Seismic Behavior of Precast Piers On High Speed Railway Bridges

D 2017
SEISMIC BEHAVIOR OF PRECAST PIERS ON

HIGH SPEED RAILWAY BRIDGES
ANDRÉ MANUEL BESSA MONTEIRO
A thesis submitted to the Faculty of Engineering of the University of Porto, in accordance with
the requirements of the degree of Doctor in Civil Engineering
Porto, 2017
Supervisor: António José Coelho Dias Arêde

Co-Supervisor: Nelson Saraiva Vila Pouca
This work received financial support from project SIPAV – Soluções Inovadoras Pré-
Fabricadas para Vias Férreas de Alta Velocidade, on behalf of Portuguese Agência de
Inovação (AdI), and FEDER SI-IDT-3440/2008 on account of QREN (2009-2012);
This work was also supported by: Project POCI-01-0145-FEDER-007457 – CONSTRUCT -

Institute of R&D In Structures and Construction funded by FEDER funds through
COMPETE2020 - Programa Operacional Competitividade e Internacionalização (POCI) – and
by national funds through FCT - Fundação para a Ciência e a Tecnologia;
ACKNOWLEDGEMENTS
To Professor António Arêde, for accepting me as a member of the LESE research

group, nearly a decade ago, for introducing me to the framework of seismic engineering, and for
sparking in me the will and determination to aim for the academic level which I am now seeking
to fulfill. He accepted the challenge of supervising my work, and has been my mentor and friend
for a long time, and I could not be more grateful for the opportunity of learning with someone
whose wisdom, knowledge and experience on these subjects still keeps surprising me. For being
inexhaustible with helping me on the main parts of my work, for the patience and guidance on
all the others and while always on a busy schedule, I express my deepest gratitude.
To Professor Nelson Vila Pouca, for being the most knowledgeable and competent
Engineer that I know of and had the pleasure to work with. The innumerous work discussions
with him, regardless of context, constitute some of the most fruitful learning moments that I can
remember, and also one of the main supporting pillars of several strategies developed for this
work. For always being available, whether over a long conversation or a five minute coffee talk,
I express my deepest gratitude.
I would also like to extend a special appreciation to my family, namely my parents and
sister, who supported me through so much untold and unimaginable stress and difficulties until I
managed to reach this stage of the work. It warms my heart to know that, by concluding it, I am
also fulfilling some of their hopes and dreams.
To Professors Rui Faria and Mário Pimentel, for their help in understanding and
overcoming several numerical simulation difficulties, enabling me to present the results herein
included. To Professor Miguel Castro for the suggestion and help with using FEUP’s cluster
computation capabilities, significantly reducing the effort associated with the dynamic analyses.
To my colleague Luís Macedo, for his invaluable help on the framework of ground motion
selection. To several others, like Nuno Pereira, Pedro Montenegro, Joel Malveiro or João Rocha
for the opportunity of exchanging fruitful advice, discussion and general friendship.
And finally, to all the other friends and close persons who, whether by actively listening
to all my doubts and complaints or by just providing me with a cheerful and friendly outlet for
all my frustrations, helped me deal with this journey on a day-by-day basis
Thank you all!
André Monteiro
ABSTRACT
The context of bridge engineering and, in particular, High Speed Railway Line (HSRL)
bridges, has always been one of the most demanding frameworks for structural engineers. In
that regard, it is common for technological breakthroughs to be achieved as a result of solving
the numerous challenges that high-end structures create, acting as a driving force for constant
updates of the common engineering practices. Precast solutions are a prime example of that,
regarding economic objectives and strict deadlines, as they strive for optimization of
geometrical layouts, reinforcement designs and material compositions, in order to achieve a
competitive synergy between design and construction while reducing costs and overall building
times.
Precast systems for the construction of bridge decks have undoubtedly become a
widespread solution, however full-scale applications for bridge piers have not been considered
as much, particularly in the HSRL framework. As far as it is possible to assess, important
operational constraints influencing the cost-effectiveness of precast solutions relative to the
more common monolithic alternatives, and also critical performance requirements such as those
presented by seismic loading, leading to weaker or unreliable designs, are some of the main
reasons preventing a larger prevalence of precast for bridge piers.
It is within that context that the current work aimed to provide further insight, by studying
a precast bridge pier solution in comparison with monolithic alternatives. In that regard, a
double column bridge pier layout is presented, based on a previous design of the foreseen
Portuguese venture into HSRL. The pier structure is constituted by two large columns and a
short span beam to couple the column heads for displacement compatibility, resulting in a high
stiffness bent frame capable of addressing the strict HSRL deformation limits, which is studied
in a three stage experimental campaign (for monolithic specimens, precast specimens and single
column foundation specimens) under cyclic loading and reduced scale (1:4) conditions.
The experimental data is further used to calibrate refined numerical modelling strategies for
evaluation of different and not tested pier conditions, using a 2D FEM based methodology. In
addition, it is also used to calibrate global bridge modelling tools within a concentrated
plasticity approach, for the comparative assessment of the seismic performance of different
bridges designed for monolithic and precast piers, subjected to a variety of increasing intensity
ground motions under incremental dynamic analysis procedures.
i
SUMÁRIO
No âmbito da engenharia de estruturas, o contexto das pontes e, em particular, das pontes
inseridas em Linhas de Alta Velocidade (LAV) ferroviária, é reconhecidamente um dos campos
de aplicação mais exigentes. Nesse contexto, os desafios criados pelas estruturas mais
complexas constituem oportunidades regulares para a utilização de sistemas e processos na
vanguarda da tecnologia, servindo como elemento dinamizador para uma atualização constante
das práticas mais comuns. As soluções prefabricadas assumem-se como exemplos típicos do
referido, no contexto dos objetivos económicos e logísticos a cumprir pelas estruturas, uma vez
que permitem a otimização de geometrias, armaduras de reforço e composições materiais, de
forma a fomentar sinergias competitivas entre o dimensionamento e a construção, ao reduzir
custos e tempos de construção.
Ainda assim, sendo já vasta a experiência de aplicação de sistemas prefabricados em

tabuleiros de pontes, o mesmo não se pode dizer para uma utilização em larga escala aplicada
aos respetivos pilares, particularmente no que diz respeito a estruturas de LAV. Nesse sentido,
sabe-se que não só existem condicionalismos operacionais importantes, capazes de influenciar o
rácio custo-benefício das soluções prefabricadas relativamente a soluções monolíticas
alternativas, como também questões de exigência estrutural, como por exemplo o carregamento
sísmico, que facilmente conduzem a soluções fracas ou pouco fiáveis e, dessa forma, pouco
desejadas.
Esse é o contexto do presente trabalho, que visa estudar uma solução prefabricada para
pilares de pontes em comparação com soluções monolíticas, através de uma campanha
experimental sob condições de carregamento cíclico e escala reduzida (1:4) organizada em três
fases (modelos monolíticos, modelos prefabricados e modelos de fundação). A estrutura
estudada foi baseada no projeto, entretanto cancelado, para implementação de LAV em
Portugal, e é constituída por dois grandes fustes verticais e uma viga de vão curto para
compatibilidade de deslocamentos, resultando em elevada rigidez transversal capaz de garantir
o cumprimento dos rígidos limites de deformação aplicáveis a LAV.
A informação experimental é posteriormente utilizada para calibrar ferramentas refinadas

de modelação numérica, para avaliação de condições de carregamento diferentes das testadas
em ambiente laboratorial, através de uma metodologia 2D baseada no MEF. Por fim, realiza-se
também o estudo comparativo da performance sísmica de diferentes pontes com sistemas
monolíticos e prefabricados, quando sujeitas a carregamentos sísmicos de intensidade crescente
e através de procedimentos de análise dinâmica incremental, utilizando modelos globais com
plasticidade concentrada calibrada de acordo com a informação experimental.
ii
TABLE OF CONTENTS
1. INTRODUCTION .................................................................................................. 1.1
1.1. BACKGROUND OVERVIEW ....................................................................................... 1.1
1.2. MOTIVATION AND OBJECTIVES................................................................................. 1.3
1.3. THESIS OUTLINE..................................................................................................... 1.4
2. PIERS FOR HSRL BRIDGES AND VIADUCTS ................................................... 2.1
2.1. INTRODUCTION ....................................................................................................... 2.1
2.2. COMMON RAILWAY BRIDGE LAYOUTS ...................................................................... 2.2
2.3. RAILWAY BRIDGE SEISMIC DESIGN .......................................................................... 2.5
2.3.1. General remarks ............................................................................................ 2.5
2.3.2. Railway Performance Requirements ............................................................ 2.10
2.3.3. Seismic Load ............................................................................................... 2.13
2.4. PIER TYPES ON RAILWAY BRIDGES ........................................................................ 2.15
2.4.1. Overview ..................................................................................................... 2.15
2.4.2. Single Columns ........................................................................................... 2.17
2.4.3. Wall-Piers .................................................................................................... 2.22
2.4.4. Multiple Column Piers .................................................................................. 2.25
2.5. FINAL REMARKS ................................................................................................... 2.31
3. PRECAST TECHNOLOGY FOR BRIDGE PIERS ............................................... 3.1
3.1. ADVANTAGES AND DISADVANTAGES OF PRECAST ..................................................... 3.1
3.2. PRECAST CONNECTIONS......................................................................................... 3.4
3.2.1. Foundation-to-element Connections .............................................................. 3.5
3.2.2. Element-to-element Connections ................................................................. 3.11
3.3. BRIDGE PIER PRECAST SYSTEMS AND DESIGN ....................................................... 3.13
3.3.1. Layouts for non-seismic zones .................................................................... 3.14
3.3.2. Layouts for seismic zones............................................................................ 3.19
3.4. FINAL REMARKS ................................................................................................... 3.27
iii
4. CONCEPTION AND SEISMIC DESIGN OF TEST SPECIMENS ........................ 4.1
4.1. REFERENCE PIER MODEL: THE RAV POCEIRÃO-CAIA CONCESSION PROJECT ............ 4.1
4.2. PIER SEISMIC DESIGN............................................................................................. 4.5
4.2.1. General Considerations and Design Methodology ......................................... 4.5
4.2.2. Implications of the structural system on the design strategy .......................... 4.7
4.2.3. Design strategy............................................................................................ 4.14
4.2.4. Case Study Viaducts ................................................................................... 4.18
4.2.5. Reinforcement and Detailing Layouts .......................................................... 4.24
4.2.6. Precast Conception ..................................................................................... 4.30
4.3. TEST SPECIMENS ................................................................................................. 4.43
4.3.1. Phase 1 – Monolithic Specimens ................................................................. 4.45
4.3.2. Phase 2 – Precast Specimens ..................................................................... 4.52
4.3.3. Phase 3 – Foundation-to-element connections ............................................ 4.58
4.4. FINAL REMARKS ................................................................................................... 4.62
5. EXPERIMENTAL CAMPAIGN.............................................................................. 5.1
5.1. TEST SETUP........................................................................................................... 5.1
5.1.1. Phase 1 and Phase 2 – Bent Pier Tests ........................................................ 5.1
5.1.2. Phase 3 - Single Column Tests ................................................................... 5.16
5.2. EXPERIMENTAL RESULTS – BENT PIER TESTS ........................................................ 5.18
5.2.1. Phase 1 Observations ................................................................................. 5.18
5.2.3. Results Discussion ...................................................................................... 5.51
5.3. EXPERIMENTAL RESULTS – SINGLE COLUMN TESTS ............................................... 5.67
5.3.2. Result Discussion ........................................................................................ 5.73
5.4. FINAL REMARKS ................................................................................................... 5.80
6. NUMERICAL SIMULATIONS ............................................................................... 6.1
6.1. INTRODUCTION ....................................................................................................... 6.1
iv
6.2. REFINED MODELING ............................................................................................... 6.1
6.2.1. Modeling Strategy .......................................................................................... 6.2
6.2.2. Constitutive Models ..................................................................................... 6.10
6.3. EXPERIMENTAL RESULTS VS. NUMERICAL SIMULATIONS .......................................... 6.24
6.3.1. Monolithic Specimens: SP_M02, SP_M03 and SP_M04 ............................. 6.25
6.3.2. Precast Specimens: SP_PC02A and SP_PC02B ........................................ 6.32
6.3.3. Single Column Specimens ........................................................................... 6.40
6.4. NUMERICAL SIMULATIONS OF UNTESTED PIER SYSTEMS ........................................ 6.43
6.4.1. Simulations with Full Height Piers ................................................................ 6.43
6.4.2. Shorter Pier Scenarios................................................................................. 6.48
6.5. FINAL REMARKS ................................................................................................... 6.51
7. SEISMIC PERFORMANCE ASSESSMENT ........................................................ 7.1
7.1. OBJECTIVES AND MAIN METHODOLOGY ................................................................... 7.1
7.2. GROUND MOTION CHARACTERIZATION .................................................................... 7.2
7.3. STRUCTURE CHARACTERIZATION ............................................................................ 7.5
7.3.1. Pier Behavior Calibration ............................................................................... 7.6
7.3.2. Non-Linear Static Pushover Analyses .......................................................... 7.10
7.4. DAMAGE CHARACTERIZATION................................................................................ 7.12
7.4.1. DM #1 Structural Performance .................................................................... 7.13
7.4.2. DM #2 Lateral Deflection ............................................................................. 7.14
7.4.3. DM #3 Derailment Conditions ...................................................................... 7.15
7.5. SEISMIC PERFORMANCE OUTPUT .......................................................................... 7.17
7.5.1. DM#1: Structural Performance Results ........................................................ 7.18
7.5.2. DM#2: Lateral Deflection Results................................................................. 7.20
7.5.3. DM#3: Derailment Conditions Results ......................................................... 7.21
7.6. FINAL REMARKS ................................................................................................... 7.22
8. CONCLUSIONS ................................................................................................... 8.1
8.1. MAIN REMARKS ...................................................................................................... 8.1
v
8.2. FUTURE DEVELOPMENTS ........................................................................................ 8.6
vi
LIST OF FIGURES
Figure 2.1 – Rail buckling (from http://www.railtemperature.com/) .................................................. 2.3

