
GD1286/86.
Lundberg, S. (1995) Design for fire resistance. In Training in
Aluminium Application Technology (TALAT) EU-
COMETT Program, F. Ostermann Ed., Aluminim Training
Partnership, Brussels: Section 2500.
Mazzolani, F.M. (1995) Aluminium Alloy Structures - 2nd
Edition. E&FN Spon, London.
Newman, G.M., Robinson, J.T., Bailey, CG.: (2000) A New
Approach to Multi-Storey Steel-Framed Buildings (SCI-
P288). Ascot, The Steel Construction Institute.
O'Connor, M. A. & Martin, M. D. (1998) Behaviour of a
Multi-storey Steel Framed Building Subjected to Fire
Attack, J. Construct. Steel Research, 46 (1-3), Paper No.
169.
Ponticelli, L. (1999). Methods for fire safety assessment of
aluminium Structures (in Italian). Graduation thesis,
University of Naples Federico II (Supervisors: F.M.
Mazzolani, R. Landolfo, G. De Matteis).
Schleich J.B., Kmpa J., Newman G., Twilt L. (2001) Model
Code on Fire Engineering, ECCS No. Ill, Brussels,
p. 165, ISBN 92-9147-000-65.
Schleich, J.B. at al (1999) Competitive Steel Buildings through
Natural Fire Safety Concept, CEC 7210-SA / 125, 126,
213, 323, 423, 522, 623, 839, 937, Brussels.
Simões da Silva, L.A.P., Santiago, A. & Vila Real, P. (2001) A
component model for the behaviour of steel joint at
elevated temperatures, Journal Constructional Steel
Research 57(11), 1169-1195.
STEP (1995) Timber Engineering, Centrum Hout, Amsterdam.
Vila Real, P. M. M. & Franssen, J.M. (1999) Lateral buckling
of steel I beams at room temperature - Comparison between
the EUROCODE 3 and the SAFIR code considering or not
 the residual stresses", internal report No. 99/01, University
of Liege.
Vila Real, P. M. M. & Franssen, J.M. (2001) Numerical
Modelling of Lateral Buckling of Steel I Beams Under Fire
Conditions - Comparison with Eurocode 3", Vol. 11, No.
2, Journal of Fire Protection Engineering, USA.

154
Evaluation of Structural Robustness of Members and Connections

L. Simões da Silva, L. Neves

University of Coimbra, Portugal
L. Baniotopoulos1, P. Perdikaris2, M. Zygomalas1
Aristotle University of Thessaloniki, University ofThessaly, Greece
V. Bosiljkov
Slovenian National Building and Civil Engineering Institute, Slovenia
H. Bouchair
CUST- Lermes Blaise Pascal University, France
G. De Matteis
University of Naples Federico II, Italy
D. Dubina
"Politehnica" University of Timisoara, Romania
P. Haller1, U. Kuhlmann2, F. Kühnemund2, H. Stangenberg3
TU Dresden, University of Stuttgart, University of Aachen, Germany
G. Huber
Aste Engineering, Innsbruck, Austria
A. Koslowski
Rzeszow University of Technology/University of Beira Interior, Poland/Portugal
F. Wald
Czech Technical University in Prague, Czech Republic

ABSTRACT: A state-of-the-art review of the evaluation of the ductility of structural members and
connections in view of robustness is presented in this paper, focussing on the specific contributions of the
COST C12 members. Firstly, a general methodology common to all structural materials is presented for both
members and connections. Next, specific aspects for the various materials are described, together with
reference to available experimental, numerical and analytical work.

1 INTRODUCTION systems was predominant in that research work.

Robustness of a structure represents the ability of
the structure not to suffer a major collapse due to
minor accidental damage. According to Bertero
(1997), a robust steel structure should be provided
with balanced stiffness, strength and ductility
between its members, connections and supports.
Because the issue of ductility is still the most
difficult to establish, special emphasis in this paper
will be devoted to the description of the various
methodologies and current developments (analytical,
numerical and experimental) for the evaluation of
ductility for the various structural materials.
2 GENERAL METHODOLOGY
ο ι ι/ ζ.
2.1 Members
2 1 1 T t d t'
. The ductility of structural members was
The evaluation of the ductility of structural
members was a topic of several research programs in
the last years. The determination of available
rotation capacity for plastic distribution m structural
155
These results are important for the examination of
structural robustness both in general and also for
exceptional loading.
Subsequently the testing procedure for the
determination ofthe ductility of structural members
is illustrated, followed by the respective definition
for the available deformation capacity. Different
authors recalculated the test behaviour using
numerical or mechanical models. These models are
introduced with references for further study,
Different concepts for the check of sufficient
rotation capacity are followed by the codes or
researchers, respectively. A short overview over
these procedures closes this state-of-the-art review.
2.1.2 Tests
2Λ.2Λ Testing procedure and test results
investigated in three-point-bending tests. Figure 1
shows the prmciple of the corresponding test set up.
From the measured beam end rotation the angle of
rotation φ as shown in Figure 1 could be derived.
 The typical result of these tests is a non-linear
moment-rotation curve with a stable increasing
branch, a maximum bearing moment capacity and a
post critical decreasing part. The plotted rotation is
not a physical value, but derived from the measured
end rotation. It characterises the ductility of the
beam including the plastic curvature in the yield
zone.
Figure 1. Principle oftest set up and test result of a three point
bending test.
2.1 .2.2 Definition of the members ductility
Up to now, the ductility of members is mainly
dealt in view of the available rotation capacity. It
characterises the ability of a plastified section to
rotate while maintaining its design moment
resistance. Due to strain-hardening effects the real
moment exceeds the ideal plastic moment Mpi. With
increasing rotations the moment graph passes a
maximum value and decreases until it reaches the
level ofthe ideal plastic moment Mpi again.
The available rotation capacity <j>avaii of a member
is defined as the rotation from the point Mpi is first
reached (under assumption of ideal-elastic/ideal-
plastic behaviour) to the point the real --curve
reaches the level of Mpi again (see Figure 1).
2. 1 .3 Models for the determination of
ductility
2. ι .3. ι Models for I-shape sections
Kuhlmann (1986, 1987a,b) developed a numerical
procedure to calculate the moment-rotation-
behaviour of I-shaped beams considering local
buckling in the plastic range. The behaviour of the
beams in the post critical part of the moment-
rotation-relationship is considered by a yield line
model for the buckled web and flange parts (see
Figure 2). Feldmann (1994) uses a similar model for
representation of the post critical buckling behaviour
of I-shape beams. Spangemacher bases his
investigation on numerical simulations of I-shaped
beams. Axhag (1998) presented in his thesis a cross
section model, which predicts the moment-rotation
relationship for slender I-shaped girders made of
high strength steel.
156
Figure 2. Yield line model for the post buckling behaviour of a
beam flange from Kuhlmann (1987a).
2.1.3.2 Models for hollow sections
Research work of Stranghöhner (1995) focuses on
the ductility behaviour of hollow sections. She
follows a similar procedure as the researches
mentioned above. Tests are compared to results of
numerical and mechanical modelling. The
decreasing branch of the respective moment-
rotation-curve is represented by a corresponding
yield line model for the webs ofthe hollow section.
2. 1 .4 Concepts for check of sufficient ductility
2.1.4.1 General demand
The application of plastic hinge theory requires
sufficient rotation capacity of those sections, where
plastic hinges occurs, except for the section forming
the final hinge. Consequently the available rotation
capacity of the section φβνβΐΐ must be larger than or
equal to the required rotation φ^ as demanded from
the structural system. Equation (1) has to be
fulfilled.
Ψavail πreq (1)
2.1.4.2 Deemed-to-satisfy-criteria
For practical design rules should be simple to use.
Eurocode 3 (1992) replaces the explicit check of
sufficient rotation capacity by the check of profile
slenderness values. From thorough scientific
investigations boundary values are derived due to
flange and web slenderness and stress distribution in
the cross section. The rotation capacity of members
is limited among others by local buckling. In
dependence of the slenderness values of its. parts
Eurocode 3 (1992) classifies the sections to allow for
different kinds of design methods. In case the
slenderness values according to class 1 are satisfied
no further check of rotation capacity is necessary.
2.1 .4.3 Explicit check of sufficient rotation capacity
Feldmann (1994) and Spangemacher (1992)
follow a different path for the check ofthe sufficient
rotation capacity. For the determination of the
available rotation capacity of I-Shaped sections a
formula is given, which is based on tests as well as
on numerical and mechanical parameter studies. By
that the field of application is extended, e.g. also on contributing components, followed by their
the check of the ductility capacity for plastic design characterisation (determination of their structural
of structures under cyclic loading from earthquakes. properties as force-deformation curves) and ending
However for each case the required rotation capacity up with the assembly ofthe component properties or
has to be determined in dependence of the structural curves to those of the joint.
system and on the relevant level of sophistication.
Further investigations on the required values of
rotation capacity also incorporating the joint
behaviour are necessary.

2.2 Connections
2.2.1 Introduction
Conventionally, connections are treated either as
nominally pinned or as fully continuous. A modern
approach is to develop efficient connection types
first and then to take their realistic behaviour into
consideration within the frame analysis. An accurate
representation of the joint behaviour forms the basis
for a correct and safe structural design, where the
available joint properties have to be compared with
the requirements from the structure at serviceability
and ultimate limit states.
Connections have to transfer moments and forces
between members with an adequate margin of safety.
Their behaviour obviously influences the distribution
of forces within the structure. The connection's
representation can be subdivided into the four steps:
characterisation, classification, idealisation and
modelling. This complex process allows finally to
establish adequate procedures to ensure that equation
(1) is satisfied, and is described below in more
detail.
actual curve statistical «valuation
ι quasi-static test from »»veril tasts
Figure 3. Joint characteristics
Attention should be paid to the delicate wording
when talking about rotations (Figure 4). φρι is the
(overall) rotation capacity, whereas (φρι - φβι) should
be named the plastic rotation capacity (necessary for
global redistribution). So this plastic rotation
capacity of joints corresponds to the rotation
capacity of members (beams and columns). However
whereas in the case of plastic hinges within members
only this plastic rotation appears between the axis of
the adjacent members (the elastic rotation only leads
to a common curvature), within a joint both the
elastic and the plastic rotation can be observed as an
angle between the connected elements.
φ ρ) ... (overall) rotation capacity
τ rodisi =
Φ οι "Φ* Pestio rotation capacity
φ ,/φ., ...rotation capacity factor
Φ

2.2.2 Characterisation
Φ.. *
Figure 4. Proposed terms in view of rotations
That means the determination of the joint's
moment-rotation curves (Figure 3); or, at least, the 2.2.2.1 Component identification
determination of the corresponding key values According to the Eurocodes a (basic) component
(initial rotational stiffness, moment resistance and of a joint is a specific part that makes an identified
rotation capacity). This can be achieved with contribution to one or more of its structural
different approaches: (i) experimental, using full-
scale joint tests (Huber 2000, chapter 3.6), (ii)
numerical, based on finite element calculations and
(iii) analytical.
In the following the component method will be
used as the analytical tool to determine the joint
response, especially its ductility. An analytical
description of the behaviour of a joint has to cover
all sources of deformabilities, local plastifications, Not cofnpofMitt

plastic redistribution of forces within the joint itself

1 Marlor .»led web panal
2 concret· ancaaamant

and local instabilities. Due to the multitude of

3 axtanor «Ml wtb pan·, (oatumn ttar-^) aflad*)
4 «>ff«^o<c«nwtU»wMiwiwrtoi.«ti»>T.qr*p>TWg

influencing parameters, a macroscopic inspection of 3

β
beam flange (toa. effade). oonlact piale, and plata
ittidvvApaft^bìclpavtofAirtga.faMndlui

the complex joint by subdividing it into components a >x.>jrnn1*:w^iErib*rto1ng(tdinm*d)

has proved to be most appropriate. The components 9

10
arrdpiata.hb*rrdrig,b«amvi^htenikin
bolUintafu]«.

can be modelled by translational springs with non¬ 11

12
nlnlonmmMt (M^topen*) ht t**ton
>]<.pofc>iM*pMHabMm(dMtolnc»ai^

linear force-deformation response. The joint 13

14
i^b«ctton of urtatane^
itM. web pan»! In «rtaar
foroe·

characterisation basing on the component method 16 concrete «ncMMntxtt In «haar

therefore starts with the identification of the Figure 5. Component model (spring model)

157
properties. When identifying the contributing
components within a joint one can distinguish
between components loaded in tension (or bending),
compression and shear. Apart from the type of
loading one can distinguish between components
linked to the connecting elements, those linked to
load-introduction into the column web panel and the
component "column web panel in shear". The nodal
subdivision leads to the component model (spring
model, Figure 5).
2.2.2.2 Component characterisation
The component model is based on known force-
deformation curves of the individual contributing
components, which have to be derived within the
component characterisation (components' key values
or even full curves). Again this can be done at
different levels of accuracy using different tools,
such as component tests (experimental approach),
finite element simulations (numerical approach) or
analytical mechanical models.
Sophisticated mechanical models describing the
r_u
a»
Figure 6.
rzn
Different accuracy levels of assembly
E3
Two different tools have been provided by the
authors for the assembly of component F- curves to
the joint's Μ-φ curve. In (Silva and Girão, 2000)
formulae are developed on the basis of an energy
formulation that enable the evaluation of the Μ-φ
curve. Alternatively, a computational tool is
described in (Huber 2000, chapter 4.3) which is
based on an iterative approach for the non-linear
branches (Figure 7) covering both the sophisticated
and the simplified component model.
i} BALANCED P=0
Γ»« - On .0BXWI N

basic components' response have been developed at cv M nn-vomen**!

¡-muml *} tu»
several research centres and have been validated
.-."ί*"
against component test results and numerical
simulations. With the help of comprehensive
parameter studies the sophisticated and relatively
complex formulae describing the stiffness, resistance
and deformation capacity of each basic component -Lkj *m|»^ Innot ^
VH · 1« fimdl
could be reduced to easy-to-handle formats. Such tá ·* 1M*7 |<it»m]
«t"*JM (mm]
ι>ί'7ΛΙ1 1anâ\
simplified formulae are integrated into the codes. H »*v-(IÄ>4|Wl
I* tofPWfl pt*t¡
»onvc-n » öm fv»!]

2.2.2.3 Component assembly WtïlN y-íi.3

Figure 7. CoBeJo program for component assembly

The transfer from force-deformation curves ofthe
individual joint components to the moment-rotation 2.2.3 Classification, Idealisation
curves of the full joint has to be done based on the
The classification is a possible but not obligatory
component model fulfilling the requirements of
tool for simplification of modelling providing
compatibility and equilibrium. Doing so it is assured
boundary conditions defining the limits for the use
that the joint model behaves exactly in the same way
of conventional types of modelling (nominally
as the complex component model with respect to
continuous or nominally pinned). In view of the
applied moments.
drastic increase of frame software capability the use
Depending on the intended level of accuracy a
of advanced joint models will become more and
more sophisticated or a simplified component model
more natural. When still aiming at conventional rigid
(different modelling of the components' interplay)
or hinged joint models the joint characteristics in
can be chosen (Figure 6). Furthermore the assembly
view of stiffness, strength and rotation capacity have
can be done for the main rotational key values only
to be compared to classification boundaries.
(initial rotational stiffness, plastic moment
Depending on the method of global analysis either
resistance, rotation capacity) or for the full shape of
all or only one of the classification checks becomes
the resulting Μ-φ curves. Depending on those
decisive.
choices the assembly either has to be done by an
In the following the required rotation for full
iterative computational procedure or can be solved
redistribution within the frame ("ductile" joints) will
by a simple calculation adding up the individual
be provided: In systems where the first plastic hinges
springs step by step parallel or in series. Especially
form in the joints (due to partial strength MjiR(j<MRd,
for the determination of the joint's failure mode and simultaneously at both beam ends), the joints have to
the resulting rotation capacity (ductility) a simplified
provide sufficient overall rotation capacity φρι for
analytical approach for a hand calculation could not
plastic redistribution within the beam. This required
yet be provided. The codes are therefore only giving
overall rotation φρι can be shared into a part φει (ideal
rough guidelines.
elastic=MjjRd/S), the elastic rotation until the hinge in
158
 the joint has formed, and a part φΓε<ϋ8ΐ, the further global analysis and to the classification of the joint:
rotation necessary to allow for a final plastic hinge at A continuous joint ensures full rotational continuity,
midspan due to global redistribution. The ratio a semi-continuous joint provides only partial
between φβΐ and φπ^ obviously depends on the joint rotational continuity and a simple joint prevents any
stiffness S, however the overall required rotation rotational continuity between the connected
φρΐ= φεί + φκαΐβΐ is independent of the joint's stiffness members (Figure 10).
(Fig. 6).
In contrast to its independence from the joint's
stiffness the required rotation indeed depends on the
joint's strength: φρι decreases - following nearly a
h rigid, full-strength continuous

h
rigid or semi-rigid,
full or partial strength, seml-contlnuous
linear function - with increasing moment resistance specific rotation capacity
of the joints. In other words the required rotation of a
high strength joint is lower than that of a hinge
(Figure 9). Details can be taken from (Huber 2000,
h pinned (hinged)
(sufficient rotation capacity)
FigurelO. Types of joint modelling
simple

chapter 3.10)
EI
τ = SL 2.2.5 Conclusion
Comprehensive knowledge is available on the
(m^-1|mj) = ^
field of the analytical derivation of the joint's
stiffness and resistance whereas the determination of
1 + 10γ
384 ΕΙ 1 + 2γ
the rotation capacity still needs further research
work. On the one hand only little information exists
at present in view of the components' deformation
IX, 1] Λ * 12-(1+2γ) capacity and on the other hand comprehensive
M,
parameter studies will then be necessary to evaluate
P-L2
M- = |îi.(l+6y) the interplay of different components in view of the
overall rotation capacity. Special attention will have
S-0
"-"« M-0 Φ--0 ~-èM·»-
to be paid to the fact that already a slight
3B
overstrength of a ductile component may change the
«. 1
tt>mt "M,^ Μ·Μ, .-¿;(mw-0.5.mJ ^.0
failure mode to the second weakest and maybe brittle
ΟίΜμ,,ίΚν, (panisi »Iranglh joint) M«,íM„,
component! That can result in a completely different
,-«+_ -^--^^«,-Ο.δ-Μ^) IndipMidiiitiToniS
rotation capacity ofthe whole joint.
Figure 8. Required rotation capacity for full redistribution These effects are being studied at different
research centres under the umbrella of the actual
Fe 360 Fe 510 COST-C12 project. The comparison between
M S235 M S355 required and available ductility values should finally
25 35 lead to a classification system for joints hopefully as
MRdd MRd
¡o
simple as that for beams where width-to-thickness
ratios lead to 4 cross section classes.

!
1
AH.
3 3 CONCRETE STRUCTURES
Recently, in relation to the "displacement-based"
.s
,. φ ImradJ
50 m« Φΐ""«« 70», seismic design of new R/C structures and the
MRdambi(MhagiRdi?MMn , 2M|m
evaluation and retrofitting of existing R/C structures,
Figure 9. Proposal for ductility boundaries there has been an increased interest in estimating the
deformation-capacity of R/C members. The 1997
Idealisation means the conversion of non-linear NEHRP Guidelines for the Seismic Rehabilitation of
moment-rotation curves into simplified (linearised) Buildings (FEMA-273/274 1997, FEMA-356 2000)
ones. Depending on the available software either the evaluate a member by comparing the seismic
full non-linear shape of the joint-curves or demand and the member's capacity in terms of
simplifications of them can be assigned to the deformations. In order the performance-based design
respective joint springs. of reinforced concrete structures subjected to
earthquake loads to result in improved structural
2.2.4 Modelling performance, aside of other criteria such as
specifying realistic demands, the capacity and
Modelling is the reproduction of the connection ductility of component members and connections
behaviour within the structural analysis. Linked to need to be evaluated and/or predicted with as much
the joint characteristics there are three types of joint certainty as possible. The local moment-, rotational¬
modelling which are closely linked to the method of and ductility-capacity of R/C beam and column
159
components (and connections) or the global drift-
capacity of the structural system under earthquake-
type loads on the one hand, and their shear strength
on the other hand, are of major importance. This is
true for either designing new structures or repairing
existing ones. This phase of the design process,
which may be associated with either primary or
secondary structural components, refers to either
strength or stiffness (deformation) evaluation at
different performance levels. The force reduction or
response modification factors, R, assigned to
different structural and material types by different
design codes effectively reflect the perceived
available ductility of a structural system (Paulay &
Priestley 1992), although sometimes account for
effects additional to ductility. Models which
consider the effects of inclined cracking, bond-slip
law at the concrete-steel interface, the contribution
of concrete in tension and a more appropriate σ-ε
diagram for the steel reinforcement have been for
example proposed to determine the plastic rotation
capacity of beams under monotonie loading (CEB
1993, 1998). Because of the rather complex
phenomena involved during the inelastic response of
R/C members, in addition to the on-going needed
research investigations on the development of more
advanced models for determining the load- and
deformation- capacity, further experimental research
is needed especially under cyclic loading to
determine their true inelastic response and validate
the proposed models.
Based on a large database of mainly cyclic test
results on R/C beams, columns and walls it was
concluded (Panagiotakos & Farthis 2001) that: (a)
steel reinforcement ductility (high-ductility steel vs.
cold-worked steel or tempcore steel) influence
strongly the R/C member's deformation capacity, (b)
load cycling appears to be important in reducing
substantially the deformation capacity only if it is
applied fully at the maximum deformation, while the
number and magnitude of deformation cycles before
ultimate is unimportant, (c) the most important
parameter in increasing the member's deformation
capacity appears to be the shear-span ratio, (d) the
effect of confining reinforcement on ductility is not
as high as expected, (e) for values of concrete
strength up to 120 MPa, the increase in ductility is
equally positive
Figure 11. Experimental setup for the beam-column specimen.
160
as the compression-to-tension steel ratio, (f)
increasing the applied axial load from zero to that at
balance results in an almost linear decrease in
deformation capacity to about 50% and (g) all other
geometric and mechanical parameters being equal,
shear walls failing either in flexure or shear-flexure
exhibit deformation capacities of about 30% of that
of a beam or a column member.
A current experimental research effort (Kotsovos
& Perdikaris 2001) at the National Technical
University in Athens and the University of Thessaly
in Volos is underway to investigate the available
strength and ductility in R/C members at the so-
called "critical locations" and determine the
appropriateness of the design requirements and
procedures of various design codes, which may
result in grossly under- or over-estimating the load
and ductility capacity of R/C members. The main
objective of this project is not only an improved
design procedure for new R/C structures that insures
the desired ductility but also for repairing and
stiengthening damaged R/C structures. Two-span
(1.20-m and 1.95-m long) continuous 15-cm (wide)
χ 30-cm (high) R/C prismatic beam-column
specimens are tested under a concentrated
monotonie and cyclic transverse loading applied at
midspan ofthe 1.95-m long span for various levels
of constant axial load (see Figures 11-12). The
concrete specimens are reinforced with 8500-2φ14
steel rebars on each top and bottom face and
appropriate number and distribution of 8320-φ6
closed steel stirrups according to the current Greek
design code for R/C structures (EKOS 2000) and the
"compressive-force path" concept (Kotsovos &
Pavlovic 1999). The applied transverse load and
Figure setup.
support system used for the tests result in the
formation of two plastic hinges (underneath the
applied load and at the middle support) and an
inflection point between the load-point and the
middle support, a common condition for a column
member of a building subjected to lateral seismic
loads. Representative load-displacement loops under
monotonie and cyclic transverse load application for
a constant axial load equal to N=315 kN (about 44%
of the axial load at the balanced condition) are
shown in Figures 13a and 13b.
(Jd) Cyclic Loading
Figure 13. Load-displacement loops for a specimen designed
according the current design code for R/C structures in Greece.
4 STEEL STRUCTURES
4.1 Robustness: Redundancy, Overstrength and
Ductility
A robust steel structure should be provided with
balanced stiffness, strength and ductility between its
members, connections and supports (Bertero, 1997).
Conceptually, Seismic Resistant Steel (SRS)
structures are always redundant, because redundancy
is the inherent condition of the reliability of
structural systems. Therefore, in the American
Design Codes (UBC, 1997, AISC, 1997),
redundancy (or reliability) coefficient is a
component of reduction factor R (corresponding to
q-factor in Eurocode 8). Bertero R & Bertero V
(1999) have shown the redundancy is dependent of
overstrength and ductility of structural system. In
fact, to take advantage of redundancy, it is necessary:
1) to decrease the coefficient of variation, (COV), of
the demand relatively to the COV of the supplied
capacity (Nakashima, 1998); 2) to provide the
necessary overstrength of structural components in
order to assure the yielding of dissipative-designed
zones (Calderoni et al, 1996, Ohi, 1998); 3) to
increase the plastic rotation capacity, i.e. ductility
(Gioncu, 2000); 4) to guarantee a minimal rotation
capacity in all members of the structural system so
that they can follow the displacement ofthe structure
without failure and allow other elements to dissipate
the earthquake input energy. Therefore, structural
robustness means redundancy, overstrength and
ductility.
As a measure of redundancy (overall overstrength)
can be taken the ratio ccu/a,; au- the ultimate
multiplier of seismic action; a, - the multiplier of
seismic action corresponding to the first plastic
hinge. This definition is used in Eurocode 8(1994) to
define the reduction factor; for MRS frames, for
instance, q = 5xau/a,. Dubina et al (1999)
suggested as a measureof structural redundancy the
Seismic Performance Factor (SPF):
SPF-EPA' (2)
EPA.
EPAU = Ultimate Effective Peak Acceleration,
corresponding to the seismic input, evaluated for the
most unfavorable failure criteria: drift, plastic
rotation capacity for members and joints and plastic
instability (collapse);
EPAS = Effective Peak Acceleration induced by
seismic excitation
SPF > 1, and can take different values, according to
the performance objectives of design.
A special MRS frame designed according to SCWB
philosophy (Strong Column Weak Beam), and based
on the AISC- 1997 provisions, is an example of a
redundant structure. Such a frame, well designed, is
capable to withstand ground motions of twice the
design level very little likelihood of collapse,
whereas a frame designed according to the WCSB
approach (Weak Column Strong Beam), is ill-
conditioned and may develop a collapse mechanism
at an excitation level well below twice the design
level (Biddah A & Heidebrecht, 1999). However, the
WCSB traditional philosophy must nowadays be
revised to account for the performances of Partial
Moment (PR) Frames (Elnashai, 1998 and Salazar &
Halder, 2001) and also for the Reduced Beam
Section Moment (RBSM) Frames (Uang & Fan,
2000). In fact, in these cases, the redundancy and
ductility of structure is obtained by weakness (thus
not overstrength, obviously!) of beam-to-column
joints and beam-to-joint zones, respectively.
4.2 Performance based design: Methodology
and Modelling
Robustness of a SRS structure, e.g.
redundancy/overstrength and ductility, must be
designed to satisfy specific performance objectives.
SEAOC-Vision 2000 (1995) and FEMA 273 (1996)
documents proposed four performance levels
expressed in terms of interstorey drift limits (Table
Ο
ΙΟΙ
 Multilinear Μ-φ models for beam-to-column
Table 1. Performance level limits
joints calibrated with test results (Figure 14).
Performance level Damage state Drift JJDDR -300-
[3-5] [%]
Fully operational, No damage 0.2 1
Immediate occupancy
Operational, Damage Reparaible 0.5 0.3
Control, Moderate
Life safe-Damage state Irreparable 1.5 0.6 C 35

Near collapse, Severe <2.5 0.8

Limited safety, - - Experimental
DRAIN 2DX model
Hazard reduced, >2.5 1.0
Collapse -300-
Rotation [rad.]
For comparison, in Eurocode 8 (1994) the Figure 14. Calibration of joint model
allowable drift takes values between 0.004h and

μΐ
0.00h, while UBC 97 allows for values between
0.02h and 0.025h; h is the height of the storey.
Alternatively, SEAOC-Vision 2000 defines the
structural performance in terms of displacement
ductility demand, measured by the Inelastic
Displacement Demand Ratio (ÏÏ0DR). The values are
also shown in Table 1: IDDR=0 corresponds to an
elastic performance; IDDR=1.0 corresponds to an
instability limit state and which the useable ductility
is fully expended; TDDR=0.6 corresponds
approximately to the upper limit for "life safe" level.
The IDDR criterion can be used in Displacement
Design approach (Court & Kowalsky, 1998), while
drift criterion is a target for Capacity Design
approach actually included in available design codes.
The design and analysis procedures (Ghabarah,
2001) play a very important role in the correct
prediction of robustness parameters. Foutch et al
(1998) shown that the calculated drifts at any
Figure 15 shows the difference in capacities of
members and joints, corresponding to the two
modeling techniques.