Figure 2.2 – Different longitudinal static schemes for continuous span railway bridges ....................... 2.4
Figure 2.3 – Lock-up Device: Schematic view (Taylor, Taylor Devices Inc.) ..................................... 2.5
Figure 2.4 – Performance assessment strategy for railway bridge service, prescribed in the Japanese RTRI
standard (RTRI (2007b)) ............................................................................................. 2.9
Figure 2.5 – Illustration of horizontal movement limit (DGF (2007)) ............................................... 2.10
Figure 2.6 – Lateral vibration displacement limits (adapted from RTRI (2007b)) .............................. 2.13
Figure 2.7 – Seismic zoning map for Portugal (adapted from NP EN1998-1 (2010)) ........................ 2.14
Figure 2.8 – EC8 elastic acceleration response spectra (NP EN1998-1 (2010)) ................................ 2.15
Figure 2.9 – Flow chart for pier types in bridge structures (Carmichael and Desrosiers (2008)) .......... 2.16
Figure 2.10 – Representation of a wall cross-section geometry ........................................................ 2.16
Figure 2.11 – Single column flare/cap on HSRL viaducts .............................................................. 2.18
Figure 2.12 – Special solutions for horizontal stiffness ................................................................... 2.19
Figure 2.13 – Pier layout variation in the HSRL Vernégues Viaduct in France.................................. 2.20
Figure 2.14 – Composite deck bridges .......................................................................................... 2.21
Figure 2.15 – Composite deck bridges .......................................................................................... 2.21
Figure 2.16 – Wall-piers' examples on low-rise viaducts................................................................. 2.22
Figure 2.17 – TGV junction bridges – France (from http://en.structurae.de/) .................................... 2.23
Figure 2.18 – Auxonne Viaduct – France (from http://en.structurae.de/) ........................................... 2.23
Figure 2.19 – Wall-piers in composite steel girder viaducts (from http://en.structurae.de/) ................. 2.24
Figure 2.20 – Box girder railway bridges supported by single column piers with similar width. ......... 2.25
Figure 2.21 – Bent-type columns with multiple girder bridges ........................................................ 2.26
Figure 2.22 – RC box girder decks supported by multiple column frames ........................................ 2.27
Figure 2.23 – Anguera Viaduct - Spain ......................................................................................... 2.27
Figure 2.24 – Viaduct over the Guadalete River– Spain (from Cutillas (2007)) ................................ 2.28
Figure 2.25 – Stöbnitz Viaduct - Germany ..................................................................................... 2.28
Figure 2.26 – Japanese Shinkansen rigid frame characteristics (adapted from Tamai (2014)) ............. 2.29
Figure 2.27 – Japanese Shinkansen frame unit connections (adapted from Tamai (2014)) .................. 2.29
Figure 2.28 – Visual overview of different HSRL viaducts. ............................................................ 2.30
Figure 2.29 – Piers in the TGV Rhine-Rhone HSRL ....................................................................... 2.30
Figure 3.1 – Material savings using composite precast designs (Yee and Eng (2001)) ......................... 3.2
Figure 3.2 – Schematic pocket connection ("fib Bulletin 43" (2008)) ................................................ 3.5
Figure 3.3 – Socket connection ...................................................................................................... 3.6
Figure 3.4 – Column with base plate and welded rebars ("fib Bulletin 43" (2008)) ............................. 3.7
Figure 3.5 – Shoe column connection (http://www.peikko.ca/).......................................................... 3.8
Figure 3.6 – CIP footing connections .............................................................................................. 3.9
Figure 3.7 – Grouted connection types according to Matsumoto et al. (2008) ................................... 3.10
Figure 3.8 – Mechanical coupling connections (Haber et al. (2013)) ............................................... 3.10
vii
Figure 3.9 – Corrugated grout sleeves (Matsumoto et al. (2008)) .................................................... 3.11
Figure 3.10 – Hybrid duct and socket connection for post-tensioned piers (Davis et al. (2012)). ........ 3.11
Figure 3.11 – Match-cast manufacture process by Billington et al. (1999a) ...................................... 3.12
Figure 3.12 – Match-cast surface and shear keys ........................................................................... 3.13
Figure 3.13 – Precast pier according to Billington et al. (1999a)...................................................... 3.14
Figure 3.14 – Possible precast bent configurations (Billington et al. (1999a)) ................................... 3.15
Figure 3.15 – Longitudinal view of the Ayuntamento 2000 bridge (Cruz Lesbros et al. (2003)) ......... 3.16
Figure 3.16 – Precast bent-pier assembly (Cruz Lesbros et al. (2003)) ............................................. 3.17
Figure 3.17 – Precast pier in the Sorell Causeway Channel Bridge (Gibbens and Smith (2004)) ........ 3.17
Figure 3.18 – Assembly of precast cap elements ............................................................................ 3.18
Figure 3.19 – General substructure concept for the José, Battice, Ruyff and Hervé viaducts
(Couchard and Detandt (2003)) ................................................................................ 3.18
Figure 3.20 – Precast segmental bridge pier connected by unbonded post-tension
(Hewes and Priestley (2002)) ................................................................................... 3.20
Figure 3.21 – Segmental precast pier seismic performance as reported by Hewes and Priestley (2002)3.21
Figure 3.22 – Hybrid rocking mechanism as reported by Palermo et al. (2007)................................. 3.21
Figure 3.23 – “Flag-shape” hysteretic curves for hybrid rocking piers (Ou et al. (2010)) ................. 3.22
Figure 3.24 – Experimentally tested segmental pier variants by Wang et al. (2008). ......................... 3.22
Figure 3.25 – Concrete-filled fiber tube segmental piers (Elgawady and Sha'lan (2011)) ................... 3.23
Figure 3.26 – Ductile Fiber Reinforced Cement Composite segmental pier
(Billington and Yoon (2004)) .................................................................................. 3.23
Figure 3.27 – Composite precast segmental pier layout by Sumitomo Mitsui Co. Lda
(Hoshikuma et al. (2009)). ...................................................................................... 3.24
Figure 3.28 – CIP emulated Reinforced Concrete system by Hieber et al. (2005).............................. 3.25
Figure 3.29 – Integral columns precast system proposed by Khaleghi (2005) ................................... 3.26
Figure 3.30 – Large diameter rebar connections for bridge piers by Pang et al. (2010). ..................... 3.27
Figure 4.1 – Illustration of the Poceirão-Caia segment within the full Portugal and Spain HSRL
expected layouts Altavia (2009) .................................................................................. 4.1
Figure 4.2 – Pier layouts on the Poceirão-Caia project Altavia (2009) ............................................... 4.2
Figure 4.3 – Longitudinal connection layouts on Poceirão-Caia HSRL structures ............................... 4.4
Figure 4.4 – Plastic hinge placement on different structural systems .................................................. 4.6
Figure 4.5 – Double Column Bent-pier geometry ............................................................................. 4.8
Figure 4.6 – Stiffness ratios for lateral displacements ....................................................................... 4.8
Figure 4.7 – Comparisons of structure stiffness relative element ratios and absolute variation for pier
heights in the assumed range and different analysis scenarios ...................................... 4.10
Figure 4.8 – Moment curvature analyses ....................................................................................... 4.11
Figure 4.9 – Beam curvature φb demand variation (with column height) for node rotations associated
with column yielding for different column-beam elastic stiffness ratios. ....................... 4.12
Figure 4.10 – Coupled shear wall systems (Kumar Subedi (1991a), Kumar Subedi (1991b)) ............. 4.13
Figure 4.11 – Beam shear dominated response. Diagonal splitting example (Kumar Subedi (1991a)) . 4.14
viii
Figure 4.12 – Seismic zoning of the Poceirão-Caia HSRL line ........................................................ 4.15
Figure 4.13 – Elastic ground motion response spectra for soil type C ............................................... 4.16
Figure 4.14 – Maximum behavior factors for the Response Spectrum Method (EN1998-2 (2005)) .... 4.17
Figure 4.15 – Case Study Viaducts ............................................................................................... 4.19
Figure 4.16 – Support layouts on the case study viaducts (illustrative scale). .................................... 4.20
Figure 4.17 – Macheda Viaduct model in SAP2000 ....................................................................... 4.21
Figure 4.18 – First vibration mode shapes, period and MPMR values .............................................. 4.23
Figure 4.19 – Spectral Analysis Results: Maximum base shear values among all the piers for each
viaduct .................................................................................................................. 4.24
Figure 4.20 – Beam strains during differential movement ............................................................... 4.25
Figure 4.21 – Sliding shear failure at beam-wall interface (Paulay and Binney (1974)) ..................... 4.26
Figure 4.22 – Cyclic loading results on conventionally reinforced coupling beams ........................... 4.26
Figure 4.23 – Bi-diagonal Reinforcement Layouts ......................................................................... 4.27
Figure 4.24 – Cyclic loading results on diagonally reinforced coupling beams ................................. 4.27
Figure 4.25 – Rhombic truss reinforcement layout ......................................................................... 4.28
Figure 4.26 – Rebar density in coupling beams of shear walls (Parra-Montesinos et al. (2010)) ......... 4.29
Figure 4.27 – Pier design strategy ................................................................................................. 4.30
Figure 4.28 – Precast system: Option A ........................................................................................ 4.32
Figure 4.29 – Beam-column precast connection mechanism ........................................................... 4.33
Figure 4.30 – Liebherr LTM 1350 lifting capacity calculation procedure (http://www.liebherr.com/) . 4.34
Figure 4.31 – Hollow sections for precast column elements ............................................................ 4.35
Figure 4.32 – Precast system: Option B......................................................................................... 4.36
Figure 4.33 – Precast system: Option C......................................................................................... 4.38
Figure 4.34 – Pier vs. Wall stiffness comparison ............................................................................ 4.39
Figure 4.35 – Bent pier to Wall pier stiffness ratio ......................................................................... 4.39
Figure 4.36 – Precast system: Option D ........................................................................................ 4.41
Figure 4.37 – WP element functionality ........................................................................................ 4.42
Figure 4.38 – Experimental phases in correspondence with elastic moments distribution................... 4.44
Figure 4.39 – 1:4 Reduced scale model geometry .......................................................................... 4.45
Figure 4.40 – General column reinforcement for all specimens. ...................................................... 4.46
Figure 4.41 – Specimen SP_M01 reinforcement layout .................................................................. 4.47
Figure 4.45 – Beam-column node rebar force transfers ................................................................... 4.52
Figure 4.46 – Specimen SP_PC02A overview ............................................................................... 4.54
Figure 4.47 – Specimen SP_PC02A assembly operations ............................................................... 4.54
Figure 4.48 – Specimen SP_PC02B overview ............................................................................... 4.55
Figure 4.49 – Specimen SP_PC02B assembly operations ............................................................... 4.56
Figure 4.50 – Specimen SP_PC02C overview ............................................................................... 4.57
ix
Figure 4.51 – Specimen SP_PC02C assembly operations ............................................................... 4.57
Figure 4.52 – Specimen SP_F01 overview .................................................................................... 4.59
Figure 4.53 – Specimen SP_F02 construction process .................................................................... 4.60
Figure 4.54 – SP_F03 overview ................................................................................................... 4.61
Figure 4.55 – SP_F03 assembly process ........................................................................................ 4.62
Figure 5.1 – Phase 1 and Phase 2 test setup ..................................................................................... 5.2
Figure 5.2 – Test setup examples including deck masses .................................................................. 5.3
Figure 5.3 – Test setup examples including vertical prestress for axial load application ....................... 5.4
Figure 5.4 – Axial loading system details ........................................................................................ 5.5
Figure 5.5 – Mechanical hinge and free rotation steel plates ............................................................. 5.6
Figure 5.6 – Mechanical hinge and threadbar cap relative positions ................................................... 5.7
Figure 5.7 – Lateral load application examples for single columns .................................................... 5.7
Figure 5.8 – Displacement compatibility device ............................................................................... 5.8
Figure 5.9 – Numerical simulation including constant displacement between coupled points ............... 5.9
Figure 5.10 – Strut-and-tie representation of beam-column nodes under moment loading .................. 5.10
Figure 5.11 – Lateral shear loading system .................................................................................... 5.10
Figure 5.12 – Base supporting system with force retention beams ................................................... 5.11
Figure 5.13 – Hinged force retention system ................................................................................. 5.12
Figure 5.14 – Out-of-plane bracing system .................................................................................... 5.13
Figure 5.15 – Loading displacement histories ................................................................................ 5.14
Figure 5.16 – Schematic overview of Phase 1 and 2 monitoring features .......................................... 5.15
Figure 5.17 – Phase 3 test setup.................................................................................................... 5.16
Figure 5.18 – Schematic overview of Phase 3 monitoring features and overall dimensions ................ 5.18
Figure 5.19 – Equilibrium model .................................................................................................. 5.19
Figure 5.20 – SP_M01 force-drift results ...................................................................................... 5.20
Figure 5.21 – SP_M01 Beam shear: comparison of experimental and theoretical estimates ............... 5.21
Figure 5.22 – SP_M01 after testing .............................................................................................. 5.21
Figure 5.23 – SP_M01 Beam cracking pattern ............................................................................... 5.22
Figure 5.24 – SP_M01 Column cracking patterns .......................................................................... 5.22
Figure 5.25 – SP_M01 Beam-Column node interface ..................................................................... 5.23
Figure 5.31 – SP_M02 Beam-column interface .............................................................................. 5.27
Figure 5.34 – SP_M03 Beam after testing ..................................................................................... 5.29
x
Figure 5.42 – Strength results for Phase 2 specimens ..................................................................... 5.36
Figure 5.43 – SP_PC02A force-drift results................................................................................... 5.36
Figure 5.44 – SP_PC02A Beam shear: comparison of experimental and theoretical estimates............ 5.37
Figure 5.45 – SP_PC02A at the final stage of the test ..................................................................... 5.37
Figure 5.46 – SP_PC02A Beam cracking pattern ........................................................................... 5.38
Figure 5.47 – SP_PC02A Column cracking patterns ...................................................................... 5.39
Figure 5.48 – SP_PC02A North precast joint opening progression .................................................. 5.39
Figure 5.49 – SP_PC02A progression into failure .......................................................................... 5.40
Figure 5.50 – SP_PC02B force-drift results ................................................................................... 5.41
Figure 5.51 – SP_PC02B Beam shear: comparison of experimental and theoretical estimates ............ 5.41
Figure 5.52 – SP_PC02B after testing ........................................................................................... 5.42
Figure 5.53 – SP_PC02B Beam cracking pattern ........................................................................... 5.42
Figure 5.54 – SP_PC02B Column cracking patterns ....................................................................... 5.43
Figure 5.55 – SP_PC02B: Progression of north precast joint opening .............................................. 5.44
Figure 5.56 – SP_PC02C force-drift results ................................................................................... 5.44
Figure 5.57 – SP_PC02C Beam shear: comparison of experimental and theoretical estimates ............ 5.45
Figure 5.58 – SP_PC02C after testing ........................................................................................... 5.45
Figure 5.59 – SP_PC02C Beam cracking pattern ........................................................................... 5.46
Figure 5.60 – SP_PC02C Column cracking patterns ....................................................................... 5.47
Figure 5.61 – SP_PC02C precast joint opening progression ............................................................ 5.48
Figure 5.62 – SP_M02C force-drift results .................................................................................... 5.48
Figure 5.63 – SP_M02C Beam shear: comparison of experimental and theoretical estimates ............. 5.49
Figure 5.64 – SP_M02C after testing ............................................................................................ 5.49
Figure 5.65 – SP_M02C Beam cracking pattern ............................................................................ 5.50
Figure 5.66 – SP_M02C South beam-column joint crack width evolution ........................................ 5.50
Figure 5.67 – Procedure for definition of equivalent bilinear systems .............................................. 5.52
Figure 5.68 – Performance level comparison between Phase 1 and Phase 2 specimens ...................... 5.53
Figure 5.69 – Cumulative ductility at the collapse threshold (for the positive loading direction) and
yielding drifts (for both directions) for Phase 1 specimens. ........................................ 5.54
Figure 5.70 – Cumulative ductility at the collapse threshold (for the positive loading direction) and
yielding drifts (for both directions) for Phase 2 specimens. ........................................ 5.55
Figure 5.71 – Beam shear vs. Cumulative ductility for Phase 1 specimens ....................................... 5.56
Figure 5.72 – Beam shear vs. Cumulative ductility for Phase 2 specimens ....................................... 5.58
Figure 5.73 – Stiffness degradation vs. Cumulative ductility for Phase 1 specimens.......................... 5.60
xi
Figure 5.74 – Stiffness degradation vs. Cumulative ductility for Phase 2 specimens.......................... 5.61
Figure 5.75 – Energy dissipation vs. Loading half-cycles for Phase 1 specimens .............................. 5.62
Figure 5.76 – Energy dissipation vs. Loading half-cycles for Phase 2 specimens .............................. 5.63
Figure 5.77 – Illustration of the parameters used for calculating the ratios used in this work for
comparison of beam-column deformations ............................................................... 5.64
Figure 5.78 – R1 deformation ratio vs. drift evolution for Phase 1 specimens.................................... 5.65
Figure 5.82 – Moment – curvature analysis results for Phase 3 specimens........................................ 5.67
Figure 5.83 – SP_F01 force-drift results ........................................................................................ 5.68
Figure 5.84 – SP_F01: Evolution of column cracking pattern.......................................................... 5.69
Figure 5.85 – SP_F01: South side onset of spalling, buckling and fracture ....................................... 5.69
Figure 5.87 – SP_F02: Evolution of column cracking pattern .......................................................... 5.70
Figure 5.88 – SP_F02: North side onset of spalling, buckling and fracture ....................................... 5.71
Figure 5.90 – SP_F03: Evolution column cracking pattern ............................................................. 5.72
Figure 5.91 – SP_F03: North side spalling, buckling and fracture ................................................... 5.73
Figure 5.92 – Performance level comparison between Phase 3 specimens ........................................ 5.74
Figure 5.93 – Cumulative ductility at the collapse threshold for the positive loading direction ........... 5.75
Figure 5.94 – Stiffness Degradation vs. Cumulative Ductility ......................................................... 5.76
Figure 5.95 – Energy Dissipation vs. Loading Half-Cycles ............................................................. 5.77
Figure 5.96 – SP_F01 lateral displacement profiles ........................................................................ 5.78
Figure 5.97 – SP_F01 to precast specimens lateral displacements ratio ............................................ 5.79
Figure 5.98 – Mean curvature profile ............................................................................................ 5.80
Figure 6.1 – Geometric definition for numerical modeling ................................................................ 6.3
Figure 6.2 – Adopted generic layout for the FEM mesh .................................................................... 6.4
Figure 6.3 – Outer finite element layer and ductility demand ............................................................ 6.5
Figure 6.4 – Joint element in Cast3m (Pegon), Costa (2009)) ............................................................ 6.6
Figure 6.5 – FEM rebar meshes for Phase 1 and Phase 2 analyses ..................................................... 6.7
Figure 6.6 – Binding strategy at the precast joints ............................................................................ 6.8
Figure 6.7 – Geometric 3D envelope definition for columns. ............................................................ 6.9
Figure 6.8 – Phase 3 FEM mesh ..................................................................................................... 6.9
Figure 6.9 – Steel reinforcement finite element mesh ..................................................................... 6.10
Figure 6.10 – Concrete models Gauss point monotonic response ..................................................... 6.13
Figure 6.11 – Concrete models Gauss point cyclic response ............................................................ 6.14
Figure 6.12 – Example of DAMAGE_TC uniaxial cyclic behavior ................................................. 6.15
Figure 6.13 – Fracture energy in smeared crack based models ........................................................ 6.16
Figure 6.14 – Different domains for concrete behavior laws ........................................................... 6.17
xii
Figure 6.15 – Effectively confined concrete assumption ................................................................. 6.18
Figure 6.16 – Menegotto-Pinto uniaxial cyclic behavior ................................................................. 6.19
Figure 6.17 – Eligehausen bond model as implemented with ACIER_ANCRAGE in Cast3m............ 6.20
Figure 6.18 – JOINT_DILATANT model in Cast3m (Costa (2009)) ............................................... 6.22
Figure 6.19 – Methodology for determination of Kn ....................................................................... 6.24
Figure 6.20 – Numerical vs. Experimental Force – Drift results for monolithic specimens ................. 6.25
Figure 6.21 – Maps of compressive damage variable 𝑑 − maps on monolithic specimens ................. 6.26
Figure 6.22 – Principal compressions σ22 on monolithic specimens ............................................... 6.28
Figure 6.23 – Beam shear results for monolithic specimens: numerical vs. experimental ................... 6.28
Figure 6.24 – Maps of tensile damage variable 𝑑 + maps on monolithic specimens .......................... 6.29
Figure 6.25 – Principal strain ε11 maps over the deformed shape on monolithic specimens ................. 6.30
Figure 6.26 – Maps of longitudinal reinforcement stress ratios on monolithic specimens ................... 6.31
Figure 6.27 – Yielding drift for monolithic specimens: numerical vs. experimental .......................... 6.31
Figure 6.28 – Maps of transverse reinforcement stress ratios on monolithic specimens...................... 6.32
Figure 6.29 – Numerical vs. Experimental Force – Drift results for precast specimens ...................... 6.34
Figure 6.30 – Maps of compressive damage variable 𝑑 − maps on precast specimens ....................... 6.35
Figure 6.31 – Principal compressions σ22 on precast specimens ..................................................... 6.36
Figure 6.32 – Beam shear results for precast specimens: numerical vs. experimental ........................ 6.36
Figure 6.33 – Maps of tensile damage variable 𝑑 + maps on precast specimens ............................... 6.37
Figure 6.34 – Principal strain ε11 maps over the deformed shape on precast specimens ...................... 6.37
Figure 6.35 – Precast joint maximum opening width results: numerical vs. experimental .................. 6.38
Figure 6.36 – Maps of longitudinal reinforcement stress ratios on precast specimens ........................ 6.39
Figure 6.37 – Yielding drift for precast specimens: numerical vs. experimental ................................ 6.39
Figure 6.38 – Maps of transverse reinforcement stress ratios on precast specimens ........................... 6.40
Figure 6.39 – Numerical vs. Experimental Force – Drift results for SP_F01 ..................................... 6.41
Figure 6.40 – Numerical results representing typical column bending behavior at 3.50% drift ........... 6.42
Figure 6.41 – Numerical results for the tensile behavior of concrete at 3.50% drift ........................... 6.42
Figure 6.42 – Finite element mesh layout for full height pier numerical simulations. ........................ 6.44
Figure 6.43 – Force – drift results for full pier tests ........................................................................ 6.45
Figure 6.44 – σ22 compressions on SP_PC02A_full ....................................................................... 6.46
Figure 6.45 – Maximum beam shear value comparisons for full structure analyses ........................... 6.46
Figure 6.46 – Maps of damage variables for full structure analyses (at failure for SP_M02_full and at
3.50% drift for SP_PC02A_full) ............................................................................... 6.47
Figure 6.47 – Map of longitudinal rebar stress ratios from full height pier analyses .......................... 6.48
Figure 6.48 – Illustration of the procedure for obtaining scaled results from SP_PC02A ................... 6.49
Figure 6.49 – Force – drift results for reduced height pier scenarios ................................................ 6.50
Figure 6.50 – Map of longitudinal rebar stress ratios from reduced height pier analyses .................... 6.50
Figure 6.51 – Maps of damage variables for reduced height pier analyses ........................................ 6.51
Figure 7.1 – SelEQ ground motion selection output ......................................................................... 7.4
Figure 7.2 – Structural modeling scheme for a generic viaduct .......................................................... 7.5
xiii
Figure 7.3 – Sample pier calibration against experimental model results ............................................ 7.7
Figure 7.4 – Comparison of the moment – drift relationship calibration results ................................... 7.8
Figure 7.5 – Comparison of the Energy dissipation per cycle results .................................................. 7.9
Figure 7.6 – Lateral stiffness ratio between method B and A ........................................................... 7.10
Figure 7.7 – First lateral vibration mode for the case study viaducts ................................................ 7.11
Figure 7.8 – Ductility demand on the piers of the case study viaducts .............................................. 7.12
Figure 7.9 – Lateral deflection evaluation procedure. ..................................................................... 7.15
Figure 7.10 – Procedure for running safety assessment using SI (Luo (2005)). ................................. 7.17
Figure 7.11 – Example results obtained for the Macheda Viaduct with Precast behavior. .................. 7.18
Figure 7.12 – Fragility curves for DM#1 – Structural Performance results ....................................... 7.19
Figure 7.13 – Fragility curves for DM#2 – Lateral Deflection results .............................................. 7.20
Figure 7.14 – Fragility curves for DM#3 – Derailment Conditions results ........................................ 7.21
xiv
LIST OF TABLES
Table 2.1 - Limits for the maximum transverse deformation (EN1990-1, +A1:2005 (2002)) ............ 2.11
Table 2.2 – Seismic condition displacement limits on the Japanese RTRI Standard (RTRI (2007b)) ... 2.12
Table 2.3 – Seismic Importance factors ......................................................................................... 2.14
Table 2.4 – Reference PGA values for Portugal (NP EN1998-1 (2010)) .......................................... 2.14
Table 3.1 - Comparison between Reinforced Concrete and Hybrid systems (adapted from Hieber et al.
(2005)) ....................................................................................................................................... 3.26
Table 4.1 – Poceirão – Caia structures overview .............................................................................. 4.3
Table 4.2 – Structures with double column bent-type piers ............................................................... 4.3
Table 4.3 – Poceirão-Caia Response Spectra Parameters ................................................................ 4.16
Table 4.4 – Main characteristics of viaducts .................................................................................. 4.19
Table 4.5 – Node Constraints between pier-head and deck nodes .................................................... 4.20
Table 4.6 – HSRL Bridge dead-loads ............................................................................................ 4.20
Table 4.7 – Element properties ..................................................................................................... 4.21
Table 4.8 – Reference design parameters for test models ................................................................ 4.30
Table 4.9 – Precast element geometric characteristics..................................................................... 4.33
Table 4.10 – Precast systems overview ......................................................................................... 4.43
Table 4.11 – Cauchy’s similitude relationships for a 1:4 reduced scale ............................................ 4.44
Table 4.12 – General properties for the full experimental campaign................................................. 4.45
Table 5.1 – Loading displacement histories characterization – Phase 1 and 2 ................................... 5.13
Table 5.2 – Loading time-history characterization – Phase 3 ........................................................... 5.17
Table 5.3 – Peak Beam Shear results for Phase 1 specimens ........................................................... 5.56
Table 5.4 – Peak Beam Shear results for Phase 2 specimens ........................................................... 5.57
Table 6.1 – Confinement factors ................................................................................................... 6.18
Table 7.1 – Scaling factors for the Ground Motion records for all viaducts......................................... 7.4
Table 7.2 – EDP limits for DM#1 ................................................................................................. 7.13
Table 7.3 – EDP limits for DM#2 (according to RTRI (2007) ) ....................................................... 7.15
Table 7.4 – EDP limits for DM#3 ................................................................................................. 7.16
xv
LIST OF MAIN SYMBOLS
Ab Cross-section area of beams;

Ac Cross-section area of columns;
Asl Longitudinal reinforcement;
Ast Transverse reinforcement;
𝛼𝑔𝑅 Reference peak ground motion acceleration;
αs Shear span-to-depth ratio;
C Stiffness coefficient;
𝜉 Critical damping ratio;
𝑑− Concrete compressive damage variable;
+
𝑑 Concrete tensile damage variable;
dy Yielding displacement;
∆𝑑𝑖 Displacement increment for post-yielding half-cycle i;
∆𝜇𝑖 Ductility increment for post-yielding half-cycle i;
δj Normal joint displacement;
E Young Modulus;
Ecm Mean concrete Young Modulus;
ε11 Principal tensile strains on Plane stress analyses;
fcc Concrete confined strength;
fcm Mean concrete compressive strength;
fuc Concrete unconfined strength;
F Lateral force;
Fp Peak lateral force;
ϕ Rebar diameter;
𝑔𝑓 , 𝐺𝑓 Fracture energy;
𝛾𝐼 Importance factor;
γj Tangential joint displacement;
ℎ Cross-section height;
I Flexural moment of inertia;
J Torsional moment of inertia;
K Secant stiffness;
Ky Yielding stiffness;
Kn Normal joint stiffness;
Ks Shear joint stiffness;
Lh Plastic hinge length;
xvi
lw Wall length;
M Moment;
MEd Design moment;
𝑀𝑠 Surface wave magnitude;
N Axial load;
ν Axial load ratio;
𝜃 Slope angle of the compressive struts of shear in concrete;
PDL Probability of exceedence of the damage limitation criterion;
PNCR Probability of exceedence of the non collapse criterion;
φc Curvature on columns;
φb Curvature on beams;
φcy Yielding curvature on columns;
φby Yielding curvature on beams;
q Behavior factor;
Rv Vertical reaction at a column base load cell;
ρ Longitudinal reinforcement ratio;
s Concrete-steel bond slip;
Sa Spectral acceleration;
Sv Spectral pseudo-velocity;
σ22 Principal compression stresses on Plane stress analyses;
σn Normal joint stress;
σn0 +
Tensile normal joint stress threshold;
𝜎𝑠𝑙 Longitudinal reinforcement stress;
𝜎𝑠𝑙𝑦 Longitudinal reinforcement yield strength;
𝜎𝑠𝑙𝑢 Longitudinal reinforcement ultimate strength;
𝜎𝑠𝑡 Transverse reinforcement stress;
𝜎𝑠𝑡𝑦 Transverse reinforcement yield strength;
𝜎𝑠𝑡𝑢 Transverse reinforcement ultimate strength;
tw Wall thickness;
T Vibration period;
T1 Fundamental vibration period of a structure;
TDL Return period of the seismic load for the damage limitation criterion;
TNCR Return period of the design seismic load for the non collapse criterion;
τb Concrete-steel bond stress;
τs Tangential joint stress;
𝜇 Displacement ductility;
xvii
𝜇𝑐𝑢𝑚 Cumulative ductility;
VEd Design shear;
Vc Shear capacity component carried by concrete struts;
Vsd Shear capacity component carried by main reinforcement;
Vst Shear capacity component carried by transverse reinforcement;
VRd Design shear strength;
xviii
LIST OF ABBREVIATIONS
HSRL High Speed Railway Lines;

RC Reinforced Concrete;
FEUP Faculty of Engineering of University of Porto;
LESE Laboratory of Earthquake and Structural Engineering;
SIPAV Soluções Inovadoras Pré-Fabricadas para Vias Férreas de Alta Velocidade;
MEBEP Mota Engil – Betões e Prefabricados;
RAV Rede de Alta Velocidade;
LUD Lock-up Devices;
RTRI Railway Technical Research Institute;
ULS Ultimate Limit States;
SLS Service Limit States;
SI Spectral Intensity;
PGA Peak Ground Acceleration;
EQ Earthquake;
CIP Cast-in-place;
PRESS PREcast Seismic Structural System;
DFRCC Ductile Fiber Reinforced Cement Composite;
STU Shock Transmission Unit;
FBD Force Based Design;
DBD Displacement Based Design;
FEM Finite Element Method;
WC Wall-pier column segment;
WP Wall-pier panel segment;
LVDT Linear Displacement Voltage Transducers;
DWT Draw Wire Transducers;
LC Load Cell;
PL Performance Level;
IM Intensity Measure;
EDP Engineering Demand Parameter;
DM Damage Measure;
DV Decision making Variables;
PBEE Performance Based Earthquake Engineering;
IMK Modified Ibarra-Medina-Krawinkler;
xix
Introduction
1
INTRODUCTION
1.1. BACKGROUND OVERVIEW
The permanent technological evolution supporting man’s standing against the multiple
challenges posed by history has truly been promoting globalization in our world. Individual
social scopes are broadening, people are changing how they think regarding family, business
and career management, and the economical background of the XXI century has been growing
support on the concept of multinational companies and populations.
One of the main reasons behind that evolution has been the refinement of the international
travelling and transportation routes, nowadays enabling the establishment of efficient
connections that were, in the past, either too costly or time-consuming. On that context, the
importance of Railway and, particularly, High Speed Railway Lines (HSRL) is duly noted,
benefitting from high technological advances and an increase in territorial coverage which
confirmed them as a clear alternative to more common means of transportation, presenting an
adequate balance between costs and travel time while usually associated with significant levels
of comfort. Unsurprisingly, it can be seen that more and more countries are improving their
railway networks, with more than a dozen now promoting dedicated high speed lines.
From a technological perspective, railway lines are usually quite long, developing over
hundreds of kilometers and presenting difficult morphological challenges that are often
addressed with the construction of bridges and viaducts. In that regard, since the control of high
initial investments is usually a common concern for the railway promoters, using optimized
structural layouts and construction methods can strongly improve the economic viability of the
whole project. Therefore, Reinforced Concrete (RC) precast techniques can prove to be a good
1.1
Introduction
option, providing economic solutions and reduced construction time periods, while still
guaranteeing adequate performance under safety and service conditions.
In the specific case of HSRL, there are multiple examples of structures built using full or
partial precast solutions for bridge decks. One can observe cases with “U-shaped” beam
elements (e.g. Spain), cases where the decks are supported by multiple prestressed beams (e.g.
France), or even where full-span box girders are used (e.g. Italy, Taiwan), thus highlighting
precast solutions applicable to a wide array of bridge layouts. On the other hand, precast
elements are rarely considered for the construction of full railway bridge piers, particularly in
HSRL. In fact, apart from a few cases in Belgium and Netherlands where the bridge
substructure was designed using an unconventional layout where some precast elements were
included, the generalized application of that technology for bridge piers still seems to be a
non-reality, at least to the extent that was possible to investigate. There are a number of reasons
that can help on explaining that circumstance but still, in the author’s opinion, extending the
general use of precast solutions to bridge piers, as well as to decks, could potentially lead to
substantial economic advantages and increased production capacity of construction companies,
provided the challenge is technically viable.
From a purely technical perspective, structural solutions for railway bridge and viaduct
piers are designed for compliance against very strict performance criteria, particularly
considering applications on high speed lines. Train nosing, traction and braking forces,
horizontal deflection limits and, additionally, seismic loads are some of the more impactful
aspects of railway bridge substructure design. Moreover, although the seismic performance of
bridge piers is a well documented area involving numerous experimental and numerical works
performed by scientific community members in the latest decades, the specific context of
railway bridges and, particularly, those designed for high-speed trains is not yet as thoroughly
addressed. Also, this fact is further enhanced when discussing the possibility of adopting precast
applications, raising other concerns related with durability and on-site workability issues that
are important for the design of precast RC structures.
In this framework, the present thesis aims at studying, discussing and providing relevant
contribution to general knowledge on that area and, hopefully, to foster further discussion of the
above mentioned subjects.
1.2
Introduction
1.2. MOTIVATION AND OBJECTIVES
The main motivation for this work originated from a research project carried out at the
Faculty of Engineering of University of Porto (FEUP) between 2009 and 2013, named SIPAV –
Soluções Inovadoras Pré-Fabricadas para Vias Férreas de Alta Velocidade (Innovative
Precast Solutions for High Speed Lines), where the author actively participated. The project
aimed to study the application of precast techniques on the context of High Speed Railway
Lines, and to evaluate the potential for technical innovation regarding new and/or existing
improved design solutions. With that in mind, FEUP and the external construction contractor
MAPREL/MEBEP (Mota Engil – Betões e Prefabricados, now a part of Mota Engil –
Engenharia e Construção) joined efforts hoping to profit from the technical knowledge and
analytical capabilities of the former, together with the practical insight of the latter concerning
common practices and design misconceptions. This joint endeavor engaged two worlds that take
significant benefit in learning from one another, since academic studies ultimately aim to
provide practical solutions to the real world requirements. Therefore, the whole project was
highly appealing to the author, as he firmly believes that the development of new skills and
knowledge in this area is greatly enhanced by industry interest, and that its technological
outlook should be a main driving force for related research objectives.
Part of the work presented in this thesis was initially developed for the SIPAV project,
from which some guidelines were established. Considering the main area of expertise of
MAPREL/MEBEP, only RC construction was addressed; in addition, fully precast pier
elements were sought instead of partial precast applications. Furthermore, some decisions taken
during the development of the research project, which affected the general outline of this thesis,
were also influenced by political and economic objectives at that time, strongly supporting
HSRL implementation in Portugal starting with the Poceirão-Caia line portion (whose
construction was initiated but halted at mid-course). Nonetheless, owing to the complexity of
the different thematic areas likely to influence the technical viability of a precast application for
HSRL bridges, such as structural behavior (e.g. strength, ductility and durability), building
process (e.g. joint connections execution, on-site element handling and accessibility) and also
logistics (e.g. formwork reuse capability, precast tables setup, transportation, storage capacity),
the approach followed in the present work narrowed the study mostly to the issues related with
structural behavior and, particularly, those on the seismic performance framework.
Therefore, this thesis, as expressed by its title “Seismic Behavior of Precast Piers on High
Speed Railway Bridges”, addresses the main objective of promoting the study and possible use
of precast solutions designed for railway bridge structures, particularly focusing on the piers, in
1.3
Introduction
areas of moderate to high seismicity, where concerns regarding stable mechanisms of

earthquake energy dissipation and ductile deformation capacity are of critical importance. In
that context, the precast applications studied within this work aim to provide good seismic
performance, accounting for the difficulty of ensuring ductile behavior with potentially large
inelastic incursions, particularly in the precast joints. In order to assess different design options
for precast systems and to identify their most significant strengths and weaknesses relative to
equivalent monolithic construction, the elaboration of an experimental campaign capable of
providing such information was also a significant objective of this work.
Finally, while the local demand and global response of specific precast system components
can be adequately assessed from experimental testing, the influence of such findings on the
performance of a real bridge is still one of the most important aspects to consider when
full-scale application is desired. Therefore, the complementary link between those two issues
(system/component analysis and full bridge behavior) constitutes the final objective of this
work, intended to be addressed resorting to experimentally calibrated numerical applications,
where the performance of the pier systems is tested on both a local and global perspective.
1.3. THESIS OUTLINE
The organization of the present document aims to reflect the strategy adopted to
accomplish the proposed objectives. For that purpose, it should be acknowledged that while
each chapter tackles fundamentally distinct parts of the work, a common guideline is still
followed throughout, which can roughly be linked to the thought process for designing,
analyzing and validating the structural solutions herein presented. Thus, the thesis is comprised
of eight chapters, which are presented as follows:
▪ Chapter 1, of which this section is a part of, essentially aims at briefly describing
the framework of the thesis, its general outline as well as its objectives;
▪ Chapter 2 presents the first stage of the work, mostly focusing on understanding
the design challenges that HSRL piers are required to overcome. For that purpose,
railway specific loading and design criteria are briefly discussed, particularly
addressing aspects that impact the pier design. Furthermore, pier layouts
commonly used in HSRL structures are also reviewed, aiming at the definition of a
“typical use” profile by taking into account the geometric and structural
characteristics of the bridges and viaducts they were designed for;
1.4
Introduction
▪ Chapter 3 consists of a state-of-art review of the main precast solutions for RC

columns, with special emphasis on bridge piers. Aspects related with the
manufacturing process, transportation, handling and joint connections are also
herein discussed. Additionally, some of the most recent design strategies for
improving the seismic performance of bridge piers on high seismic demand
situations are also discussed in order to evaluate their potential for application with
precast systems;
▪ Chapter 4 aims to describe the studies performed to define the conceptual solution
proposed for experimental testing. The RAV1 Poceirão-Caia design proposal is
presented as the basis for this work, taking into account its importance in the
development on the SIPAV project. Also, the design and detailing strategy for the
structural solution of the proposed pier concept is object of several analysis and
discussions, including a numerical study made for a selection of viaducts from the
previously mentioned Poceirão-Caia design project, where seismic response
parameters are estimated for the test specimen design.
▪ Chapter 5 encloses all the activities related with the experimental campaign carried
out within this work. Thus, the process of designing and testing the laboratory
setup is thoroughly described, and experimental observations for all the tests are
also presented. Moreover, the obtained results are discussed and compared for a
variety of different response parameters, aiming at further characterizing
differences between monolithic and precast models.
▪ Chapter 6 presents the first stage of numerical applications performed on this work,
essentially aiming to use the experimental data for calibration of numerical models
suitable for exploring and studying further non-tested scenarios. For that purpose, a
refined FEM modeling strategy was prepared for simulation and validation against
a selection of pier tests, which was then used to simulate similar applications with
different constraints and geometric characteristics, aiming at representing the
expected demand range defined for the pier systems.
▪ Chapter 7 describes the second stage of numerical applications, based on the global
simulation of a set of idealized case study viaducts for seismic performance
assessment regarding several damage measures. The experimental data was used to
1
RAV stands for “Rede de Alta Velocidade” (in portuguese), which translates into “High Speed Network”.
1.5
Introduction
calibrate the behavior of piers on each structure, according to monolithic and