I
"5
!
The increase of capacity is on about 40-50%,
which lead to a 30% increase of capacity in pushover
analysis.
Obviously, the robustness supply of a given
structure is significantly dependent by the design
methodology, including both analysis procedure and
quality of models used in computation

200 '

1O0
Columns

^~ Fiber ·ι·ιτι Mit

Paitad «Η almod·!

rot lud Τ

specific storey in MRS frames can differ by more I 0.01 0.02 0.09 0.04 0.0S 0.00

than 50% depending on the analysis method used

and the ground motion considered. Also the Beams
computation models used for members and joints
may significantly affect the structural response. In a
recent study by Chitina et al (2001), the influence of iA
correct modeling of members and joints capacities is
analyzed. The seismic response of a three bay-four Fiber «Umani
story SCWB frame, subjected to far- and near-field so -
~ PsMftct «l-pl »mod·!

seismic actions is studied. The columns are X Γ «(rad.]

shaped welded steel sections, while beams are 0 0.01 0.02 0.03 0.04 0.03 Ο.Οβ 0. D7

double Τ welded sections. Beam-to-column joints

Joints
are of bolted extended end plate, their performance
being experimentally established (Dubina et al,
2001).
Two different sets of computation models for A
/" '
members and joints have been used in DRAIN 2DX
analysis, i.e.:
150

100
Ί? PtHct ·Η>Ι mod·!
1 ~"~ ΜιιΚΙΙπμγ mod·! -«xp*rim.
- Elastic-plastic bilinear Μ-φ models so

and I rot Ind.]

< 0.01 0.02 0.0} 0.04 0.05 0. D6

- Nonlinear inelastic fiber models for members, Figure 15. Μ-φ curves used in DRAIN 2DX simulation models
calibrated for given sections with DUCTROT
program (Gioncu & Petcu, 1999)

162
 4.3 Design Criteria and Detailing advantages and improved fire resistance. In this
paragraph some examples of these facts are reported
Structural robustness of steel structures must be
for steel and composite structures under monotonie
based on proper design criteria and accurate
and cyclic loading.
detailing, in accordance with material properties and
Tn seismic design composite frame structures are
type of loading effects (high strain rate, fatigue etc.).
designed to provide dissipative zones in the regions
If discussion remains in the field of MRS frames of maximum bending moments. This implies the
(but the problems are of similar type for braced and application of high behaviour factors taking account
dual steel frames), then the robustness depends on
of the amount of energy dissipated by the structure
the performance of (SAC Phase Π Project, Sanders,
when forming plastic hinges to achieve the full
1998): plastic chain. Thus, members and connections have
- SCWB design principle to be prevented from sudden failure as buckling or
- Panel zone strength cracking which is covered by the so-called capacity
- Connection strength and degradation
design such that dissipative zones are mainly located
characteristics
in the beam cross-sections.
- Ρ-Δ effects Considering recent research results in the field of
- Member local buckling composite it can be stated that composite members
The detailing analysis must mainly focus the and semi-continuous connections generally have
joints. The following design conditions are capability to perform quite well in static and cyclic
considered to be generic irrespective the connection loading. This aspect affects current capacity design
type:
which excludes connections and columns from
- Welded joints:
contributing to the global behaviour.
1. through-thickness strength
2. base material notch-toughness
3. weld wire notch-toughness 5.2 Minor axis joints
4. weld backing and ran out tabs
5.2.1. Experimental behaviour and analitycal
5. reinforcing fillet welds
model of semi-rigid minor axis composite seat and
6. cope hole size, shape, workmanship
web site plate joints
- Bolted joints:
1. Bolt sizing, hole type, tightening The activities of Rzeszow University of
2. Net section strength Technology in the field of composite joints are
In Europe a large COPERNICUS Project, called described below. Collection of tested composite minor
"RECOS", focused the above mentioned problems axis specimens was shown in Table 2.
and others, like influence of loading asymmetry, Table 2. Collection oftest specimens
strain rate effect and the global performances of SPECIMEN COLUMN BEAM REINFORCEMENT
MRS frames accounting for connection behaviour. ΓΡΕ240 6T10;p = 0,5%
CP-1 HEB 200
The results have been summarized in a book edited 10T10;p = 0,8%
CP-2.1 HEB 200 IPE240
by the coordinator of international research team, 10T10;p = 0,8%
CP-2.2 HEB 200 ΠΈ240
Prof. F.M. Mazzolani (2000).
CP-3 HEB 200 IPE240 14T10;p=l,l%
Connected to this subject the special issue of the
Journal of Structural Engineering (2000), entitled
Figure 16 shows a general view of the tests, and
"Steel Moment Frames After Northridge, must be
Figure 17 shows the comparison of bare steel and
also referred.
composite joints behaviour.
Robustness of steel structures is function of
structural redundancy, overstrength and ductility and
is a matter of design processus. It depends on the
quality of conceptual design, design philosophy and
performance criteria. Design methodology, analysis
IJ r
and modeling play a very important role in correct
evaluation of robustness parameters. Design
detailing and material properties are also key factors
of robustly designed structures.
» ,-

5 STEEL-CONCRETE COMPOSITE
STRUCTURES
5.1 Introduction
The behaviour of steel-concrete composite joints, &
compared with the bare steel structures, has enhanced
resistance and stiffness, besides other non-structural 6. General view of test

163
 5.2.2. Experimental behaviour of minor axis semi¬
An analytical joint model based on component rigid steel and composite end plate joints
method has been developed for prediction of the
moment resistance (Figure 18), initial stiffness Tests of semi-rigid minor axis steel and
(Figure 19) and rotation capacity. composite end plate joints, performed at the
g-120 University of Coimbra are referred below.
Ζ Two column and two beam types were considered,
2100 in two different joint configurations. Each of these
80
configurations was tested in monotonie loading
, _i_

under positive and under negative moment, and

60 CP-1 under cyclic loading, for the bare steel structure in a
*- CP-2 total of six tests. Six more similar tests were
40 --CP-3 performed adding concrete to the structure to obtain
20 the composite joint of Figure 21, and thus evaluating
the composite action in this particular joint layout -
0 I" 1

Table 3
10 20 30 40 50 60
φ [mrad] Table 3. Collection oftest specimens
Figure 17. Μ-φ curves for composite and bare steel specimen TEST COLUMN BEAM TYPE LOADING
El Monotonic M+
'"«?" t~*~< F, HEA220 ΓΡΕ200 STEEL Monotonic M-
E2
*· Τ
.aa·"' " Sb«: E3 Cyclic
E4 Monotonie M+
2 «ί
% a
E5 ΓΡΕ330 ΓΡΕ240 STEEL Monotonic M-
2M16(5.8)
Fe
E6 Cyclic
-"f E7 Monotonie M+
JPE240
4Μ1β(5.8)
¡j a, ιι a, E8 HEA220 ΓΡΕ200 COMPOSITE Monotonic M-
\ ii / E9 Cyclic
E10 Monotonie M+
Ell ΓΡΕ330 ΓΡΕ240 COMPOSITE Monotonie M-
Figure 18. Model of joint for predicting the moment resistance E12 Cyclic

X F, .

~fT

Figure 19. Spring model of joint for predicting the initial

stiffness

Initial joint stiffness prediction was developed in

the form of Equation 3.

»ϊ(±+±)-2ηΛΊ-+»2ΛΊ-+Ί-+τ-
(3)
SjJni -' Λ ι Κ,! ι ν ι
(λ, +
λγ
+ Χ
kc kb
+
kc
) τ
Kc

Figure 20. Geometry of the joints

164
As in the previous paragraph, joint stiffness and rot (rad)
resistance are strongly enhanced by the addition of
concrete (Figure 22).
The plastic moment of the steel joint in Figure 22
(obtained for a rotation of 20 mrad) is about 18
kN.m. A final moment of 50 kN.m is attained for a
large rotation of 200 mrad (limited by the stroke of
the hydraulic jack). This rotation clearly shows the
excellent ductility of this joint component and the
importance of membrane action.
In the equivalent composite joint, the maximum 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
moment attained was 90 kN.m, and the critical 22 steel and behaviour
component was not the column web loaded out of
the plan. Yielding and instability of the compressed
80
flange in the beam (IPE 220) led to the collapse of
60
the joint. It is to be noted, however, that in cyclic
ΊΟ
tests the degradation of joint characteristics is much
stronger in composite joints than in the bare steel
joints with the same geometry, with the former Έ 2
0.08 0.1

tending to the characteristics of the bare steel φ

ε
configuration after a significant number of cycles. ο
ES

This behaviour is illustrated in Figure 23. It is to be -Ε9 -composite

-Ε1 -steel
noted, however, that in cyclic tests the degradation
of joint characteristics is much stronger in composite
Figure 23. Comparison between steel and composite behaviour
joints than in the bare steel joints with the same under cyclic loading
geometry, with the former tending to the
characteristics of the bare steel configuration after a
significant number of cycles. This behaviour is 5.3 Major axis joints under static and cyclic
illustrated in Figure 23. loading
The current ECSC research-project "Applicability of
·> IT. π · ·
composite structures to sway frames" addresses
these aspects. Beside a static and cyclic full scale
frame test a test series on composite joints is being
IS performed (Figures 24 e 25). The series comprises
static and cyclic joint tests taking account of the
geometries and the distribution of internal forces of
the reference full scale tests.
Supplemented with numerical analysis based on
lui the component method guidance can be developed to
evaluate the available robustness of a given joint in a
sm
U's« certain load situation.
'»A w

">' .

Figure 24. Setup for static test on unsymmetrically loadad

beam to column joints

165
 ihMr connectoni
/
I

«e//10mm
Λ- XI
Figure 25. Setup for cyclic test on beam to column joints

Therefore ductility assessment developed for some

joint components (AiF-Project: Komponeten-
verformung steifenloser Anschlüsse) and already
verified for steel joints in static loading (SIF-
Project: Vorhände Rotationskapazität wirt¬ Figure 27. External node, composite column (Tests E5, E6
schaftlicher Anschlußkonstruktionen) will be taken (monotonie) and E10 (cyclic))
into account and checked for their applicability for
composite joints. Table 4, four main geometries were considered:
Experimental work was also carried out at the external node and internal node (Figures 26 and 27)
University of Coimbra (Simões, 2000) on steel and both for the bare steel and for the composite
composite structures with RC slab. As explained in structure. In addition, each of these geometries was
loaded under monotonie and cyclic conditions.
Table 4. Collection oftest specimens
Figure 28 reports the behaviour of the internal node
TEST NODE TYPE LOADING under (a) monotonie and (b) under cyclic loading.
El Monotonic M- / M- Figure 29 reports the correspondent behaviour in
E2 INTERNAL STEEL Monotonic M+ / M- external joints
Ell Cyclic τ
I
E3 Monotonie M+
E4 EXTERNAL STEEL Monotonic M-
E9 Cyclic I ι

E7 Monotonie M- / M- i -20

E8 INTERNAL COMPOSITE Monotonic M+ / M-

E12 Cyclic -o-T<¡etE1
E5 Monotonie M+ . TeetE2
· T«etE7
E6 EXTERNAL COMPOSITE Monotonic M- . TeslES

E10 Cyclic Rotation 4t,m (miad)

Figure 28a. Moment-rotation curves in monotonie tests El, E2,

E7 and E8

8 a hear conntctore

-ire 270 0 -40 -30 / 30 40

-*=--
i
i 0
COLUMN SECTION \

-150 T«stE11
Ta«tE12
-200
Rotstlon (mrad)

Figure 28b. Moment-rotation curves in cyclic tests Ell and

E12

Figure 26 Internal node, composite column (Tests E7, E8

(monotonie) and E12 (cyclic))

166
 has been established.Most natural loading condition
-M SO
for the masonry is compressive loading, nevertheless
..| -100 when exposed to accidental loading (Figure 30),
masonry structural elements has to have enough
resistance to prevent sudden collapse of the whole
-io 10 20 30 sb <b
-J s
' Γ
structure.
1

.it so -Tait 63
-Taal 64 6.2 Robustness of masonry elements
S-200-
-TeetE5

-460 -Τ··ΙΕβ In loadbearing masonry buildings, whether they are

Rotation iTcM (mrad) single or multistory, the walls serve as structural
elements to support or resist loads, as architectural
Figure 29a. Moment-rotation curves in monotonie tests E3, E4, elements to divide or enclose space, or as a finish
E5 and E6
material. According to EC 6 (Eurocode 6, 1995), the
main types of structural walls are distinguished as:
single-leaf wall, double-leaf wall, cavity wall and
150-
grouted cavity wall.
.100
When defining the term 'robustness' most
designers assume the structure's ability not to suffer
a major collapse due to minor accidental damage.
-40 / 20 30 40
Factors, which are influencing the robustness of
-SO .

masonry buildings, are primarily the elevation, plan,

SO
-Tes! ES wall layout as well as connections (wall to roof, wall
-200 -TestEIO to floor, wall to wall, ties etc.).
Rotation (mrad)
There are two main aspects of determination of
Fig. 29b Moment-rotation curves in cyclic tests E9 and E10 the robustness of masonry elements:
- due to the blast, impact forces, gas explosions,
6 MASONRY STRUCTURES traffic accidents (flexural loading) and
- due to the earthquakes (shear loading).

6. 1 Introduction 6.2. 1 Out of plane loaded masonry

Masonry is a composite, heterogeneous, nonlinear Most common way of designing masonry walls
structural material. As with other composite imposed to lateral loading is by taking into account
materials, also with masonry the mechanical its flexural strength both parallel (fx¡) and
properties are conditioned with the properties of perpendicular (fid) to the direction of the bed joints.
composite components, their volume ratio and the In the case of accidental loading (ultimate state) it is
properties of bond between the bricks and the layers much more advisable to design the masonry
of mortar -joints. elements according to the predicted mechanism of
failure (Figure 31).
Compressive loading

Load

I Flexural
loading
Maximurp
load

Ultimate
load

Shear loading

Worlcing
load Mechanism
or failure
Shear loading
(normal to the direction ofthe wall)

Deflection at center
Figure 30. Types of loading conditions in masonry.
Figure 31. Flexural loaded masonry structural element.
The type of the masonry within the Europe
strongly differs in dependence from the most
unfavorable thermal and loading conditions that
could be expected (Figure 30). From that point of
view the tradition of bricklaying in particular region
167
 6.2.2 In-plane loaded masonry 7 TIMBER STRUCTURES
7.1 Generalities

opa
This chapter deals with wood as a structural
material. Engineers and architects are often critical
of its material properties particularly with respect to
strength, anisotropy and durability.
ΐ52529 Research has to focus on these three
characteristics to enhance the acceptance of wood
for structural applications. The objective of this
++
.umili chapter is to show a perspective for building with
JZDED
CjEZ3
wood where drawbacks can be overcome using
CZltZDD
sea wood modification and new technologies.

Figure 32. Shear tests for masonry structural elements. *,.

i£ mr
There is no standardized test method for I so
determination of the robustness and shear resistance
of the masonry elements due to the earthquake
loading (shear loading). Some of the most frequent » ; ;α aE 3 * 'ιΐ β *a

ways of testing the large masonry elements are maaaains ««lanf esaianninirir. c*mft*t*m
Fani.
Tanate

presented in Figure 32. I lapmct aiSpnjce»omt* OPeM.Spng« OPres. Sutura τ OiWeat t

Neither of them can be considered as the one

which response to the reality, but they are all chosen Figure 34. Mechanical characteristics of clear spruce, densified
because they reproduce undoubted static or spruce and oil heated densified spruce.
kinematics boundary conditions which are easy to be
interpreted. According to our experience and the 7.2 Modification of wood properties -
analysis of the results of other authors, only the test densification and heat treatment
set-up presented in Figure 32a & 32f, with cyclic
loading history can give us all the behavioral The design of a structure is decisively determined
parameters such are: global forces, global by the properties ofthe building material. Therefore
displacements, deformations and response the modification of the properties of the building
characteristics (mode of failure, shape of hysteresis material has the strongest impact on the structure.
(Figure 33), cracking patterns, ductility capacity, The above mentioned detrimental characteristics -
energy absorption, damage evolution etc.) for the strength, anisotropy and durability - can be improved
seismic assessment of the masonry structural by means of existing technologies and knowledge.
elements. Strength, for instance, can be increased using better
grades or species with better mechanical
characteristics (Natterer et al, 1991). Durability
120
-M Wt4t44t44>
depends to a great extent on the natural resistance of
90
"Γ+
the species, which is very good for topical hardwood
Η
for instance. Chemical preservation (Willeitner and
60
Η
imr Schwab, 1981) provides good results for outdoors
30 applications, however, the toxicity has a harmful
: t effect on the usual environmental friendliness of
0
wood.
LV2
-30 For a further improvement of strength and
-Η, durability various techniques have been proposed in
-60
ã the past (Kollmann et al, 1975). It is well known that
-90
wood is a porous material that can be easily
-H -Hpuu, Displacement [mm] compressed at a temperature of about 140°C. The
-120 density of domestic softwood, mostly spruce and
-25 -15 -s 15 25 pine, is increased in this process from about 450
kilogram per cubic meter to a value of over 1000,
Figure 33. Shear tests for masonry structural elements. which results roughly in a doubling of the
mechanical characteristics.