precast results, and incremental dynamic analysis were performed for the
calculation of fragility curves, which are the statistical representation of the
probability of exceedence of a given damage or performance state;
▪ Chapter 8 concludes this thesis and, as such, describes the main conclusions
extracted from multiple areas of the present document, as well as the future
development proposals relevant to the addressed frameworks. Due to its
preponderance on the work development, the relative performance of precast and
monolithic systems is given particular attention;
1.6
Piers for HSRL Bridges and Viaducts
2
PIERS FOR HSRL BRIDGES
AND VIADUCTS
2.1. INTRODUCTION
The main objective established for the present work, integrated in the previously mentioned
SIPAV research project, involves three thematically different areas addressing the following
topics: railway bridge design, pier seismic performance and precast mechanisms. Moreover,
while its focus is directed to the seismic performance of bridge piers, it is undeniable that each
of those frameworks can present varied design challenges that need to be taken into account
when discussing a precast application for HSRL structures.
In that regard, a global overview of existing railway bridges and viaducts enables the
following conclusion: the use of precast elements in the construction of pier structures is
limited, especially in the context of HSRL. Some cases can be observed where precast pieces
were used as casting forms for the footing and deck connections, or where unconventional
design options were adopted to allow industrialized construction procedures, but virtually no
application could be identified where the main body of the pier structure was fully precast.
In a way, that circumstance highlights the innovative nature associated with the SIPAV
project, considering that a precast construction is sought on the framework of structures that are
usually built with in situ methods. More importantly, the knowledge of the common layout
solutions used for those situations can help on guiding the design for precast, because similar
demands can be expected assuming that the general behavior patterns of the structure can be
maintained for the present application. Additionally, there are important railway bridge
performance requirements for compliance with safety and service conditions that are not
2.1
considered for other structure types (namely motorway bridges), which have a relevant impact
in structure design and, therefore, provide valuable insight for the current discussion.
With that in mind, this chapter essentially aims to review the most common railway bridge
design options and associated performance criteria, by observing typical application cases
which may help on understanding some of the main design challenges set for the present
endeavor.
2.2. COMMON RAILWAY BRIDGE LAYOUTS
In general, decisions regarding the construction of new bridges and viaducts address
multiple thematically distinct areas. Concerns related to politics and economics (initial
investment/maintenance costs, expropriations, economic potential of affected locations),
environmental issues (existence of water courses, impact on biological activity) and, naturally,
technical difficulties are some of the most relevant. In that regard, the applicable performance
requirements are heavily dependent of the specific use that the structures are to be designed for,
which, for bridges and viaducts, generally refers to motorway or railway purposes.
Moreover, while this often leads to concerns of similar nature, the different characteristics
of the travelling stock of each type of traffic lead to distinct design challenges. For example, on
railways the trains move through a rigorously defined path whose positioning is strictly set by
the rail track, and the train operators solely control the longitudinal movement issues, such as
speed and direction. Additionally, the physical dimension of trains is larger than road vehicles
(AREMA (2003)). By contrast, on motorways there is more freedom of transverse movements
associated with the steering capacity of vehicles, which also have generally higher ratios of
power per mass unit.
As a consequence, railway bridge design is more rigorously defined by track related safety
and operational issues than motorway structures, where structural performance is often the
critical factor. For example, a common characteristic of some railway lines is the use of
continuous welded rail (particularly on HSRL); in that context, the rail stress levels must be
controlled, especially under traction and braking forces as well as thermal loading, to prevent
the occurrence of instability phenomena such as rail buckling, which is illustrated in Figure 2.1.
2.2
Figure 2.1 – Rail buckling (from http://www.railtemperature.com/)
The potential for the occurrence of track instability can force the use of rail expansion
devices, in order to reduce the stresses produced due to longitudinal bridge movement, but that
can be an undesired design decision since it may end up reducing overall track durability
(Figueiredo et al. (2009)). Moreover, the increasing design speed for railway lines brought to
light severe dynamic performance problems, related to deck accelerations capable of causing
loss of wheel-rail contact or track side-resistance (Zacher and Baeßler (2005)), which also
require rigorous structure stiffness and displacement control.
In this context, it is understandable that railway bridge design options require adequate
structural layouts capable of providing generally low displacements. Concerning the
longitudinal direction, simply supported deck spans or continuous decks are often some of the
most common options. The former may ensure track stability without the need of track
expansion devices, benefitting line durability and is also simpler to evaluate. However, it may
also present a significant limitation to the structural stiffness contribution provided by the
bridge deck for longitudinal movement, therefore increasing the need for strong piers capable of
controlling horizontal deformations. It is usually observed in low rise bridges with short spans
between 25m to 35m, where decks can provide sufficient stiffness to vertical deflections while
horizontal movement is largely controlled by the columns.
By contrast, continuous span decks can ensure higher overall stiffness and generally
perform better under dynamic loading. When used with rail expansion devices, the length of
continuous structures can go up to 1200m (Manterola and Cutillas (2004)), otherwise a
maximum value of 90m should be respected for concrete structures. This type of deck is
frequently seen with longer spans and in high rise bridges, often designed to cross deep valleys.
Therefore, the longer spans are often a considerable challenge for design, requiring the use of
advanced construction technologies such as the balanced cantilever method.
2.3
Other options include design of an association of smaller multi-span continuous decks,

where the length of each segment can be adjusted according to the needs of any particular
project. That means this approach can avoid using rail expansion devices by having each deck
part with less than 90m of length and separated by a structural joint or, alternatively, single span
frames, as illustrated in Figure 2.2 - a). However, just like in the case of simply supported spans
(and for the same reasons), it also means that each continuous deck segment should present
sufficient longitudinal stiffness; when that is not possible, it is usual to provide a certain number
of higher stiffness fixation zones, where the structure can mobilize horizontal reactions. Those
zones can be simply constituted by columns that are designed for higher stiffness, or by using a
different support bearing scheme (possibly including some type of rigid shear key device).
Continuous Deck Neutral Frame Continuous Deck
a) Single neutral frame in between continuous decks
b) Continuous deck using LUDs (Marioni (2006))
Figure 2.2 – Different longitudinal static schemes for continuous span railway bridges
Additionally, the use of Lock-up Devices (LUD, as shown in Figure 2.2 – b)) is also a
possibility that enables a good compromise between performance due to creep, shrinkage and
thermal related movements, and dynamic loading. These devices are usually constituted by a
hollow cylinder with two chambers, filled with a viscous silicon compound and separated by a
movable piston (Figure 2.3). When the two anchor points of the device are displaced due to a
slow-velocity action such as thermal load or creep/shrinkage effects, the piston slowly moves
through the compound with very little friction, therefore generating low reactions, comparable
to sliding bearing behavior. However, when high-speed loads are applied such as train traction,
braking or earthquake loads, the compound is squeezed through the piston, generating high
friction that blocks movement between the two anchor points, therefore enabling fixed support
2.4
behavior. This is the main difference between LUDs and regular Viscous Dampers, as the
former provides considerably higher damping and stiffness.
Figure 2.3 – Lock-up Device: Schematic view (Taylor, Taylor Devices Inc.)
2.3. RAILWAY BRIDGE SEISMIC DESIGN
2.3.1. GENERAL REMARKS
In order to understand some of the most common seismic design concerns associated with
railway bridges and viaducts (particularly with the piers), a brief review of relevant performance
requirements is presented. For applications in Portugal, the Eurocodes are usually considered,
and their main railway specific recommendations can be found in Eurocode 0 (EC0 - EN1990-1,
+A1:2005 (2002)) and Eurocode 1 (EC1 - EN1991-2 (2003)), whereas general seismic design
guidelines are included in Eurocode 8 (EC8 - NP EN1998-1 (2010) and EN1998-2 (2005)).
For further guidance on the subject, other relevant literature can also be consulted, such as the
Japanese standard issued by the Railway Technical Research Institute (RTRI), for example
(RTRI (2007a) and RTRI (2007b)), which provide additional insight onto the seismic
performance problem.
The design principles considered for structure compliance regarding Ultimate Limit States
(ULS) are generally related to structural safety and users’ safety. The corresponding main
concern is to prevent collapse, which is defined for loss of equilibrium of the structure or any
part of the structure, loss of stability due to excessive deformations or member failure, and/or
failure due to time-dependent effects (e.g. fatigue). Regular practice involves the combination of
loads in such a way to obtain the most unfavorable effects for each structural element. Within
the context of Eurocode 0 and Eurocode 8, applicable ULS combinations are represented by
equation 2.1 for persistent action design and 2.2 for seismic action design.
𝐸𝑑 = ∑ 𝛾𝐺,𝑗 ∙ 𝐺𝑘,𝑗 " + "𝛾𝑃 ∙ 𝑃" + "𝛾𝑄,1 ∙ 𝑄𝑘,1 " + " ∑ 𝛾𝑄,𝑖 ∙ 𝜓0,𝑖 ∙ 𝑄𝑘,𝑖
2.1
𝑗≥1 𝑖>1
𝐸𝑑 = ∑ 𝐺𝑘,𝑗 " + "𝑃𝑘 " + "𝐴𝐸𝑑 " + " ∑ 𝜓2,𝑖 ∙ 𝑄𝑘,𝑖 " + "𝑄2
2.2
𝑗≥1 𝑖>1
2.5
where:
▪ 𝛾 - Partial safety factors for actions;
▪ 𝐺 - Permanent loads;
▪ 𝑃 - Prestress loads;
▪ 𝑄 - Variable loads;
▪ 𝑄2 – Long duration load effects;
▪ 𝐴𝐸𝑑 – Design seismic load;
▪ 𝜓 – Combination coefficient;
Furthermore, Service Limit States (SLS) are also considered to ensure that, for example,
deformation limits are compatible with normal structure use. Within that context, ULS design
tends to lead to strength capacity checks, implying that the effects of the actions are inferior to
structural capacity within a small margin of probability of exceedence for extreme occurrences
(as represented in equation 2.3), while SLS design generally involves explicit or implicit control
of structure displacements (and stress levels) under loading scenarios correspondent to higher
probability of occurrence.
𝐸𝑑 ≤ 𝑅𝑑 2.3
The condition of railway traffic, however, introduces a few performance requirements that
challenge the previous notion by involving several checks of structure deformations that aim to
provide safety to the circulation of trains. In that regard, those can effectively be considered as
ULS and not SLS (Goicolea (2007)), for the purpose of rail traffic safety checks, despite them
being introduced as such. Additionally, there are other requirements related with ensuring
passenger comfort, which are also relevant for SLS design. The following list includes most of
the applicable performance checks according to EC0/EC1, for which the associated limits tend
to be stricter for higher design speed values:
Performance checks related with running safety of trains:

▪ Vertical accelerations of the deck;
▪ Vertical deflection of the deck;
▪ Twist of the deck measured along the centre line of each track;
▪ Rotation of the deck extremities about a transverse axis;
▪ Vertical displacement between deck end sections;
2.6
▪ Longitudinal displacement of the upper surface deck extremities;
▪ Transverse deck deflection;
▪ Horizontal rotation of the deck about a vertical axis;
▪ First natural frequency of lateral vibration of the spans;
▪ Unrestrained uplift at the bearings;
Performance requirements relative to traveling comfort of passengers:

▪ Vertical accelerations inside the coach;
Among all the previous, most of the listed criteria influence the design in order to provide
control of not only the associated displacements, but also the rail stress levels, as previously
discussed. Moreover, the criteria that influence pier design are essentially those that suggest a
limitation of horizontal deformations, namely the relative longitudinal movement between upper
surface deck extremities and the horizontal rotation of decks about a vertical axis, which can be
linked to global transverse displacements of pier and deck. In this regard, the consideration of
large train traction and braking forces, nosing forces and centrifugal forces on curved viaduct
segments lead to some of the most relevant design challenges.
Within that context, it is understandable the increased stiffness of railway bridge and
viaduct design layouts relative to equivalent motorway structures. However, increased structural
stiffness can often induce larger seismic forces, involving a different major concern for bridges
and viaducts, particularly on areas prone to strong earthquake activity. In that regard, according
to the design philosophy of EC8, seismic performance must be assessed for two different
intensity levels, corresponding to the following requirements:
▪ No collapse
This demand level requires the structure to withstand the action of a design seismic
load while retaining structural integrity and some residual capacity, even if local or
global collapse occurs. The design seismic load is defined for a reference
probability of exceedence PNCR of 10% in 50 years, corresponding to a reference
return period TNCR of 475 years.
▪ Damage limitation
The damage limitation requirement represents the need to account for earthquake
events that happen more frequently than the design seismic action. For economic
reasons, the structure is expected to remain fully functional, with little or no display
2.7
of structural damage requiring immediate attention and repair. The seismic event
related to this criterion has a probability of exceedence PDL of 10% in 10 years,
corresponding to a reference return period TDL of 95 years.
One of the most relevant aspects of seismic design with interest to pier applications is that
the formation of flexural plastic hinges is allowed, as long as specific detailing rules are adopted
for provision of sufficient ductile deformation capacity. In addition, bridge decks are expected
to remain essentially elastic, with only local damage allowed in secondary components such as
expansion joints. Within this context, it is possible to understand that this methodology
incorporates a tradeoff between strength capacity and deformation capacity, since the piers can
be designed for a reduced strength demand but involving the accommodation of displacements
above than the elastic levels.
Therefore, it can casually be said that railway and seismic performance criteria are
associated with nearly opposite perspectives concerning pier design: the former leads to
increased stiffness while the latter accepts increased displacements. Finding the right balance
between the two is not an easy task, and it is made harder by the fact that there is no distinction
between running safety check limits proposed for regular railway travels and those that may be
applicable to seismic events (constituting a much more severe and rare occurrence).
In fact, other sources show that usually larger values are considered for rail traffic safety
during an earthquake event. The Japanese standards, for example, provide a more
straightforward integration of these issues, as they include a clear distinction between ordinary
railway travels and those under seismic loading conditions. Furthermore, a performance based
approach is adopted, where different objectives are established for varying levels of structure
response. For ordinary travel conditions, riding comfort (serviceability) and running safety
(safety) are checked, while for seismic conditions running safety is the main concern. An
additional restorability performance level is also introduced, associated with expectations of no
or minimal need for repair. Restorability is checked for rail traffic operating under both ordinary
and seismic conditions. The associated performance items are illustrated in Figure 2.4.
2.8
Figure 2.4 – Performance assessment strategy for railway bridge service, prescribed in
the Japanese RTRI standard (RTRI (2007b))
According to the previous, the set of requirements for railway performance within seismic
events’ framework, are slightly less strict when compared to those applicable for ordinary
operating conditions, which is a reasonable compromise. Likewise, according to Dutoit et al.
(2004), it is a usual practice to associate the displacement verifications and serviceability
requirements to a seismic event of reduced intensity, which is similar to how EC8 considers a
damage limitation requirement comparatively to the no collapse requirement. However,
eurocode’s provisions do not account for different performance targets according to ordinary
travel conditions or those under seismic loading, and single limit values are proposed for each
criteria.
Within this context, the following sub-sections aim to briefly present the seismic action as
considered for this work, in addition to the verification limits associated with the relevant
performance requirements for pier design, according to the previous discussion.
2.9
2.3.2. RAILWAY PERFORMANCE REQUIREMENTS
2.3.2.1. Longitudinal deformations
The main concerns associated with large longitudinal deck displacements are related to the
increased stresses accumulated in the rails due to thermal and variable loads which, according to
EC1, require careful assessment and limitation to 72 MPa in compression and 90 MPa in
tension. Therefore, the horizontal displacement δ2 between deck parts (or relative to the
abutments) due to traction or braking (Figure 2.5 – a)) is limited to:
▪ 5 mm for continuous welded rails without rail expansion devices or with a rail
expansion device at one end of the deck;
▪ 30 mm for rail expansion devices at both ends of the deck;
▪ More than 30 mm only if both expansion devices and ballast movement gaps
are considered;
As a reference, the values presented by Dutoit et al. (2004) on account of the design of
HSRL in Mediterranean France and Asia for continuous welded rail structures, without rail
expansion devices and under the moderate earthquake, were 20mm and 25mm, respectively.
Additionally, the horizontal movement between deck parts (or relative to the abutments) due to
vertical loading (Figure 2.5 – b)) is limited to:
▪ 8 mm when track-structure interaction is taken into account;
▪ 10 mm when track-structure interaction is ignored;
a) Limit due to traction and braking b) Limit due to vertical loading
Figure 2.5 – Illustration of horizontal movement limit (DGF (2007))
The previously presented conditions are a significant challenge to the longitudinal stiffness
of bridges and viaducts. Moreover, the limits are small enough so that the combined effect of at
least pier bending, foundation rotation and displacement, as well as bearing displacements must
2.10
be taken into account in the global stiffness, in order to obtain realistic results. For that reason,
track-structure interaction analyses are usually mandatory, because simplified procedures
generally lead to results that are too conservative (Dutoit (2007)).
2.3.2.2. Transverse deformations
With respect to transverse deformations, lateral deflection is checked to ensure an

appropriate track radius for rail traffic. The main performance checks imply the assessment of
the maximum horizontal rotation between deck parts (or between the deck and the abutments),
as well as the maximum change in curvature radius. EC0 suggests the following limits,
presented in Table 2.1.
Table 2.1 - Limits for the maximum transverse deformation (EN1990-1, +A1:2005 (2002))
δh – transverse deflection of the deck;