168
 The application of densified wood has been
recently investigated in structural connections
(Haller and Wehsener, 1999) (Figure 35). The joints
from densified wood showed a significant
improvement ofthe load bearing capacity.
Heat also affects the hygroscopic stability and
durability. Heat treatments - also in natural oil - at
about 200°C result in a dimensionally stable material
with excellent biological resistance. The
combination of a simple heat treatment and
densification (Rapp and Haller, 2001) leads to a new
material which is in many cases better suited than
ordinary wood (figure 34).
- f *'t. .... ',
^¿*¡£^ -
made of
7.3 Anisotropy and reinforcement
Anisotropy of wood results in an excellent ratio
of strength and weight that surpasses metals,
minerals and plastics to a great extent. However, it
may get even experienced designers into trouble
because many details in timber construction require
high strength in more than one direction.
Today's wood based materials represent a
development where the ratio of anisotropy in the
longitudinal and transversal direction is reduced.
This reduction leads to a compromise for strength
and stiffness in both directions.
Solid wood is characterized by low strengths in
shear and parallel to grain . Unlike bending stresses
in the fibre direction, the corresponding loadings
cannot be transferred economically by choosing a
larger cross section. Therefore various techniques
have been developed in order to reinforce wood
perpendicular to grain by means of glued-in rods,
wood based panels for instance. These measures
demand a profound knowledge from the designer
and are limited to a few specific problems.
As the design for high shear stresses and stresses
parallel to grain cannot be done economically by a
wider section the strength has to be increased by
means of reinforcement techniques. Technical
169
textiles represent an effective and versatile
technology to reinforce timber structures (Haller and
Birk, 2001). The reinforcement can be applied
locally or entirely to the construction element where
the textile fibres may be oriented towards the
occurring stresses and the shape. Moreover, the
textile layer protects wood from weathering and
hence affects its durability to a great extent.
As stresses are difficult to control within
connections e.g. at the vicinity of dowel type
fasteners the load bearing behaviour can be tailored
according to stiffness, strength and ductility. The
influence of stress oriented textile reinforcements
can be seen from (Figure 36). It also appears that
knitted textile placements behave best with an
increase of the load bearing capacity for the single
dowel of about 4 times.
so
45
40
35
30
25 Lochleibungsfestigkeit (hândische Strukturen)
20
Figure 36. Tailor made reinforcements of dowel type fasteners.
7.4 Material efficiency
In wood construction solid rectangular cross
sections predominate. These sections dispose of a
very low material efficiency compared to technical
profiles which need less than 20% of the wooden
section to achieve the required moment of inertia. If
one further considers the low yield of about 50% for
the saw milling of the logs, it is evident that the
material efficiency is dramatic and that wood
construction loses competitiveness at this stage.
The condition for the manufacturing of technical
profiles is a plastic material behaviour to permit
rolling or extrusion. Wood is regarded as an easy
workable material. Quite the reverse! Wood can only
be worked by cutting processes which are more
expensive than forming processes. However, if one
consider the above thermal densification of wood as
a result of a compression strain of about 50%, it
becomes obvious that wood is very deformable. In
deed, a tensile strain of 100% can be reached on
densified wood by pulling it in hot and moist
atmosphere to its original, undensified state.
With this principle it becomes possible to
manufacture open and closed prismatic sections, the
material efficiency of which is not inferior to
technical profiles (Figure 37). The availability of
such cross sections on the market would promote
wood as a structural material in civil engineering and
other disciplines to a great extent.
Figure 37. Thermo-mechanically formed cross section.
7.5 Wood compounds
Wood can be combined with other materials to
compound elements. If one disregards here the
compounds of two or more wood cross sections any
load bearing building material could be used for a
suitable mixed compound.
Timber concrete composites have been
successfully used in the past for residential houses,
bridges and in rehabilitation due to load bearing
performance, fire resistance and sound insulation
etc. This compound has been mentioned in the
literature of the 40's and rediscovered and
investigated in the 80's (Natterer and Hoeft, 1987)
(Figure 38).
Figure 38. Example of a wood concrete cross section for
floors.
The compound of wood and fibre reinforced
plastic or technical textiles is recent. First
unidirectional fibre reinforced plastic was used to
increase the bending moment of gluelam beams.
However, this method do not solve a severe design
problem as the load bearing capacity can be
efficiently increased by choosing a higher cross
section where the height is raised to a power of two
or three respectively.
Compounds between thermo-mechanically
formed wood profiles and textiles are much more
170
interesting since the ratio of the areas wood and
textile is much better than in the example above.
Theoretical considerations show that such
compounds have extremely good structural and
economical characteristics (Ziegler, 2001).
8 ALUMINIUM AND STAINLESS STEEL
STRUCTURES
8. 1 Aluminium members and connections
In the field of aluminium alloy structures the need
for adopting special calculation methods for
evaluating the structural integrity of members and
connections is felt, especially owing to the necessity
to take into account the actual mechanical features of
the material (Mazzolani 1995). This is due to several
reasons, such as: (1) the complex mechanical
behaviour of the material in post-elastic range,
which exhibits continuous hardening and limited
available ductility; (2) the great variety of aluminium
alloys having mechanical features substantially
different from each other; (3) the possibility to
combine different basic materials with different
types of mechanical fasteners (for instance, in case
of bolted connections, both aluminium and steel
bolts are used in the current practice); (4) the strong
effect of welding on the mechanical features of heat
treatable aluminium alloys. As a result, the direct
application to aluminium structures of design rales
calibrated for steel and other traditional materials is
often inappropriate.
Many of these aspects have being discussed within
the activity of the CEN-TC250/SC9 Committee
chaired by Prof. F.M. Mazzolani, which has worked
out the first edition of Eurocode 9 "Aluminium
Alloy Structures" »(Mazzolani 1999), which includes
many methods and procedures accounting for the
above features of the material. On the other hand,
several research programs are ongoing in order to
verify and extend the validation range ofthe existing
procedures. Among many others, some of these
projects are dealing with stability problems
(Landolfo 2000, Rasmussen & Rondal 1999), plastic
design (Mazzolani et al. 1999), cross-sectional
classification (De Matteis et al. 2001, Faella et al.
2000).
In particular, as far as structural joints are
concerned, a large research project has been recently
carried out at the University of Naples Federico Π,
sponsored by the Italian Ministry of University and
Scientific and Technological Research (MURST). It
is mainly concerned with the behaviour of bolted T-
stub joint components, which has been deeply
examined under analytical, numerical and
experimental points of view. Therefore, firstly, an
analytical method to determine both the collapse
mechanism type and the corresponding load bearing
capacity of aluminium T-stubs based on the existing
formulations adopted into EC3-Annex J has been
proposed, emphasizing that for aluminium the joint
collapse mechanism cannot be so easily predicted as
for steel, it being strongly conditioned by the actual
mechanical features of flange and bolt materials,
namely strain hardening and available ductility (De
Matteis et al. 2000). Then, a wide experimental
campaign has been carried out aiming at validating
the above model. Such experimental investigation,
which represents the largest developed activity in the
field of aluminium bolted connections in tension and
bending, comprises 26 different specimens tested
under monotonie and cyclic loading and allows
many of the major influencing parameters to be
worked out (De Matteis et al. 2001α). Τη particular, 4
different T-stub types (varying in-plane dimensions,
plate thickness, number and location of bolts), 3
different aluminium alloys as flange material, 3
different bolt materials and two different types of
coupling systems have been considered. The
obtained results allow the influence of the above
parameters to be pointed out in terms of initial
stiffness, ultimate strength, deformation capacity as
well as observed failure plastic mechanisms (De
Matteis et al. 20016).
Another research project, which has as scope the
definition of deformation and load-carrying capacity
of aluminium riveted connections, was recently
carried out at the Institute of Steel Structures of the
Department of Civil Engineering, Aristotle
University of Thessaloniki (Zygomalas et al. 2001).
In particular, aluminium strips of 1 mm thickness,
60mm width and 250 mm length have been tested.
For each experiment two pieces were connected by
applying 8 rivets 04mm. The obtained results ofthe
tests include a linear part of deformation and a part
where the deformation is almost constant. The latter
"plateau" phenomenon is due to the rotation of the
rivets during the experiment, a behavior that changes
the static response of the connection. Further
experiments have been just started to be performed
aiming to define the deformation in both aluminium
pieces and the whole connection as a continuous unit
within an aluminium structure and to clarify the
previous phenomenon. In parallel to the
aforementioned laboratory testing, the behavior of
butt- welded aluminium joints have been numerically
studied by applying the F.E.M. aiming to investigate
in details the HAZ effect and the ultimate tension
strength of such welds (Kontoleon et al. 2000).
As far as members are concerned, an extended
parametric nonlinear numerical study of the
structural response of the structural members of a
typical aluminum structure having as scope to
investigate their serviceability and strength were
recently performed at the Institute of Steel Structures
171
of the Department of Civil Engineering, Aristotle
University of Thessaloniki (Preftitsi et al. 1998).
Concerning the material law of the investigated
aluminium members, the Ramberg-Osgood isotropic
stress-strain law and the nonlinear evolution law
proposed by F. Mazzolani have been both used. At
the nonlinear folded thin-shell structural model
applied, the aluminium members were subjected to a
variety of loading conditions (as dictated by
Eurocode 1 and 9) that correspond to the loading
conditions of a typical aluminium structure (as is e.g.
an aluminium curtain-wall system). The computer
software CASTEM 2000 has been used for the
treatment of the structures under investigation. All
the results obtained are presented in (Preftitsi et al.
1998, Baniotopulos et al. 1999). In the forthcoming
months, the obtained numerical results are to be
checked by performing at the laboratory of the
Institute of Steel Structures in Thessaloniki an
already started program of tests on aluminium
members of the same geometry and elasticity
characteristics as those of the numerical
experimentation.
8.2 Stainless steel members and connections
Stainless steel used for structural applications is
covered by ENV 1993-1.4. The austenitic steel is the
most widely used in construction. Its stress-strain
curve shows a non-linear and ductile behaviour with
a high ultimate to yield strength ratio. This gives a
large capacity of resistance and deformation which
let suppose a good dissipation of energy under cyclic
loading and a good redistribution of plastic loads
without failure (ASCE 1990, NidL-Eurofriox 1994).
It has a good fire resistance and large possibilities of
strain hardening. However, for bolted connections, a
design for the ULS loads may not be sufficient to
avoid unacceptable plastic deformations in the SLS.
In fact, the available design codes are focused on
structural carbon steel. The formulas and design
approaches must be modified to take into account
the non-linear behaviour of stainless steel. This non¬
linear behaviour is usually modelled by the
Ramberg-Osgood law using a conventional elastic
limit (0.2% plastic deformation). A ECSC project,
covering all major aspects of the design of stainless
steel elements in buildings, was initiated in 1998.
One of its principal objectives was to verify, and
propose improvements to the EC3-1.4. An
experimental study for the cover-plate joints, was
conducted at Clermont Ferrand and the results were
compared to the predictions of EC3-1.4 (Bouchair et
al. 2001). The tests show that the various ultimate
state joint strengths are safely predicted with a very
large ductility of all the joints under static loads.
This will be a good guarantee of safety under cyclic
or dynamic loads. The study is now being
complimented by the use of finite element modelling
in order to better understand joint behaviour. It
should also help to extend the study to the design of
other geometrical configurations and other types of
joints (beam-to-column joints for example). Also,
the evaluation of a beam deflections and a plastic
loads redistribution needs to be more analysed. For
the members, the secant modulus could be used in
substitution to the elastic modulus (Van Den Berg
2000). Also, attention has to be made to ensure that
calculated plastic moment transmitted to the
connections do not exceed their own strength.
9 CONCLUSIONS
This paper presents a state-of-the-art in the
evaluation of ductility in view of ensuring adequate
robustness of mixed structures under exceptional
loading. It focuses on the contributions from within
the COST C 12 project and highlights the research
needs for a comprehensive treatment of this subject.
REFERENCES
General methodology: members
Axhag, F. 1998. Plastic Design of Slender Steel Bridge
Girders. Doctoral Thesis, Lulea University of Technology.
Eurocode 3. 1992. Design of steel structures. Part 1-1: General
rules and rules for buildings. European prestandard.
Feldmann, M. 1994. Zur Rotationskapazität von I-Profilen
statsich und dynamisch belasteter Träger. Dissertation,
RWTH Aachen.
Kuhlrnann, U. 1986. Rotationskapazität biegebeanspruchter I-
Profile unter Berücksichtigung des plastischen Beulens.
Mitteilung Nr. 86-5, Dissertation, Ruhr-Universität
Bochum.
Kuhlmann 1987a. Rechnerische Ermittlung der
Rotationslcapazität biegebeansprichter I-Profile. Der
Stahlbau. 56:321-327.
Kuhlmann 1987b. Experimentelle Ermittlung der
Rotationskapazität biegebeansprichter I-Profile. Der
Stahlbau. 56: 353 - 358.
Spangenmacher, R. 1992. Zum Rotationsnachweis von
Stahlkonstruktionen, die nach dem Traglastverfahren
berechnet werden. Dissertation, RWTH Aachen.
Stranghöner, N. 1995. Untersuchungen zum Rotationsverhalten
von Trägern aus Hohlprofilen. Dissertation, RWTH
Aachen.
General methodology: connections
COST Cl 1999. Recent advances in the field of structural steel
joints and their representation in the building frame analysis
and design process. Brussels/Luxembourg.
COST CI 1992. Semi rigid behaviour of civil engineering
structural connections. Proceedings ofthe 1st state ofthe art
workshop, Straßbourg.
COST CI 1994. Semi rigid behaviour of civil engineering
structural connections. Proceedings of the 2nd state of the
art workshop, Prague.
COST CI 1997: Semi rigid behaviour of civil engineering
structural connections. Proceedings of the International
Conference, Liège.
COST CI 1996. Composite steel-concrete joints in braced
frames for buildings. Brussels/Luxembourg.
172
Commission ofthe European Communities 1997. Frame design
including joint behaviour (Vol. 1-3). Executive Committe
Ref.no. 93-F6.05, ECSC no.7210-SA/212+320.
ECCS 1992. Analysis and design of steel frames with semi¬
rigid joints. ECCS TC8, report no.67.
ECCS 1999. Design of composite joints for buildings (incl.
Model code provisions for composite joints in building
frames). TCI 1, Brussels.
ENV-1993-l-l/pr A2 1994. Revised annex J (version Nov.94):
Joints in building frames.
prEN-1993-1-8 2001. Design of steel structures, Design of
joints, Final draft.
prEN-1994-1-1 2002. Design of composite steel and concrete
structures, Section 8 Composite joints in frames for
buildings, Final draft.
Huber 2000. Non-linear Calculations of Composite Sections
and Semi-continuous Joints, Ernst&Sohn, Berlin, ISBN 3-
433-01250-4.
IABSE 1997. International Conference "Composite Construc¬
tion - Conventional and Innovative" in Innsbruck. Zurich.
Jaspart 1991. Etude de la semi-rigidite des noeuds poutre-
colonne et son influence sur la resistance et la stabilite des
ossatures en acier. Doctoral thesis, Liège.
Silva, L and Coelho, A 2001. A ductility model for steel
connections", Journal of Constructional Steel Research,
57(1), 45-70.
Tschemmernegg 1997. Connection of floor systems to columns
- Conventional and Advanced. International Conference
"Composite Construction - Conventional and Innovative" in
Innsbruck, Report p.445-450, IABSE Zurich.
Tschemmernegg 1999. Neue Innsbrucker Mischbautechnologie
mit Erfolg beim Millennium Tower in Wien angewandt.
Stahlbau, 68. Jahrgang, Heft 8.
Weynand 1997. Sicherheits- und Wirtschaftlichkeitsunter¬
suchungen zur Anwendung nachgiebiger Anschlüsse im
Stahlbau. Doctoral thesis, Aachen.
Concrete structures
ASCE. 2000. FEMA 356 Prestandard and Commentary for the
Seismic Rehabilitation of Buildings, ASCE for the Federal
Emergency Management Agency, Washington, D.C,
November.
ATC. 1997. NEHRP Guidelines for the Seismic Rehabilitation
of Buildings, FEMA Report 273, Applied Technology
Council for the Building Seismic Safety Council,
Washington, D.C.
ATC. 1997. NEHRP Commentary on the Guidelines for the
Seismic Rehabilitation of Buildings, FEMA Report 274,
Applied Technology Council for the Building Seismic
Safety Council, Washington, D.C.
Comité Euro-International du Beton. 1998. Ductility of
Reinforced Concrete Structures, Bulletin d' Information
242, T. Telford, ed., London, May.
Comité Euro-International du Beton. 1993. Ductility-
Reinforcement, Bulletin d'Information 218, Lausanne.
Comité Euro-International du Beton. 1993. CEB/FIP Model
Codes 1990, T. Telford, (éd.), London.
Eurocode 2. 1991. Design of Concrete Structures, Part 1:
General Rules for Buildings, European Prestandard, CEN,
ENV 1992-1-1, Brussels.
EKOS. 2000. Greek Code for Design and Construction of
Reinforced Concrete Structures, Organization for Seismic
Design and Building Protection, Athens.
Kotsovos, M.D., and Pavlovic, M.N. 1999. Ultirnate Limit-
design of Concrete structures: A new approach, T. Telford,
London.
Kotsovos, M.D. and Perdikaris, P.C. 2001. Ensuring Adequate
Ductility Capacity in New and Repaired/Strengthened
Existing R/C Structures, Report to the Organization for
 Seismic Design and Building Protection, Athens, Greece,
July.
Krawinkler, H. 1999. Challenges and Processes in
Performance-Based Earthquake Engineering, International
Seminar on Seismic Engineering for Tomorrow (In Honor
of Professor H. Akiyama), Tokyo, Japan, November.
Panagiotakos, T. B., and Fardis, M. N. 2001. Deformations of
Reinforced Concrete Members at Yielding and Ultimate,
ACI Structural Journal, V. 98(2), 135-148.
Paulay, T., and Priestley, M.J.N. 1992. Seismic Design of
Reinforced Concrete and Masonry Buildings, J. Wiley, New
York.
Steel structures
Bertero, V.V. 1997. Earthquake Engineering. In Golden W.G.
(ed.), Structural Engineering Slide Library (Golden W.G.
Ed.), The University of Berkeley. California, USA.
Bertero, R. D., Bertero, V.V. 1999. Redundancy in Earthquake-
Resistant Design. Journal of Structural Engineering, ASCE,
Vol. 125(1): 81-88.
Biddah, Α., Heidebrecht, A.C. 1999. Evaluation of the seismic
level of protection afforded to steel moment resisting frame
structures designed for different design philosophies. Can.
Journal of Civil Engineering, Vol. 26: 35-54.
Calderoni, Β., Ghersi, Α., Rinaldi, Ζ. 1996. Stability Analysis
of Seismic Behaviour of Steel Frames: Influence of
overstrength. Journal of Constructional Steel Research, Vol.
39, No.2: 137-161.
Chitina, Α., Stratan, Α., Dubina, D. 2001. Seismic response of
MRS frames accounting for Μ-φ models used for members
and joints (in Romanian). 2nd national Conference on
Earthquake Engineering, Bucharest, Romania, 8-9
November 2001: 3.113-3.123.
Court. A.B., Kowalsky, M.J. 1998. Performance-based
engineering of buildings - a Displacement Design
Approach. Structural Engineering Worldwide, Elsevier
Science (CD-ROM), Paper Ref. T109-1.
Dubina, D., Chitina, Α., Stratan, A. 2002. Cyclic tests on
Bolted steel and composite double-sided beam-to-column
joints, Steel and Composite Structures, (in print).
Dubina, D. et al 2000. Ductility demand for semi-rigid joint
frames. In Moment Resistant Connections of Steel Building
Frames in Seismic Areas (F.M. Mazzolani, Ed.) : London:
E&FN SPON: 371-408.
Elnashai, A.S., Elghazouli, A.V., Denesh-Astiani, F.A. 1998.
Response of steel frames to cyclic and Earthquake loads.
Journal of Structural Engineering, ASCE, Vol. 124(8): 857-
867.
FEMA 273 1997. NEHRP Guidelines for Seismic
Rehabilitation of Buildings. BSSC, Washington DC, USA.
Fouten, D.A., Yun, S. 1998. Comparative reliability of Linear
and Nonlinear Approaches to demand characterization in
steel structures. Structural Engineering Worldwide, Elsevier
Science (CD-ROM), Paper Ref. T166-2.
Ghabarah, A. 2001. Performance-based design in earthquake
engineering: State of development. Engineering Structures,
Vol. 23: 878-884.
Gioncu, V. 2000. Framed Structures: Ductility and Seismic
Response. General Report. In Dubina D. (ed.), Journal of
Constructional Steel Research, Special Issue on Stability
and Ductility of Steel Structures, SDSS '99, Vol. 55, No.l:
3125-154.
Nakashima, M. 1998. Variation associated with trade-off
between Strength and Ductility in Seismic Design of Steel
Building Structure. Structural Engineering Worldwide,
Elsevier Science (CD-ROM), Paper Ref. Tl 13-3.
Ohi, Κ 1998. Resistance versus Resistance Reliability Concept
in the Design of Steel Frames and Connections. Structural
Engineering Worldwide, Elsevier Science (CD-ROM),
Paper Ref. T104-4.
Petcu, D., Gioncu, V. 1999. DUCTROT MSN Computer
Program Guide for users, INCERC Timisoara, Romania.
Sanders, C.M. 1998. Design criteria for steel moment frame
buildings. Structural Engineering Worldwide, Elsevier
Science (CD-ROM), Paper Ref. T166-3.
SEAOC-Vision 2000 (1995). Performance Based Seismic
Engineering of Buildings: Conceptual Framework.
Sacramento, California, USA.
Uang, C.-M., Fan, C.-C. 2000. Cyclic instability of steel
moment connections with reduced beam section. In
Mazzolani and Tremblay (eds.), Behaviour of Steel
Structures in Seismic Areas (STESSA 2000): 747-754.
Rotterdam: Balkema.
Steel-concrete composite structures
Altmann R., Maquoi R., Jaspart J.P.: Experimental Study ofthe
Non-Linear Behaviour of Beam-to-Column Composite
Joints. Journal of Constructional Steel Research. Vol. 18,
1991.
Anderson D., Najafi A.A.; Performance of Composite
Connections: Major Axis End Plate Joints. Journal of
Constructional Steel Research. Vol. 31, 1994.
Carvalho C.V., de Andrade S.A.L., da S. Vellasco P.C.G.:
Experimental Analysis of Bolted Semi-Rigid Steel
Connections. Journal of Constructional Steel Research. Vol.
46, No 1-3, 1998.
Composite Steel-Concrete Joints in Braced Frames for
Buildings. Ed. by D. Anderson. Brussels, Luxembourg,
1996.
Connections in Steel Structures. Behaviour, Strength and
Design. Ed. by R. Bjorhovde, J. Brozetti, A. Colson.
Elsevier Science Publishers, London - New York 1988.
Connections in Steel Structures II: Behaviour, Strength and
Design. Ed. by R. Bjorhovde, A. Colson, G. Haaijer, J.
Stark. AISC, 1992.
Connections in Steel Structures ΠΙ: Behaviour, Strength and
Design. Ed. by R. Bjorhovde, A. Colson, R. Zandonini,
Elsevier Science Publishers, 1996.
Davison J. B., Kirby P. Α., Nethercot D. Α.: Rational Stiffness
Characteristics of Steel Beam-to-Column Connections. -
Journal of Constructional Steel Research. Vol. 8, 1987.
Davison J.B., Lam D., Nethercot D.A.: Semi-rigid action of
composite joints. The Structural Engineer. Vol. 68, No 24,
Dec. 1990.
Li T.Q., Nethercot D.A., Choo B.S.: Behaviour of Flush End¬
plate Composite Connections with Unbalanced Moment and
Variable Shear/Moment Ratios - I. Experimental
Behaviour.. Journal of Constructional Steel Research, Vol.
38, No. 2,1996.
Neves, L.F.C., Semi-rigid joints in steel structures: evaluation
ofthe stiffness in minor axis configurations (in Portuguese),
Msc thesis, University of Coimbra, May 1996
Neves, L.F.C., Gomes, F.C.T.Semi-rigid behaviour of beam-to-
column minor axis joints, Proceedings IABSE conference,
Istanbul, 1996
Ribeiro L.F.L., Gonsalves R.M., Castiglioni CA.: Beam-to-
Column End Plate Connections - An Experimental
Analysis. Journal of Constructional Steel Research. Journal
of Constructional Steel Research. Vol. 46, No 1-3, 1998.
Silva, L.A.P.S, Neves, L.F.C., Gomes, F.C.T, Rotational
stiffness of RHS composite connections, ASCE Journal of
Structural Engineering, in printing
Simões, R.A.D.S., Behaviour of Beam-to-column Composite
Joints under Static and Cyclic Loading (in Portuguese), Phd
Thesis, University of Coimbra, 2000
173
Tschemmernegg F.: Testing and Test Results of Composite
Joints. COST PROJECT CI. Proceedings ofthe First State
ofthe Art Workshop. Strasbourg 1992.
Zandonini R.: Semi-rigid Composite Joints, in Structural
Connection. Stability and Strength. Ed. by R. Narayanan.
Elsevier Applied Science. London - New York. 1989.
Analysis ofthe stiffness, stability and carrying capacity of steel
structures with semi-rigid joints, (in Polish). Research
project KBN nr 7T07E06910. Rzeszow University of
Technology. 1998.
Kozlowski Α.: Shaping of the steel and composite skeletons
with semi-rigid joints. Rzeszow University of Technology.
Monograph (in Polish). Rzeszow, 1999.
Masonry structures
Eurocode 6: Design of masonry structures - Part 1-1: General
rules for buildings - Rules for reinforced and unreinforced
masonry. ENV 1996-1-1.: 1995. CEN Brussels. July 1995
Bosiljkov, V. & Zarnic, R. 2000. Brickwork from its
components to assemblage: an overview of some different
destructive techniques for assessment of the mechanical
properties. In Proc. of te International workshop on urban
heritage and building maintenance V, G.W. Verhoef & F.H.
Wittman (eds): 25-35. Aedificatio Pubi., Germany
Timber structures
Natterer, J.; Herzog, T.; Volz, M.; Holzbautatlas Π; Institut fur
internationale Architektur-Dokumentation, München, 1991
Willeitner, H.; Schwab, E.; Holz-Außenverwendung im
Hochbau; Verlagsanstalt Alexander Koch, Stuttgart, 1981;
Kollmann, F.; Kuenzi, E.; Stamm, Α.; Principles of Wood
Science and Technology, Vol. Π: Wood Based Materials;
Springer Verlag, Berlin/Heidelberg , 1975;
Haller, P.; Wehsener, J.; Use of technical textiles and densified
wood for timber joints; Proceedings, RILEM Symposium
on Timber Engineering, Stockholm, 1999
Rapp, Haller, P.; Physikalische und mechanische Eigenschaften
von öl-hitze-behandeltem Preßholz; Interner
Forschungsbericht; Institut fur Baukonstruktionen und
Holzbau; Technische Universität Dresden; 2001
Haller, P.; Birk, T.; Physikalische und mechanische
Untersuchungen an textilbewehrtem Holz und
Holzbauteilen; in: Arbeitsbericht; SFB 528 Textile
Bewehrungen zur bautechnischen Verstärkung und
Instandsetzung"; Fakultät Bauingenieurwesen; Technische
Universität Dresden; 2001
Natterer, Hoeft; Zum Tragverhalten von Holz-Beton-
Verbundkonstruktionen; Report; CERS Nr 1345; Chair de
Construction en Bois - D30IS; Ecole Polytechnique
Fédérale de Lausanne; Switzerland; 1987
Ziegler, S.; Herstellung und Machbarkeit gekrümmter
Querschnitte aus flächig, einachsig verdichtetem Nadelholz;
diploma thesis; Institut für Baukonstruktionen und Holzbau;
Technische Universität Dresden; 2001
Aluminium and stainless steel structures
ASCE (1990). Specification for the design of cold-formed
stainless steel structural members. ASCE Standards, 114
pages.
Baniotopoulos, CC, Koltsakis, E., Preftitsi, F. Panagiotopou-
los, P.D., (1999). Aluminium Mullion-Transom Curtain-
Wall Systems: 3-D F.E.M. Modelling of their Structural
Behaviour. Leight-Weight Steel and Aluminium Structures
(P.Maekelainen and P.Hassinen eds.), Elsevier, Amsterdam,
Lausanne: 433-440.
Bouchair, Α., Ryan, I., Kaitila, O. and Muzeau, J. P. (2001).
Some aspects of the analysis of bearing-type stainless steel
bolted joints. 9* Nordic Steel Const. Conf, Helsinki., 755-
762.
174
De Matteis, G., Della Corte, G., Mazzolani F.M., (2001b).
EExperimental analysis of aluminium T-stubs: tests under
monotonie loading.. XVUI Congresso C.T.A., Venezia
(Italy), 26-28 Settembre 2001 : Vol. 2, 29-40.
De Matteis, G., Landolfo, R., Mazzolani, F.M., (2001a).
Experimental Analysis of Aluminium T-Stubs: Framing of
the Research Activity. 8,h Int. Conf. on Joints in Aluminium
(INALCO 2001), Munich, Germany, 28-30 March.
De Matteis, G., Mandara, Α., Mazzolani, F.M., (2000). T-stub
Aluminium Joints: Influence of Behavioural Parameters.
Computers and Structures, 78:1-3: 3 1 1-327.
De Matteis, G., Moen, L.A., Langseth, M., Landolfo, R.,
Hopperstad, O.S., Mazzolani, F.M. (2001). Cross-Sectional
Classification for Aluminium Beams - A Parametric Study.
Journal of Structural Engineering (ASCE), 127 (3): 271-
279.
Paella, C, Mazzolani, F.M., Piluso, V., Rizzano, G. (2000).
Local Buckling of Aluminium Members: Testing and
Classification. Journal of Structural Engineering, ASCE,
126 (3): 353-360.
Kontoleon, M.J., Preftitsi, F., Baniotopoulos, CC (2000).
Butt-Welded AJuminium Joints: A Numerical Study of the
HAZ Effect on the Ultimate Tension Strength. The
Paramount Role of Joints into the Reliable Response of
Structures (CC Baniotopoulos & F. Wald eds.), Kluwer,
Dordrecht, Boston: 337-346.
Landolfo, R. (2000). Coupled Instabilities in non-linear
materials. 3 Int. Conf. on Coupled Instability in Metal
Structures, Lisbon: 263-272.
Mazzolani, F.M. (1995). Alviminium Alloy Structures, E & FN
Spon, London.
Mazzolani, F.M. (1999). The structural use of aluminium:
design and applications. Leight-Weight Steel and
Aluminium Structures (P. Maekelainen and P.Hassinen
eds.), Elsevier, Amsterdam, Lausanne: 475-486.
Mazzolani, F.M., Mandara, Α., Langseth, M., (1999). Plastic
design of aluminium members according to EC9. Leight-
Weight Steel and AJuminium Structures (P. Maekelainen
and P.Hassinen eds.), Elsevier, Amsterdam, Lausanne: 457-
464.
NidL-EuroInox (1994). Design Manual for Structural Stainless
Steel. Nickel Develpt. Institute.
Preftitsi, F., Baniotopoulos, CC, Koltsakis, E.,
Panagiotopoulos, P.D. (1998). A Nonlinear Numerical
Analysis of the Structural Response of Aluminium Curtain-
Wall Systems. 3th Greek Conf. on Steel Structures
(K.Thomopoulos, CCBaniotopoulos & A.Avdelas eds.),
Aristotle University of Thessaloniki, Thessaloniki: 92-98.
Rasmussen, K.J.R., Rondal, J. (1999). Column curve
formulation for alviminium alloys. Leight-Weight Steel and
Aluminium Structures (P. Maekelainen and P.Hassinen
eds.), Elsevier, Amsterdam, Lausanne: 441-448.
Van Den Berg, G. J. (2000). The effect ofthe non-linear stress-
strain behaviour of stainless steels on member capacity.
Journal of Constr. Steel Research, 54:135-160.
Zygomalas, M., Kontoleon, M.J., Baniotopoulos, CC (2001).
A Hemivariational Inequality Approach to the Resistance of
Aluminium Riveted Connections. 6* Greek Conf. on
Mechanics, Aristotle University of Thessaloniki,
Thessaloniki: 70-77.
WG3 - Urban Design
A Generative Design System for Low-Energy Architecture Design
L. G. Caldas
Instituto Superior Técnico, Portugal

ABSTRACT: This paper presents a novel Generative Design System that applies goal-oriented techniques to
help architects in the creation of low-energy architectural solutions. The system uses a Genetic Algorithm as a
search engine, and the DOE2.1E building simulation software as the evaluation module. Three different
applications are discussed. The first section researches the incorporation of architecture design intentions into
the generative system, using a building by architect Alvaro Siza as an experimental platform. Tn the second
part, the system capabilities are expanded to handle multicriteria problems, resorting to Pareto optimization
concepts. Finally, the system is applied to three-dimensional shape generation, to investigate formal
configurations with a high degree of adaptation to the exterior environment.

based on triangles, which can be divided, joined, etc.

1 INTRODUCTION
The field of Artificial Intelligence has provided
designers from different areas with new possibilities
that reach far beyond the mere use of computer
simulations to assess design performance prior to
actual construction. Goal-based design allows
researchers and practitioners to invert the standard
development direction of defining a solution,
simulating its performance, and progressively
refining it by means of sequential, parametric
simulations. Using inverse-design methods, it is
possible to establish performance goals, and apply
search procedures to backtrack and determine which
design features may lead to the closest achievement
ofthe desired targets.
Architecture and structural engineering have only
recently started to embrace the advantages offered by
diverse AI search methods, mostly reverting to the
application of Genetic Algorithms and Simulated
Annealing, which significantly improve on
previously available techniques like linear and non¬
linear optimization, dynamic programming, and
others.
In the area of structural analysis, Shea (1997,
1998) applies Simulated Annealing [SA] to the
optimization of geodesic-like domes. In this
approach, the representation of structures is based on
an analogy with a network, where design elements
are joints and the relations between them represent
the flow of forces through the structural members.
Finite-element analysis is used to calculate the
performance of the structure. Dome designs are
177
Shea et al. (1997) first proposed a planar truss
grammar, and only later Shea & Cagan (1997)
developed a space truss grammar, which is used to
generate geodesic-like domes. The system applies
shape-modifying rales in a dynamic, probabilistic
way, and also size modification rales that alter, for
example, the cross section of structural elements.
Initial user specifications include number and
location of supports, materials properties, and if
symmetry is required or not. Constraints relate to
structural stress, Euler buckling and geometric
obstacles, among others. The objective functions that
can used include efficiency [achieving minimum
mass], economy [minimum number of distinct cross-
sections and distinct lengths], utility [maximum
enclosure space with minimum surface area], and
even aesthetics [achieving a uniform metric or a
golden ration metric]. The final objective function
value is calculated as a weighted sum of the
objective functions chosen, plus a weighted sum of
constraints violation. The main flaw of this method
is that it stills allows the generation of solutions that
are not structurally viable.
Monks et al. (1998) used a combination of SA
and steepest descent to optimize the acoustical
performance of architectural spaces such as
auditoria, by setting acoustic performance targets
and then backtracking from there. Variables used
relate to physical characteristics of the space, like
room geometry [tilt of reflectors and other surfaces],
and materials properties. The objective function is a
 combination of six acoustical measures.
Performance is calculated by a simulation software
that uses an hybrid method of simulating early sound
with beam tracing and late sound with a statistical
approximation. The desired acoustic performance is
set a priori, and the material and geometric
parameters ofthe environment are determined by the
optimization algorithm in order to approach that
goal.
Within the area of low-energy architecture, work
by others includes the GenOpt software for
optimization purposes (Wetter 2000), which uses the
Simplex method of Neider and Mead with an
extension by O'Neill. The main drawback of this
search method is that it works well only in small
problem sizes, up to about 10 variables. Radford
used dynamic programming as early as 1978 to
select window size and glazing materials to
minimize energy consumption in a space, but the
method was ineffective because the functions were
non-separable. More recently, previous attempts to
use DOE-2.1E to optimize design parameters of
buildings include using regression analysis on data
created by parametric DOE-2 runs (Sullivan et al.
1992). Simplified procedures like the LT method
(Baker & Steemers 2000) use energy graphs to
explore the impact of a few key building-design
parameters but are unable to handle the level of
detail and complexity enabled by the system
presented in this paper. GAs have been used in
building applications related to energy consumption
mostly to optimize the sizing and control of HVAC
systems (Huang et al. 1997, Wright 1996, Dickinson
et al. 1995).
2 SYSTEM DESCRIPTION
The Generative Design System presented [Plank]
applies inverse-design methods to the search of low-
energy architectural solutions, by considering such
key factors as the use of natural lighting, subsequent
savings in artificial lighting, thermal behavior of the
building, and its annual energy consumption.
Development and testing of the system has been
described elsewhere (Caldas & Norford 2000;
Caldas & Rocha 2001). Plank couples a search
procedure [Genetic Algorithm] with a building
simulation software [DOE-2. IE], allowing hundreds
of design alternatives to be evaluated, while
requiring from the user only the effort of creating the
initial DOE-2 input file. The evolution of design
alternatives is guided by objective function values
[in the simplest case, decreasing annual energy
consumption]. Plank has both a method for creating
new designs, a procedure to evaluate them, and a
mechanism to evolve solutions towards improved
performance.
A GA was chosen as the search mechanism due
to its flexibility regarding problem structure, and
178
because it manipulates in parallel a population of
solutions. This is particularly appropriate to
architectural design, as the final output is not a
single solution, but a number of high-performance
alternatives that the architect can further develop by
considering other criteria not included in the initial
search process. A GA starts by generating a
population of random solutions, evaluates their
fitness [objective function], and subsequently applies
the basic genetic operators of reproduction,
crossover and mutation. This generates a new
population with higher average fitness than the
previous one, which will in turn be evaluated. The
cycle is repeated for a user-defined number of
generations. For more information on GAs, see
Goldberg (1989).
DOE2-1E was chosen as the simulation engine
because it provides a good compromise between
accuracy of results and reasonable computational
time, and because it incorporates both lighting and
thermal analysis. Those aspects are interdependent
when designing elements such as fenestrations, and a
software that considers only one of them, even ifin a
more detailed way, would not be as useful a design
tool as one that does a more holistic building
analysis. In this study, even when annual energy
consumption is used as the single objective function,
that value incorporates both space conditioning and
lighting energy. Better daylighting use is expressed
as lower artificial lighting consumption.
3 EXPERIMENTS
3. 1 Incorporating language constraints
From the early stages of system development, it was
clear that Plank would only be accepted by
practitioners if it could handle the complexity of
real-word applications, and if it could provide
architects with the possibility of incorporating
design intentions and architecture language into the
system. The latter was a particularly sensitive point,
and was the basis for the research described in this
section.
To investigate the integration of architectural
design intentions into Plank, the first application
area was the design of building facades. Studying an
existing building was the approach chosen for these
experiments. The School of Architecture at Oporto,
designed by architect Alvaro Siza, was used as a test
bed because the clear but complex composition rales
used in the elevations provided an excellent
framework to work upon. Due to the large dimension
ofthe project, the study focused solely on one ofthe
studio buildings, tower H [see figure 1].

HC·
Figure 1. Southeast view of studio towers. Tower H is in the
foreground
Tower H was chosen for this study for its rich
spatial configurations and use of a variety of
architectural light sources: fenestrations of different
proportions and sizes facing distinct orientations
[some including overhangs], zenithal light from roof
monitors in the top floor, and a loggia in the south
façade [see figure 1]. From a computational
perspective, tower H also presented some
challenging features. The internal relations between
the different spaces and their light sources give rise
to a multiplicity of interactions that are hard to
predict, and make the resort to computational
analysis an interesting option. The fact that Tower H
mainly houses studio teaching rooms also makes a
strong case for the careful control of natural light in
order to maintain adequate daylighting levels for
drawing tasks while precluding direct sun over the
drafting tables and excessive solar gains in the
rooms.
The objectives of these experiments were
twofold: first, to study the incorporation of language
constraints into the generative system, so that
solutions generated are within certain design
intentions; second, to examine the results generated
by Plank from the perspective of the existing design
by Siza, an architect well known for his control of
light, and to analyze to what extent the inclusion of
factors other than light [like the thermal performance
of the building] could make solutions follow a
different path.
The generative system works on a complete three-
dimensional description ofthe building, including its
geometry, orientation, spatial organization,
construction materials, etc. In this study, building
geometry, space layout and construction materials
were left unchanged, and the algorithm's search
space related only to elevation design solutions. For
the existing building layout, Plank generates a
population of façade solutions that take into account
the use of daylighting, artificial lighting, and energy
used to condition the building. Although maximum
use of natural lighting is a desirable goal, the control
of heat gains and losses introduces a balance point to
be achieved. It is this elusive balance point that the
computer tries to locate.
179
For each individual space in Tower H where
daylighting is available, two lighting reference
points were selected [typically the furthest points
from the windows where a certain light level was to
be achieved] and desired illuminances values were
specified according to the type of occupation and
tasks performed. Generally, 500 lux were used for
studios and other working spaces and 150 lux for
service areas. The artificial lighting system is
supposed to be continuously dimmable [even though
in Siza' s existing building such a system is not
implemented], as this provides an efficient method
to measure daylight use.
The GS was run for two climates with distinct
characteristics, to test its capability of adapting
architectural solutions to different climatic
requirements while subject to the same language
constraints. Apart from Oporto's mild climate,
where the existing building is located, the other
climate chosen was Chicago, USA, a climate
dominated by extremely low winter temperatures.
3. 1. 1 ANALYSIS OF EXISTING BUILDING
To incorporate architectural design intentions into
the GS, rules derived from Siza's original design
were used. Analysis of drawings and visits to the
building allowed us to infer design rules that we
consider to be applicable to the existing elevations.
Those rales relate both to compositional axes of the
facades and to general proportions of the openings.
In tower H, different rules seem to apply to each
elevation, while maintaining a strong coherence in
the overall design ofthe building and in the relations
with internal spaces [for example, long horizontal
strip windows are always used in the studios]. This
interpretation of existing design rales was followed
by the determination of search areas for the
generative mechanism, implemented as constraints
to the algorithm. Those are bounded by maximum
and minimum dimensions the openings can assume,
and those limits were made sufficiently broad to
allow for a significant search space that could
promote the emergence of a rich variety of solutions.
Other constraints implement the compositional axes
determined during the analysis stage.
In figure 3, the upper row represents the
constraints applied. Compositional axes are shown
in lighter lines. For each opening, the smaller area
represents the lower bounds to the algorithm, and the
larger area the upper bounds. For horizontal
windows, the constraints are specified in a way that
prevents the appearance of vertical openings. This
set of constraints was proposed by us to control the
generation of solutions within certain architectural
intentions that we relate to Siza's design. Changing
the constraints would allow for the exploration of
many different design solutions, a path not pursued
in this work.
 Once the constraints were graphically determined,
they were used as inputs to the generative system.
After the GS finished running, results from the
search process could be inspected using an existing
visualization program, to relate each design solution
to a corresponding energy consumption profile.
3.1.2 RESULTS
Results from the GS ranged from an almost exact
coincidence with Siza's solutions to some radical
departures from the existing design. Figure 2
displays three-dimensional models of both Siza's
and the best solution found by the GS. Figure 3
shows the elevations of some representative GS
solutions, including constraints used and the existing
design.
admittance of daylight into the room, and to
counteract that effect the GS increases the openings
size. When overhang depth is a variable
[oporto_shading], the algorithm reduces it to 0.5m,
and also reduces window sizes to a dimension closer
to that used by Siza. The shallower overhangs still
manage to block direct sun and high solar gains,
since in the hottest months the sun is high in the
south quadrant and can be controlled with smaller
overhangs. On cold winter months, when the sun is
lower in the sky, useful solar gains are still admitted
into the rooms, reducing the need for heating. It
should be added that the oporto_shading solution has
lower energy consumption than oporto_best.

:o =o
ι ι ι
ÆE 6
Figure 2. 3D models of Siza's and GS solutions. The images on
the left show northeast views, with Siza's on the left and GS on
the right. The images on the right show southwest views, with y/\ ,

Siza's on the right and GS on the left. Τ

U_l CD
1 1=1 ι 1 1

In the north façade, the large horizontal stripes 1 cn ι 1 1

generated by the algorithm very approximately

1

ι 1 1
CD c=> =3
ιι r
6 i_u ' 1 o-
resemble those created by Siza [except for the
melodic variations in height in the original design],
demonstrating that in Oporto's mild climate the use
of natural light in the studios clearly offsets heat n n
Pa
losses through the large glazing areas, as Siza may ΤΓΓ Oa
have predicted. It can be observed in figure 3 that as iD = -ffib- Pd
the quality of solutions decreases [oporto_best,
oporto_average, oporto_worst], north window sizes
decrease too.
Towards the west, the GS used small window
I-n ' I
sizes as Siza did, even further reducing them. This
I

I I
I I I

was due to the lower illuminance levels that the

zrczt
bSd £ "=>
service areas [stairs and restrooms] require and to
the reduced size of the spaces. As the openings get
larger, the quality of solutions decreases.
In the south orientation, the generative system Β I <=»
solutions present more significant modifications in I

I=D Πα
relation to the existent. In Siza's design, the second izzicz Ξ. f cd
and third floors have south facing studios with long i ι η ι-

horizontal windows shaded by 2-meter deep

overhangs. The algorithm solutions tend to suggest
these overhangs may be too deep. When the
overhang depth is kept as 2 meters [oporto_best],
window sizes assume the largest dimensions allowed
by the constraints. The deep overhangs block the
Figure 3. Oporto solutions
In the 6th floor, the GS solution for the south-facing
loggia must be analyzed in conjunction with the roof
monitors. The 6th floor is basically occupied by a

180
single space, lit from above by two roof monitors,
from the south by a loggia window, and with blank
walls in all other directions. The GS increases the
south-facing loggia window to the maximum
allowed by the constraints, and reduces the glazed
area of the roof monitor that lights the space closer
to the loggia (see figure 4). The roof monitor faces
north and is a large source of heat losses in winter,
particularly because warm air rises to the glazed
areas. Increasing the south opening permits reducing
the roof monitor without losing too much daylight in
the studios. On the other hand, the second roof
monitor assumes the largest dimensions possible in
the GS solution [as in Siza's design], since that area
of the sixth floor has no other light source. This
result suggests the tilt of the roof could be varied to
allow for a larger roof monitor in that location, and
is the basis for further experiments described later in
this section.
Figure 4. GS solution for the larger roof monitor, viewed from
the outside [left]. Existing solution viewed from the inside
[right].
The 4th and 5th floor south solutions must be
analyzed together with east results, since in those
floors the studios share both south and east
openings. The GS increases south-facing windows in
relation to the existing design, and simultaneously
reduces east-facing ones. East orientation is
unfavorable due to high solar gains during the
morning in summer months and reduced daylighting
levels during the afternoon for most of the year.
South-facing openings perform better both in terms
of natural light admission and control of heat gains.
When the algorithm has the possibility of trading
between the two options, it consistently favors south.
Figure 3 shows that as the size of east facing
windows increase, the quality of solutions decreases.
However, when the dgorithm was allowed to place
overhangs in the east façade too [oporto_shading], it
significantly increased window sizes in the second
floor, while placing quite deep overhangs to shade
the low morning sun. It should be noted that the
studio in the second floor has only east-facing
windows. For the studios on the 4th and 5th floors,
which have both east- and south-facing windows, the
GS kept east openings small [although slightly larger
than in the unshaded case] with shallow overhangs,
and again privileged south-facing openings.
181
Table 1 shows annual energy consumption levels
for the several solutions represented in figure 3. For
Oporto's climate, the worst solution found by the GS
has about 26% higher energy consumption than the
best solution with shading as a variable. Siza's
design consumes about 10% more energy than the
best GS solution with shading added.
Finally, Figure 5 shows results broken down by
energy end-use: lighting, space heat, space cooling,
ventilation fans and others (pumps, etc.).
Table 1. Annual energy consumption for Oporto solutions
Solution MWh
oporto_best_shading 87.58
oporto_best 89.99
oporto_average 96.22
oporto_existent 96.45
oporto worst 110.55
oporto_baK^«hao%ig opwto_«v«igt opono_»»it)tf»
3.1.3 LIGHTING SIMULATIONS
To further analyze daylight patterns in the space,
a lighting simulation program that combines
radiosity with ray-tracing was used to visualize both
the existing and GS solutions at some
representatives days of the year [solstices and
equinoxes].
The room simulated was the 4th floor studio, a
space where the GS could trade off between south
and east-facing windows, to try to understand the
mechanisms behind the final choice. Figure 6 shows
the renderings for the summer solstice at 9am, 10am,
noon and 3pm. The GS solution shown is the best
solution without using shading as a variable.
From the figure it can be seen that, during the
summer, the large unshaded east-facing windows in
the existing solution allow direct sun penetration in
most of the room during the morning. Tn the GS
solution, although both windows are still unshaded,
the south-facing one allows significantly less direct
sun into the room, and the east-facing window is
used only to light the back of the room. In the
afternoon, the existing solution becomes quite dark,
while the GS solution presents higher and more
evenly distributed light levels.
ejtisteflt_9 03am GS_9 00am exelent_10.00am
exis!ent_noon ejdstent_3.00pm
Figure 6. Comparison of existing and GS solutions [summer
solstice, 9am, 10am, noon and 3pm]
To further investigate daylight patterns during the
afternoon, illuminance contours in lux were
produced (figure 7). In Siza's solution, daylighting
levels never achieve the specified setpoint of 500
lux. In the furthest corner from the windows, light
levels are only around 200 lux. The same location, in
the GS solution, has a daylight illuminance level
about 4 times higher. In general, the GS solution
achieves quite high luminance levels throughout the
space. Close to the south-facing windows,
illuminance levels may even be too high, but this
could be solved by placing a shallow overhang over
the south window, as the GS did when it was
allowed to use shading devices as variables.
1>ί I*«
Lighting levels contours for existent [left] and GS
7.
[right] solutions. Scale goes from 0 to 2500 lux.
3.1.4 SHAPE MANIPULATION
A first attempt to introduce shape manipulation into
the generative system is described in this section.
The GS was allowed to vary the roof tilt of Tower H,
and thus control the size of the rooftop monitors.
This experiment was a consequence of the results
previously described, which suggested that the GS
would tend to change the size of the roof monitors if
allowed to do so by the roof configurations. To
simplify the experiments, it was assumed that the
roof monitors would always cover the entire width
of the building, as in Siza's original design. The
height would be determined by the tilt of the
corresponding roof, as the glazed opening would
have the same height as the wall. The roof tilt was
allowed to vary between 10° and 45°. The
northernmost rooftop had to be at least 2 meters set
back from the north façade so that it could not be
read as part ofthe elevation.
GSJO.OOam
Figure 8 represents some solutions extracted from
those generated by the GS. The best solution is in the
middle ofthe top row.
GS_3 OOpm
EaV-
Figure 8. Roof tilt solutions generated by the GA.
3.1.5 CHICAGO EXPERIMENTS
The extremely cold Chicago climate originated some
interesting departures in relation to Oporto solutions
(fig. 9). For north-facing studios, the windows were
reduced to the minimum dimensions allowed by the
constraints, due to high heat losses through the
glazing and to the absence of solar gains that would
be beneficial in Chicago's cold climate. This facade-
level solution may allow for an extrapolation in
terms of spatial organization, suggesting that north-
facing studios should be avoided in this type of
climate. It is also interesting to observe that, for the
north orientation, the best solution for Chicago is
very similar to the worst solution for Oporto (fig. 4).
=fÆ f_
Figure 9. Chicago results
Towards the south, unshaded windows were made
quite large, since they couple daylight admission
with useful solar gains. However, shaded windows
were reduced to minimum dimensions, as both
natural light and solar gains are blocked, and heat
182
losses prevail. When overhang depth was used as a
variable, the algorithm reduced it to the minimum
allowed and simultaneously increased window sizes
[result not shown in the image]. It can be concluded
that south shading may be undesirable in this
climate. Towards the east, rooms that have only east-
facing windows received average-sized openings
[1st and 2nd floors], a compromise between such
positive factors as daylight admission and morning
solar gains, and such negative factors as high heat
losses through the glazing. For studios with both
east- and south-facing windows, east openings were
made quite small because once again south was
preferred. West fenestration received minimum
dimensions.
3.2 Applications to multicriteria problems
This section presents an extension of Plank to handle
multiple objective functions. Most decisions in
architecture design are made considering different
criteria. More often than not, those criteria are in
conflict with each other. Architects are trained to
consider many simultaneous factors in decision¬
making, but when quantitative aspects are at stake,
computer approaches that help providing relevant
information can prove to be particularly useful tools.
The most common multicriteria optimization
methods are plain aggregating approaches,
combining by means of weighting factors the several
criteria under assessment, to provide a single figure
of merit that aims at completely characterizing the
quality of a solution. The main flaw with this
approach is that the final result is heavily dependent
on the weights attributed to each factor, and provides
little insight into performance according to each of
the objectives.
Pareto optimization is based on the work of
Italian economist Vilfredo Pareto (1848-1923). It
rejects the search for single, optimal solutions, and
avoids artificial aggregations using weighting
factors. Instead, Pareto optimization supplies
decision-makers with information on the best trade¬
offs achievable for a specific problem formulation
and constraints. This information is presented under
the form of Pareto fronts. The decision-maker will
then chose where in the front will the final solution
be located, that is, what compromises will be made
in the design. This is a particularly suitable approach
to handle multicriteria problems in architecture,
since there is usually no single solution that
performs best in terms of all the criteria, meaning
there is no Optimal' solution. Optimal performance
according to one objective often implies
unacceptably low performance in one or more of the
other objective dimensions, creating the need for a
compromise to be reached. In any case, there are
solutions that represent better trade-offs than others.
183
For example, if cost of a solution is being considered
against performance, the Pareto front will provide
information on the best possible performance for a
given cost. However, for a higher cost, a better
performance may be achieved. On the other hand, it
may be possible to find a more economical solution,
but performance will most likely degrade. What is
undesirable, in any situation, is that for the same cost
one may chose a worse solution, and the
determination of Pareto fronts can be instrumental in
avoiding this situation.
3. 2. 1 PARETO GENETIC ALGORITHMS
Pareto optimality makes use of the concept of
dominated and non-dominated solutions. To explain
the idea of a dominated solution, χ dominates y if χ
is better than y for at least one objective function,
and at least as good on all the others (Tamaki 1996).
A solution is Pareto optimal if it is not dominated by
any other solution. In graphical terms, this concept
can be observed in figure 10, for a maximization
problem with two objective functions.
In figure 10, A is clearly a better solution than C,
but between A and Β it is not possible to
immediately say which one is a better solution. A
performs better than Β in terms ofthe χ axis, but the
opposite is true for the y axis. A Pareto search will
exclude all points of type C (dominated ones), and
find only points of the type of A and B, which will
form what is known as a Pareto front. To decide for
one of the points in the front, the designer will later
have to exert judgment and preference. In Pareto
optimization, no preference information is included
in the search, contrarily to plain aggregating
approaches, where complete preference information
is given.
y
5 -
< - 1-OVI)
3 -
1 I #Λ-<4,2)
4 «c-cj,«
Figure 10. Dominated and non-dominated [Pareto] solutions. C
is dominated by A. Both A and Β are Pareto solutions
Although GAs are most commonly used in single-
criterion problems, their implicit parallelism makes
them particularly suitable for multicriteria problems,
since a GA searches a population of solutions in
parallel. As Pareto fronts are populations of optimal
solutions, GAs prove to be adequate procedures to
locate them. There are several implementations of
Pareto GAs described in the literature. The one used
in these experiments is a Nondominated Sorting
Genetic Algorithm [NSGA] (Srinivas 1995), and
includes, apart from the most common features, are not to be considered particularly accurate ones,
sophisticated ranking and niche induction techniques but they give a relative measure of the comparative
that ensure the whole front of Pareto solutions for a costs of the different materials. In a real-world
given problem formulation to be sampled, providing application, care should be taken in obtaining
the user with a high degree of confidence that the accurate values. If quantity discounts exist, they can
best trade-offs are being located and presented. For be handled using simple if-then rules, ofthe form 'if
more detailed information on these methods, see material χ area is greater than y, then cost of material
Goldberg (1987) and Srinivas (1995). χ equals z, else if ... '.In general, costs per unit area
of insulation materials were lower than those for
3.2.2 EXPERIMENTS WITHNSGA concrete block. Expanded polyurethane was more
expensive than expanded polystyrene. Air layer costs
This section presents the use of a NSGA with Plank were set to zero. Window costs were significantly
to help selecting construction materials for external higher per unit area than any other materials.
walls, and for fenestration sizing in a building. It
assumes the external finishes have been chosen, and Table 2. Library of DOE2 materials
only the materials constituting the interior of the
u DOE2 Material Thickness
wall are under study. The criteria considered are: code Jem]
initial construction cost, the building's annual energy 1 ALU Air layer 5 1.9
consumption, and environmental impact of the 2 AL21 Air layer 1.9 to 10.2
3 AL31 Air layer E>10.2
materials applied. Expanded polystyrene 2.54
4 IN33
The building used for the experiments was a 5 IN35 Expanded polystyrene 5.08
simplified model which can be seen in figure 11. It 6 IN36 Expanded polystyrene 7.62
has four similar rooms, facing the four cardinal 7 IN43 Expanded polyurethane 2.54
8 ΓΝ45 Expanded polyurethane 5.08
directions. There is only one window per room, on CMU, medium weight, hollow 10.2
9 CB21
the longest wall. Window dimensions are variables 10 CB23 CMU, medium weight, perlite filled 10.2
to the program. External wall materials are similar 11 CB26 CMU, medium weight, hollow 15.2
for the entire building. In the first set of experiments, 12 CB28 CMU, medium weight, perlite filled 15.2
CB41 CMU, lightweight, hollow 10.2
two criteria were used: cost of construction 13
CMU, lightweight, perlite filled 10.2
14 CB43
materials, both for walls and windows, and annual 15 CB46 CMU, lightweight, hollow 15.2
energy consumption. 16 CB48 CMU, lightweight, perlite filled 15.2

Figure 12 illustrates the layout for exterior wall

Figure 11. Schematic building layout for Pareto experiments
The list of construction materials for the GS to
choose from is presented in table 2. Those materials
were selected from the materials library of
DOE2.1E, which contains an extensive pre-defined
list, and allows the user to add ones. Sixteen
materials were selected, eight with insulating
characteristics (polystyrene, polyurethane and air
construction. The interior finish was gypsum board,
and the exterior had 2.5 cm of mortar. Materials for
the three internal layers were chosen by the GS.
Material 1 represents the layer closest to the exterior.
Material 2 the intermediate layer, and Material 3 the
innermost one. Because some ofthe material options
are air layers, constraints prevented materials 1 and 3
to be air layers, due to buildability issues.
«nati «iß nto Interior

layers with varying thickness), and eight masonry

types (concrete blocks with different densities,
thickness, and some perlite-filled). Many other
materials could be included, the choice belonging to Figure 12. Wall construction scheme
the architect.
Even though the glazing type could be variable
too, it was kept constant to simplify result
interpretation. The double-glazing unit used had two
layers of 6mm clear glass, with a 6mm air layer. The
U-value was 3.16 W/m2-k, the shading coefficient
was 0.81, and the visual transmittance 0.78.
Costs for the different materials were obtained by
telephoning several retailers in the US, and
averaging the different prices obtained. Those values

184
3.2.3 RESULTS apparent that the more expensive solutions have
larger window areas, mainly towards the south.
Figure 13 shows the search progression of the
Pareto front for the Phoenix climate [Arizona, USA],
from the initial population [gen 1] with points south east north west

randomly scattered around the graph, to generation

100, where the Pareto front is already clearly visible.
Running another 100 generations [gen 200] made
only some minor improvements in the front.

a g · n 100
f « n 200

Figure 13. Pareto front results for Phoenix climate

The following step was to analyze which design

configurations led to the appearance of this frontier
of best trade-offs. Table 3 shows the solutions
belonging to the Pareto front. The area of each
window is indicated in square meters. Materials are
represented using DOE2 codes from table 1. Results
are displayed in ascending order according to energy
consumption [KWh], with lowest energy levels first.
Note that solution 2 is not Pareto optimal.

Table 3. Pareto front results for Phoenix climate

Window areas Mataria« Obj. functions
South East North West Total Mat1 Mat 2 Mat 3 KWh Cost I$]
1 8 5 7 3 23 IN45 IN36 IN36 21682 7276
2 13 5 5 3 26 IN45 IN36 IN36 21787 7917
3 12 5 8 3 27 IN45 AL31 IN36 21793 7286
4 8 3 7 3 22 IN45 IN33 IN36 21824 6653
5 11 3 5 2 21 IN33 IN36 IN36 21986 6331
6 11 4 3 3 21 IN35 IN35 IN36 22051 6294
7 12 3 4 2 21 IN33 IN35 IN36 22158 6127
8 8 3 3 3 17 IN45 IN36 IN36 22175 6093
9 8 5 3 3 20 IN33 IN36 IN36 22215 6019
10 12 3 4 2 21 IN33 IN35 IN36 22241 5971
11 11 3 3 2 19 IN33 IN35 IN36 22278 5736
12 11 3 1 3 18 IN33 IN36 IN36 22426 5619
13 8 3 1 3 14 IN36 IN36 IN36 22464 5507
14
15
8
8
4
4
1
1
3
3
16
16
IN36
IN36
IN35
IN43
IN43
IN43
22499
22550
5388
5318
D
16 11 2 3 1 17 IN33 IN35 IN36 22607 5256
17 8 3 1 3 16 IN33 IN35 IN36 22623 4962
18 8 3 1 2 14 IN33 IN36 IN36 22647 4834
19 8 3 3 3 16 IN33 IN35 IN33 22791 4606
20 8 2 1 2 14 IN33 IN43 IN36 22819 4527
21 8 2 1 1 12 1N33 1N35 IN36 22915 4284
22 8 3 1 2 14 IN33 AL.31 IN36 22983 4038
23 6 1 0 9 IN33 IN43 IN36 23698 3509
24 4 3 1 10 IN45 AL31 IN33 23879 3315
25 4 1 0 7 IN36 IN35 IN33 24026 3248
26 β 1 0 8 IN33 IN43 1N33 24269 2829
27 1 3 1 6 IN36 AL11 IN33 25430 2517
28 1 1 1 4 IN45 AL31 IN36 25438 2752
29 1 1 1 4 IN33 AL31 IN36 25622 2461
30 1 1 2 5 IN33 AL31 IN43 25857 2278

One significant point to analyze is that window

costs dominate the overall cost. Thus, a main
strategy of the GA to reduce costs is to decrease Figure 14. Pareto front for Phoenix. Solution 1 is in the top row
window size, as illustrated in figure 14, where it is

185
 However, those solutions are also the ones with
better energy performance. There is a strong relation
between reducing fenestration size and a decrease in
the building's performance. West seems to be an
exception, since even in the best solutions window
sizes are small.
In terms of wall materials, even though in the first
random solutions many instances of heavyweight,
masonry elements appear, in the final Pareto
solutions only lightweight, insulation materials are
used (in the wall sections in figure 14, the lighter
hatch represents polystyrene, the darker hatch is
polyurethane, and white represents an air layer). One
determining factor is probably that the cost per unit
area of insulation materials is lower. However, it
might be expected that in a hot climate like Phoenix
some thermal mass would be applied in the walls,
leading to the choice of masonry materials. Instead,
the GS may be using the thermal mass from the
concrete floors and roofs as heat storage to dampen
peak loads, allowing the walls to become
lightweight, insulating elements. Finally, it is
possible to see that the lowest energy solutions use
better insulation materials, like #6 and #8. To reduce
construction costs, the GS progressively starts using
large air layers and lower quality insulation
materials, combining that with a reduction in
window sizes.
It is worth noting that one of the individuals the
initial random generation [geni] already performed
almost as well as the best Pareto solution in terms of
energy, but its construction costs were about 33%
higher. Using the Pareto front search, it was possible
to achieve a similar performance while saving about
1/3 in costs. In terms ofthe average performance of
the populations, the energy reduction from geni to
geni 00 was only 6% [from about 24.7 MWh to 23
MWh], but cost reduction was about 41% [from
$8,434 to $4,965], suggesting that Pareto searches
may be an effective measure for achieving similar
performance levels at lower costs. Finally, there is a
19% difference between the best and worst solutions
in terms of energy consumption, and a 69%
difference between the highest and lowest cost,
suggesting that there is a large margin for decision¬
Materials used are mostly the same as in the
previous experiments. The main difference is that
material #4 was replaced from 2.54cm of expanded
polystyrene to 8.9cm of cellulose fill, as this is
considered to be an environmental-friendly
insulation material, and including it in this
experiment could possibly yield interesting results.
The data used for GGE emissions [table 4] were
appropriate for Switzerland, but it was considered
acceptable to apply it in the context of this study.
Once again, emissions data are not intended to be
highly accurate, but to provide an approximate
measure of the environmental impact of each
material.
Table 4. GGE emissions [kg C02 eq./ kg of building material],
Switzerland data
Material Air layer Cellulose Expanded Expanded Cement Perlite
fill polystyrene polyurethane
GGE 0 0.113 2.31 13.75 0.0737 0.67
Figure 15 presents the results obtained for the
Oporto climate, and displays very low GGE levels
associated with materials used in the Pareto-front
solutions. This happens because the GS uses mainly
cellulose insulation for the external walls, due to its
reduced GGE rate. The two solutions in the graph
with the lowest energy levels use other insulation
materials, like lightweight concrete blocks and
expanded polystyrene, but the corresponding
decrease in energy use is negligible, while the
increase in GGE is significant. An additional
consideration is that most wall configurations
proposed by the GS have buildability and structural
problems, as they are not self-supporting. Although
the cellulose insulation walls could be built in a
composite panel configuration, with exterior rigid
panes and interior insulation, they would require an
internal frame, whose cost and GGE emissions
should be included in the analysis for a more correct
comparison with the masonry walls.
»7000
4
o5
o
£
,6000

making within the Pareto results. 5000

4000 -D õ"
3.2.4 INCLUDING MATERIALS EMISSIONS
3000 s
This section expands the work on Pareto fronts to 2000
include the Global Warming Potential [GWP] of 1000 »··-. agen 1
materials used. Two objective functions are *·*... gen 200
considered: annual energy spent for space 0
22 24 26 28 30
conditioning [MWh], and Greenhouse Gas MWh

Emissions [GGE, measured in kg CO2 eq.] related to

Figure 15. Pareto front for Oporto climate, relating energy
the use of a material. GGE emissions can also serve
consumption with materials GGE emissions
as an indicator of the embodied energy of the
material.