L – deck span length;
The values from the previous figure highlight a critical dependence of railway line design
speeds, as the performance limits for its highest values are more than two times as severe as for
speeds lower than 120 km/h. Therefore, the maximum horizontal rotation would be limited to
0.0015 rad for bridges and viaducts designed for train speeds greater than 200 km/h.
In this regard, the limits considered by Dutoit et al. (2004) for performance under the
moderate intensity earthquake for the design of the French Mediterranean and Asian HSRL
structures were slightly larger, 0.0030 rad for the former, and up to 0.0017 rad for the latter.
2.11
Additionally, the Japanese lateral deflection and corresponding angular rotation limits, for
running safety under seismic conditions, vary according to the type of track deformation
experienced (parallel shift or folding). For design speeds greater than 360 km/h, the minimum
angular rotation limit considered is 0.002 rad, according to Table 2.2, which is still larger than
the EC0 value. For the restorability performance level, even larger values of 0.006 rad or 0.008
rad are considered, for slab track or ballast track, respectively. Nonetheless, it should be noted
that these values are associated with the Japanese JIS 50N and JIS 60 specifications, and not the
UIC54 or UIC60 rail types that are usually observed in European HSRL.
Table 2.2 – Seismic condition displacement limits on the Japanese RTRI Standard (RTRI
(2007b))
a) Running safety performance level
b) Restorability performance level
In addition, maximum lateral vibration displacements are also checked for the seismic
condition, resorting to the Spectral Intensity (SI) calculated from the pseudo-velocity of an
equivalent period structure at the track level. The corresponding verification is made by
comparison against values of Figure 2.6, where the design SI (in mm) should, for the associated
structural period range, stand below the threshold represented by the blue line.
2.12
Figure 2.6 – Lateral vibration displacement limits (adapted from RTRI (2007b))
2.3.3. SEISMIC LOAD
In addition to the horizontal train loads, seismic loading is widely acknowledged as a

critical factor in bridge pier design. Regarding applications in Portugal, the Eurocode
requirements of no collapse and damage limitation are associated with the definition of two
seismic intensity levels, identified by reference peak ground acceleration values (PGA) αgR of
earthquake (EQ) events reflecting the local seismicity of the construction site under analysis.
Furthermore, EC8 suggests the consideration of two different sets of EQ characteristics for each
reference PGA, in order to reflect the influence of distinct seismic events regarding for example
magnitude and distance to epicenter:
▪ Type 1 EQ: High and moderate seismicity regions (𝑀𝑠 > 5.5);
▪ Type 2 EQ: Low seismicity regions (𝑀𝑠 ≤ 5.5) and near-field earthquakes;
Structures are also classified according to the importance factor γI , enabling the
characterization of seismic intensity levels that are different than the reference values, reflecting
the different evaluation of the importance of specific bridges, regarding consequences for
human life in case of failure, or “for maintaining communications, especially in the immediate
post-earthquake period, and on the economic consequences of collapse”. Three importance
classes are established for bridges, according to Table 2.3, which can be related to the
consequence classes defined in EN1990 (2002).
2.13
Table 2.3 – Seismic Importance factors

Importance class Importance factor 𝛾𝐼
I 0.85
II 1.00
III 1.30
The reference seismic PGA for either type of EQ and return period TNCR of 475 years can be
determined from the national zoning maps, which are a representation of the local seismic
hazard assumed to be constant within each zone, found in each country’s National Annex.
Figure 2.7 illustrates the seismic zoning maps for mainland Portugal, whereas the corresponding
reference PGA values can be obtained from Table 2.4. Those values can be adjusted for the
return period TNCR of 95 years correspondent to the damage limitation requirement by using the
recommended reduction factors ν of 0.40 and 0.55 for Type 1 and Type 2 EQ, respectively.
Zones Zones
Figure 2.7 – Seismic zoning map for Portugal (adapted from NP EN1998-1 (2010))
Table 2.4 – Reference PGA values for Portugal (NP EN1998-1 (2010))
Type 1 EQ Type 2 EQ
Seismic a gR Seismic a gR
2
Zone (m/s ) Zone (m/s2)
1.1 2.50 2.1 2.50
1.2 2.00 2.2 2.00
1.3 1.50 2.3 1.70
1.4 1.00 2.4 1.10
1.5 0.60 2.5 0.80
1.6 0.35 - -
2.14
Characterization of multi components for the seismic action according to different loading
directions should be provided in order to mobilize the capacity of structures that account for
significant differences between the longitudinal, transverse and vertical performances. However,
bridge pier design is usually not critically influenced by vertical seismic loading, which is why,
according to EC8, the effects of the vertical component should only be taken into account in
zones of high seismicity, when the piers are “subjected to high bending stresses due to vertical
permanent actions of the deck, or when the bridge is located within 5 km of an active
seismotectonic fault”. The EQ motions relative to both horizontal loading directions are
described by the elastic ground motion acceleration response spectra illustrated in Figure 2.8,
notwithstanding the fact that different spectra may actually be obtained according to distinct
combinations of site dependent parameters (such as soil type).
a) Type 1 EQ b) Type 2 EQ
Figure 2.8 – EC8 elastic acceleration response spectra (NP EN1998-1 (2010))
2.4. PIER TYPES ON RAILWAY BRIDGES
2.4.1. OVERVIEW
A detailed characterization of different pier types for a bridge substructure is presented in

the form of a flow chart in Figure 2.9.
2.15
Figure 2.9 – Flow chart for pier types in bridge structures (Carmichael and Desrosiers (2008))
An attentive review of many concrete railway bridge structures, however, shows that,
although piers can be found within a wide variety of shapes and forms, most of the actual
layouts can be included in one of three distinct categories:
▪ Single column;
▪ Wall-pier;
▪ Multiple column pier (with or without transverse connection, e.g. bent-type columns);
The main distinction between single columns and wall-piers (which may be solid or hollow)
reports to the cross-section dimensions of the elements and their respective thickness (tw) to
length (lw) ratios (Figure 2.10). This is a key issue for substructure design, because usually
single columns are expected to perform as mainly flexural elements, while wall-piers develop
important shear deformations that must be taken into account in the design.
lw
tw
Figure 2.10 – Representation of a wall cross-section geometry
2.16
According to Eurocode 2 (NP EN1992-1-1 (2010)), the following should be considered for
wall segments:
“...reinforced concrete walls with a length-to-thickness ratio of 4 or more...”
The latter is a broad characterization of walls as structural elements, and is not specific to
wall-piers. Caltrans (Caltrans (2000)) bridge design manual indicates some design principles
applicable to wall-piers with a clear height to length ratio higher than 2.0, while on the other
hand, both the ACI 318-14 (ACI 318 (2014)) and the International building code (International
Building Code (2009)) define a wall-pier as “a wall segment with a horizontal
length-to-thickness ratio of at least 2.5, but not exceeding 6, whose clear height is at least two
times its horizontal length.”. This last definition is clearer and in line with the geometrical
dispositions of actual bridge wall-piers. As such, this work adopts the ACI318 and IBC
𝑙𝑤
definition and the 𝑡𝑤
ratio of 2.50 for distinction between columns and wall-piers.
Multiple column pier layouts are simpler to describe, involving the use of more than one
vertical element to support the same bridge alignment with or without a transverse connection.
For all purposes, Caltrans bridge design manual relates the concept to that of bridge bents,
stating that “Bents are a bridge support system consisting of one or more columns supporting a
single cap” and is a suitable definition for the current purpose.
Considering the impact that several design speed performance criteria applicable to railway
structures have on pier stiffness, as previously discussed, it can be worthwhile to examine the
common layouts used within the framework of HSRL, aiming to understand some of its merits
and shortcomings. With that in mind, the following section presents a brief review of common
geometrical characteristics of viaduct structures associated with each of the three previous pier
layout categories, particularly total length, main span and pier height.
2.4.2. SINGLE COLUMNS
The use of single columns for supporting railway bridge decks is a very common practice.
Typical cross-sections are solid (shorter structures) or hollow (taller structures) in nature, with a
square, rectangular, circular or octagonal geometry, and their main advantage is granting similar
stiffness on both the main horizontal loading directions (longitudinal and transverse). This is
because most single columns (with no monolithic connection) perform like a vertical cantilever,
with considerably larger restrictions of rotation at the footings than at the deck connections.
On another note, the transverse dimension of bridge decks is often quite larger than the
piers, posing a challenge regarding the implementation of support bearings on top of the shorter
2.17
cross-section of the pier heads. For that reason, bridges with single column substructure systems
commonly adopt a flared form or even a distinguishable pier cap (e.g. a hammerhead column
for caps with larger dimensions). In fact, that helps to accommodate the rotations resulting from
deck loads, and to increase the available surface for positioning the support bearings, although
at the expense of potentially increased bending moments on the columns due to eccentric
positions of vertical reactions. This particular aspect is observed in many cases over different
countries, as shown in Figure 2.11 which illustrates a few located in Spain, Taiwan, France and
Japan.
a) Jalon Viaduct – Spain b) C270 Viaduct – Taiwan

(from http://en.structurae.de/) (from http://www.lusas.com/)
c) Avignon Viaduct – France d) Ishikawa Viaduct – Japan

(from http://www.rff.fr/) (from http://www.nishimatsu.co.jp/)
Figure 2.11 – Single column flare/cap on HSRL viaducts
The use of single column piers tends to be more prevalent in high rise structures, or those
with longer spans (typically beyond 30m to 35m), since these key parameters are generally
related and influenced by one another regarding design options. In the interest of economic
structures, the overall height of the deck should dictate the main span length, with longer spans
being adopted for taller bridges, in order to minimize the number of pier alignments. That is
because regular on site construction of tall columns often requires the use of special and costly
technologies, such as climbing formwork, when compared with regular solutions. On the other
2.18
hand, deck height directly impacts the stiffness contribution that can be expected from single
columns for longitudinal and transverse movements, but the main advantages of multi column
or wall pier alternatives also lose effectiveness in that regard. In this case, the use of hollow-
sections for single columns can be valuable, as they provide a higher ratio of stiffness per mass
unit and, generally, a cheaper structure.
On tall bridges of short length, the use of a continuous deck and the influence of the
abutments can be sufficient to control the deformation levels, while longer bridges need to
explore more creative solutions to ensure small displacements under service loads. Some
bridges in Germany incorporate inverted “V-shape” alignments designed to address this
problem (see Figure 2.12 a) and b)) because the longitudinal stiffness provided by these
elements is considerably higher. Another layout that incorporates a similar idea (although with
multi column piers) can also be found in the Gänsebach Viaduct (Figure 2.12 c)), where sets of
shorter “V-shape” piers along both the longitudinal and transverse directions grants additional
horizontal stiffness to the bridge. All these cases relate to long viaduct structures, with lengths
around 1000m.
a) Fulda Viaduct – Germany b) Pfieffe Viaduct – Germany

(from http://en.structurae.de/) (from http://en.structurae.de/)
c) Gänsebach Viaduct – Germany

(from http://cms.asce.org/)
Figure 2.12 – Special solutions for horizontal stiffness
2.19
On bridges with irregular height development, it can be beneficial to combine adequate

structural solutions for short and tall zones. The design strategy of the Vérnegues Viaduct, from
the French Mediterranean HSRL, which is presented in Figure 2.13, is a clear example of that.
As it is possible to observe, the bent-type frame solution adopted in the lower rise zones (a)) is
gradually replaced by a single column layout on taller zones (b)).
a) Lower rise zone b) Taller zone
Figure 2.13 – Pier layout variation in the HSRL Vernégues Viaduct in France
(from http://en.structurae.de/)
The number of railway tracks of the HSRL lines is another important detail for design
decisions regarding bridge pier layouts. In fact, most lines consider ongoing train traffic on both
ways, therefore it is common to see two track bridges with decks around 8 to 14 meters wide.
Single track structures are also a possibility, mainly when considering the construction of two
sideway bridges (one for each traffic direction), although that is not very commonly observed.
With that in mind and considering that the transverse dimension of single columns is generally
considerably smaller than the upper deck surface width, the possibility for strong torsion
moments can be an issue of concern. Therefore, it is not surprising that single columns are
typically used to support box girder decks, which are more suitable to provide a good
performance under such loading conditions, as well as optimal configurations to use with
advanced construction methods for spans longer than around 40 meters.
There are also multiple examples of single column supported bridges with composite decks
using strong steel girders and concrete slabs. The most common configuration is a plate girder
deck that uses two or more main steel girders, and transverse bracing with steel trusses or
precast concrete elements. As long as the transverse bracing is designed to account for force
transfer between the main girders, the behavior of this type of deck can also be similar to that of
a reinforced concrete (RC) box girder, while benefitting from a generally lighter structure. It is a
structural solution very common in France, and Figure 2.14 illustrates some of those cases. Just
2.20
like in RC box girders, the bottom width of the plate girder deck is larger than the single
columns, requiring the use of a pier cap to accommodate its support.
a) Orgon Viaduct – France b) Tech Viaduct – France

(from http://en.structurae.de/) (from http://www.ioa.fr/)
Figure 2.14 – Composite deck bridges
Single column piers were also observed supporting steel truss decks. This is a structure type
that is especially suited for cases where ground conditions advise a reduction in structure
weight, according to Millanes Mato (2004)). A typical construction is the “Warren truss”,
which was very common in the first Tokkaido Shinkansen Viaducts as illustrated in Figure
2.15 - a) (Konishi (2012)), where the track is located on the bottom of the truss. On the other
hand, Figure 2.15 b) is an example of a half-lenticular design, where the track is located on top
of the supporting steel truss.
a) Tokkaido Shinkansen Viaduct - Japan b) Viaduc de L’Arc - France

(from Konishi (2012)) (from http://en.structurae.de/)
Figure 2.15 – Composite deck bridges
Overall, the use of single columns in HSRL viaducts is common and associated with several
different structure designs. However, one of the defining traits of a single column seems to be
the use of a flare or pier cap, as most structures support two-track lines and there is the need to
2.21
provide support over the wider deck sizes. Single columns are also more prevalent in high rise
bridges and viaducts, that usually also correspond to longer span structures.
2.4.3. WALL-PIERS
As mentioned before, wall-piers are pier structures that have a transverse-to-longitudinal

dimension ratio higher than 2.5. Usually, the longitudinal thickness is between 1.0 to 2.0 meters,
while the transverse length can be as high as the supported deck widths. Due to this, wall-piers
have high lateral stiffness and are heavily influenced by shear, especially on low-rise structures,
which is a very common occurrence for wall-pier supported viaducts (deck heights between
10m to 25m), as illustrated in Figure 2.16.
a) Innerste Viaduct - Germany b) Padulicella Viaduct – Italy

(from http://en.structurae.de/) (from Calçada et al. (2008))
Figure 2.16 – Wall-piers' examples on low-rise viaducts
The higher width that wall-piers provide is a clear advantage of this type of structure, in
contrast to single columns, since their transverse length is well suited to accommodate support
bearings for wide box girder and multiple girder deck solutions. Some cases were identified,
however, where a longitudinal enlargement (see Figure 2.16 – b)) is considered, especially with
simply supported decks, because of the increased number of bearing devices needed for
supporting two different spans.
In that regard, the close relation between the transverse dimension of the decks and the
wall-pier bearing length (which are typically similar) is a common characteristic of the observed
bridges and viaducts. On multiple girder bridges, the bearing width corresponds to the full
transverse dimension of the deck, as seen, for example, in the “TGV East-Europe” and “TGV
Eastern” junction bridges, illustrated in Figure 2.17.
2.22
Figure 2.17 – TGV junction bridges – France (from http://en.structurae.de/)
In the previous example, each deck was built with seven precast concrete beams (PRAD
system, Vavel (2004)) tightly fit within the deck width, which is nearly the same as the wall-
pier length. Support is provided by four bearing devices placed along that same length, beneath
stiff beams that are cast on site on the transverse direction of the deck. This is a widely
acknowledged and common design strategy, and can be observed on multiple structures (even
regardless of HSRL application). Another example with a different girder deck can be observed
in Figure 2.18, where the bearing devices were placed directly beneath the main girders.
Figure 2.18 – Auxonne Viaduct – France (from http://en.structurae.de/)
The wall-pier layout is also observed to support other deck types, such as RC box girders
and composite plate girders. In these cases, a reduced number of bearing supports is typically
considered. Figure 2.19 illustrates some examples on French TGV HSRL viaducts where the
decks were built with steel girders including two clear support bearing zones.
2.23
a) Mosel Viaduct b) Mondragon-Vénéjan c) Ardre Viaduct

Viaduct
Figure 2.19 – Wall-piers in composite steel girder viaducts (from
http://en.structurae.de/)
Additionally, although most wall-piers are associated with low-rise structures, some cases
can also be observed in tall bridges and viaducts, which are typically related to box/plate girder
decks and longer spans. Moreover, the length-to-thickness ratio of these taller wall-piers is
usually small, often around the lower limits of 2.5 to 3.0, as evidenced in Figure 2.20.
To summarize, several characteristics of viaducts constructed using wall-pier layouts were

observed, namely regarding span length, deck type and deck height. For example, the deck of
the TGV junction bridges (French East-Europe HSRL) is constituted by several multiple girder
short spans of 20m-30m. On the other hand, the Meuse Viaduct, on the same railway line, is a
composite plate girder deck with a main span of around 50m, while the deck height of both
structures is around 10m. A logical conclusion is that the wall-pier layout is a very adaptable
design, although low-rise bridges are typically where the structural advantages of a wall-pier
seem to prevail.
2.24
a) Baüerbach Viaduct – Germany b) Glems Viaduct – Germany

c) O Eixo Viaduct – Spain d) Saubach Viaduct – Germany

(from http://horsost.blogs.upv.es/) (from http://www.vde8.de/)
e) Arroyo Espinazo Viaduct – Spain f) Ricardell Viaduct – Spain

(from http://www.ideam.es/) (from http://www.tucrail.be/)
Figure 2.20 – Box girder railway bridges supported by single column piers with similar width.
2.4.4. MULTIPLE COLUMN PIERS
One can identify mainly two types of multiple column layouts used for supporting railway
bridges and viaducts: those that focus on taking advantage of a frame-like behavior resorting to
some type of transverse connection for force transfer and displacement compatibility purposes,
and those that mostly focus on the behavior of each column as an individual unit. In that regard,
the former is the most commonly found one, typically considering the so-called bent-type
columns.
As far as it was possible to observe, multiple column piers are generally found in structures
whose characteristics are, in general, quite similar to those described before for wall-piers and,
2.25
particularly, in low-rise viaducts (around 10m to 20m high). In particular, bent-type piers seem
well-suited to accommodate the large support widths of multiple girder decks, as the cap beam
can be designed to the required bearing length. Just like in wall-pier structures, this usually
leads to the use of several support bearings, placed beneath the main girders, which are well
served by the extra space provided by the former. This can be observed, for example, in the
viaducts from French and Turkish HSRL that are illustrated in Figure 2.21, showing a strong
and wide cap beam, where a minimum of seven bearing devices for deck support is considered.
a) Thérouanne Viaduct – France b) Vandiéres Viaduct - France c) Viaduct 4 – Turkey

(from http://en.structurae.de/) (from http://en.structurae.de/) (from Millanes Mato and
Ortega Cornejo (2007))
Figure 2.21 – Bent-type columns with multiple girder bridges
When used with other deck layouts, such as steel plate girders or RC box girders, where the
transverse bearing length is shorter, the maximum effective length for the cap beams is also
reduced. In this case, it was possible to observe cases where the pier design included a tall cap
beam, in order to provide large stiffness under horizontal load (expecting significant shear
demand), and cases where the cap beam was mostly adopted for a displacement compatibility
function, focusing on providing strong column stiffness instead. An example of the former is the
Crould Viaduct, from the French TGV North HSRL, where the cap beam is constituted by a tall
element with short span, and is illustrated in Figure 2.22 - a). By contrast, Figure 2.22 – b)
illustrates a part of the chinese Danyang Kunshan Grand Bridge, where the transverse beam is
considerably more slender in relation to the corresponding vertical columns. Both of these
structures are associated with RC box girder decks, as observed, highlighting different
approaches to pier design.
2.26
a) Crould Viaduct – France b) Danyang Kunshan Bridge – China

(from http://en.structurae.de/) (from Zhao, E. - https://www.flickr.com)
Figure 2.22 – RC box girder decks supported by multiple column frames
Another case of a similar strategy to that observed in Figure 2.22 – b), where horizontal
transverse stiffness is almost exclusively dependent on the strong columns’ behavior and a small
transverse beam is included for displacement compatibility purposes, is the Anguera viaduct, in
the Spanish HSRL (Sobrino and Murcia (2007)), where a steel tubular element is used instead
of a reinforced concrete beam, but with similar design purposes.
a) Viaduct overview b) Pier alignment cross-section

(from http://www.pedelta.es/) (from Sobrino and Murcia (2007))
Figure 2.23 – Anguera Viaduct - Spain
Extreme cases of the strong column design strategy correspond to those where no cap beam
is considered, which are not as common, as previously mentioned. Figure 2.24 shows one such
example, the Viaduct over the Guadalete River, where columns have a skewed shape with
larger cross-sections near the footing and thinner cross-sections at the deck level.
2.27
Figure 2.24 – Viaduct over the Guadalete River– Spain (from Cutillas (2007))
The two previously presented viaducts both involve twin box girder decks, where the
position of each column is generally determined by the geometrical layout of the two girders,
for optimal support and avoidance of eccentric loading. On that regard, a different design for a
twin box girder deck is observed in the Stöbnitz Viaduct (Figure 2.25). In this case, the structure
is designed as a monolithic Vierendeel girder, with horizontal load transfer between the
superstructure, a concrete slab, concrete supports and pile caps, according to Schlaich (2012).
a) Viaduct overview b) Pier alignment cross-section

(from http://en.structurae.de/) (from Schlaich (2012))
Figure 2.25 – Stöbnitz Viaduct - Germany
Within the context of frame solutions for HSRL bridges, the classic Japanese rigid frame
should also be mentioned. It is a structural solution used since the first Shinkansen lines,
consisting of a series of monolithic frames with a rigid set of columns and stiffening beams, and
a slab serving as railway track support. As stated by Koyama (1997), that was considered the
most economic substructure layout also capable of exhibiting good seismic performance. The
overall bridge and viaduct behavior with this substructure layout depends on the arrangement of
the rigid frame units. The characteristics of each unit can vary, but they are usually less than
60m long and less than 20m tall. The longitudinal distribution of the vertical elements is often
around 10m, while for frames higher than 15m, stiffening cross beams are used, as illustrated in
Figure 2.26 (Tamai (2014)).
2.28
d = 5 to 15m; Standard is 10m 5 < h < 20m; Cross beams are desirable for h > 15m
a) Interval of columns b) Frame height

Figure 2.26 – Japanese Shinkansen rigid frame characteristics (adapted from Tamai
(2014))
Additionally, the longitudinal interaction between the rigid frame units can be designed in
different ways. The three main ones are, as indicated by Tamai (2014), the butt type connection,
the girder connection and the integral frame behavior, shown in Figure 2.27.
Figure 2.27 – Japanese Shinkansen frame unit connections (adapted from Tamai (2014))
This structural layout gives the Shinkansen structures a different visual aspect than most of
other HSRL bridges and viaducts, as the main span is generally quite smaller and the rigid
frames make the substructure seem more visually condensed than single columns, wall-piers or
bent-frames, as Figure 2.28 aims to illustrate.
2.29
a) Shinkansen Viaduct - Japan b) Grenette Viaduct – France

(from Tamai (2014)) (from http://en.structurae.de/)
c) Loire Viaduct – France d) Piacenza Viaduct – Italy

Figure 2.28 – Visual overview of different HSRL viaducts.
The tallest multiple column substructures were found in the French TGV Rhine-Rhone
HSRL, illustrated in Figure 2.29, where in some zones the piers are above 30m high. The pier
design from the Lizaine Viaduct (Figure 2.29 – a)) corresponds to a monolithic frame structure
with a large cap beam, in comparison to the thinner vertical elements. It is interesting to note,
however, that the cross sections of the vertical elements are wider in the longitudinal direction
and, therefore, provide higher stiffness for longitudinal loading. As for the Linotte Viaduct
(Figure 2.29 – b)), the layout provides high transverse stiffness mostly through the inclined
columns, and their connection node cannot be interpreted as a classic cap beam.
a) Lizaine Viaduct b) Linotte Viaduct

Figure 2.29 – Piers in the TGV Rhine-Rhone HSRL
(from http://www.lgvrhinrhone.com/)
2.30
Despite the cases like the previous ones, it is possible to observe that the vast majority of
multiple column substructures are used in low-rise viaducts (up to 20m). As previously
discussed, there is not an absolute reason in favor of such application; nonetheless it seems to
benefit more from the structural advantages of multiple column and, particularly, bent piers.
Additionally, it can also be observed that these structures are typically quite long, frequently
spanning over a few hundreds of meters, while the main span length is mostly determined by the
type of the deck considered for each case.
2.5. FINAL REMARKS
According to the objectives established for the present chapter, its content focused on
analyzing design options for HSRL bridges and, in particular, of bridge piers. Within that
context, common layouts for HSRL bridges were reviewed, focusing on the structural aspects
relevant to the simply supported or continuous deck types. Regarding the substructure design
and, in particular, of the bridge piers, seismic and HSRL specific performance requirements
were presented, from which the importance of structural collapse prevention, structural and
track damage limitation, as well as the running safety of trains was emphasized. Finally,
cross-section shape options for HSRL bridge pier design were also reviewed and discussed,
according to the associated structure’s defining characteristics, where three of the most common
layout types were presented: single columns, wall-piers and multiple column piers.
2.31
Precast Technology for Bridge Piers
3
PRECAST TECHNOLOGY FOR
BRIDGE PIERS
3.1. ADVANTAGES AND DISADVANTAGES OF PRECAST
On the context of developing a precast solution for bridge piers, discussion eventually
focuses on the merits and shortcomings of the precast technology itself. In that regard, it is
widely acknowledged that precast solutions contribute on a large scale to the construction speed
of a particular project. There are a few reasons one can mention to support that claim, but
enabling the simultaneous off-site construction of multiple elements is among the main ones. In
fact, when the core part of a structure is constituted by an assembly of precast elements, the
building process can be managed through several tasks in parallel, benefitting from workload
distribution between site labor and precast plant manufacture for considerable gains in overall
time spent. That strategy is not as well suited for the traditional on site casting, known as the
cast-in-place (CIP) procedure in the construction industry, because most structures require some
type of sequential construction, where supporting elements (even if temporary) must generally
be concluded before the construction of supported elements.
Considering the rapid construction benefits, precast solutions are naturally convenient for
situations that present considerable time constraints, such as reinforcing and retrofitting
operations on active bridges. In those cases, traffic hindrance is common and the responsibility
of a contractor is to minimize the disturbance as well as avoid traffic congestion altogether, if
possible. When CIP construction is used, considerable resources must be allocated to formwork
execution, steel reinforcement preparation and concrete pouring of multiple elements, such as
foundations, columns, abutments and deck. Furthermore, construction schedules must account
for the concrete curing between operations (Freeby et al. (2003)), often leading to situations
3.1
where the workforce volume is defined by the manpower required to prepare the next operation
before curing periods are over. On the other hand, a common procedure on precast structures is
to cast the foundations on site, while columns and beams are built elsewhere and quickly placed
once brought to the construction site. In that regard, the time spent on assembling precast
elements on site is considerably smaller than the duration of the process related to equivalent
CIP construction, while also requiring less resources’ allocation, since formwork and steel
reinforcement preparation tasks are moved off-site, encompassing accountable economic gains.
A related benefit is that the construction process with precast elements becomes more
environmental healthy, because noise, air pollution, dust and debris are all reduced when most
of the casting occurs in the precast plant. Moreover, precast elements benefit from the increased
quality of factory construction, as well as higher quality materials and independence from
weather conditions. Therefore, the correspondent designs can provide significant savings over
the course of a particular project, also relating to the sustainability of the construction activity as
a whole (Yee (2001), VanGeem (2006)). This is further reinforced by the flexibility related to
defining the assembly method and its relevance in the design process. A common procedure for
beams, for example, is to use a hybrid solution where the precast element corresponds to the
beam body (web and bottom flange, possibly including prestress), acting as formwork for
combining with “in situ” concrete topping (the collaborating slab) for equivalent global
behavior and considerable material savings (Figure 3.1).
Figure 3.1 – Material savings using composite precast designs (Yee and Eng (2001))
Precast elements can also provide adequate technical solutions for situations where
different constraints affect the application of a normal construction management strategy. For
example, some bridges are built over long stretches of water, where work zones are limited and
restricted to adjacent support platforms. Using precast elements reduces the site workload, since
mostly assembly procedures are thus required. That situation can also occur on tall structures or
3.2
others where accessibility and work zone conditions are concerns, and consequently, by
reducing the amount of time manpower required to operate in potentially dangerous conditions,
the use of precast solutions also contributes to increased labour safety.
All the above describe the precast technology benefits in relation to design or construction
related issues. Nonetheless, the importance of accounting for logistics and provision
management difficulties is also paramount for achieving a good compromise between a
technically adequate precast design and an economically viable solution. With that in mind, the
production and application of precast elements in construction can be addressed in five main
phases (Castilho and Lima (2012)), all raising specific concerns:
▪ Element fabrication;
▪ Transport and storage within factory environment;
▪ External transport from the manufacturing plant to the construction site;
▪ Placement of the elements in their final positions;
▪ Implementation of the connections to the local structure;
The development of adequate factory procedures to manufacture precast elements relies

significantly on the capability to establish systematic labor circuits, imposing an
industrialization mindset (Alinaitwe et al. (2006)). As a rule of thumb, the more repetitive the
design for the precast elements is, the easier it is for production teams to achieve a certain
operating rhythm, and also to train newcomer workers for. Moreover, the moulds used for
casting are of high quality (steel moulds are frequently used), enabling first grade finishing,
textures and accurate shape definition for better architectural appearance (Manrique et al.
(2007)), and can be used several times before needing replacement. However, that potential can
only be tapped into if a project requires several precast elements of similar characteristics, since
casting beds and moulds can be used repetitively without significant time-consuming changes in
between operations.
That also raises the issue of the production space in the precast plant, as well as
transportation to storage areas. In fact, the weight of a precast unit is a common limitation for
the design and relates directly to transportation concerns. In a factory environment, the lifting
capacity can be relevant in determining whether the casting position is vertical or horizontal,
because multiple cranes can easily be used to carry the heaviest elements by having spaced out
lifting points. That is also a concern for external transportation to the construction site, since
heavy trailers and trains have limited carrying capacity. For example, PCI (1997) points to
practical limits of around 200 ton for truck shipping and 500 ton for rail shipping. Still, in the
construction site, special lifting equipment and bracing may be necessary to move the precast
3.3
units, as well as to place them correctly before definitive connections are built. Additionally,
temporary storage may be considered, if the construction site has suitable facilities.
Careful handling of the precast elements is also paramount to avoid unwanted damage
before the structure is finished. PCI (1997) states that “Precast concrete bridge products are
designed to be furnished crack-free. However, cracks should not be considered a reason for
rejection unless the product is structurally or aesthetically impaired beyond repair”. In that
regard, the main reasons for the appearance of cracks on concrete are widely known, but extra
attention must also be dedicated to prevent accidental impacts, especially because handling
precast specimens in between manufacture and installation generally involves some difficulty.
3.2. PRECAST CONNECTIONS
A precast system for bridge piers or building columns is generally characterized by two
main aspects: the structural element itself and its components, as well as the connection
mechanisms. In that regard, while the quality of the off-site manufactured pieces is undeniably
higher than what is usually obtainable “in situ”, for the above discussed reasons, the global
behavior of precast structures is dependent of the integrity achieved at the connection joints,
since the intrinsic monolithism ensured by CIP construction is not easy to replicate within
precast assemblies. Furthermore, inadequate joint detailing can lead to early structural damage,
as they are the weakest points in the overall precast system, therefore raising also durability
concerns.
There are four potential locations for precast pier connections, according to Marsh et al.
(2011):
▪ Pile-to-foundation: typically the connection between pile caps and piles, which are
regularly located below ground and also difficult to inspect and repair;
▪ Foundation-to-element: connection between the foundation system (the most
common types of which are the spread footing, pile cap or drilled shaft) and the
substructure element, which may or may not be accessible, and is a location prone
to severe damage during seismic events;
▪ Element-to-element: element connections are established between segmental pieces
or between a segment and a pier cap/cap beam. The connection itself can be
performed with several different mechanisms, and be located on a variety of
column points, but these are generally accessible to inspection and repair;
▪ Element-to-superstructure: structural layouts that require continuity between
substructure and superstructure generally involve a connection between pier caps
3.4
or cap beams to a deck girder, typically a diaphragm. Depending on the specific

detailing of the connection, severe seismic damage may occur;
Considering the scope of the present work, the following sections will mostly address
connections related with the main pier elements.
3.2.1. FOUNDATION-TO-ELEMENT CONNECTIONS
This connection type is generally designed to enable the transfer of all weight loads, as well
as lateral loads, while displaying overall behavior similar to monolithic construction and
avoiding additional limitations. Footing connections for precast columns are usually performed
as one of the following types:
▪ Pocket connection;
▪ Socket connection;
▪ Base plate connection;
▪ Cast-in-place footing with reinforcement continuity;
▪ Grouted sleeves connection;
3.2.1.1. Pocket connection
In general, pocket connections require an opening on the footing, with bigger size than the
column cross section, upon which the latter is first introduced and then adequately held and
braced. Afterwards, the gaps between the column and the footing are filled with concrete or
grout. In order to have a large enough force transfer area between the socketed part of the
column and the pocket hole, this connection can only be performed on fairly large footings
(either CIP or precast). Figure 3.2 highlights the general configuration for a pocket connection.
Similar solutions can also be adopted if pile shafts are needed instead of spread footings, as
reported in Tran (2012) for seismic regions, as an example.
Figure 3.2 – Schematic pocket connection ("fib Bulletin 43" (2008))
3.5
Vertical design loads are defined for the column cross section and the adjacent bearing area
of grout/concrete, while overturning moments are resisted by the lateral pair of reactions formed
between opposing sides of the column, as shown in Figure 3.2. Additionally, shear stresses are
developed at the interface between the column and the surrounding infill material, and rugged
textures can be adopted in both the column and the pocket to increase the corresponding shear
resistance. Moreover, according to "fib Bulletin 43" (2008), the depth of the pocket (dc) should
be calculated as follows in equations 3.1 and 3.2:
𝑀
For < 0.15 × ℎ , 𝑑𝑐 > 1.2 × ℎ 3.1
𝑁
𝑀
For < 2.00 × ℎ , 𝑑𝑐 > 2.0 × ℎ 3.2
𝑁
Where h represents the cross-section height and M and N relate to the overturning moment and
vertical load, respectively.
Considering the previous, the difficulty of using this type of connection for structures with
large moment demands is considerable (common occurrence on columns experiencing inelastic
deformations during seismic events), typically resulting in inconvenient large sized foundations.
3.2.1.2. Socket connection
Socket connections are performed when the complete footing is cast around the vertical
column instead of just the surrounding gap hole. Different authors studied the behavior of
distinct variants (e.g. Marsh et al. (2010); Haraldsson et al. (2013)), but generally all involve
previous column precasting and adequate placement on a pre-excavated site, according to
Figure 3.3 - a). Afterwards, the footing is cast following the preparation of the reinforcement
steel around the column.
a) Schematic illustration (Marsh et al. (2010)) b) Lateral loading (Haraldsson et al. (2013))
Figure 3.3 – Socket connection
3.6
There are two noteworthy aspects regarding socket connections. First, the vertical load
transfer from column to footing depends of the shear friction in the footing-column interface,
often requiring the adoption of rugged textures on the precast element. Second, the fact that
longitudinal rebars cannot be bent into the footing, therefore increasing the difficulty to develop
adequate bond stresses on the tensile strained rebars and, consequently, to achieve the
theoretical bending capacity of the column. As a result, the bending behavior of the socketed
column can also rely on anchorage devices to mobilize the equilibrium of compressive forces
between the footing and the column, as illustrated in Figure 3.3 – b). A common solution for
those devices is to incorporate a steel or precast concrete plate at the column base, in order to
facilitate the casting of the precast element, which also simplifies transportation and handling.
3.2.1.3. Base plate and shoe connections
There are also layouts that incorporate an end steel plate as a moment resisting element,
which is an immediate solution for element stability during column placement and helps on
reducing the required depth of the footings. The lack of reinforcement continuity usually
requires welding of the rebars to enable adequate bond behavior, and both the steel plate
thickness and dimensions should be determined in accordance with moment induced stresses
and the position of the anchor bolts (see Figure 3.4).
Figure 3.4 – Column with base plate and welded rebars ("fib Bulletin 43" (2008))
Shoe connections can also be a particular case of steel plate connections, where openings
are considered at the outer perimeter of the column to allow anchor bolting the vertical element
to the footing without requiring additional space, as shown in Figure 3.5. The required devices
are commercialized by several companies (e.g. http://www.peikko.ca/; http://www.pfeifer.de/),
but generally their application is limited to moderately loaded columns, which is frequently not
the case of bridge piers, particularly under seismic demands.
3.7
Figure 3.5 – Shoe column connection (http://www.peikko.ca/)
3.2.1.4. CIP footing with reinforcement continuity
When footings are cast in place, a very common solution for establishing the
foundation-to-element connection consists on extending the longitudinal rebars out of the
precast element and into the designated space for the footing. Afterwards, complementing steel
reinforcement is prepared and the footing is cast, encompassing a manufacture procedure
similar to that of socketed columns. The main difference between them is that the present
methodology ensures stronger moment capacity and less dependency on mobilizing contact
surface shear forces.
The main difficulty associated with performing this connection tends to be the temporary
placement of the precast unit before casting. Since the longitudinal rebars are extended out,
temporary leveling pads can be used (Figure 3.6 – a)). Additionally, the foundations can be cast
in two phases, considering adequate lap-splicing lengths for the second one (Figure 3.6 – b)).
Nonetheless, the casting itself may be more complex due to the protruding nature of the column
rebars, requiring specially adapted formwork and accessibility, in order to provide good
concrete vibration. The potential for worker safety issues due to the danger of handling heavy
elements with protruding reinforcement also comes to mind.
3.8
a) Temporary steel supporting pad b) Preparation of the second casting phase