186
 In terms of window areas, and because glass has a
high GGE level, the GS privileges south windows to
achieve low energy consumption in the building, and
progressively reduces fenestration areas in all the
other facades. Only when minimum areas are
reached towards the other directions does the GS
make a reduction in the south openings. This
strategy is repeated until minimum areas are reached
in all orientations simultaneously.
3.3 3D SHAPE GENERATION EXPERIMENTS
This section presents the application of Plank to the
manipulation of three-dimensional architectural
shape, guided by goals of daylighting use and energy
conservation. Departing from an initial schematic
design and a set of rules and constraints that encode
the architect's intentions, the system searches the
solution space for design configurations that respond
to the initial targets. Methodological questions are
first presented, concerning the encoding of initial
layouts and the emergence of new design features.
Results from experiments are then presented, using
Energy Use Intensity as the objective function, and
also considering the use of penalty functions.
To use this method, the architect defines the
search space or universe of possible solutions
through the use of a set of formal symbolic rales,
constraints, relations, etc. Then, the GS manipulates
the three dimensional geometry of the building, its
space layout and window positioning and size, to
approach the desired targets.
3. 3. 1 Parametric shape manipulation
Finding simplified initial layouts that could allow for
the emergence of a rich variety of formal solutions
when manipulated by the GS was one of our first
research questions. Determining adjacencies was
perceived as a major problem, so a simple
experimental layout was chosen where all spatial
adjacencies were predetermined.
The basic layout from where the new solutions
could emerge was simply, in plan, a square divided
in four similar squares [see figure 16]. In the 1st
floor, this corresponded to four rooms [Rl, R2, R3
and R4], which could vary in their length and width,
but were constrained to have the same height. In the
2nd floor, there were four other rooms [R5, R6, R7,
and R8] that could vary in height too. The tilt of the
roof could vary from a flat roof to a maximum of
45°, and the roof azimuth could also vary from 0° to
90°, as shown by the arrows in figure 16. Whenever
there is a tilted roof, a roof monitor with length
equal to the corresponding wall is also generated.
This apparently simple problem has nevertheless
44 independent variables and generates about 350
dependent variables, which must all be correctly
187
coded for the program to run without mistakes. By
looking at the specificities of passing geometrical
information into a heat transfer simulation program,
it will be possible to understand this large number of
dependent variables.
DO 0
/s Λ
0 j
R4 R3 R8 R7
so j
RI R2 R5S, R8,
o
2nd floor
Figure 16. Basic layout for 1st and 2nd floor. Arrows show the
possible directions for each roof tilt.
The Building Description Language [BDL] used
by DOE2 has three coordinate systems, namely the
building, room and wall systems. To locate a new
space in the overall building layout, it is necessary to
correctly determine its coordinates in the building
coordinate system. The building coordinate system
origin (0,0,0) was fixed as the intersection point of
the two axes defining the four rooms. From that
reference point, it is possible to correctly locate each
of the spaces, even though their dimensions are
variables generated by the genetic algorithm. Using
the insertion point of each room as its lower left
corner, and considering the room's azimuth [North =
0, East = 90, South = 180, West = 270], the insertion
point of each space is calculated. A similar
procedure was used to locate the 2nd floor spaces,
with the difference that the Ζ coordinate was equal
to the height ofthe 1st floor [2.8 m]. Since the height
of 2nd floor rooms was allowed to vary, the Ζ
coordinate for the roofs was variable too. Because
roof azimuths were not fixed either, roof insertion
point coordinates would be different if the azimuth
was set to 0° or 90°.
This method for handling a variable building
geometry represents one of the simplest parts of the
problem. When building geometry is fed into a
program like DOE2, that calculates heat transfer
across surfaces, other layers of information [e.g.
spatial adjacencies and other topological
characteristics] become crucial. In the example used,
the problem was simplified by choosing a layout
where space adjacencies are fixed. However,
adjacencies to the external environment are not
entirely described once the spaces start to be
dynamically parameterized by the GA.
Figure 17 illustrates the simplest example of
altering just one room dimension. While in the first
case there is only an internal wall between Rl and
R2, in the second case a previously nonexistent
exterior wall has appeared in Rl . In the third case, a
new exterior wall also appears, but now belonging to
R2. Within BDL, the interior wall must be declared
for both RI and R2, but the new exterior wall should
only be declared for the room it belongs to.
R4 R3 R< R3 R4 R3
IW IrV IW
R2 R1
R1 R2
R1 ew EVV m
Figure 17. Example of external walls appearing due to
parametric variation of room dimensions
While in geometrical terms the interior and
exterior walls may be the same vertical planes, in
construction terms those entities are usually built
with different materials, and that is coded into
DOE2. In terms of heat transfer they are even further
different. An interior wall is just a boundary between
two interior environments that are usually not very
different, and thus heat transfer across the wall is
small. In contrast, an exterior wall is a boundary
between indoor and outdoor environments that can
be very different, and thus the wall would have
significant heat transfer across it. Furthermore, for
the exterior wall the azimuth it faces is important,
due to its relation to sun position. Another
characteristic of exterior walls is that they may have
windows, which are determinant factors in the
performance ofthe building.
In the 2nd floor, the generation of new external
walls becomes more complex, since the rooms are
also allowed to vary in height. For each possible
adjacency situation, four different possibilities have
to be predicted, which are depicted in figure 18. In
the first case, no new external wall is created for Rl .
In the second case, a vertical external wall appears
[El]. In the third case, a horizontal one appears [E2].
In the forth case, an L shaped external wall appears,
which has to be decomposed in two pieces [El and
E2], as shown in the figure, due to BDL format
specificities.
These four possibilities, that in geometric terms
are quite simple, pose nevertheless a data
representation problem for the generative system.
Because the GS does not have a CAD interface,
from which it could read information on building
geometry and generate the respective BDL file, the
system needs to work with a fixed-format BDL file.
This file must include from the beginning all the
possible geometric occurrences within the problem
setup.
In this example, the most complex case would be
the L shaped wall, with occurrence of both El and
E2 components. Given this, all rooms must predict
the possible existence of El and E2 entities. If a
given entity does not occur when the genetic
algorithm generates a new individual, that entity
parameters and coordinates must be driven to 0. If
188
that entity exists in the current geometric
formulation, its parameter values [like width, height,
and other possible information about construction
materials, wall solar absorptivity, internal light
reflection, etc.] must be calculated from the
independent variable values, and its insertion point
coordinates determined for correct positioning in
relation to the space, now using the room coordinate
system. For simplicity, the code prevents the
appearance of windows in these external walls,
whose existence is uncertain, as this would add
another layer of complexity to the problem.
However, in a more complex problem formulation it
would also be possible to predict the appearance of
new openings in the newly generated facades.
/
/- .-1\ <-T \
"";.<-' \ <"" I \>
\a; x"'
\
\ fcr^ V
^v
s"
/ \ ^,''
\
Figure 18. New walls in Rl, according to R2 dimensions
A similar procedure to that used for 2nd floor
external walls was applied to 1st floor roofs, and
exterior floors slabs of 2n floor rooms. In all of
them, the most complex case is the existence of an L
shaped element, and thus all rooms must incorporate
the possibility of the occurrence of this case. If any
elements do not exist in a given solution, the
program must drive their values to 0. Forgetting to
include any of these surfaces can introduce
significant errors into the search procedure, as they
represent important heat transfer areas. Finally,
interior walls between adjacent spaces must also be
calculated.
As for window size and positioning, a number of
issues emerge when dealing with variable building
shapes. When the building shape is fixed, it is
possible to easily determine the upper bounds for
window size, as those are limited by the dimensions
of the exterior wall. However, if the wall size is not
known in advance, it is not possible to determine
that upper bound. This represents a major drawback
in terms of the standard genetic algorithm
functioning. In common GA implementations, the
constraints for each variable are determined prior to
running the program. To overcome the fact that
constraint bounds are not known in advance, the
constraints would have to change dynamically
during the course ofthe program. Since this dynamic
constraints GA has not yet been implemented, it was
necessary to find a simplified solution to realize the
experiments. This solution was to make the window
width equal to wall width minus external walls
thickness, thus becoming a dependent variable. In
terms of height, 1st floor windows posed no
problems, as wall height was fixed and constraints
could be determined in advance. For the 2nd floor,
the maximum window height was set equal to the
minimum wall height, to ensure windows would
always fit into the respective wall, no matter what
their height might be.
These simplified rules have the drawback of
allowing little variation in façade design. Windows
always stretch from wall to wall, and can only vary
in height. This led to a certain standardization of
generated window solutions, which is nevertheless
counteracted by the great variety of shapes that are
generated by the GS. To introduce more diversity
into the experiments, and also as an useful
environmental analysis strategy, window height can
be driven to 0, meaning that if the GS finds that
excluding a window introduces benefits in terms of
overall building performance, it is allowed to do so.
The location of daylighting reference points has
to be calculated by the program for each generated
space geometry. The rale for placing the sensors
were: one sensor in the center of the space, and the
other 2 meters away from the innermost walls, that
is, the walls that have no windows. This tries to
ensure that natural light is used throughout the space,
and that it penetrates into the deeper areas of the
rooms. The Ζ coordinate of the sensors is 0.75m,
approximately desktop height.
Other parameters, like room area and volume,
have to be calculated after the independent variable
values are generated. In calculating the volume, it
must not be forgotten that when a tilted roof exists,
the corresponding volume under it must be added to
the basic parallelepiped volume. Air volume can
significantly influence the energy consumption of
HVAC systems.
A remark should be added about self-shading
calculations. DOE2 will not calculate building self-
shading unless appropriate surfaces are explicitly
declared as shading surfaces. Since from the outset it
is unknown what surfaces will be shading others, all
exterior planes [wall, exterior floors, roofs] should
be declared as shading surfaces so that they are
considered as such. This can be particularly
important in cases where a 2nd floor space projects
over a 1st floor one, and acts as an overhang for it,
influencing both solar gain and daylighting levels, or
for rooms with different heights, tilted roofs, and
recessed external walls in relation to adjacent ones.
189
3.3.2 Experiments
Some initial random configurations generated by the
GS using the previously described layout are shown
in figure 19.
Figure 19 - Some initial random configurations.
A major concern in these experiments was that
the main strategy the GS would use to reduce
building energy Consumption would be reducing
building area. No matter how efficient and adapted
to the outdoor environment a large building is, it will
always use more energy than a very small building.
So, the predicted outcome from experiments using
energy consumption as the objective function would
be a population of minimum possible dimensions
buildings, with some variation in façade design.
It was thus evident from the outset that building
area would have to be included in the fitness
function. This was implemented in two different
ways. One implied the use of penalty functions
related to area requirements violations, and the other
used Energy Use Intensity [EUT] as the objective
function, which translates energy use per unit area.
3.3.3 Penalty functions
The use of penalty functions was first
experimented. The strategy used was that each ofthe
building floors should have a certain area, related to
functional and programmatic requirements. The GS
could assign different areas to each of the four
spaces in that floor, in the best way it found in terms
of environmental performance, but the total area of
the floor would have to equal a given number of
square meters. The penalties were then calculated
according to the amount by which that area
requirement had been violated by a solution.
Penalties were equally applied both for too small and
too large spaces. These penalties would then be
added to the original fitness value of the solution
[annual energy consumption], and would degrade
that fitness according to the extent the area violation.
This method intended to ensure that if the GS
tried to reduce floor area to a minimum so that the
energy consumption would be low, a high penalty
function would significantly degrade that solution's
performance and make the GS move away from it.
The penalty values could vary by a large extent,
since they are based on floor area calculations, and
floor area is allow to vary significantly in this
problem set. Each room dimension can vary between reduced ones towards west. In general, east/west
3 m and 15 m, so room area can range from 9m2 to facades are smaller that south/north ones. Towards
225 m . This represents already a large variation, but the northern side on the building, it avoids the
if those areas are aggregated into the total area for northeast orientation, which is always unfavorable
each floor, combining the four rooms, each floor due to reduced lighting levels for most of the day,
plan can range from 36 m2 to 900 m2, a very placing very small spaces in that corner. Towards
significant variation. For this reason, the penalty northwest, it used larger volumes [even if smaller
values for solutions that seriously violated the area than those towards the south], and used the 2nd floor
requirements could be quite high. The penalties were volume to shade the 1st floor west-facing opening, to
based on a required area for each floor of control solar gains, as it did with the southwest
approximately 470 m2, which is about half of the room. In the overall, this seems a balanced and
allowed range. Following this method, the penalty reasonable solution, and demonstrates the GS is able
for each floor area was: to create appropriate geometries for a given problem.
penalty= (abs (470-floorarea) / 470) * 70
The factor of 70 was adopted after parametric
experiments, as it was found that too small penalty
V·
factors would still make area reduction the best
strategy for energy savings. The penalty function
method was developed because there is no formal
way in genetic algorithms to constraint the outcome
from a combination of variables. For example, it is
possible to place upper and lower bounds in a / y \
room's length and width, but not in a room's area, /'/\
since it is the result of a multiplication of two
independent variables.
Figures 20 and 21 show results for experiments
using penalty functions, for Oporto's climate. A "_Î25 2JZ23
population size of 30 individuals was used, from
which only 9 are represented. The GS was run for
200 generations. Figure 20 shows six good results,
and figure 21 shows three poor-performance ones.
^
From figure 20 and table 5, some conclusions can be
drawn. The values in the table show that the best
solutions use different strategies to achieve low 5 45
fitness values. While some stay closer to the required
areas, increasing energy consumption but decreasing /
penalty values [#1, for example], others prefer to
reduce their energy consumption levels by cutting on
floor area, while still achieving a good final fitness ¥ ' '
^
value [#2].

Table 5 - Initial fitness values in MWh [solution numbers

correspond to those in figure 20], penalty for area constraints
Figure 20. Best solutions for Oporto using penalty functions. 1
violation, and final fitness values
is the best solution. For each building there is a SW [45°] and a
Solution # Fitness with no Penalty Final fitness NE view [225°]. Drawings not to scale.
penalty [MWh] value
1 96 1 97 Solution 2 is not very dissimilar to 1 towards the
2 82 26 108 south side, but towards northwest it substantially
3 98 11 109 reduced room areas. This generated an energy
4 92 19 111 reduction, but also a quite large area penalty, which
5 105 9 114 lead to a solution over 10% worse than 1. Solution 3
6 85 30 115 is not too dissimilar to 1, but has higher energy
consumption even though its area has been reduced
Solution 1 is the best. Even though it complies [as shown by the penalty function value], probably
with area requirements, it manages to do it without because of the exposed south-facing 1st floor roofs,
greatly increasing energy consumption values, what since the 2nd floor is recessed into a terrace
reveals a good degree of adaptiveness to the configuration. It also has quite larger west-facing
environment. It creates elongated spaces facing openings. Solution 4 is interesting in that it plays
south, with generous openings to this orientation, but with the tilts ofthe roofs to generate south and north

190
lighting sources, privileges south-facing spaces, and
uses projected volumes to shade 1st floor west facing
windows. However, it reduces too much the north
side volumes, and is penalized in terms of areas.
Solutions 5 and 6 start to be more hybrid. 5 stays
close to the area requirements, but has high energy
consumption levels, probably due to the large
glazing areas towards east and west, namely due to
steep roof tilts facing those orientations. Solution 6
has a very high area penalty, and probably too big
east-facing windows.
Raising the issue of the interactivity between the
architect's intentions and the GS, after being
presented with these initial solutions, the architect
can decide what paths of exploration he is willing to
pursue to achieve solutions closer to his intentions.
He can decide to change some constraints, run a
MicroGA to explore the neighborhood of good
solutions like 1, or manually perform same changes
and simply do a DOE2 simulation of the modified
design to assess the impact of those changes.
Looking at poor performance solutions is also an
useful exercise, to assess which design feature have
a negative impact on a solutions performance. In
figure 21 and table 6, the worst solutions in the final
population are presented. In general, these solutions
not only have large penalty values for area
violations, but they also show high energy
consumption levels. This is mainly due to large west
and east facing glazing areas from roof monitors
generated by the steep roofs, and deep overhangs
shading south and north openings, thus blocking
daylight and useful solar gains. It should be added
that the external floors were modeled with no
insulation [they are just concrete slabs with an
interior finish] so in general the appearance of large
projected volumes is not encouraged.
x>
./
W2_45
\ A
\W2_225
Figure 21. Worst solutions in the last generation, using penalty
functions. Wl is the worst solution. For each building there is
a SW [45°] and a NE view [225°].
191
Table 6. Initial fitness values in MWh of the worst solutions
[numbers correspond to those in image 21], penalties for areas
violation, and final fitness values
Solution # Fitness with Penalty for Final fitness
no penalty area violation value
[MWhl
Wl 109 44 153
W2 113 29 142
W3 119 22 141
The use of penalty functions introduced another
level of complexity into the interpretation of results.
The two factors [energy use and floor area violation]
become difficult to isolate for result analysis. For
this reason, another set of experiments was done
using a different objective function: Energy Use
Intensity [EUI], which expresses energy
consumption per unit floor area. This way it was
easier to evaluate the relative environmental
performance of different solutions, independently of
the overall floor area.
3. 3. 4 ENERGY USE INTENSITY [EUI] FOR
OPORTO EXPERIMENTS
Figure 22 shows results for Oporto climate using
EUI as the final objective function. The best
individual is solution 1, which is somehow different
from the one found using penalty functions. Since in
this case the GS is not asked to assign a given total
area to the building, it sizes and distributes spaces in
a different way. Slim, shallow, all-glazed elements
are used towards the south, in a configuration that
resembles a sunspace. Larger spaces, of more bulky
proportions, are used towards the north of the
building. Those spaces have lower surface to volume
ratio, and thus have less heat transfer surfaces to lose
energy from. However, these deep spaces become
more difficult to bring daylight into. For that reason,
the GS generates quite large openings, mainly north-
facing ones, [including roof monitors], but also
towards east and west. In general, the proportions of
the building tend to orientate the widest facades
towards south and north, and the shorter ones to east
and west. The best performing shapes are quite
compact, with reduced exposed roof areas in the 1st
floor, and few overhangs or projecting elements.
This overall layout is kept in solutions 2 and 3,
with some minor variations, but starts to suffer more
significant changes as the fitness of the solutions
decreases, with results becoming more difficult to
interpret those intermediate solutions.

..-\
/ matches these configurations, they should be
/>N.
V
ν
\
V 3_ΝΕ
V .·
\ / V"
, /
-./ \
/ >v
y
/
Figure 22. Best solutions for Oporto using EUI. #1 is the best
solution. For each building there is a SW and a NE view.
Table 7 shows the fitness function values for each
of the solutions shown in figure 22, in terms of EUI
[kWh per square foot]. Among the six best solutions,
the variation in EUI values is over 10%.
Table 7. Fitness values in EUI of best solutions [solution
numbers correspond to figure 22]. EUI is in kWh/sq. ft
Solution # Final fitness
value [EUI]
1 9.02
2 9.57
3 9.69
4 9.79
5 9.88
6 10.04
Worst-performing solutions help finding patterns
of elements leading to poor behavior. Fig. 23 shows
three poor-performance individuals, with the worst
one on the left [#1]. The main characteristics of
these solutions are: high surface-to-volume ratios,
caused by slim, elongated shapes; exposed external
surfaces, like external walls, 1st floor roofs, and
floors from projected 2nd floor elements; large
glazing areas facing unfavorable orientations like
east and west.
Knowing that these configurations lead to lower
performance can either make the designer move
away from them, or if the type of architectural
language the architect is looking for somehow
counteracted by properly insulated external surfaces,
more sophisticated glazing systems, and appropriate
use of shading. These solutions will usually imply
higher construction costs, so if low-cost construction
is an issue, the architect may try to infer from the GS
results the type of shapes that lead to better
performance by adaptation to the environment.
\
/
,α W /. \ /
Figure 23. Worst solutions for Oporto, using EUI. Wl is the worst
solution.
Table 8 . Worst solutions for Oporto using Energy Use
Intensity [numbers correspond to those in figure 23].
Solution # Fitness value [EUIÌ
Wl 13.18
W2 11.85
W3 11.79
4 CONCLUSIONS
This paper presents an overview of several
applications of a new Generative Design System that
aims at operating in early conceptual phases of the
design process, to help architects in developing
architecture designs that are less energy-intensive
and more adapted to the exterior environment.
Solutions generated by this GS must not be
interpreted as definite or optimal answers, but as
suggestions for further architectural explorations,
building thus an innovative and promising
interaction between architecture and computation.
The experiments using an existing building by
Architect Alvaro Siza proved the Generative System
to be flexible enough to incorporate constraints that
allow the architect to manipulate certain
architectural design intentions, while reducing the
energy consumption levels of the final solution. The
close coincidence between GS and Siza's solutions
192
in some situations was of particular interest in
showing that a close control of architectural
language is possible within the GS. On the other
hand, the departures from the existing design
proposed by the algorithm suggest that this
generative system may be a useful tool in exploring
multiple paths during the design process to achieve
lower energy designs.
Another interesting dimension of the GS is its
capability to account for interactions between
different elements of the building, and to make the
design for each specific element dependent on its
integrated role in the architectural whole. The system
also proved to be able to provide alternative
solutions adapted to different climatic conditions,
within similar architectural language constraints.
Results from the Pareto-based studies proved in
general to valuable in understanding how the trade¬
offs between conflicting objectives influence design
solutions located by the Generative System. The
final algorithm implementation did successfully
locate spread-out, well-defined Pareto fronts, what
provided enough confidence on the results obtained.
The Pareto front was usually quickly found by the
GS [a similar behavior to that documented in the
literature about NSGAs], and most of the
computational effort was spent in small refinements
of the front. This suggests computational time could
be reduced if the algorithm was ran for less
generations, without significant loss in the quality of
solutions.
The materials studies were valuable in suggesting
that using highly insulated walls and leaving thermal
mass for floors and roofs slabs may be a cost-
effective and energy-efficient means of construction.
The experiments using GGE as the objective
function proved cellulose insulation to be an
excellent alternative to conventional insulation, both
in terms of environmental impact and to achieve
good thermal resistance levels. Wall thickness tends
to become much higher when cellulose fill is used,
but that can be turned to architectural benefit too, for
example by having recessed windows in deeper
facades, which would provide some free shading,
and give the architect the opportunity of plastically
exploiting these type of solutions.
The expansion of the Pareto search method to
include more than two objective functions could lead
to other interesting results, and remains as future
work to be developed.
Finally, the shape generation experiments showed
how, departing from identical initial schematic
layouts, the Generative System was able to create a
variety of architectural shapes that respond to the
climate where they were located, both in terms of
daylighting use and control of heat losses and solar
gains.
The use of penalty functions for area requirement
violations proved to be a somewhat unstable method,
193
as the solutions generated adopted different
strategies to achieve low annual energy consumption
levels. Despite the high penalty factors used, some
solutions still used reduced areas to maintain small
fitness function values. Energy Intensity Use
provides a more reliable measure to assess the
effectiveness of different building configurations,
combined with specific façade solutions. Future
work will also address the issue of incorporating
dynamic constraints into the system, so that an extra
degree of flexibility is added to design elements like
windows, roof monitors and other light sources.
This GS is not to be regarded as an optimization
tool, but instead as a generative mechanism whose
goals are not only to reduce energy consumption in
buildings, but also to suggest alternative building
configurations and work as an augmented design aid.
The particular shapes generated in these experiments
are a result ofthe initial layout, rales and constraints
applied. Different initial conditions would lead to the
emergence of other design solutions, suggesting this
Generative System can be a powerful tool for
architects to quickly study alternative low-energy
designs and understand which architectural features
are more decisive towards achieving desired
performance targets.
REFERENCES
Baker, N. & Steemers, K. 2000. Energy and Environment in
Architecture: A Technical Design Guide, E & FN Spon,
London
Caldas, L. & Norford, L., 2000, Energy Design Optimization
Using a Genetic Algorithm, Automation in Construction,
Special Issue 2000, Elsevier
Caldas, L. & Rocha, J., 2001, A Generative Design System
Applied to Siza's School of Architecture at Oporto,
Proceedings of CAADRIA '01, Sydney, April 19-21, pp.
253-264
Deb, K. & Goldberg D., 1989, An Investigation of Niche and
Species Formation in Genetic Function Optimization.
ICGA
Dickinson, S. & Bradshaw, A. 1995. Genetic Algorithm
Optimization and Scheduling for Building Heating Systems,
In Genetic Algorithms in Engineering Systems: Innovations
and Applications, 12-14 September 1995, pp. 106-111,
University of Sheffield: Conference Publication No. 414,
Institution of Electrical Engineers
Goldberg, D. & Richardson, J., 1987, "Genetic Algorithms with
Sharing for Multimodal Function Optimization."
Goldberg, D. 1989. Genetic Algorithms in Search,
Optimization and Machine Learning, Addison-Wesley
Publishing Company
Huang, W. & Lam, H. 1997. Using genetic algorithms to
optimize controller parameters for HVAC systems, Energy
and Buildings, 26, 277-282
Monks, M., Oh, B. & Dorsey, J., 1998, Audioptirnization: Goal
based acoustic design, MIT Technical Report MTT-LCS-
TM-588
Radford, Α., 1978, Some Room / Environment Optimization
Models Using Dynamic Programming, Computer Report
CR30, Department of Architectural Science, University of
Sydney
Shea, K., Cagan, J. & Fenves, S., 1997, A shape annealing
approach to optimal truss design with dynamic grouping of
members, ASME Journal of Mechanical Design, V.199,
N.3, pp.338-394
Shea, K. & Cagan J., 1998, Generating Structural Essays from
Languages of Discrete Structures. In Artificial Intelligence
in Design '98, Gero, J. S. and Sudweeks, F. (Eds). Kluwer
Academic Publishers. London, pp. 365-404
Srinivas, N. & Kalyanmoy D., 1995, Multiobjective
Optimization Using Nondominated Sorting in Genetic
Algorithms, Evolutionary Computation 2(3):22 1-248. MIT
Press
Sullivan, R., Lee, E. & Selkowitz, S, 1992. A Method for
Optimizing Solar Control and Daylighting Performance in
Commercial Office Buildings, ASHRAE/DOE/BTECC
Conference on the Thermal Performance of the Exterior
Envelopes of Buildings V, Dec 7-10, Clearwater Beach, FL
Tamalci, H., Kita, H. & Kobayashi, S., 1996, Multi-Objective
Optimization by Genetic Algorithms: A Review, IEEE
Wetter, M., 2000, Design Optimization with GenOpt, Building
Energy Simulation User News, Vol. 21, September/October
2000
Wright, J., 1996, HVAC optimization studies: Sizing by genetic
algorithm, Building Services Engineering Research and
Technology, Vol.17, No.l, pp.7-14

194
The existing buildings in the future

Jesper Engelmark
Technical University of Denmark, Denmark

ABSTRACT: With a still growing material welfare, the Danes call for still more building area per capita; but
at the same time, the future Danish labour force will have comparatively less hands available in the building
sector. This situation calls for a radical different way of acting in new building, as well as in maintenance and
rebuilding, as the size, the age and the quality of the existing Danish building stock makes it a major part of
tomorrows man made environment ofthe country.

50.000.000 and 75.000.000 m2, giving a "need of

buildings" of 20 to 30 m2 per capita.
1 INTRODUCTION A growth in just one century with a factor 4-5!

1st of January2001 the total building stock of
Denmark had passed 650.000.000 m2 gross area. Of
Which 350.000.000 m2 is housing and the rest -
300.000.000 m2 - is buildings for production,
administration, institutions e.g., in other words: all
other buildings needed to make the community
function to present standard ofthe Danes living.
At the same time the Danish population was just
beyond 5.3 mill.
This means that for every Dane there is above 120
m2 of gross area of buildings, of which about 65 m2
is housing.
The number of dwellings was a little more than 2,7
mill, giving an average of 130 m2 per dwelling and
an average of 2 persons per dwelling/household.
In total the building stock of Denmark grew from
about 200.000.000 m2 in 1950 to the present days
650.000.000 m2.
In 1950 the population was 4.3 mill. So over just
half a century we have experienced a boom in "need
of buildings" from approximately 50 m2 to 120 m2
per capita.
Tn 1900 the population was 2.5 mill, and the building
stock at that time is estimated to be between
Parallel to growth in building comes the growth of
construction works - from the numerous invisible
underground constructions like water-, sewer-, gas-,
district heating-, electricity- and all kinds of
communication-systems to all visible parts of to
days modern town and landscape.
All this started up in a very modest way only a 150
years ago and has during the 20th century been
growing and accelerating much more than building it
self - a growth in a century with a factor manifold
more than that of 4-5 !
This growth has of cause had a similar effect on land
use in Denmark.
At the beginning of the 20th century towns and
infrastructure, corresponding to about 500 m2 per
capita occupied 3% ofthe country.
At the end of the century the same figures were
1.200 m2, respectively 15%.
The professional life of the Danes has also changed
remarkably over the century.
In 1900 Denmark was still a farming country, but
with a growing industrial sector and with a total
labour force of close to 80% of the population
occupied with agriculture and industry, including
building.

195
Trade and transportation counting for 15% and
services (of all kind, public as well as private) being
the rest: 5 %.
To day the occupation in the service sector is nearly
50% of the labour force, app. 20% is in trade and
transportation, and only about 30% is occupied in
agriculture and industry, including building.
Of a special interest in this connection: the building
trade includes now a days only about 120.000
workers, of which app. 25.000 - 30.000 are
occupied with construction works, and the rest with
building in general - maintenance/renovation and
new-building, and the working force here about
equally divided between the two.
New-building is at present, and has in the recent
decade been, about 10.000.000 m2 a year, meaning
that a workers annual output in this field is in round
figures just beyond 200 m2.
But in maintenance/renovation the same worker has
been able to handle about 14.000 m2 - according to
statistics at least. Even if this field is characterised
by a lot of "do-it-your-self-jobs", and also that a lot
of "black" economy is involved, still the skilled
professional worker is officially only this little
involved. It might be telling us, that we in general
have fallen behind in maintaining and renovation,
even if the majority of buildings are not very old.
The building stock is relatively new: 90% of it is
built within the last 100 years, and it goes for both
the residential and the non-residential part.
The last 10% include all buildings built before year
1900, of which the oldest have an age of about 800
years.
Approximately 35% of the dwellings are built in
1960's and -70's. The 1980's and -90's only count
for 15%, so roughly 40% are built between 1900 and
1960.
As the gross area ofthe dwellings in the second part
ofthe century is bigger in general than earlier - and
during the period growing - the major part of the
housing stock is built over the last 40 years.
For the non-residential part of the building stock a
similar situation applies.
The standard of housing is fairly good, as 95% of all
dwellings are furnished with toilet, bath and central
heating and only less than 2% are without any of
these (modern) facilities. It should be noticed that
these 2% not necessarily is located in the
older/oldest part of the buildings, because of town-
renewal and renovation in general in the recent
years.
196
Ofthe housing stock is about 1 mill. Dwellings, with
a gross area of up around 100.000.000m2, situated in
multi-storey buildings.
The rest, and far the biggest part, is single-family
houses - detached houses, terraced houses and
similar. Mostly all of these are private owned,
whereas the situation is the opposite in multi-family
housing (which in general is identical to multi-storey
housing), here only about 20% of the apartments is
owned by the occupant, and the majority of the
remaining part is social housing.
(In the figures above summerhouses are included,
partly because around 15% of them are inhabited
permanently and partly because almost all of them
have a standard very close to or like dwellings.
They are therefore very often used all year around,
functioning like a second home. They don't count
for a major part, as they are a little more than
200.000 by number and 20.000.000 m2 by gross
area)
2 BUILDING TECHNIQUES
The technique applied in building up till around
1950 is in general described as "traditional",
meaning use of the two major materials: wood and
brickwork, performed by almost only two kind of
skilled workers: carpenter/joiner and bricklayer.
Predominant produced at the site and without any or
very little help from mechanical equipment in now a
day's sense.
Roughly speaking the situation was like this through
the first half of the 20th century, but in the decades
ahead more skills had joined the team: the plumber
and the electrician started to be members from
around 1900.
Mechanisation in production of building materials
started in the later part of 1800's.
For woodworks it meant access to e.g. factory-made
tongue-and-grooved boards, mouldings and later on
also whole building-parts as doorplates for example.
For the bricklayer and his assistant it didn't mean
much, that the fabrication of bricks in most tile
works were all mechanized in the first decades ofthe
1900's - their jobs were still the same
Only the new people at the site - the plumber and
the electrician - started out as "fitters" in a more
modern sense.
Bearing facades and main division wall, all massive
brick built characterise the structural system of the
majority of these buildings. The floors are in general
made of wood, spanning from facade to facade.
Stability ensured by facades, gables and staircase
walls anchored to the floors.
Partitions, mainly made of plastered boards, are
secondary used for the final arrangement of the
apartments.
It doesn't change much in the overall description of
the traditional building technique that some new
materials and/or constructions appeared as time went
by.
It started with the use of concrete in foundations,
then as flooring cast on iron beams in (the new)
bathrooms and over basements, and was followed by
a more widespread use of reinforced concrete as
flooring cast on site.
Concrete was also used in construction of balconies.
Followed by stairs made of (prefab) concrete
instead of wood, and the appearance of other types Figure 1. A closed building system - "CON-BOX", a box-
frame system with all components specially developed by one
of lightweight partitions, than those made of the
producer and with very limited possibilities in further design.
plastered boards. The system was only used once for an estate of 200 apartments
in Aalborg, Jutland. The non-success despite of good economy
This was the general situation in housing as well as was of cause the poor appearance, looking like containers
in building of offices, institutions and similar kind of stacked, (ref. Nissen, 1972)
buildings.