(Cruz Lesbros et al. (2003)) (Billington et al. (1999b))
Figure 3.6 – CIP footing connections
3.2.1.5. Grouted sleeves connection
An important issue must be previously stated regarding grouted sleeves’ connections: this
is a common methodology in every type of joint connections, such as foundation-to-element,
element-to-element, or element-to-superstructure. Therefore, although this section reports to the
foundation-to-element joints, it is also generally applicable to other locations.
Grouted sleeve connections involve leaving duct openings on either the foundation or the
vertical precast element (or both), in order to introduce continuity rebars during assembly
procedures. The length of the ducts is related to the required bond lengths, as they are generally
filled with grout, enveloping the rebars and enabling full bond mechanism. This procedure is
simple to execute, doesn’t require strenuous job site preparations and enables fairly rapid
construction. Despite that, it can be susceptible to several shortcomings. For example, since the
duct space is generally small, it can be challenging to ensure that the grout fully envelops the
reinforcement bars. Larger duct diameters can be used to prevent this issue, but that can also be
a detrimental solution. In fact, the inclusion of several sleeves can lead to steel congestion, as
they occupy a large space in the cross-section (Stanton et al. (2006)), which is further
aggravated if large diameter ducts are used. Furthermore, when specimens include protruding
bars, careful positioning of the ducts must be ensured, in order to prevent gross misalignments
and all the additional work and delays that would be required to overcome such problems.
According to Matsumoto et al. (2008), three different grout connection types can be
performed: grouted pockets, grouted ducts and bolted connections, as shown in Figure 3.7 - a),
b) and c), respectively. Bolted connections are difficult to perform in foundation-to-element
joints due to inaccessibility for bolting, which is the reason why they’re usually only considered
3.9
in the uppermost element-to-cap connections. Moreover, the main difference between grouted
pocket and grouted duct connections is that the former is established for multiple rebars, while
ducts are generally for individual bars.
a) Grouted pocket b) Grouted duct c) Bolted connection
Figure 3.7 – Grouted connection types according to Matsumoto et al. (2008)
Alternatively, when available embedment lengths are short, mechanical couplers can also
be considered. Haber et al. (2013) explored different solutions in the context of a precast footing
including protruding bars, using conventional grout sleeves with either mechanical couplers or
transition bars, as illustrated in Figure 3.8. It is also interesting to note the use of a concrete
pedestal to facilitate the placement of the precast unit before establishing the grouted
connection.
Headed Coupler (HC) connection Grout Coupler (GC) connection GC with Precast Pedestal (GCPP)
Figure 3.8 – Mechanical coupling connections (Haber et al. (2013))
For increased bond between the lap splices and the precast pieces, this type of connection is
generally performed using corrugated sleeves, as illustrated in Figure 3.9. Therefore, careful
handling and, if needed, temporary protection of precast units (including ducts) should be
accounted for, because dirt, water and other construction residues can accumulate inside and
impair the connection.
3.10
Figure 3.9 – Corrugated grout sleeves (Matsumoto et al. (2008))
As mentioned before, grouted sleeves can also be used in combination with a variety of
other connection types. For example, Davis et al. (2012) presented a new layout for foundation-
to-element and element-to-cap joints when vertical post-tension is used. In this variant, mild
steel rebars are introduced through corrugated sleeves, while a socket connection is used for the
reduced cross-section part of the column that includes the post-tension tendons, as illustrated in
Figure 3.10.
Figure 3.10 – Hybrid duct and socket connection for post-tensioned piers (Davis et al. (2012)).
3.2.2. ELEMENT-TO-ELEMENT CONNECTIONS
Precast element-to-element connections are found in segmental precast piers and can
generally be included in one of two main categories: loose-fit joints and match-cast joints. The
first type consists of unifying subsequent precast elements by lap splicing reinforcement bars
through the joint and filling the space with grout, mortar or CIP concrete. It is a simple
technique associated with low requirements for precast element manufacture, for which the
productivity rate is only limited by the capacity of the precast plant. However, the difficulty of
creating the connection tends to increase with the associated element dimensions, because
careful suspension of the precast pieces in geometrically aligned positions is paramount to
ensure optimal pier performance, thus generally leading to time-consuming operations.
Additionally, it is also difficult to ensure an even distribution of the concrete/mortar, which
3.11
increases the risk of partially filled joints, stress concentrations, cracking and possible corrosion
exposure of the reinforcement steel (Billington et al. (1999b)).
Match-cast joints avoid many of the inconvenients and difficulties associated with a loose-
fit joint. This type of connection can be summarized as requiring precast element pieces to be
cast against one another (or against pre-shaped formwork), ensuring a strong fit between them.
Figure 3.11 presents an illustrative scheme of precast column match-cast elements proposed by
Billington et al. (1999a), where the vertical casting of subsequent elements is performed on two
levels, enabling the previously cast piece to serve as bottom formwork.
Figure 3.11 – Match-cast manufacture process by Billington et al. (1999a)
3.12
The joint surface created in the match-cast process can be either a dry joint or an epoxy or
grout-filled joint. Dry joints are easier to perform, but lack protection against freezing or salt-
waters. They usually also have some rough edges that are prone to crushing and, therefore, can
potentially increase the fragility of the connection.
The structural integrity of match-cast products does not rely on the continuity of regular
steel rebars. Instead, post-tension prestress steel is generally used to compress precast segments
against one another, enabling adequate interaction on the fully assembled element. Moreover,
while the shear friction induced by post-tension compressions can often be sufficient, shear keys
can also be provided to further increase the shear capacity of the connection, as illustrated for
segmental precast piers in Figure 3.12.
Shear Key
a) Segmental pier illustration b) Segmental pier element photo

(Billington et al. (2001)) (Schokker et al. (1999))
Figure 3.12 – Match-cast surface and shear keys
In conclusion, loose-fit and match-cast connections can be adopted depending on the

situation, as they have a clear difference regarding advantages and disadvantages. It can be said
that the former could be considered when the time and resources spent in performing the
connections is not a project limiting constraint, and an increased precast piece productivity rate
can translate into economic gains or faster construction. Such is the case where, for example, a
large number of simultaneous work fronts are established, creating higher demand for precast
elements that may not be easy to meet by adopting match-cast manufacture procedures. On the
other hand, if pier erection time is a clear constraint, then match-cast products legitimately have
the potential to be a better option.
3.3. BRIDGE PIER PRECAST SYSTEMS AND DESIGN
Over the years, precast systems for applications in bridge substructures have been gaining
acceptance as a rapid construction focused alternative. Despite that, precast bridge piers are not
as commonly observed as precast girders are for bridge decks. Several technologies and
construction methods have been developed and improved, but CIP bridge piers are still
3.13
regularly accounted for as the most cost-effective alternative. On the scope of understanding the
available technology to use in the precast application for the present work, this section aims to
review some examples of design layouts studied for real applications of precast bridge piers, as
well as relevant scientific developments related with performance assessment and possible
improvement suggestions.
3.3.1. LAYOUTS FOR NON-SEISMIC ZONES
Billington et al. (1999a) presented one of the first fully integrated precast solutions for
roadway bridge piers in non-seismic zones. In that regard, the corresponding development
framework was established for compatibility with the most common superstructure
configurations as well as existing precast plant equipment and infrastructures, leading to a
limitation of the maximum element weight to the range between 700kN to 750kN.
The general layout defined a segmental construction comprising three basic precast
elements: column segments, a template flared segment and an inverted T-cap element,
illustrated in Figure 3.13 – a). The column segments were match-cast and the corresponding
joints were epoxy-filled. Additionally, for adequate site geometry control, the connections of
column segments to the foundations and the template element were expected to be cast-in-place
with high strength concrete.
a) General assembly layout b) Column segment sizes
Figure 3.13 – Precast pier according to Billington et al. (1999a).
Four different segments were designed with heights between 0.60m and 2.40m, focused on
a hollow cross-section configuration, where both post-tension strands and bars were included to
help on achieving structural integrity, according to Figure 3.13 – b). The precast elements could
then be combined to establish flexible technical solutions for supporting bridge superstructures,
according to the specific needs of a given project (e.g. deck width, span length, pier height…),
from single column layouts to multi-column frame bents as tall as 18m, as illustrated in Figure
3.14. This feature can relate well with standardization procedures, as the initial cost of preparing
3.14
the formwork for the four different sizes may translate into a reduced impact on overall project
costs due to continuous use of the same layouts.
Figure 3.14 – Possible precast bent configurations (Billington et al. (1999a))
According to Billington et al. (1999a), the most critical design conditions report to service
loads, where both maximum concrete stresses and zero tensile stress limit should be checked,
due to existence of post-tensioning prestress. A minimum passive reinforcement ratio should
also be considered, in order to control creep and shrinkage effects. However, the passive
reinforcement was not continuous, and post-tension was the only mechanism expected to
provide structural integrity. Therefore, although vertical prestress could contribute to a possible
reduction of residual displacements and improvement of joint shear resistance, concerns were
raised about the ductility and overall energy dissipation it provides, which has been a strong
reason against the use of this type of solution in high seismicity areas.
Another fully precast segmental bridge pier system was presented by Cruz Lesbros et al.
(2003), developed for the Ayuntamento 2000 bridge, located in a low-intensity EQ region in
Mexico. That bridge is a six span structure with over 160m of length, which was built in four
and a half months thanks to extensive use of precast, both in the superstructure and the
substructure. Interestingly, the author himself indicates that the number of bridges built using
fully integral precast systems amounts to less than 1 percent of all bridges in Mexico, which is a
statement for the innovative nature of the endeavor. The bridge crosses a deep valley with a
longitudinal slope of 7.5%, leading to pier heights from 12m to 42m, according to Figure 3.15.
3.15
Figure 3.15 – Longitudinal view of the Ayuntamento 2000 bridge (Cruz Lesbros et al. (2003))
The substructure was constituted by two abutments and five bent-piers which, as evidenced
in Figure 3.15, have considerably different heights. Additionally, hard accessibility to a deep
valley limited the operational conditions of trucks and elevation equipment. In order to take that
into account, the bent piers were designed as multi-column assemblies of up to three precast
unit levels, corresponding to a maximum precast element size of 15m and maximum single
element weight of around 60 ton, which required the use of hollow sections. Furthermore, for
increased lateral stiffness and improved seismic behavior, piers included both an intermediate
transverse beam and a cap beam, involving horizontal prestress to ensure adequate displacement
compatibility.
The construction procedure of the Ayuntamento 2000 bridge piers was based on loose-fit
CIP connections of the precast units. Therefore, it involved the use of a temporary supporting
pad for the first vertical elements, in order to correctly establish full lap splicing between the
precast elements and the CIP foundation, as shown before in Figure 3.6 – a). The
column-to-column connection was performed with a similar procedure, where upper precast
elements were placed on top of previously installed units, using corrugated sleeves for lap
splicing bars (as no vertical prestress was used) and providing a void on the column cross
section for posterior placement of the intermediate beams and to enable sufficient space for the
CIP connection concrete pouring. This is illustrated in Figure 3.16 – b).
3.16
a) Protruding rebars on fixed element b) Placement of subsequent elements
Figure 3.16 – Precast bent-pier assembly (Cruz Lesbros et al. (2003))
The construction of the Sorell Causeway Channel Bridge in Australia also evidenced the
potential of precast solutions (Gibbens and Smith (2004)). The structure was meant to replace
an old bridge that was in a high degree of deterioration due to sea water exposure and heavy
chloride reactions, therefore enforcing the rapid construction requirements of the new bridge.
Additionally, almost all the structure length developed over the sea, for 18 spans of around 25m.
Using precast solutions enabled not only the reduction of construction times, but also the
amount of supporting off-structure gear.
The precast system used in the new Sorell Causeway Channel Bridge involved a deck
supported by twin piers, each erected from CIP pile caps. These pile caps were constructed by
using precast formwork shells, according to Figure 3.17 - a). From there, match-cast pier
sections were placed, prestressed and grouted, according to Figure 3.17 - b). This procedure
enabled fast construction over sea by committing most of the heavy construction work to a land
based precast plant, while also minimizing the accessibility requirement beside the sea
structures.
a) Pile cap precast shells b) Precast pier segments
Figure 3.17 – Precast pier in the Sorell Causeway Channel Bridge (Gibbens and Smith (2004))
3.17
Despite the examples of the previously presented precast applications, the most common
utility for precast solutions on bridge substructures is related to bent and pier caps. In this
regard, it is usual to see vertical columns with protruding bars serving as guiding elements for
the introduction of the precast cap beam, as illustrated in Figure 3.18.
a) Multi column bent cap beam b) Single column pier cap

(Fouad et al. (2006)) (NCHRP (2003))
Figure 3.18 – Assembly of precast cap elements
To conclude this sub-section, there is a special case of precast usage that is worth
mentioning. In fact, to the author’s best knowledge, it is the only application of precast elements
for substructure construction of high speed railway bridges that was identified. As described by
Couchard and Detandt (2003), the José, Battice, Ruyff and Hervé viaducts were constructed
using a concept based on inclined portal frames with sloping strut elements, illustrated in Figure
3.19 – a). Precast elements formed the inclined part of the rigid frames, which were then
connected using prestress, according to Figure 3.19 – b). This configuration is capable of
achieving high longitudinal stiffness through axial deformations of the inclined struts, while
transverse stiffness of the frame is ensured by using adequately placed steel bracing along the
sloping strut plane.
a) Inclined portal frame b) Precast strut post-tensioning

Figure 3.19 – General substructure concept for the José, Battice, Ruyff and Hervé viaducts
(Couchard and Detandt (2003))
3.18
3.3.2. LAYOUTS FOR SEISMIC ZONES
The main difference of precast layout designs from non-seismic zones to seismic zones can
be linked to the increased demand that the joints are subjected to during an earthquake event.
For example, match-cast solutions as described by Billington et al. (1999b) tend to have high
compressive stresses resulting from the vertical post-tensioning of their segments. When the
seismic motion is further applied to the structure, causing the occurrence of horizontal
displacements and joint deformations, it is resisted by additional forces provided by the
prestress steel, holding the precast units together. Those additional forces are associated with
exploring the full capacity of the prestress steel and, essentially, result in increased tensile
stresses which can be associated with dangerous compression levels on the concrete, especially
around the foundations. A potential shortcoming is that the overall ductility of the piers can be
diminished, particularly if concrete crushing cannot be prevented, and the overall energy
dissipation capacity of the prestressed layouts may also be an issue of concern. Furthermore,
adequate detailing for providing a ductile pier response and the structural integrity of precast
connections can easily lead to large congestion of reinforcement steel.
These issues have been some of the main technical reasons preventing a more generalized
application of precast solutions for bridge piers, thus providing an open framework for research
activity focused on improving knowledge over the seismic performance issues of precast piers,
and also on developing solutions to address the associated shortcomings. For building
applications, however, several authors have addressed this topic over the years (e.g. Yee (1991),
Proença et al. (2002) or Pampanin (2003)), and the PRESS programme (PREcast Seismic
Structural System, Priestley (1991)), in particular, enabled some of the most relevant
technological advances. The general concept of the approach was based on setting the precast
connections on the usual plastic hinge regions, in a lumped ductility design supported by
prestress, aiming to reduce the usual CIP damage due to inelastic incursions.
Several results of that programme were also adapted for the context of bridges, where the
use of unbonded prestress enabled designers to take advantage of the innate concentration of
rotations on the precast joints, while disregarding the permanent effects of large inelastic
deformations. One such work was that of Hewes and Priestley (2002), where a precast
segmental bridge pier, which is illustrated in Figure 3.20, was studied to determine appropriate
design detailing for good seismic performance. In that work, prestress was the only continuous
reinforcement, providing structural integrity between the vertical segments of the structure.
3.19
Figure 3.20 – Precast segmental bridge pier connected by unbonded post-tension (Hewes and
Priestley (2002))
As shown in the previous illustration, the expected seismic behavior of this segmental pier
involved, by design, a “rocking joint” mechanism, where the introduction of a steel jacket in
the adjacent segment meant to provide increased concrete confinement to help on protecting
against excessive base compression forces due to rotations. From a structural behavior
perspective, this is a fundamentally different approach from usual CIP piers, where the designer
normally selects and carefully details the specific zones (namely, the column bases) for intended
flexural yielding of the longitudinal reinforcement and concentration of inelastic deformations.
In this case, the plastic incursion of the materials is substantially lower, and large lateral
displacements are associated with mostly rigid rotation of the pier segments around the
compression toe, when dead-load induced moment strength is overcome. The post-tension
contribution also acts as a self-centering element, since overturning moments produce increased
elastic strains on the prestress steel that are naturally recovered after seismic motion. The
overall solution was capable of achieving a low level of damage and strength deterioration
(Figure 3.21 – a)), but it was found to lead to generally thin cyclic force-displacement loops
(Figure 3.21 – b)), which can be associated with limited capacity of energy dissipation.
3.20
a) Precast segment damage b) Force-displacement behavior
Figure 3.21 – Segmental precast pier seismic performance as reported by Hewes and Priestley
(2002)
As rocking behavior was found not to be sufficiently dissipative, other authors such as
Palermo et al. (2005), Palermo et al. (2007) or Ou et al. (2008) aimed to explore additional
options. A very common approach was to design the precast connections for a hybrid,
controlled rocking behavior. The general concept involved introducing conventional
reinforcement or external dissipators across the “rocking joints” to increase the passive energy
dissipation during rocking movements, as illustrated in Figure 3.22.
Figure 3.22 – Hybrid rocking mechanism as reported by Palermo et al. (2007).
The general seismic behavior of hybrid rocking solutions revealed considerably increased
energy dissipation paired with low residual displacements, with the hysteresis loops forming a
“flag-shape”, as evidenced in Figure 3.23. Furthermore, unbonding the passive reinforcement
3.21
by introducing the corresponding bars through corrugated ducts was also seen to improve the
global performance of the pier by delaying bar fracture. However, according to Ou et al. (2010),
despite the potential benefits of that approach, it might not be desirable due to increased labor
work associated with unbonding those rebars and also because it weakens their protection
against corrosion.
Figure 3.23 – “Flag-shape” hysteretic curves for hybrid rocking piers (Ou et al. (2010))
Wang et al. (2008) studied the adoption of high-strength steel bars crossing the precast
joints as the main energy dissipating source, and different sizing of the plastic hinge segment,
according to Figure 3.24. His work revealed that the increase in the height of the first precast
segment enabled a more distributed cracking pattern to form in the plastic hinge region, in
contrast with the regular pier segments where most of the deformation is concentrated in the
joints.
Figure 3.24 – Experimentally tested segmental pier variants by Wang et al. (2008).
Elgawady and Sha'lan (2011) studied the seismic performance of precast segmental bents,
where the columns were fabricated within concrete-filled fiber tubes, and presented two
additional and noteworthy details. One of the layouts included the isolation of the precast
elements by introducing neoprene sheets in the foundation-to-element and element-to-cap
3.22
connections (Figure 3.25 - a)). The other included external energy dissipators located beside the
previously referred connections (Figure 3.25 - b)). The overall results were satisfactory, but
considerable differences were observed between the different proposals. In fact, using neoprene
isolation drastically reduced the initial lateral stiffness of the bents, potentially leading to lower
seismic forces but also to larger displacements. The use of external energy dissipators was
successful in improving that aspect of the bent system seismic behavior, but, unfortunately, it
was observed to lead to larger residual displacements and increased damage as well.
a) Neoprene isolation b) External energy dissipaters