Up till the mid 1900' s reinforced concrete were in
fact seldom seen in building except for factories,
warehouses e.g., in which case the design of such
buildings called for special properties.
The beginning adoption of (some of) the ideas from
functionalism in the 1930's was in fact based on
well-known building technique trying to look like a
turn in construction as well.
The 2 World War and post-war situation, with a
very little building production, had in the 1950's
resulted in shortage of mainly dwellings. And in
combination with a rising growth of population and
- later on - also an increase in material welfare, it
could be foreseen that the present labour force and
way of building was far from being able to produce
the needed buildings in the future.
Therefore the scene was set for a radical change, and
a lot of, but small-scale experiments to raise
productivity in housing construction were carried
out in the 1950's.
In this early non-traditional or "prefab building"
numerous experiments in the use of structural
systems, materials and building techniques was
applied.
The efforts were mainly concentrated on bringing
down labour force at site-works. This was partly
done by mechanisation and partly by moving
production to factories. Also in substituting skilled
labour force by "non-skilled" and by bringing in
"new" materials and processes. And by also using
the winter for site-works.
Many kinds of structural systems are represented:
bearing and bracing external as well as internal walls
and column-beam systems. Slabs spanning
transversely, longituditionally or mixed.
The materials being used were preferably concrete,
but also brickwork as bearing and bracing transverse
internal walls occurred.
Factory-made elements were quickly introduced, but
only after trying out different ways of using
standardised and/or reusable formwork at site in a
more or less traditional way or in (rather primitive)
fabrication on site.
It turned out almost from the start, that the market in
a little country like Denmark generally was too
small for using "closed building systems", being a
total product coming from just one producer (figures
land 2).
The further development was based on the
appearance of a new and common building code.
The code substituted the many local former ones
(being as many as the number of communes), and
expressed the demands to buildings in relation to
functions instead of materials (as had been the case
for centuries), to give possibilities to the use of new

197
Figure 2. The TVP-System" was introduced as a closed system
and as such only used in 2 estates with a total of around 500
apartments located in Næstved respectively Hundested, both
Sealand. The resemblance to The Jespersen System partly
substituting walls with frames or beams/columns is close and it
seems as a minor problem to have combined the two giving
flexibility to the standardisation. But at the time flexibility was
not asked for considering the extra costs involved, (ref. Nissen,
1972)
materials and techniques on a nation-wide scale.
Parallel to this a modular system was introduced to
back up the design and production of building parts
independent of the single project, and on the other
hand enabling the single project to take advantage of
variety within a mass production.
In the end this would not have had the effect it had,
if not the government to a start had been the
guarantee for the production of 7.000 dwelling-units.
This made it economical profitable to set up the
needed factories, based on a common agreement on
modular sizes and joints of elements - using an
"open building system".
These dwellings were situated in two bigger areas in
Copenhagen (the Ballerup- and the Høje Gladsaxe-
schemes) and in several scattered, smaller areas in
Jutland (the Southern Jutland Scheme), all of them
being multi-storey housing and built on the principle
of the open building system: "The Jespersen
System" (figure 3).
To prove the validity ofthe system to be able to
198
Figure 3. An open building system - "The Jespersen System",
the all over used in prefab multi-storey housing. The
standardisation of joints and modular based sizes of
components was accepted as common in planning and
fabrication, leaving the final design to the single project. The
system is here shown with slabs spanning between bearing
transverse walls and with non-bearing facades of timber
construction, but the facade elements might as well be a
concrete sandwich construction. The Jespersen System is also
applied in buildings with slabs spanning transversely between
bearing facades and main division wall. (ref. Nissen, 1972)
ensure different designs, the Ballerup scheme was
designed by 4 different architects, the Høje Gladsaxe
scheme by just one architect. So was also the case in
the over-all design of the Southern Jutland Scheme,
but here the variety was supposed to be caused by
the different locations.
(Anyhow the difference in design is not very
significant in the layout of the dwellings in the
different schemes, or in a bigger variation in sizes of
the dwellings. And the facades are rather uniform,
despite ofthe efforts done to avoid it).
All of this required mechanisation in the site works -
above all by having the crane as the new "member
of the team" to move around with the much heavier
elements.
In construction of housing the use of concrete only
for bearing and bracing elements very soon became
standard, and above all in the shape of buildings
with slabs spanning in between transversal bearing
and bracing walls. Bracing longitudinally was
preferably placed in connection with staircases, thus
leaving the facades to be constructed only according
to demands ofthe envelope.
The main reasons for this was of economical nature:
combining bearing and bracing with demands
primarily to fire-safety and sound insulation was the
optimal solution, but with no regard to future
flexibility and an eventual different way in living.

In building where greater demands to flexibility was

or could be foreseen, the use of column-beam
systems with bracing-walls around cores for lifts,
staircases, toilettes etc. was preferred
This was - and is - mostly the case in building of
offices, institutions etc. And also here concrete was
the material over-all used for the bearing/bracing
system.

The new way of building introduced in the 1950's

had very little impact on the layout of dwellings in
multi-storey buildings. The traditional access from a
stairway to two - in some cases 3 apartments - per
storey was dominating, not even the alternative and
new found use of balcony-access made a difference,
thus giving the same layout in all stories.
In fact: flats in Danish multi-storey housing have
had the same design for centuries, except for the
later coming modern facilities like toilets and
bathrooms and the use of one concrete staircase
instead of two made of wood (fig. 4).
Figure 4. The layout of Danish apartments in multi-storey
housing has been practically the same for centuries. The two
plans above represent a span of about a century, but could be
more. The upper one, being a massive brick-built building from
the late 1900 s, could be found anywhere in the centres of
Danish towns. The other could be any ofthe many multi-storey
concrete panel buildings built according to the principles ofthe
preferred open building system, and situated in the outskirts of
all Danish towns (ref. Skaarup & Jespersen, 1987; Nissen,
1972).

The building of multi-storey housing as introduced

in the 1960' s is, with minor corrections, in principles
Factory made elements are in general restricted to
still done the same way to day but a little more
delivery of completing parts like windows, doors,
efficient compared to earlier.
cupboards etc. Mechanisation of site works was -
In general the single project tends to be of a smaller
and is - limited compared to building of multi-storey
scale, and there is a trend towards "hiding the prefab
housing.
look". But experiments with different building
techniques the way that was seen half a century a go
Despite this, the boom in building in the 1960's and
do not occur.
-70' s was much bigger in construction of one-family
houses, than multi-storey housing: Almost 540.000
The "traditional" way of building has to some
units compared to about 240.000. And with a bigger
degree survived in construction of one-family
gross area per unit in one-family houses the ratio is
houses.
more like 3 to 1, comparing the quantity by area.
After experiments in the 1950's similar to those in
multi-storey housing, the building of one-family
In the last two decades ofthe century construction of
houses turned into a kind of standardisation in the
housing has decreased total to less than half the size
exterior design and the lay-out of plans, rather than
ofthe former, but with the same ratio as above.
in new ways of construction.

199
3 THE FUTURE
With a size ofthe existing stock of housing being 65
m2 - and buildings in general corresponding to 120
m2 - per Dane;
With a building stock, where the majority is built
within the last 40 years and where only 10% is older
than 100 years;
With a housing stock where less than 2% of the
dwellings are without modern facilities in sanitation;
- One would not believe Danes to have problems in
these fields - but we have:
We have a shortage of dwellings here and now (!).
We have a (future) problem in maintaining and
renovation of the existing building stock, not just in
housing, but also as a whole.
We have a (future) problem in general in the size
and layout ofthe existing dwellings.
The shortage of dwellings is a result of an increase
in 1 -persons households, partly caused by the
breaking up of the former family patterns and partly
because young people want to have their own
apartment much earlier than before. The relatively
small production of preferably apartments in the last
decades has added to the problem.
One could say, that this is a minor problem to be
solved just by building more - and so we do (but
hereby just increasing the problems listed below, as
both the design and the techniques practised are
those of to day).
But the next (future) problems are not that easily
solved, as they incorporate both the existing building
stock and our way of living.
Projecting the trends from the 20th century to the
present, the Danes of tomorrow will increasingly be
working as "white collars" or "with the brain".
"Blue collars" - people working with their hands -
will be a still smaller part.
This calls for even more mechanisation and
automation in all trades.
Tn building it will have the effect that more and more
ofthe site works will move to factories.
In new building we just have to live with it, and get
a different approach to the meaning of architecture
as being more like an industrial product than the
individual formed "objet d'art".
The traditional repair situation will gradually change
to replacing - and with still bigger parts - to reduce
the need of manpower at site.
The skilled builder, as we know him as a bricklayer,
a carpenter, a plumber etc., will be seldom seen -
200
and here we have a grave problem with the
traditional built part ofthe existing building stock.
There will be a (big) gap between the technique
applied not only to the traditional built houses, but
also between most of the existing buildings and the
future coming.
Even if this situation will not come "tomorrow" - or
"the day after" - we have to be aware of this in due
time.
The third problem (as a future one) is closer to
present days.
Tn the future one might experience a (much) broader
range in fulfilling people's needs/wishes to their
homes, e.g. the following:
Size: from very little to very much. The minimum
size related to persons, merely seeing their home (or:
homes, as having more than one home all ready has
been introduced) as a hotel, taking advantage of the
surrounding city at its most. The maximum size
leaves no limits....
Space: from the standard height of rooms, just fitted
to stand upright, to rooms with double- or triple
heights (the ideas from the start of the former
century brought forward through institutions like the
Bauhaus and single persons like le Corbusier, has
not yet been carried out more general).
Sanitation: from the absolute minimum of to days
standard (toilet and a shower) to the standard of a
"roman emperor".
Cooking: from just being able to prepare coffee/tea
and a micro-wave-oven dinner to the big, shining
laboratory for experiments in preparing dishes from
all over the world (the basic needs are the same:
cold/hot water, an outlet and electricity/gas).
Flexibility: possibility to expand or reduce dwellings
fit for the actual situation.
Variability: possibility to easy modifying of the
interior regarding e.g. sanitation, cooking facilities
within the apartments - and maybe even on a day-to¬
day basis.
The problems listed above are all directly connected
to the buildings as such.
At another level we have the general problems
related to the modern (yesterdays) urban planning
with the (strong) separation of functions: areas for
living, areas for working, areas for amusement etc.
We also have (beginning) segregation in different
living areas, not only by social standard or income
level, as we have experienced before, but also by
nativity.
The future - not even the present - Danish society
does not call for such a separation, nor segregation.
A change in this situation is far from just done by have lesser problems and may be the best potential
new building or under a new "code" for future in adaptability to the future use all ready now,
building, considering not only the size and the despite their original as well as actual reputation in
quality of the existing building stock, but also our general.
wishes/demands to the standard of life in the future
society.
Our awareness of the impact of human activities on
the environment has just recently been recognised,
but not yet really understood at its full extent.
yt^zv
It seems as if we have to have a look at and act with
the existing building stock - as well as the rest of our
man-made physical surroundings - as being more or
less like the nature given resources.

We will quite simple be asked for another attitude to

the built environment.
This means in broader terms: seeing the existing
' .'· -·' -Α."*β»-.ιΙ.*:'--·
buildings/structures as "three-dimensional building
sites" open to future operations according to their
all-over properties and the coming future building
Figure 5. A medium size area with about 800 apartments in
techniques. blocks of 4 and 7 storeys. "Bispehaven", location: Aarhus,
Jutland. Built 1969/73, renovated in the late 1980's. (ref.
Christiansen et al, 1993)
4 TOMORROW
á *-.E^ rií" ;7~α*-*>
Having an over-all look at the existing building '..¿V? -τ·*
stock, the traditional built part, regardless of
category of building, at first represents a major
problem in a future with fewer hands available in
maintenance and renovation.
Next - and with respect to category of buildings - '^sS^L ^
the newer part of the one-family houses will be a
problem, partly because they in general are built in a
rather traditional way, and partly by their locked and
uniform design tailored to (yesterdays) standard
family pattern.

Together the traditional built stock and the newer Figure 6. A smaller area with about 300 apartments in 4 storey
one-family houses count for far the biggest part of blocks is typical for the majority of the estates.
the existing buildings - but not yet recognised as a "Vangdalparken", location: Randers, Jutland. Built 1969/70,
coming problem in terms as described ahead. renovated in the mid 1980's. (ref. Christiansen et al., 1993)
One could say that this is somehow understandable
with regard to the age and quality of Danish housing
Right from the start these buildings and areas were
in general. Adding to this is also the recent decades
not very popular and they were named "concrete
massive efforts laid in bringing down energy
slums".
consumption in the existing buildings by upgrading
This negative image was not connected to the
the envelopes with extra insulation, replacing
apartments as such, as they in general were bigger
windows etc.
and better equipped than the older. But the location
Finally there exist in general a widespread
and the size of the new estates, the rapid
acceptance of deficiencies in the traditional built
proliferation of the building complexes, and the
part, quite simple because of its age and visual
often-homogeneous composition of tenants, socially
appearance interior as well as exterior.
and economically, called for criticism.
Finally the buildings as a rale had - and have - a
Contrary to this the majority of the multi-storey
sheer visual monotony.
housing from the second half of the 1900' s seems to

201

a^lií-l- UJJLtói
τκΓ^-
W^.i.'.îiU;
Figure 7. "Hørgården": 140 apartments in 3 storey bloclcs.
Built 1972, renovated 1993: new extra insulated facades, new
windows and new (flat) roofs - no internal changes, just a
different look. Location: Åbenrå, Jutland (ref. Bech-Danielsen
& Vanning, 1997).
The size of the new estates was big compared to
earlier, but in general far from so big as seen
elsewhere in Europe.
Tn fact we only have 1 (one) very big: The Gjellerup
Scheme (Aarhus, Jutland) of about 6.000 dwellings.
Then a handful being as big as 2.000 to 3.000
dwellings, while the rest - and the majority - rooms
from 200 to 1.000 dwellings (figure 5 and 6).
The layout of these areas does not differ from the
ideas introduced in town planning some decades
earlier: blocks of in general 4 storeys placed very
uniform and preferably lengthwise north-south.
These areas were also named "sleeping towns", as
they contained nothing else than housing and a
shopping centre - and this situation is still the same.
Also problems with the structural quality in general
became apparent rather soon. The accelerated tempo
and large scale production compared to earlier was a
part of the problem, and the introduction of new
materials and structural techniques taken into wide
202
1' ·
*tì\
<&
Figure 8. "Vangdalparken": 320 apartments in 4 storey blocks.
Built 1970, renovated 1987: new extra insulated facades, glass
clad balconies, new windows and new hipped roofs no
internal changes except for splitting up a few of the. bigger
apartments, just a different look. Location: Randers, Jutland
(ref. Christiansen et al., 1993).
spread use without being adequately tested was
another.
However, most of these buildings are constructed
with a bearing system more or less independent of
the finishing elements like the envelope, and it is
with the latter that technical problems lie.
For example, there are problems with both
lightweight and heavy non-bearing external walls, as
well as with flat roofs. The durability of the
envelopes in general has caused difficulties, and
extensive maintenance work has been required far
sooner than expected.
The combination of maintenance problems and
wishes of bringing down energy consumption has
resulted in a fairly comprehensive effort to renovate
the envelopes of such buildings. This has mainly
been done by adding an extra layer of insulation on
the facades and new windows, glass covering of
balconies and putting on hipped roofs. At the same
time giving a new look to the buildings and their
 I «

* »'»»""Ο?*"*

«f.'

Figure 9. "Tåstrupgård": 950 apartments in 3-4 storey blocks.
Built 1972, renovated 1983 and 1991: Facades painted, new
windows and new, hipped roofs. The areas between the blocks
being partly reorganized - no internal changes to mention, as
the variety of different sizes of apartments were fairly good,
just a different look. Location: Tåstrup, Sealand (ref.
Christiansen et al., 1993).
Figure 10. "Ladegårdsparken": 900 apartments in 3-4 storey
blocks. Built 1972, renovated 1997: Facades repaired, gables
extra insulated, glass clad balconies, new windows and in
combination with new hipped roofs, addition of an extra storey
no other internal changes, just a different look. Location:
Holbæk, Sealand (ref. Bech-Danielsen & Vanning, 1993).

surroundings, and in this way trying to escape the

the traditional stairway serving only 2 or 3 dwellings
negative image (figure 7, 8 and 9, illustrating the
per storey.
general situation as built, and after renovation).
Tn some cases financing of such renovations has
However, precisely a different and more varied
been aided by the simultaneous addition of an extra
composition of apartments, as well as other and
storey (figure 10).
(much) more varied types of apartments, including
Anyhow, the new "make-up" of these areas has not
juxtaposition and sequence of rooms, would perhaps
yet produced a situation very far from the original;
be of importance for the future greater success of
problems with the bad image still exist.
these buildings.
It would be additionally supportive of this future
The types and sizes of apartments in the "concrete
success, to the same extent that interior design and
slums" have not seriously been questioned since
standards of equipment might be adapted
they were built. Apparently there has been overall
individually in the future.
satisfaction with the relatively large dwellings and
their level of equipment.
In this regard it should be remembered that the
This has probably been largely supported by a belief
different parts of a building have greatly varying life
that these buildings are fixed in their original form
expectancies, and that the interior placed, bearing
once and for all, mostly due to their structural
concrete structures in buildings like these have a life
system with bearing transverse walls, but also
expectancy to be calculated in centuries.
because ofthe most common access conditions via

203
Buildings with bearing transverse walls, stripped to
consist of the bare structural elements alone, can be
compared to open bookshelves - ref. figure 3.
Combining or dividing of apartments on a horizontal
level can be done with no great difficulty by creating
or closing off door openings in the bearing
transverse walls - ultimately by turning the
structural system into a beam-column system -
possibly supplemented by erecting a number of
longitudinal light partitions, as dictated by the
specific requirement.
With access via the traditional stairway, however,
the possibility for altering the division of the areas
between stairways is limited. In other words,
creating a smaller apartment is automatically linked
to the creation of a bigger one.
But the ground floor can always be reached more or
less directly from the surrounding terrain, and each
of the lowest sections of the "bookshelf' can
become an independent dwelling.
The second floor can be reached also directly from
the terrain via outdoor stairways, as it often in seen
in the lower housing complexes of recent years.
Even here, individual dwellings can be created,
independent ofthe existing stairway system.
Finally there is also the more traditional possibility
of establishing new access facilities by adding them
to the existing building, e.g. by balconies or alike.
It should also be possible to combine apartments
vertically by making holes in or removing entire
decks - thus the number and design of various
dwellings will be considerable, both with regard to
size and layout.
If additional installations in the individual stories are
made entirely above deck in a raised floor and
(preferably) with vertical assembly outside the
building, then layout of the individual stories and
dwellings will be completely free - and this freedom
will be so complete as to permit entirely individual
solutions, not only of a more permanent but also as a
day-to-day character.
The advantages that hereby arise with regard to
acoustics, fire protection and future maintenance and
rebuilding are obvious.
As sketched above, and with a more or less high
degree of randomness in the horizontal and vertical
permutations, and combinations in erasing and
adding of structures, and all according to how this is
done and the desired influence, there will be a move
away from the monotony for which the buildings
204
have been criticized - as well in their interior as in
their exterior appearances.
As has been shown earlier, the renovations of the
Danish panel buildings of the 1960's and -70's has
not included a more radical change. Problems have
been handled only by "make-up" ofthe buildings, as
they were designed originally - not as elsewhere in
Europe, where other ideas have been practiced if
not very often carried out in reality, shown in
projects (fig. 11).
But even with the foresight of a better chance for the
newer part of the multi-storey housing complexes to
survive in the future as living up to standards of to¬
morrow, the "concrete slums" also need to be
inhabited during the day, and not only being to days
"sleeping cities".
With buildings being multi-usable - as the concrete
panel buildings tend to be, if treated right - the
future situation in these suburbs should not turn out
to be different from what has happened over the
years in the central areas ofthe cities.
V-Jűi'
, "t ^;.
'<f
'»»,
<·>
rC(<»
Figure 11. The problems of "concrete slums" are not just a
Danish one. The architect Lucien Kroll shows in the sketch
above a solution to avoid the monotony: Redesign of such an
area, built 1960 in Alencon, France. The renovation project is
from the late 1970's: Additions are showed plenty to avoid the
eternal repetition in the exterior, but interior alterations are not
accounted for (ref. Bech-Danielsen & Vanning, 1997).
Tn the central areas we have experienced for a long
time a mixture of living and working, not just within
the areas, but also in the buildings.
We have also seen a reverse use of buildings over
time, and not just nowadays.
Over the years residential buildings have been
(partly) turned into offices, shops, institutions etc.,
and (partly) back again. Lately also industrial
buildings of nearly all kinds have been rebuild to
other purposes (very) far from the original use.
This should of cause be possible elsewhere,
especially with multi-usable structures, as the
concrete panel buildings in reality are.
And in a future, where far the majority of jobs will
be in the service sector, and probably taking
advantage of ΓΤ to an extend far beyond our present
imagination, the former disintegration of cities will
not have to exist.
5 CONCLUDING REMARKS
Maintaining and rebuilding of the existing building
stock accomplished with the present technique
seems to be impossible in a future with still growing
material welfare and less hands to do the manual
jobs.
Future new building will call for quite a different
way in construction and maintenance for the same
reasons.
At the same time the size, the age and the quality of
The situation however is recognisable from country
to country if one puts in a "space/time" standard:
seeing the problems of today in the richer states as
the problems of tomorrow in the poorer, taking in
account, that the development will take a similar
direction - learning from what could have been done
different in these richer states over the period.
REFERENCES
Bech-Danielsen , C. & Vanning, M. (eds.) 1997. SBI-
byplanlægning 75: Smukkere renoveringer. Arkitektoniske
kvaliteter ved renovering af nyere boligområder. Hørsholm:
Statens Byggeforskningsinstitut (SBI).
Christiansen, U., Jensen, M.K., Kristensen, H., Lindhardtsen,
H., Vanning, M. & Vestergaard, H. 1993. SBI-
byplanlægning 65: Bedre bebyggelser bedre liv?
Hørsholm: Statens Byggeforskningsinstitut (SBI).
Danmarks Statistik. Statistisk Årbog. 2001. København:
Danmarks Statistik.
Engelmark, J. 2001. Housing renovation is more than new
kitchens and bathrooms. Paper presented to WG 3 in
Prague 20-21 April.
Jørgensen, U. & Pedersen, L.S. 1983. IFH-rapport 151:
Kompetanceopbygningen i industrialiseringsprocessen.
Lyngby: Instituttet for Husbygning (TFH), Danmarks
Tekniske Højskole (DTU).
Koch, S. 1981. IFH-rapport 150: Arkitektens rolle i
byggeteknologiens udvikling efter den anden Verdenskrig.
Lyngby: Instituttet for Husbygning (TFH), Danmarks
Tekniske Højskole (DTU).
Nissen, H. 1972. Industrialized Building and Modular Design.
London: Cement and Concrete Association. .

the existing buildings call for further use many years Skaarup & Jespersen, 1987, Ældre etageejendomme -
ahead. byggeskik og renovering, København: Boligstyrelsen.
So, to avoid a gap coming up between new building
and rebuilding, we gradually have to bring nearer to
each other the techniques applied in the two cases.

Mixed building technique as being just an

improvement of structural quality in building is
obvious in connection with rebuilding of existing
structures under new circumstances - it so to speak:
lies in the words.
But in a wider sense mixed building technique also
includes methods of building - and again: in the case
of rebuilding, this gives a good meaning for reasons
listed ahead.

Now, one could argue that problems described are

just Danish, and that the situation is (very) different
from country to country depending on the standard
of material living. Tn many countries the overall
problem is just space to full fill the populations
demand for dwellings here and now. And to these
people, problems like the Danes' seem as luxurious.

205
Large-panel dwelling building, current state - ways of construction and
architectural modifications
Adam Rybka
Rzeszow University of Technology, Poland
ABSTRACT: In this paper was inserted statistic date of dwellings stocks in Poland, prefabricated dwellings
systems using in Poland between 1950-1995, present - day requirements of the existing large-size panel
building, ways of construction and architectural modifications of this buildings.
1 INTRODUCTION
Building in the people clusters is aimed to create the
environment for man life and his needs. Man life in
the huge clusters and economical issues have
influences into rising huge urban agglomerations. Tn
Poland, growing expanse of urban developed areas
and reduction ofthe costs of housing caused from 50's
to 90's creating of characteristics for this period
perpendicular's buildings shapes - "large-panel
building". The new settlements of multifamily
buildings with characteristics "box's shapes houses"
have been rising.
2 STATISTIC DATA OF DWELLING STOCKS
IN POLAND
Poland is located in the Central Europe.
Poland covers an area of 312 690 km 2. The longest
distance from East to West is about 800 km and from
North to South about 750 km. Poland has a
population of 38 654000 million averaging 124
persons per 1 km2. 23 908 000 people live in cities
and urban areas.
Poland has not a high standard of housing. On an
average, the population has at its disposal 1,04 rooms
per person. State-owned firms for non-profit housing
associations built most of the housing in Poland in
period- 1945 - 1995. These associations constitute an
instrument for realizing public housing and building
policy. The associations have played an important
role in the long-term development of industrialized
housing in Poland. At present time most of the
housing in Poland is build by private firms for
private investors.
Poland has a temperate climate with cold winters and
warm summers. The winter temperature between 15
207
- 20 °C degrees below zero the summer temperature
normally between 20 and 30°C, still with rather great
variations.
The west and east are the prevailing directions ofthe
wind. This, combined with rain and a lot of snow and
the fact that the temperature passes freezing point an
average 50 times each winter and that the ground is
freezing to 60 - 80 cm, means high requirements to
structures, design of exterior walls, joints, etc.
Building activity is not maintained all year round.
December, January and February are hard to reach
for building activity, because ofthe weather.
Total area of Poland in 1999 as of 31 December
was 312 690 km2
Area of Poland in 1999 as of 31 December
per inhabitant was 0,008 1 km 2
Population of Poland in 1999 as of 31 December
based on balances was 38 654 000
Of which
Urban areas 23 908 000
in% 61,8
Rural areas 14 746 000
in% 38,2
Population of Poland in 1999 as of 31 December
based on balances per 1 km 2 was 124.
Poland's housing stock in 1999 as of 31 December
based on balances was 1 1 763 300
Usable floor dwellings space per person 19,0 m2
Usable floor space per dwellings 61,3 m2
Table 1. Dwellings completed in Poland.

one-family multi-family
3 PREFABRICATED DWELLINGS SYSTEMS
year all units % %
houses houses USED IN POLAND BETWEEN 1950-1995
1960 142072 27519 19,4 114553 80,6
1965 170446 34321 20.1 136125 79,9 3. 1 Large size-elements system.
1970 194173 28769 14,8 165404 85,2
1975 248117 50809 20,5 197308 79,5 Dwellings system with large-size elements. That
1980 217090 39901 18,4 177189 81,6
system was used in Poland started from 1950 year.
1985 189625 37648 19,9 151977 80,1
1990 134215 15434 11,5 118781 88,5
In this system building structure was arrangement as
1991 136790 10718 7,8 126072 92,2 crosswise, lengthwise and mixed configuration.
1992 132969 8190 6,2 124779 93,8 Crosswise configuration was the main arrangement
1993 94449 5922 6,3 88527 93,7 used in this system. Modular grid 60 χ 60 cm was put
1994 76080 3842 5,0 72238 95,0
into use in horizontal projection of storey. Span
1995 67072 2531 3,8 64541 96,2
1996 62130 1612 2,6 60518 97,4
ceilings were 600, 480, 360, 240 cm. Ceilings were
1997 73706 1380 1,9 72326 98,1 22 cm fat. Main walls were 15 cm fat. Storey height
1998 80594 1438 1.8 79156 98,2 was 280 cm, in cellar 250 cm.
1999 81979 964 1.2 81015 98,8

Table 2. Dwellings stocks based on census data.

December December December Mai 1995

1970 1978 1988
Usable floor space in 410002 502649 633107 706097
inhabited dwelling in thous.
m2
Dwellings build in large size 430719
elements in thous. m2

in% 61

Dwellings build in other 275378

method in thous. m2
dwelling 8 295 000 9 528 000 10 875 000 11056000
dwelling build in large size 3924880
elements
dwelling build in other 7131120
method
in% 35,5
Figure 1. Large size-elements system. Building structure was
usable floor space in m2 par 50,7 53,9 59,1 63,9
dwelling arrangement as lengthwise configuration, (réf. Lewicki B.,
1962)

12000000

10000000

β 000 000
Imidin
w

β 000 000
*

4000000

buWIn

2000000
f --V-

Decanto·«- 1970 OftcamtMT 1fl?a DocwtAmt 1M8 IM1M5 Figure 2. Large size-elements system. Building structure was
arrangement as crosswise configuration, (réf. Lewicki B., 1962)

Graph 1. Dwellings stocks based on census data. Mai 1995.

208
 a)

εΞαασοαοο

π/ο-
\*

*
°2*
o *
° aTf
*;

+
!Sl
-lil¬
le

* ·

Figure 5. Large panel elements system. Building structure was

arrangement as:
Figure 3. Large size-elements system. Example of facade and a) two-way configuration, b) cross configuration
repeatable storey. New Town Nowa Huta, Poland (ref. Lewicki (ref. Lewicki B., 1962)
B., 1962)
4
o) b)

D3=

1 I 4- I F^
c) 1)

J
oo D

b).
Figure 4. Large size-elements system. Wind membrane,
building structure- crosswise configuration: a), b) in inside
walls, c), d) in outside walls
(ref. Lewicki B., 1962)
U
3. 2 Large panel elements system.

Dwellings system with large-panel elements. That

system was used in Poland started from 1955 year.
Modular grid 60 χ 60 cm was put into use in
horizontal projection of storey. Span ceilings were Figure 6. Large panel elements system. Repeatable storey in
600, 480, 360, 240 cm. Ceilings were 25 cm fat. two-way construction system. One ofthe settlements in
Main walls were 15 cm fat. Storey height was 280 Warsaw, Poland, (ref. Lewicki B., 1962)
cm, in cellar 250 cm.

209
 5 ι I

c;

Β
Uaj
" !

i '
3β0 SL_ J¡» -Ä1 ^ JSÍ Λ»

Figure 7. Large panel elements system. Example ofthe

repeatable storey in cross construction system, (ref. Lewicki B.,
1962)

3.3 Skeleton construction system. Figure 9. Skeleton construction system. Building structure was
arranged as cross configuration, (ref. Lewicki B., 1962)
Open dwellings system with pillar and horsehead
elements. Skeleton construction system was used in
Poland between 1954-1958 year. Tn this system
building structure was arrangement as plate-skeleton IB
configuration. Crosswise configuration was the main
arrangement used in this system. Modular grid 60 χ
60 cm was put into use in horizontal projection of
storey. Span ceilings were 480, 240 cm. Ceilings
were 14 cm fat. Main walls were 15 cm fat. Storey
height was 280 cm, in cellar 250 cm.

a)

s : ' î
a ' . ' a

»ρ e ' è :

SJ JIB 1 4O0 1 «» ' 4tX -ØJ 1 «x- I

1

Figure 10. Skeleton construction system. Example of facade

and repeatable storey. Nowe Tychy, Poland (ref. Lewicki B.,
1962)

3.4 W-70 System.

Open dwellings system with large-size elements. W -

Figure 8. Skeleton construction system. Configuration of
element building structure, (ref. Lewicki B., 1962)
70 system was used in Poland started from 1970
year. In this system building structure was
arrangement as crosswise, lengthwise and mixed
configuration. Crosswise configuration was the main

210
arrangement used in this system- Modular grid 60 χ
60 cm was put into use in horizontal projection of 600 480
storey. Span ceilings were 600, 480, 360, 240 cm.
Ceilings were 22 cm fat. Main walls were 15 cm fat.
Storey height was 280 cm, in cellar 250 cm.

1
SS
1 1 1 1 1 1

600 480
M5 Pu*5e¿29mz M5' Pu=57;79m2

360
¿80

600 240 480

M3 Pu*447im2 MS Pu =56 89m2

Figure 12. W -70 System. Example of flats forming, (ref.

a WçglarzM., 1972)

ΐϋ Y J

Figure 1 1. W -70 System. Methods of houses forming, (ref.

Wçglarz M., 1972)

Figure 13. Example of facades in W-70 system. Poland

(réf. WçglarzM., 1972)

211
3.5 Szczecin System.

Open dwellings system with large-size elements.

Szczecin system was used in Poland started from
1971 year. Tn this system building structure was
arrangement as crosswise, lengthwise and mixed
configuration. Crosswise configuration was the main
arrangement used in this system. Modular grid 60 χ
60 cm was put into use in horizontal projection of
storey. Span ceilings were 480, 240 cm. Ceilings
were 14 cm fat. Main walls were 15 cm fat. Storey
height was 280 cm, in cellar 250 cm.

Figure 16. Szczecin system. Building structure was arranged

as two-way configuration (réf. Wçglarz M., 1972)

£Ζ
1 \ 1 t t \ 1

* \
M ι
\ TI
\ \ \\\
Figure 17. Szczecin System. Building structure was arranged as
crosswise configuration (ref. Wçglarz M., 1972)

a . ...ι

Figure 14. Szczecin System. Configuration of building '^ » ""i

structure elements, (ref. Wfglarz M., 1972)

Figure 18. Example of facades in Szczecin system. Poland

(réf. Wçglarz M., 1972)

Figure 15. Building structure was arrangement as cross

configuration (réf. Wçglarz M., 1972)

212
4 CHARACTERISTICS OF THE LARGE-PANEL
BUILDING
The large-panel building consisted in an initial
prefabricating of the building elements. The
buildings were built with mix or crosswise
constructional system. Basement made from
prefabricating elements founded on continuous
footing poured out on the building site. Ceiling above
the last storey usually was used as ventilated roof
ceiling. The buildings were made with height: to five
storeys without the passenger's lift shafts, to eleven
storeys or higher with passenger's lifts and rubbish
chutes. In the prefabricating elements plants there
have been produced in the industrial way, on ready
prepared technological lines the elements like:
ceilings, inside walls with door's holes, outside
shelter walls with ready made textures, outside walls
with window's holes and balcony's door holes,
balcony and loggia panel's elements, staircase
landings and flights of staircases.
5 EXISTING PROBLEMS OF THE
PREFABRICATED MULTI-STOREY BUILDINGS
5. 1 Requirements ofthe present day building.
The requirements of the present day building have
changed a lot in last years.
Architectural requirements.
The present day apartment house has increased the
requirements for using modem forms. Varied
buildings solids, implementation of the new
materials, determine a new level of those buildings.
Quantity of the architectural forms and spatial
compositions joined with usage of materials and
surface textures requires implementing of different
constructional solutions.
Usable requirements.
Change of the life standards makes building of
the flats in the way, which gives the possibility of
easy change of the flats development shape. Easy of
forming the living space with possibility of free
adapting to own needs is a very important factor,
which have influence to the shape of present day
building. The requirements for ground floor
213
development, which are usually dedicated for
services are even bigger. Development area and
flexibility of its usage by the users. It requires to use
a big spans usually founded on the pillars net made
out in the way which allows easy exchange and
arrangement of the elevation and usage area
developing and their quick adopting for the running
activity.
Buildings made of the large-panel - possibility of
adapting it to present-day architectural requirements.
Present-days housing and service building results
from social needs. New community needs for bigger
living area allows the architects forming with bigger
momentum and implementing the variety of forms
and unique of shapes.
The question arises: what to do with the buildings
made from large-panel, which form monotonie
housing estate development built in previous years?
By the aesthetic matters and new needs for living
area they can be demolish. Looking through the
economic side - are we able to afford for such a
drastic move?
Compromise seems to be the only reasonable
solution.
Adapting the large-panel building to the present-day
social and architectonic needs requires introducing
constructional changes. Construction strengthening
for roof shape change, change of usable room spans,
adapting to heat energy consumption requirements.
All of those changes are in order to adapt of large-
panel building to the present day architectural needs
and to meet functional and aesthetic users
expectations.
5.2 Inside exchanges of existing large-panel
dwelling building, flexibility ofinside area (" bigger
/smaller").
Flats area enlarging.
Due to usable matters its is often required to enlarge
flats area. Large-panel housing offers small flats with
specialised small usable area rooms.
Building made out of large-panel technology are
usually used in mixed construction system. Flats area
enlarging can be achieved in two ways, through
joining the flat on the same level or joining
neighbourhood flats on two following storeys.
Enlarging the usable room's and flat's area in its level
involves with removal of the wall one of the
building construction element. To ensure load
transfer, construction and users safety it is required to
exchange disassembled wall with another element. It
seems rational to use beams, which are founded
under the ceiling transferring loads on the vertical
construction's elements. It is constructor skill to joint
knots shaping and calculating construction stability.
Composition ofthe roofs covers.
Tn the low buildings change of roofs shape seems to
be rational solution. Change ofthe roofs shape varies
building shape and can make at the same time its
renovation or possibility of flats development. .

Construction solutions should not constraint
architecture visions and usable room's height.
Rooms developing through joining flats on two
storeys can bring another troubles. Joining flats on
two neighbour's storeys makes transportation
problems. Easy transportation between storeys has to
be assured. The most common element are staircases
seldom passenger lifts. It makes additional troubles
because of ceilings that are construction elements.
How to put stairs in safety way into the ceiling which
originally had no holes?
Changing offirst floor area in existing large-
panel flats buildings.
First floor area in existing large-panel flats buildings
like others floors in this type of buildings are used for
dwellings. For changing the quality of live of
occupant of the houses ought to be move in new
function for example: shops, restaurants, clubs,
recreation areas, cultural service, etc. This function
needs other construction utilities.
Change of the roofs construction for rafter framing
often forcing the load increase on the last storey.
That load can increase itself because of application of
heavier roof covering, usage of attic as additional
usable floor space. Construction of the ceiling in the
shape of ferroconcrete panels unable direct load
transfer to ceiling. It seems rational to use steel or
wood construction - which is light -strengthening and
reinforcing additional loads on the building walls.
The whole construction should be light with
possibility of easy assembly. We ought to be able to
assembly such construction near the existing
building, on the ground and afterwards put it up of
the building. Such constructions should be able to
transmit the usable loads and dead weight on
responsible construction elements. Additionally the
heat insulating power and noise protection
requirements in the rooms for people should be met.
New part

5.3 Outside exchanges of existing large-panel

dwelling building

Adding vertical elements.

Outside exchanges of existing large-panel Old

dwelling building can be done by adding new
vertical elements.

Τ
Γ*Γ L5
ιΓΤ ΓΤγ
ca ΕΠΕ ι
ι

¡ sw # 3
d. C3D ET c] I H l3
Π: : Γ3 Qc ι I w 3

UL-
:Q Q: :
f Ίι1 0ι-} 'ι 3

v^f* ^t<£^<,EEjiW<3'ï;- ea^ .

o
D
.
D
D

Figure 20. Example of adding roof with new shape and new last
Figure 19. Example of adding vertical elements. (Rybka Α., storey.
Ostafijczuk Α.) (Rybka Α., Ostafijczuk A.)

214

Adding horizontal elements.
Changing of shape of existing large-panel flats
buildings by adding new elements: new parts of
building.
Changing of first floor area in existing large-panel
flats buildings plus adding new elements outside of
the building.
Old part
r-
r--
Jsy y /-> -&ä
L.
> -v ^New part
'it*
iiEiiSïttSiiïü:' ws^ãi»àeãiSÊãs§!Êm .
Figure 21. Example of adding horizontal elements. (Rybka Α.,
Ostafijczuk A.)
Improving of quality of connections of elements in
existing large-panelflats buildings.
Sometimes there are problems with connection of
elements in existing buildings. Specially with
balcony elements.
Increasing of thermal comfort.
Rising prices of heat energy and requirements for
thermal comfort makes necessity to apply additional
heat insulation on buildings. Large-panel buildings
whose outer partition walls include low layer of
insulating materials characterise high heat transfer
coefficient, which follows lowering thermal comfort
and high exploitation costs. There is necessity to
make additional thermal insulation on the existing
buildings to reduce cost of its using.
The most frequently used in Poland method of heat
insulation is light method. It consists in increasing of
wall's thermal resistance through laying the
insulating material on the outer surfaces. The
insulating material is mainly foamed polystyrene,
215
which is joint to the walls by glue materials and
mechanical elements. This elements are additionally
strengthen with plastics nets and covered by plaster
mortar.
Advantage of this method is thermal bridges total
liquidation, speed and execution simplicity.
In the buildings with higher finishing quality
requirements, there are often nobly finishing material
used. Τη that case the outer wall's surface is putted on
a framework between which the heat insulation
panels are fixed.
Elevation aesthetic raising.
The large-panel buildings aesthetic value has
devaluated a lot in last years. In order to change its
view different solutions can be applied.
The simplest one seems to be an application of a new
colouring of the elevation. It can be made together
with thermal renovation. There are new textures used
such as acrylic or mineral plasters laid after heat
insulation making or as facing of ready made
elements in the shape of panels.
More advanced one solution for elevation view
improvement is its total change. It seems to be
possible to exchange the outer walls with new
constructions with the simultaneously change of its
architecture, which enables the change of view and
possibilities for buildings adaptation. The question
arises how big influence has the change of the
elevation on the building construction's stiffness.
It seems to be possible to change the shape of the
building by its partial demolition. Disassembling of a
few storeys from the perpendicular's shape building
will vary its body.
We can add new elements which can change
elevations and help us to enlarge flats area.
. ^ . EáES^âiE ;
G
α
::-::::;
aa:--.:
à . · 'D
o-,
Q π ,-:-:
α
" Κα---,
α «iá
.,., ..: ;
-.ou:
a . . :
D · D ¡.-M-
Old part.
¿7
-" elements
Figure 22. Example of adding to facade new elements. (Rybka
Α., Ostafijczuk A.)
6 CONCLUSIONS
Tn Poland, first problem is that, we must to demolish
about 800 000 dwellings because of their technically
poor condition level.
Also, we have approximately 300 dwellings per 1000
inhabitants in Poland, nowadays. This coefficient is
one of the lowest in Europe. We must build about
1500 000 dwellings because of our social needs in
Poland. So in total, we need about 2300 000 new
dwellings in Poland in near future.
Second problem is, haw to accommodate to new
social needs dwelling-buildings which were been
done with using of prefabricate technologies in last 5
decades (between 1950-1995 year).
In Poland, large-panel building started in fifties and
continued by forty five years was in order to meet the
social flat requirements, towns' developing together
with the cost reduction. The simplest solution was
mass-production.
In advance preparation of the typed building
elements speeded up the assembly and increased the
quantity of putted to use buildings - that was the
matter.
Do the buildings which have exist for dozens of
years, must to be change? If we look at present day
flat and service development, that direction is
unavoidable.
Due to still high cost of development, the large-panel
building is going to be used. One more possibility
remains adaptation of the buildings to actual
housing and servicing needs.
What is most needed here and now?
Improving of quality of connections of elements in
existing large-panel flats buildings.
Increasing of thermal comfort in existing buildings.
Because of climate conditions we must change
composition of the existing roofs covers. In the low
buildings change of roofs shape seems to be rational
solution. While we will change ofthe roofs shape we
can varies building shape and we can make at the
same time its renovation or we have possibility of
increase number of flats in existing building.
Another thing is problem of changing first floor area
in existing large-panel dwelling buildings. Nowadays
social needs shows that we need more space for retail
trade or for handicraft in neighbourhood of the flats.
We can find the area for it on the first floor in
existing buildings.
216
What will we need tomorrow?
Flexibility of inside area in existing large-panel
dwelling buildings ("bigger / smaller"), we will need
flats area enlarging.
Adding horizontal elements to existing large-panel
dwelling buildings. We will have opportunity to
increase number of flats in existing building.
Changing of shape of existing large-panel flats
buildings by adding new elements: new parts of
building. We will have opportunity to improve
thermal comfort in existing buildings.
Changing of first floor area in existing large-panel
flats buildings plus adding new elements outside of
the building. We will have opportunity to enlarge
area for new retail trade or for handicraft in
neighbourhood ofthe flats in existing buildings.
REFERENCES
Collective work, 1998, Concrete building, Arkady, Warsaw
Wçglarz M., 1972, collective work, Dwelling construction
systems, Arkady, Warsaw
Concise Statistical Yearbook of Poland 1995, 1999. Polish
Statistical Office, Warsaw 1996,2000.
Januszaniec B., 1985, Concrete building for architects,
Biarystok University of Technology, Biarystok
Kobiak J., 1991, Ferroconcrete constructions. Arkady, Warsaw
Krol W., 1969, Ferroconcrete constructions. PWN, Warsaw
Lewicki Β., 1961, Big size elements for dwelling buildings,
Arkady, Warsaw
Parczewski W., 1995, Concrete building for architects:
finishing work elements, Warsaw University of Technology
Publishing House, Warsaw.
Rybka Α., 2001, Evaluation of MBT buildings - Modernisation
Aspects and an estimation of a heat insulation of multi-story
houses made of large-sized prefabricated units, COST C12,
Bled
Starosolski W., 1985, Ferroconcrete constructions, PWN,
Warsaw
Voivodships from 1995 till 2000, www.stat.gov.pl
Building response to the thermal, acoustical and visual comfort
functional demands - A comparative study between lightweight and
traditional constructive solutions

L. Bragança, M. Almeida & J. Mendes

University of Minho, Portugal

L. Simões da Silva
University of Coimbra, Portugal

ABSTRACT: The aim of this paper is to evaluate the potentialities of an MBT - "Mixed Building
Technology" solution under the point of view of the thermal and the acoustic performance and the visual
comfort.
For this purpose, a case study is presented, where the performance of a MBT building (an .office building in.
Coimbra) is compared with the performance of a virtual traditional building with the same geometry and
orientation. The fundamental difference between these two buildings is the weight ofthe building elements
and the solar exposure strategy.
of thermal, acoustic and visual comfort must be
tackled.
1 INTRODUCTION
To carry out the thermal, acoustic and lighting
Research and development into new construction quality quantification, as well as the overall quality
technologies should be done with a view to of the buildings, it is necessary to establish an
improving the quality of life inside buildings, the evaluation methodology that allows the
quality of buildings and the quality of the urban quantification and the classification of the physical
environment. The quality of life inside buildings is performance of the buildings, in those areas. In this
directly related to their conditions of security, work, a comparative study between lightweight
hygiene and comfort. (MBT) and traditional constructive solutions is
The basic goal of the new architectural concepts presented taking into account the building response
is to improve buildings quality by the use of new to the thermal, acoustical and visual comfort
technologies. Nowadays, this research has led to the functional demands.
use of lightweight and composite materials in the For this purpose, the results of a case study are
building construction to obtain more effectiveness in showed, where the performance of a MBT building
the relationship between weight/thickness and (an office building in Coimbra - Portugal) is
resistance. This new architectural perspective is compared with the performance of a virtual
known as MBT - Mixed Building Technology or traditional building with the same geometry and
Light Tech Architecture. orientation. The fundamental difference between
The use of lightweight materials can offer an these two buildings is the weight of the building
original response in aesthetic, functional and elements and the solar exposure strategy.
economical terms. With the present knowledge, this
can be achieved by developing new technology in
conceiving materials, improving comfort and 2 BUILDINGS CHARACTERIZATION
constructive methods, but in a way that the building
construction itself, the maintenance and operative In order to show the performance pf MBT buildings,
costs, are kept in a lower level or even reduced and it was selected a building enlarged according to an
never increased. MBT strategy. This is an office building belonging
Although quality evaluation is to some extent to the Portuguese Electricity Company (EDP) and is
subjective, depending on each person opinion, and located in Coimbra Portugal.
related to the socio-economic and cultural heritage The MBT building response to the previously
of whoever is making the assessment, it is possible defined comfort functional demands was compared
to base this evaluation on measurable physical with the one of a virtual building, with the same
criteria. geometry, but designed in a conventional
Tn this work, the question of evaluating the constructive way in what refers the structure, walls,
quality of life in buildings is approached from the windows and roof.
comfort point of view. Within this vast field, the Photo 1 shows the exterior appearance of the
subjects MBT building that has been submitted to the
217
enlargement process according to the MBT strategy. of the old part of the rehabilitated building. Figures
It must be pointed out that only the last floor is a 1, 2, 3 and 4 show schematic views of these two
new construction since the first two floors kept the buildings.
initial constructive characteristics during a
rehabilitation process. The internal layout of the
MBT floor is an open space, as it is shown in
Photo 2.
For the virtual traditional building, the adopted
architectural solution was based on the architecture

.-la'

.':Μ*Ζ\

V*'

It ht
n**

fca.

view Photo 2 - Interior view

o o

o o

MBT building Traditional building

Figure 1 - Schematic plan ofthe MBT and ofthe traditional buildings

218
 BB
BB

MBT building Traditional building

Figure 2 - Vertical scheme ofthe south facade

Β Β
Β Β

MBT building Traditional building

Figure 3 Vertical scheme ofthe north facade

D G

G G D D

α D D
Traditional building
MBT building
Figure 4 - Vertical scheme ofthe west facade

219
2.1 MBT building 2.2 Traditional building
The MBT structure is a steel skeleton that supports This virtual building was defined taking into account
the walls, the intermediate pavements and the roof. the rehabilitation strategy followed in the other two
The pavement is 12 cm thick lightweight steel floors. The pavements and roof are 25 cm thick
reinforced concrete slab, made with lost metallic beam and pot slabs. The floor finishing is wood and
moulding plates. The floor finishing are plywood the ceiling is in gypsum. The roof is a non-ventilated
rectangular plates with linoleum coating. This floor attic insulated with a 4 cm thick layer of mineral
covering is placed 10 cm above the slab in order to wool placed over the roof slab. Table 4 lists the
allow the placing of electric and network cables. floors and roof characteristics.
Under the slab there is a 13 mm thick suspended
ceiling made of gypsum.
The roof is also a 12 cm thick lightweight steel Table 4 - Characteristics of the traditional building floors and
reinforced concrete slab with a 5 cm thick thermal roof
Area Mass Overall conductance
insulation layer made of extruded expanded (m2) (kg/m2) (W/m2.K)
polystyrene. Table 1 lists the floors and roof Floor 74.12 290 1.00
characteristics. Roof 74.12 290 0.90

The exterior walls are double pane (15+11 cm)

Table 1 - Characteristics ofthe MBT building floor and roof
hollow brick walls with 2 cm of mineral wool placed
Area Mass Overall conductance
(kg/m2) (W/m2.K) in the air gap and finished with plaster on both sides.
Floor 74.12 180 0.90 Table 5 lists the exterior wall characteristics.
Roof 74.12 180 0.65

Table 5 - Characteristics of the traditional building exterior

The exterior walls are lightweight sandwich walls
wall
with a plaster layer on the outside, a 5 cm thick layer Area Mass Overall conductance
of expanded polystyrene plus another of 25 cm of (m2) (kg/m2) (W/m2.K)
mineral wool and two 13 mm thick leafs of gypsum North wall 44.71 371 0.85
placed on the inside. Table 2 lists these exterior wall East/West wall 31.34 371 085
characteristics. South wall 40.65 371 0.85

The windows are double glazed (6+12+6 mm)

Table 2 - Characteristics ofthe MBT building exterior wall with a normal metallic frame. Tn all facades the glass
Area Mass Overall conductance is clear in both sides. Tn this "conventional" solution,
NoErthwall
jsú.
44.71
(kg/m2) fW/m'.K) the solar exposure was not optimised, as it is usual in
100 0.20 Portuguese buildings, with glazed areas
East/West wall 34.34 100 0.20
South wall 5.61 100 0.20 homogeneously distributed by all the facades with
no external shading devices, as it is also common in
The windows are double glazed (6+12+6 mm) office buildings. Table 6 lists the characteristics of
with a low-leakage metallic frame. In the North the windows.
facade the glass is clear in both sides and in the
South facade the glass is clear in the inside and Table 6 - Characteristics ofthe traditional building windows
reflective in the outside. A shading device placed Area Mass Overall Solar
over the roof protects the South window facade from (m2) (kg/m2) conductance fraction
the direct solar radiation in summer. It is foreseen (W/m2.K) (%)
for all the windows an interior blind in order to North windows 10.34 30 4.20 0.75
control the solar radiation and daylight penetration. East/West windows 3.00 30 4.20 0.75
South windows 14.40 30 4.20 0.75
However, this blind (a perforated roller screen) was
not in use at the time of this study. Table 3 lists the
characteristics ofthe windows.
3 THERMAL PERFORMANCE EVALUATION

Table 3 - Characteristics ofthe MBT building windows The assessment of the buildings thermal
Area Mass Overall Solar performance can be done by experimental or
(m2) (kg/m2) conductance fraction numerical estimation methods. The experimental
(W/m2.K) (%) methods require a permanent monitoring during at
North windows 10.34 30 3.30 049
least one year, in order to get data both for winter
South windows 49.44 30 3.30 0.49
and for summer seasons. Tn most cases this approach
is not possible due to restrictions imposed by the
building users.
220
 In this study, a numerical estimation method was
used to evaluate the building thermal performance,
generally accepted as accurate enough and also
because the experimental values were not available
for an extensive period, allowing a credible analysis
of the results.
3.1 Thermal evaluation method
The assessment of the thermal performance of both
buildings is based on the evaluation of the energy
consumption using the method suggested in a
previous work (Bragança et al. 2001a). For this, the
amount of energy needed to heat or to cool the air
inside the buildings has to be determined. This
quantification is based on the thermal load concept
defined as the calorific power required to be
supplied to or to be removed from a space so that the
comfort conditions, such as the temperature and the
relative humidity, remain constant (ASHRAE,
1993).
The thermal load is obtained from the energy
balance of the space under consideration. Four
separate components are found in this balance:
- Heat exchange by conduction through the opaque
envelope, comprising roofs, floors and walls;
- Heat exchange by solar radiation and by
conduction through glazed areas;
- Heat exchange associated with the indoor air
renewal, which can be caused by forced
ventilation, natural ventilation or by infiltration;
- Heat released by internal heat sources, such as
equipment, lighting and occupants.
- Heating and cooling needs were determined for
the typical temperature comfort conditions, which
are 18°C in winter and 25°C in summer
(ANSI/ASHRAE, 1981, RCCTE, 1990, RSECE,
1998).
The considered climatic data, for both seasons,
was the one correspondent to sequences of typical
winter and summer days corresponding to the
building's location - Coimbra, denoted as 12 zone
for winter and as V2 zone for summer, according to
the Portuguese legislation (RCCTE, 1990). The
reference data for these zones considers an average
heating degrees-day, in an 18°C base, of 800°C/year,
in winter, and an exterior design temperature of 32°C
for summer with temperature amplitude of 13°C
(Mendes et al. 1989).
The heating needs (HN) are given by the
integrated value of the algebraic sum of four terms,
in the heating season:
- Heat losses through the outer envelope - Qi
- Heat losses through the inner envelope - Q2
- Heat losses associated with the entry of outside
air - Q3
- Heat gains - Q4
The energy needs, HN, per heating season, are
then calculated by the integral of the algebraic sum
221
of the instantaneous values of Qi, Q2, Q3 and Q4
throughout the heating season:
EndHeating
Enatieattng
HN= ί(Σα
ΚΣ +Σο2 +03 -Σ&)·«*
StartHeating
d)
Qi> Q2, Q3 and Q4 are calculated using the method
suggested in a previous work (Bragança et al.
2001a).
The cooling needs (CN) are given by integrating
the algebraic sum of the following portions, during
the cooling season:
- Heat gains by conduction through the outer
envelope - Qi
- Solar heat gains via glazed areas - Q2
- Heat gains from indoor heat sources - Q3
- Heat gains through replacement of indoor air by
hotter outdoor air - Q4
.
Regarding this last term, since in summer the
difference between the indoor and the outdoor air
temperature is, on average, generally small, this
contribution to the cooling needs may be
disregarded.
Regarding the internal gains caused by the
occupants, electrical appliances, lighting devices,
etc., when they are important, they always result in
the use of mechanical cooling systems. But, since
they are not a direct consequence of the envelope,
these gains are not taken into consideration for the
purposes of this calculation method.
We are thus left with the first two components
and the calculation ofthe cooling energy needs, CN,
takes the following form:
EnâCoolting
ooltmg
CN= ÍÍY.CA +
1(Σο.+Σο2)·^ (2)
StartCooling
Qi and Q2 are calculated using the method suggested
in a previous work (Bragança et al. 2001a)].
3.2 Thermal evaluation results and comments
The energy consumption of the MBT and the
Traditional buildings was calculated according to the
referred method.
Table 7 summarizes the heating and cooling
needs (partial and total values) estimated for the two
analysed buildings. Moreover that, the right column
of Table 7 expresses the net energy difference
between the MBT and the Traditional buildings.
 4 ACOUSTICAL PERFORMANCE
Table 7 - Buildings thermal performance
EVALUATION
MBT Solution Traditional MBT/
Solution Traditional
Heating Needs [kWh/year] The assessment of the acoustical performance of any
- Opaque Exterior 1250 3185 -61% building should be done through the evaluation of
Envelope each building element's noise insulation level and
- Glazing 3788 2237 69% through the room's reverberation time.
- Air Renewal 2323 2323 0% As the existing numerical estimation methods are
Useful Solar 3900 3514 11%
Gains [kWh/year]
not accurate enough to assess building acoustics
HN rkWh/yearl 3460 4230 -18% (Bragança et al. 2001b), the acoustical performance
Cooling Needs [kWh/year] ofthe MBT building was experimentally evaluated.
- Opaque Exterior 327 484 -32% Tn order to assess the performance of the two
Envelope solutions under analysis, the MBT results were
- Glazing 3233 1491 117% compared with the acoustic insulation values
CN fkWh/yearl 3569 1975 81% obtained in earlier experimental studies performed
HN + CN 7029 6205 13%
[kWh/yearl by the Building Physics Laboratory ofthe University
of Minho for several conventional constructive
solutions.
This study shows that along a typical year the overall
Tn this particular case, it was not possible to
energy needs in the MBT solution are 13% higher
evaluate the actual reverberation time in the MBT
than the energy needs in the Traditional solution.
However, in winter the MBT solution has a better building because, although at the time of the
experiments the building was almost concluded, it
performance since the heating needs are 18% lower
was not yet occupied and there was no furniture
than the ones of the traditional solution. Tn summer
inside. So, the final occupation pattern was not
the behaviour is the opposite. The MBT building has
reached yet.
cooling energy requirements 81% greater than the
Traditional solution.
This kind of performance in summer is due to the 4.1 Acoustical evaluation method
huge area of windows with no shading devices
incorporated and also to the lack of thermal inertia in The acoustical MBT voiding measurements were
done according to the ISO 140 Standards, parts TV,
the MBT building as it can be seen in Table 7.
As referred before, the MBT solution is a V and Vn (EN ISO 140/IV, 1996, EN ISO 140/V,
1996, EN ISO 140/Vn, 1996), and the ISO 717
lightweight one leading to a light inertia building in
Standards, parts 1 and 2 (EN ISO 717/1, 1996 &
opposition to the traditional building, which has
heavy inertia. The thermal inertia is responsible for
EN ISO 717/2, 1996).
The measuring equipment used was a Brael &
the delay and amplitude decrease of the heat wave
that enters into the building. The greater the thermal
Kjær rotating microphone boom, model 3923, with a
inertia is a better indoor environment is achieved, rotation time of 16 s, equipped with a Brael & Kjær
condensing microphone of Vi", type 4166, and a
preventing the risk of an overheating of the indoor
air. preamplifier type 2639. The microphone was
calibrated before and after measurements with a
About the solar exposure, the MBT building has a
Brael & Kjær calibrator, type 4231, debiting 93.8 dB
glass curtain wall in the South facade, two windows
at the frequency of 1000 Hz.
in the North facade and no openings in the East and
To produce the acoustic field to evaluate the
West facades. The shading device placed on the roof
airborne sound insulation it was used a Brael &
is not enough to protect the south glass curtain in
summer leading to large cooling needs.
Kjær sound source (class Π IEC), model 4224, with
a conic diffuser.
Tn the "conventional" solution, the solar exposure
To produce the acoustic field to evaluate the
was not optimised, as it is usual in Portuguese
buildings, with glazed areas homogeneously impact sound insulation was used a Brael & Kjær
tapping machine, model 3204.
distributed by all the facades with no external
The data registration was done with a Brael &
shading devices, as it is also common in office
Kjær real-time analyser, model 2144, with the
buildings. Tn spite of that, as the glazing area is
software VP 7342, using filters of 1/3-octave band.
much smaller than in the MBT solution and the
The measurements were done using the Brael &
thermal inertia is higher, the summer performance of
Kjær Building Acoustic Program, type 5305.
the traditional building is much better.
The airborne sound insulation was measured
To guarantee an overall good performance of the
following the usual procedures by placing the sound
MBT building, the solar exposure strategy must be
source on one side of the element, in the source
revised.
room, and measuring the sound pressure levels on
the other side of the element, in the receiving room,

222
on a 1/3-octave band. The sound pressure levels
were spatially averaged.
To measure the impact sound insulation, a
standard impact was given to the floor, with the
tapping machine, and the sound pressure levels on a
1/3-octave band were measured in the receiving
room. The sound pressure levels were also spatially
averaged.
According to the international standard EN ISO
140-V, the noise insulation of facades measured "in
situ", R'45 , is given by the following equation:
ΛςΛ
i?'45 = XM-Z2+101og dB -1,5 dB (3)
\*J
where L\¿ - the average sound pressure level on the
surface of the facade; £2 - the average sound
pressure level in the receiving room; S - the area of
the facade; A the equivalent sound absorption area
in the receiving room.
This equation is valid when the noise source is a
loudspeaker, on the assumption that the sound is
incident from an angle of 45° and that the sound
field in the receiving room is diffuse.
According to the international standard EN ISO
140/IV, the noise insulation of a partition element
measured "in situ", R' , is given by the following
equation:
R' = L,-L2+101og\^ (4)
Where L\- the average sound pressure level in the
source room; Li - the average sound pressure level
in the receiving room; S - the area of the element; A
- the equivalent sound absorption area in the
receiving room.
According to the international standard EN ISO
140/Vn, the impact sound insulation of floors
measured in situ", L' , is given by the following
equation:
L'=L2 -10 log (5)
where L% - the average sound pressure level in the
receiving room; S - the area of the element; A the
equivalent sound absorption area in the receiving
room.
The EN ISO 717-1,2 describes a rating method
that fits a standard reference curve to the measured
sound reduction index curve \R' and L'). The
resulting values are the weighted sound reduction
index (R 'w) and weighted normalized impact sound
pressure level index (L ',*)·
4.2 Acoustical evaluation results and comments
The facades noise insulation (R'45), the floor
weighted sound reduction index (R 'w) and the floor
223
weighted normalized impact sound pressure level
index (L 'iW) were determined through the "in situ"
experiments carried out in the MBT building.
To compare the acoustical performance of the
MBT solution with the traditional constructive
solution, these experimental values were compared
with others obtained in earlier experimental studies
performed by the Building Physics Laboratory ofthe
University of Minho in buildings with the same type
of constructive solutions.
The results of this comparative study are
presented on Table 8.
Table 8 - Buildings acoustical performance
Element type R'v L' n.w. R' 4L.
MBTSolution
South facade (90% glass + 10% 30
opaque)
East/West facade (0% glass + 100% 50
opaque)
North facade (19% glass + 81% 40
opaque)
Floor 53 70
Traditional Solution
South facade (26% glass + 74% 33
opaque)
East/West facade (9% glass + 91% 35
opaque)
North facade (19% glass + 81% 34
opaque)
Floor 48 77
From the study carried out it is possible to conclude
that in almost all cases the MBT solution has a better
acoustical performance than the traditional one, in
spite of having less mass. The better quality of the
glazing and the higher level of insulation of the
exterior wall allow its good acoustical performance.
However, the MBT south facade possess a noise
insulation index lower than the traditional solution
due to its large area of fenestration.
The better impact sound insulation of MBT
building floors is due to its floor finishing and also
to the existence of a separate suspended ceiling with
small stiffness, backed with mineral wool quilts.
In spite of the higher mass of the traditional
building floor, the floor finishing and the suspended
ceiling make possible to obtain a better airborne
sound insulation ofthe MBT building floor.
The same reason can be pointed out to explain the
better impact sound insulation of the MBT building
floors. Although the traditional building floor has a
higher mass, in the MBT floor, the air gap between
the slab finishing and the suspended ceiling
increases the airborne sound insulation and reduces
the impact sound pressure level.
It must be enhanced that the MBT building
acoustical performance is quite good. A good
example of this better performance is the East/West
wall. This wall has 100 Kg/m2 of mass and a
weighted sound reduction index of 50 dB. A
traditional wall with the same mass wouldn't have a
weighted sound reduction index greater than 40 dB.
Another good example is the South wall. This wall
is a glass curtain with a weighted sound reduction
index of 30 dB. The traditional buildings with very
large windows don't succeed to have more than 26
dB for the weighted sound reduction index.
5 NATURAL LIGHTING PERFORMANCE
EVALUATION
The evaluation ofthe available natural lighting in the
two constructive solutions was carried out
experimentally. The advantage of this approach is
the evaluation of natural light availability in real
conditions.
The daylight availability in buildings is measured
using the daylight factor concept, which is a
comparison between two illuminance measures. The
illuminance, measured in lux (lumens per square
meter), is the degree of concentration of light
striking a surface. The daylight factor at a given
point inside a room is defined as the illuminance at
that point expressed as a percentage of the
simultaneous horizontal illuminance under an
unobstructed overcast sky (McKennan, 1999).
For reference, the measured values were
compared with the illuminance values recommended
by the "CIE - Commission Internationale de
l'Eclairage" (CIE) for interior working conditions.
5. 1 Natural lighting evaluation method
The illuminance measurements were carried out in
accordance with the DTN Standards (DIN 5034-3,
1994) and the CIE specifications, using a photo-
radiometer with a digital display, connected by a
sign handling circuit to a photometric silicon sensor
with an optical correction filter.
The adopted measuring procedures were the
following:
It was considered the amount of light striking a
horizontal surface 75 cm above the floor level in
order to reproduce the working plane conditions.
The sensor was placed parallel to table surfaces
and all kind of shades on the photocell was avoided
unless those needed to the work place
characterization.
The measurements were carried out at different
points of the room and at different hours of the day
in order to adequately characterize the space.
Until recently it was customary to specify natural
lighting levels in terms of the minimum daylight
factor in a given room. The current trend is to
express the natural lighting of a room in terms ofthe
average (as distinct from the minimum) daylight
factor. This requires a relatively simple calculation
224
and yields, for each room, a single figure that sums
up the overall daylight level.
An average daylight factor greater than 5 per Cent
will generally give the impression of generous
daylighting (except, of course, on a dull day or in the
evening), while an average below about 2 per cent
would be judged gloomy - electric lights would be
switched on as soon as an occupant entered. CIE
recommends for comfortable natural lighting
conditions a daylight factor around 3%.
5.2 Natural lighting evaluation results and
comments
Tnside the MBT building, the measurements were
carried out in a winter clear sky day in a room with
mainly South oriented windows, between 1 pm and 2
pm.
In order to have an illuminance value
representative of the whole room it has been defined
a 170 measuring points grid. The medium
illuminance on the work plane was determined
making an area-weighted mean value of all the
measured points.
According to these procedures, a mean value of
12,899 lux was encountered for the room. Taking in
consideration that the horizontal global exterior
illuminance was 130,000 lux, the daylight factor for
this situation is 9.9%. This is an extremely high
value although expected due to the specific
measurement and room conditions.
Tn order to meet the CIE recommendations for
comfortable natural lighting conditions, the daylight
factor must be reduced to a value near 3%. This can
easily be achieved with the use of curtains in all the
windows.
Regarding the conventional building, the
measurements were carried out in a lower floor. As
said before, only the last floor of the building is a
MBT solution. The lower floors were rehabilitated
using a traditional constructive solution. Tn this
situation, a value of 1949 lux was achieved for the
average measured illuminance inside the room. This
means that the daylight factor was around 1.5%, a lot
lower than the 3% recommended by CIE. Tn this
case, the illumination can be insufficient in periods
of cloudy sky, and thus the occupants need to depend
on artificial lighting.
6 CONCLUSIONS
The MBT buildings usually have large fenestration
areas and are characterized by the use of lightweight
materials. These two aspects have a great impact on
the global building performance such as the thermal,
acoustical and visual comfort. At the design stage,
and in order to be assured the global quality of a
specific constructive solution, the designer should
know in detail its performance. This knowledge can DIN 5034-3: 1994. Daylight in interiors: Calculation. Berlin,
be achieved through specialised literature or, if Germany.
EN ISO 140/TV. 1996. Acoustics - Measurement of sound
inexistent, through measurements performed by
insulation in buildings and of building elements - Part V:
certified laboratories. Field measurements of airborne sound insulation between
This study has shown that although in winter the rooms.
MBT solution has a better performance than the EN ISO 140/V: 1996. Acoustics - Measurement of sound
traditional ones due to its heavier insulation, in insulation in buildings and of building elements - Part V:
summer, the MBT behaviour is very penalising in Field measurements of airborne sound insulation offacade
elements andfacades.
the cooling energy requirements, resulting in greater EN ISO 140/VII: 1996. Acoustics - Measurement of sound
overall energy needs. insulation in buildings and of building elements - Part ΥΠ:
This poor performance in summer is due to the Field measurements of impact sound insulation offloors.
huge area of windows with no shading control EN ISO 717/1:1996. Acoustics - Rating of sound insulation in
devices incorporated and also to the lack of thermal buildings and of building elements - Part 1: Airborne sound
inertia. To guarantee an overall good performance of insulation.
EN ISO 717/2: 1996. Acoustics - Rating of sound insulation in
the MBT building, the solar exposure strategy must buildings and of building elements - Part 2: Impact sound
be revised. insulation.
From the study carried out, it was possible to McKennan, Geoff 1999. The Manchester School of
conclude that in almost all the studied cases the Architecture.
MBT solution has a better acoustical performance Mendes, J. C. et al. 1989. Temperaturas Exteriores de Projecto
e Números de Graus-dias. Lisboa: INMG/LNEC.
than the traditional one, in spite of having less mass.
Regulamento das Características de Comportamento Térmico
The very good quality of the glazing and of the dos Edificios, 1990. Decreto-Lei n° 40/90 de 6 de
windows frames and the higher level of acoustical Fevereiro.
absorbent material placed inside the building Regulamento dos Sistemas Energéticos de Climatização em
elements allow this good acoustical performance. Edificios, 1998. Decreto-Lei n° 1 18/98 de 7 de Maio.
Illuminance values measured in the MBT solution
were generally high especially due to the specific
measurements conditions. Tn this case, the excess of
light can be easily filtered with cheap effective
shading systems. A simple roller-screen system can
solve the problem without the need of any artificial
lighting during the normal working hours. An
adjustable vertical blind for a more accurate glare
control would even be more suitable. This fact
would have a greater impact, not just in terms of
energy consumption, but also in terms of work
productivity and of occupant's health.
The conventional constructive solution, for the
same measurement conditions, doesn't guarantee a
natural illumination along the whole working
daytime period with consequent economical
implications and poor working conditions in terms
of productivity and health.

REFERENCES

ASHRAE, 1993. Handbook of Fundamentals. Atlanta,

Georgia, USA: ASHRAE.
ANSI/ASHRAE Standard 55-1981. Thermal Environmental
Conditions for Human Occupancy.
Bragança, Luís, et al. 2001a. Methodology for evaluating the
thermal performance of MBT buildings. Working Paper
presented in the WG3-Urban Design meeting; Prague,
Check Republic, 19-20 April 2001.
Bragança, Luís et al. 2001b. Acoustical Performance of
Lightweight and MBT Buildings. Working Paper presented
in the WG3-Urban Design meeting, Bled, Slovenia, 16-17
September.
CIE: Commission Internationale de l'Eclairage. Kegelgasse 27,
A-1030 Wien, AUSTRIA.

225
European Commission

EUR 20728 - COST Action C12 - Improvement of buildings' structural quality by new technologies
Luxembourg: Office for Official Publications of the European Communities

2003 xx, 255 pp. 21 χ 29.7 cm

ISBN 92-894-5684-1

Price (excluding VAT) in Luxembourg: EUR 35

 Venta · Salg · Verkauf · Πωλήσεις · Sales · Vente · Vendita · Verkoop Venda · Myynti · Försäljning
http://eur-op.eu.int/general/en/s-ad.htm
BELGIQUE/BELGIE NEDERLAND EESTI
Jean De Lannoy SDU Servicecentrum Uitgevers . Eesti Kaubandus-Tööstuskoda
Avenue du Roí 202/Koningslaan 202 (Estonian Chamberof Commerce and Industry) Les éditions La Liberté Inc.
Christoffel Plantijnstraat 2
B-1 190 Bruxelles/Brussel Postbus 20014 Toom-Kooll 17 3020, chemin Sainte-Foy
Tél. (32-2) 538 43 08 2500 EA Den Haag EE-1 01 30 Tallinn Sainte-Foy, Québec G1X 3V6
Fax (32-2) 538 08 41 Tel. (31-70)378 98 80 Tel. (372) 646 02 44 Tél. (1-418)658 37 63
E-mail: jean.de.lannoy@infoboard.be Fax (31-70) 378 97 83 Fax (372) 646 02 45 Fax (1-800) 567 54 49
URL: http://www.jean-de-lannoy.be E-mail: sdu@sdu.nl E-mail: einfo@koda.ee E-mail: liberte@medlom.qc.ca
URL: http://www.sdu.nl URL: http://www.koda.ee
La librairie européenne/
De Europese Boekhandel
Renouf Publishing Co. Ltd
Rue de la Loi 244/Wetstraat 244 HRVATSKA 5369 Chemin Canotek Road Unit 1
B-1040 Bruxelles/Brussel PORTUGAL Ottawa, Ontario KU 9J3
Tél. (32-2) 295 26 39 Mediatrade Ltd Tel. (1-613)745 26 65
Fax (32-2) 735 08 60 Distribuidora de Livros Bertrand Ld.a Strohalov Prilaz 27 Fax (1-613) 745 76 60
E-mail: mail@libeurop.be Grupo Bertrand, SA HR-1 0000 Zagreb E-mail: order.dept@ renoufbooks.com
URL: http://www.libeurop.be Rua das Terras dos Vales, 4-A Tel. (385-1)660 08 40 URL: http://www.renoufbooks.com
Apartado 60037 Fax (385-1 ¡660 21 65
Moniteur beige/Belgisch Staatsblad P-2700 Amadora E-mail: mediatrade@hi.hinet.hr
Rue de Louvain 40-42/Leuvenseweg 40-42 Tel.(351)214 95 87 87
B-1000 Bruxelles/Brussel Fax (351) 214 96 02 55 EGYPT
Tél. (32-2) 552 22 11 E-mail: dlb@ip.pt MAGYARORSZÁG
Fax (32-2) 51 1 01 84 The Middle East Observer
E-mail: eusales@iust.fgov.be Imprensa Nacional-Casa da Moeda, SA Euro Info Service
Szt. István krt.1 2 41 Sherif Street
Sector de Publicações Oficiais
Hl emelet 1/A 11111 Cairo
Rua da Escola Politécnica, 135
P-1250 -100 Lisboa Codex PO Box 1039 Tel. (20-2)392 6919
DANMARK Fax (20-2) 393 97 32
Tel.(351)213 94 57 00 H-1 137 Budapest
J. H. Schultz Information A/S Fax (351) 213 94 57 50 Tel. (36-1) 329 21 70 E-mail: meo@soficom.com.eg
E-mail: spoce@incm.pt Fax (36-1) 349 20 53 URL: http://www.meobserver.com.eg
Herstedvang 4
DK-2620 Albertslund URL: http://www.incm.pt E-mail: euroinfo@euroinfo.hu
Tif. (45)43 63 23 00 URL: http://www.euroinfo.hu
Fax (45) 43 63 19 69
MALAYSIA
E-mail: schultz@schultz.dk SUOMI/FINLAND
URL: http://www.schuitz.dk MALTA
Akateeminen Kirjakauppa/ EBIC Malaysia
Miller Distributors Ltd
Akademiska Bokhandeln Malta International Airport Suite 47.01, Level 47
DEUTSCHLAND Keskuskatu 1 /Centralgatan 1 PO Box 25 Bangunan AmFinance (letter box 47)
PL/PB 128 Luqa LOA 05 8 Jalan Yap Kwan Seng
Bundesanzeiger Vertag GmbH 50450 Kuala Lumpur
FIN-00101 Helsinki/Helsingfors Tel. (356) 21 66 44 88
Vertriebsabteilung P./tfn (358-9) 121 44 18 Fax (356) 21 67 67 99 Tel. (60-3) 21 62 62 98
Amsterdamer Straße 192 F./fax (358-9) 121 44 35 E-mail: info@millermalta.com Fax (60-3) 21 62 61 98
D-50735 Köln Sähköposti: akatilaus@akateeminen.com E-mail: ebic@tm.net.my
Tel. (49-221)97 66 80 URL: http://www.akateeminen.com
Fax (49-221) 97 66 82 78 POLSKA
E-Mail: vertrieb@bundesanzeiger.de
URL: http://www.bundesanzeiger.de Ars Polona MEXICO
Krakowskie Przedmiescie 7
BTJAB Skr. pocztowa 1001 Mundi Prensa México, SA de CV
/GREECE PL-00-950 Warszawa
Traktorvägen 11-13 Tel. (48-22)826 12 01 Río Panuco. 141
G. C Eleftheroudakis SA S-221 82 Lund Fax (48-22) 826 62 40 Colonia Cuauhtemoc
Tfn (46-46) 18 00 00 E-mail: books119@arspolona.com.pl MX-06500 México, DF
International Bookstore Tel. (52-5) 533 56 58
Panepistimiou 17 Fax (46-46) 30 79 47
E-post: btjeu-pub@btj.se Fax (52-5) 514 67 99
GR-10564Athina E-mail: 101545.2361 ©compuserve.com
Tel. (30) 21 03 25 84 40 URL: http://www.bti.se ROMANIA
Fax (30) 21 03 25 84 99 Euromedia
E-mail: elebooks@books.gr
URL: www.books.gr Str.Dionisie Lupu nr. 65, sector 1
UNITED KINGDOM RO-701 84 Bucuresti SOUTH KOREA
The Stationery Office Ltd Tel. (40-21)260 28 82
Fax (40-21) 260 27 88 The European Union Chamber of
ESPANA Customer Services E-mail: euromedia@mailcity.com
PO Box 29 Commerce in Korea
Boletín Oficial del Estado Norwich NR3 1GN Suite 2004, Kyobo Bldg.
Trafalgar, 27 Tel. (44-870) 60 05-522 SLOVAKIA 1 Chongro 1-Ga, Chongro-Gu
E-28071 Madrid Fax (44-870) 60 05-533 Seoul 110-714
Tel. (34) 915 38 21 11 (libros), 913 84 17 15 E-mail: book.orders@theso.co.uk Centrum VTI SR Tel. (82-2) 725-9880/5
(suscripción) URL: http://www.tso.co.uk Námestie Slobody 19 Fax (82-2) 725-9886
Fax (34) 915 38 21 21 (libros), 913 84 17 14 SK-81223 Bratislava 1 E-mail: eucck@eucck.org
(suscripción) Tel. (421-2)54 41 83 64 URL: http://www.eucck.org
E-mail: clientes@com.boe.es ISLAND Fax (421 -2) 54 41 83 64
URL: http://www.boe.es E-mail: europ@tbb1.cvtisr.sk
Bokabud Larusar Blonda! URL: http://www.cvtisr.sk
Mundi Prensa Libros, SA
Castellò, 37 Engjateigi 17-19
E-28001 Madrid 1S-1 05 Reykjavik
Tel. (354) 552 55 40 SLOVENIJA EBIC Sri Lanka
Tel. (34)914 36 37 00
Fax (34) 915 75 39 98 Fax (354) 552 55 60 GV Zalozba d.o.o. Trans Asia Hotel
E-mail: libreria@mundiprensa.es E-mail: bokabud@simnet.is Dunajska cesta 5 1 15 Sir Chittampalam

URL: http://www.mundiprensa.com Sl-1000 Ljubljana A. Gardiner Mawatha

Tel.(386)13 09 1800 Colombo 2
Fax (386) 13 09 1805 Tel. (94-1)074 71 50 78
FRANCE E-mail: europ@gvzalozba.si Fax (94-1) 44 87 79
Swets Blackwell AS URL: http://www.gvzalozba.si E-mail: ebicsl@sltnet.lk
Journal officiel Hans Nielsen Hauges gì. 39
Service des publications des CE Boks 4901 Nydalen
26, rue Desaix N-0423 Oslo TURKIYE
F-75727 Paris Cedex 15 Tel. (47) 23 40 00 00 TAI-WAN
Dünya Aktüel A.S
Tél. (33) 140 58 77 31 Fax (47) 23 40 00 01
Fax (33) 140 58 77 00 E-mail: info@no.swetsblackwell.com Globus Dünya Basinevi Tycoon Information Ine
E-mail: europublications® journal-offi 100, Yil Mahallessi 34440
TR-80050 Bagcilar-lstanbul PO Box 81-466
ciel .gouv.fr
Tel. (90-212)440 22 27 105 Taipei
URL: http://www.ioumal-officiel.gouv.fr
SCHWEIZ/SUISSE/SVIZZERA Fax (90-212) 440 23 67 Tel. (886-2)87 12 88 86
E-mail: aktuel.info@dunya.com Fax (886-2) 87 12 47 47
Euro Info Center Schweiz E-mail: eiutpe@ms21.hinet.net
IRELAND c/o OSEC Business Network Switzerland
Alan Hanna's Bookshop Stampfenbachstraße 85 ARGENTINA
PF 492
270 Lower Rathmines Road World Publications SA UNITED STATES OF AMERICA
CH-8035 Zürich
Dublin 6 Tel. (41-1)365 53 15 Av. Cordoba 1877
Tel. (353-1)496 73 98 Fax (41-1) 365 54 11 C1 120 AAA Buenos Aires
Fax (353-1) 496 02 28
Bernan Associates
E-mail: eics@osec.ch Tel. (54-11) 48 15 81 56
E-mail: hannas@iol.ie URL: http://www.osec.ch/eics Fax (54-11)48 1581 56 461 1 -F Assembly Drive
E-mail: wpbooks@infovia.com.ar Lanham MD 20706-4391
URL: http://www.wpbooks.com.ar Tel. (1-800) 274 44 47 (toll free telephone)
ITALIA Fax (1 -800) 865 34 50 (toll free fax)
BALGARUA E-mail: query@beman.com
Licosa SpA AUSTRALIA URL: http://www.beman.com
Europress Euromedia Ltd
Via Duca di Calabria, 1/1 Hunter Publications
Casella postale 552 59, blvd Vitosha
1-50125 Firenze BG-1 000 Sofia PO Box 404
Tel. (39) 05 56 48 31 Tel. (359-2) 980 37 66 Abbotsford, Victoria 3067 ANDERE LANDER/OTHER COUNTRIES/
Fax (39) 055 64 12 57 Fax (359-2) 980 42 30 Tel. (61-3)94 17 53 61 AUTRES PAYS
E-mail: licosa@licosa.com E-mail: Milena@mbox.cil.bg Fax (61-3) 94 19 71 54
URL: http://www.licosa.com URL: http://www.europress.bg E-mail: admin@tekimaging.com.au Bitte wenden Sie sich an ein Büro Ihrer
Wahl/Please contact the sales office of
your choice/Veuillez vous adresser au
BRASIL bureau de vente de votre choix
LUXEMBOURG CYPRUS
Livraria Camões Office for Officiai Publications
Messageries du livre SAP.L Cyprus Chamber of Commerce Rua Bittencourt da Silva, 12 C of the European Communities
5, rue Raiffeisen and Industry CEP 2, rue Mercier
L-2411 Luxembourg PO Box 21455 20043-900 Rio de Janeiro L-2985 Luxembourg
Tél. (352)40 10 20 CY-1 509 Nicosia Tel. (55-21)262 47 76 Tel. (352) 29 29-42001
Fax (352) 49 06 61 Tel. (357-22) 88 97 52 Fax (55-21) 262 47 76 Fax (352) 29 29-42700
E-mail: mail@mdl.lu Fax (357-22) 66 10 44 E-mail: llvraria.camoes@incm.com.br E-mail: info-info-opoce@cec.eu.int
URL: http:/Avww.mdl.)u E-mail: stalo@ccci.org.cy URL: http://www.incm.com.br URL: http://publications.eu.int
COST Action C1 2 entitled 'Improvement of buildings' structural quality by new technologies',
was launched in May 2000 and gathered 60 members representing 21 of the 35 COST
member countries. The main objectives of this action, relevant to the urban civil engineering
domain of COST, are to develop, combine and disseminate new engineering technologies,
to improve the quality of urban buildings, to propose new technical solutions to architects
and planners, to reduce the disturbances of the construction process in urban areas and
finally to improve the quality of living in the urban habitat. ro
o
-~j
ro
oo
The present proceedings include the scientific papers presented at the Lisbon seminar m
ZZ.

organised by the action and consist of the following three parts respectively to the WG ό
thematic topic:
- mixed building technology;
- structural integrity under exceptional actions;
- urban design.

For more information on COST C12: http://members.telering.at/geraldhuber/c12/index.html

For more information on COST and COST Actions: http://cost.cordis.lu/scr/home.cfm

Price (excluding VAT) in Luxembourg: EUR 35

* w * ISBN 'ίΕ-Α'ίΜ-ΞΙ,αΐ.-Ι
*
* * Publications Office
*
Publications.eu.ini 9 ^SÇZSÇHSÓSAS