Figure 3.25 – Concrete-filled fiber tube segmental piers (Elgawady and Sha'lan (2011))
Billington and Yoon (2004) presented a Ductile Fiber Reinforced Cement Composite
(DFRCC) with tensile strain hardening at the possible plastic hinge locations in order to
increase the plastic deformation capacity of those sections and to reduce the potential for
seismic damage as evidenced in Figure 3.26 - a). Furthermore, unlike the hybrid rocking
solution, DFRCC piers did not include any reinforcement crossing the precast joints, carrying
the load solely through the compressed region, the prestress tendons and the tensile cracked
DFRCC region. Experimental evidence revealed that the fibers’ composite enabled larger
energy dissipation by providing an increased tension-stiffening effect, although its effect was
mostly noticeable for earlier displacement cycles, as illustrated in Figure 3.26 - b).
a) Segmental pier illustration b) Energy dissipation per drift cycle

Figure 3.26 – Ductile Fiber Reinforced Cement Composite segmental pier (Billington and Yoon
(2004))
3.23
Hoshikuma et al. (2009) presented the japanese experience on precast segmental hollow
section piers. A particularly noteworthy layout was addressed, combining the concept of precast
segmental construction with an internal steel shell, according to Figure 3.27, and was initially
developed by Sumitomo Mitsui Co. Lda. (https://www.smcon.co.jp/en/). The assembly of this
layout is enabled by accurate positioning of the internal shell, which is then connected by steel
bolts and post-tension bars. In addition, the steel shells include shear key elements, in order to
help on mobilizing the concrete segments during seismic events. As presented by the author,
this configuration aims to carry the vertical dead and live loads mainly through the internal steel
elements, while earthquake force resistance requires the contribution of the connection elements
and the external concrete. In this case, considerable energy dissipation can occur in the precast
joints with yielding of the steel bolts, which can also be easily replaced due to the improved
accessibility provided by hollow-sections.
Figure 3.27 – Composite precast segmental pier layout by Sumitomo Mitsui Co. Lda
(Hoshikuma et al. (2009)).
One thing in common in all of the previously presented proposals is the focus on the
enhancement of the seismic performance of mostly precast segmental piers, usually resorting to
post-tension. That outcome may suggest that the use of post-tension for the rapid construction
of bridge piers has been gaining acceptance for applications on seismic regions. Nonetheless,
while the structural performance observed with these systems is often satisfactory, it can be
argued that viable alternatives without prestress may be preferred, in favor of cheaper
construction and less specialized operational requirements.
In that regard, adopting integral solutions with minimal element-to-element connections

may be a preferred approach, considering that the alternative segmental systems were generally
found to resort to the use of post-tension for ensuring structural integrity. Unfortunately, the
amount of research activity focused on integral column precast layouts seems to be vastly
3.24
inferior. One such work was developed by Hieber et al. (2005), corresponding to the reinforced
concrete alternative presented in that same publication, where a comparison against a hybrid
system with unbonded post-tension is also included. According to the author, the design
objective of the Reinforced concrete alternative was to “emulate traditional reinforced, cast-in-
place concrete columns”. For that purpose, structural integrity was provided by the continuity
of the longitudinal reinforcement, with protruding mild steel inserted through corrugated ducts.
Furthermore, the column segments were designed on the basis of the same geometry, material
properties and details of CIP equivalents. The seismic performance of this layout, which is
illustrated in Figure 3.28 – a), was expected to enable the formation of column base plastic
hinges (Figure 3.28 – b)), essentially relying on the same energy dissipation mechanisms of
equivalent CIP systems. Experimental confirmation of that behavior was not available,
unfortunately, as the study was essentially numerical.
a) System illustration b) Expected seismic behavior
Figure 3.28 – CIP emulated Reinforced Concrete system by Hieber et al. (2005).
The biggest difference between this reinforced concrete system and the hybrid post-tension
alternative was the displacement ductility, where the latter achieved around 50% higher values,
according to Table 3.1. However, that mostly reflected a lower cracked stiffness of the
non-prestressed system, since the overall maximum force and displacement capacity of both
structures was found to be very similar. Different damage states representative of structural
damage, such as concrete spalling or bar buckling, for example, were also found to occur for
very similar demand levels. Overall, those results suggested that adopting integral precast
solutions without prestress was a technically viable alternative.
3.25
Table 3.1 - Comparison between Reinforced Concrete and Hybrid systems (adapted from
Hieber et al. (2005))
Reinforced Hybrid Frame Percent
Concrete Frame Difference
𝐾𝑐𝑟𝑎𝑐𝑘𝑒𝑑
0.276 0.369 34%
𝐾𝑢𝑛𝑐𝑟𝑎𝑐𝑘𝑒𝑑
∆𝑚𝑎𝑥
3.69 5.51 49%
∆𝑦
𝐹𝑚𝑎𝑥 429 kips 382 kips 11%
𝑲𝒄𝒓𝒂𝒄𝒌𝒆𝒅 − Cracked stiffness; 𝑲𝒖𝒏𝒄𝒓𝒂𝒄𝒌𝒆𝒅 − Uncracked stiffness;
∆𝒎𝒂𝒙 − Maximum displacement; ∆𝒚 − Yielding displacement;
𝑭𝒎𝒂𝒙 − Peak force
Integral solutions were also addressed by Khaleghi (2005), essentially highlighting the
advantages for slanted columns, where the assembly of segments is harder to perform than for
vertical piers. Furthermore, an improvement is suggested by using a purposefully reduced size
reinforcement layout in connection zones, according to Figure 3.29. This configuration can
enable a reduction of the pier/bent cap reinforcement requirements due to a lower yielding
moment on the connection section. The previous is a relevant feature of this proposal, since
reducing rebar congestion may enable easier precast design and element manufacture. In
addition, the reduced size of the reinforcement layout at the joint section also provides
additional space for natural positioning of any required cap elements, which also represents a
big advantage of this proposal.
Figure 3.29 – Integral columns precast system proposed by Khaleghi (2005)
3.26
Alternatively, stable energy dissipation and a ductile behavior can still be achieved if no
local reduction of the moment capacity of the bent structure is adopted. In that regard, Pang et
al. (2010) studied the application of a reduced number of large diameter bars for establishment
of element-to-cap connections, achieving comparable performance to CIP regarding hysteretic
response and damage progression. Figure 3.30 illustrates the associated design, where ϕ57
rebars are introduced through 216mm diameter corrugated ducts.
Figure 3.30 – Large diameter rebar connections for bridge piers by Pang et al. (2010).
3.4. FINAL REMARKS
In this chapter, the most relevant aspects of the precast technology to the current work were
presented. Firstly, the main advantages and disadvantages of the precast technology were
discussed, aiming to provide an overview of the most important decisions to be made when a
large-scale precast application is considered, such as those related with construction schedules
or transportation and site handling constraints. Afterwards, a state-of-the-art review of the most
common precast connection types and pier systems was presented, taking into account that there
are significant differences between precast layouts designed for seismic and non-seismic zones.
Thus, the content of this chapter essentially provides detailed context of the precast technology,
regarding the structural system that is addressed in the present work and, in particular, the
decisions that influenced the conceptual design to be presented in the next chapter.
3.27
Conception and Seismic Design of Test Specimens
4
CONCEPTION AND SEISMIC
DESIGN OF TEST SPECIMENS
4.1. REFERENCE PIER MODEL: THE RAV POCEIRÃO-CAIA CONCESSION PROJECT
The precast application studied in the current work was inspired in the Portuguese attempt
at HSRL, embodied by the RAV Poceirão-Caia project (Altavia (2009)), whose construction
was halted at an early beginning. Nonetheless, the project itself was presented as the first
Portuguese venture onto the framework of high speed railways, aiming to provide fast quality
traveling between Lisbon and Madrid, and contemplated double lane UIC compliant
(http://www.uic.org/) ballasted tracks, designed for minimum and maximum speeds of 120
km/h and 350 km/h respectively. In addition, cargo transportation services were also
considered, up to a maximum travel speed of 160 km/h. Figure 4.1 illustrates the layout for the
full implementation of HSRL in Portugal and respective international connections with Spain,
where the orange shaded part refers to the Poceirão-Caia segment.
Figure 4.1 – Illustration of the Poceirão-Caia segment within the full Portugal and Spain HSRL
expected layouts Altavia (2009)
4.1
A total of 164.7 km of new line was to be constructed, involving several new structures
(around 6.3% of total length) from which 30 bridges and viaducts related to high speed railway
traffic could be identified. In that regard, two main design trends could be observed:
▪ Shorter span structures (30-35 meters) which were associated with double box
girder decks with 12.20m of width, using two precast “U-shaped” girders, CIP
slabs and smaller bent piers (Figure 4.2– a)). A solid cross-section of around 3m2 is
used for the vertical columns, while the deck amounts to a total around 6.70 m2;
▪ Longer span structures (40+ meters) that were designed with single box girders
with 12.20m of width, usually resorting to advanced construction methods such as
the launching gantry (Figure 4.2 – b)). Single column piers were adopted,
including hollow shape cross-section of approximately 5.50 m2, while the total
deck cross-section was around 9.45 m2;
a) For shorter spans b) For longer spans
Figure 4.2 – Pier layouts on the Poceirão-Caia project Altavia (2009)
On account of a possible large scale application of precast solutions for the construction of
those bridge piers, it should be acknowledged that the endeavor may only be reasonable if it is
technically viable while also striving for economic competitiveness. In that context, the design
should aim to reflect optimal conditions regarding manufacture, transportation and assembly,
where minimal costs, duration and operational difficulty are preferred. Furthermore, maximum
efficiency is generally associated with addressing repeatable layouts, considering the least
possible changes in materials and formwork. Therefore, it makes sense to try to identify an
optimal scenario for which most of these cost-efficiency driven guidelines may apply.
4.2
For that purpose, all the bridges and viaducts of the Poceirão-Caia line were further
evaluated, according to the overview presented in Table 4.1. It is possible to observe that the
majority of viaducts have a main span smaller than 35 meters, supported by structural systems
constituted by double column bents and double box girders. It should be noted that these
characteristics are in line with the findings presented in Chapter 2, as single column piers tend
to be used in longer span structures, while multi-column bents and walls are preferred for those
with smaller spans. Furthermore, the majority of the 14 double column bent structures with
main span of 30 meters represent medium to long viaducts, and around 70% of them with total
length between 100m and 600m, as illustrated by Table 4.2.
Table 4.1 – Poceirão – Caia structures overview

Main Span Technical Solution % of
Number of Viaducts
(m) Total
Piers Deck
30 Double column bent Double Box girder 14 46.67
Double column bent Double Box girder
35 + + 5 16.67
Rectangular Hollow Pier Single Box girder
40 Rectangular Hollow Pier Box girder 3 10.00
Total: 30
Table 4.2 – Structures with double column bent-type piers

Structure # of # of Pier height (m) % of
Length (m) spans structures Minimum Maximum Total
< 100 <4 3 4.83 6.43 21%
100 - 300 5 to 9 6 4.90 17.44 43%
300 - 600 10 to 18 4 6.04 19.00 29%
> 600 > 20 1 5.58 9.83 7%
14 Total: 100%
As previously discussed, this is a common scenario on the HSRL framework, since the
requirement for small horizontal radii and vertical slope structures is often fulfilled by
constructing long viaducts, which is also favorable regarding precast applications by enabling
large-scale production and increased efficiency, through replicable design and optimized
assembly procedures. Improved scheduling and resource management due to a larger number of
possible simultaneous work fronts may also come to mind. In light of these observations, the
double column bent pier layout of Figure 4.2 – a) seems to provide a favorable background for
the development of a precast solution, making it reasonable to consider partial application
focused only on the HSRL viaducts with that structural layout.
4.3
Regarding horizontal loads, the double column bent pier system provides high stiffness
against forces applied in the transverse direction. However, controlling the longitudinal
direction performance requires selection of an appropriate force anchoring layout, considering
the medium to long bridge lengths in question. The following design features were adopted on
the Poceirão-Caia line structures depending on the total bridge length:
i. Shorter structures include some variation of fixed piers around the bridge
midsection, as illustrated by the blue rectangle in Figure 4.3 – a);
ii. Longer structures concentrate the longitudinal stiffness in the pier alignments near
one of the abutments, according to the blue rectangle in Figure 4.3 – b);
iii. The installation of a STU device for restraining longitudinal high speed
movements in one of the abutments is expected, regardless of bridge length
(highlighted by the red circle in Figure 4.3);
a) Short viaduct example (U – free connection; F – pinned connection)
b) Long viaduct example (M – free connection; F – pinned connection; Mon. – monolithic connection)
Figure 4.3 – Longitudinal connection layouts on Poceirão-Caia HSRL structures
With this in mind, the pier alignments where provision of longitudinal stiffness was a
concern were generally observed to be at most 30% of the total, and in some of the cases
requiring monolithic connections. On the other hand, the main concern for the remaining
majority of alignments is the transverse behavior, corresponding to the main advantage of the
double column bent system. Therefore, it can be assumed that the longitudinal performance may
be evaluated on a case-by-case basis, enabling focusing the precast application mostly on the
transverse challenged piers with free longitudinal connections. In light of this, the main design
guidelines for this work were the following:
i. Double column bent pier and double box girder deck as the main structural layout;
ii. Medium to long viaducts, between 100m and 600m of length;
iii. Pier height range limited to values between 5.00m and 20.00m;
iv. Focus on the transverse performance of the piers, enabled by free longitudinal
connections;
4.4
4.2. PIER SEISMIC DESIGN
In order to design a precast system for this pier layout, including the core elements and
their respective connections, it is necessary to have realistic expectations for the capacity
demand on the fully assembled structures. A logical and straightforward approach to that
problem is to study equivalent monolithic solutions from which a precast alternative can be
derived. For that purpose, since mostly the transverse performance was assumed to be the main
concern, seismic events represent one of the more relevant load cases. Within those premises,
the design methodology described in the following sub-sections was established, essentially
aiming to obtain realistic design parameters for the test models, used in the experimental
campaign later presented in Chapter 5.
4.2.1. GENERAL CONSIDERATIONS AND DESIGN METHODOLOGY
Traditionally, structural design has always been related to the assessment of the applied
loads relative to resisting capacity for a given limit state. For most purposes that approach is
successful, since the loading characteristics can be determined independently from the elastic
response of the structure and the comparison between capacity and demand is straightforward:
collapse is avoided if the strength capacity available is greater than the applied loads. On the
framework of seismic events, elastic loads are often too large, preventing a cost-effective design
for strength capacity. As acknowledged by the community over the last decades, the most
adequate approach is to expect the occurrence of structural damage, associated with inelastic
deformations and ductile behavior, for which avoiding relevant strength losses becomes the
main concern. Considering this context, there are two main trends associated with current
seismic design methodologies:
▪ Force based methodologies (FBD);

▪ Displacement based methodologies (DBD);
Both of these are related to the concept of ductility, which is defined as the ratio between
the maximum and effective yield values of a chosen deformation parameter such as
displacement or rotation. FBD methodologies have associated an elastic force reduction factor,
which governs the design strength while indirectly ensuring smaller deformations than the
actual capacity through detailing. That approach is known to have some shortcomings, which
are mostly related to evaluation of the structure stiffness and the fact that it is not completely
independent from design strength, since both influence and are influenced by the initial period
estimation and the force distribution between resisting elements.
4.5
Alternatively, DBD methodologies essentially aim to design structures to be capable of

achieving a given limit state described by a maximum deformation parameter in line with the
desired hysteretic energy dissipation mechanism. In order to do that, they forego the
characterization of the effective period from the elastic properties of the system and instead rely
on the concepts of equivalent hysteretic damping, or inelastic displacement spectra, to relate the
evaluation of the effective period of the structure under design with the targeted deformation
capacity. This approach has been acknowledged to lead to more consistent results and, provided
that the characterization of the inelastic properties of the structural system is sufficiently
accurate, the resulting design process is straightforward. A more detailed discussion on these
subjects can be found in Priestley et al. (2007).
In this context, a common point of both the seismic design approaches addressed above is
the reliance on the choice of a stable energy dissipating mechanism by selecting adequate
locations for plastic hinges, as well as capacity design procedures for some specific elements
and conditions. This methodology leads to some level of predictability of structural damage,
and also helps designers on accommodating higher strains on critical zones by adopting a
reinforcement layout capable of larger deformations. For example, building structures are
normally designed to develop plastic hinges in beam ends and the base of the columns, while
avoiding soft-storey mechanisms (Figure 4.4 – a)); single column bridge structures are designed
mostly as a vertical cantilever, with a plastic hinge developing just above the footing (Figure 4.4
– b)); multi-column bents, however, are usually expected to display double bending behavior,
with high stiffness cap beams (Figure 4.4 – c)). Of course, on all cases, plastic hinges are
considered to be conditioned, by design and detailing, with sufficient ductile capacity to
withstand expected rotation demands under seismic loading.
b) Single Pier c) Multi-column bent

a) Portal frame building
bridge bridge
Figure 4.4 – Plastic hinge placement on different structural systems
4.6
In the FBD framework, the adoption of a global elastic force reduction factor can also
be difficult to evaluate for some structures, since it relies on the misleading assumption of
simultaneous formation of the plastic hinges on the main resisting elements. Furthermore,
different structure types and materials are assumed to lead to different reduction factors, for
which design codes usually provide reference and limit values. This strategy also requires the
adoption of detailing rules which are assumed to provide the necessary ductility, but a direct
relation between demand and capacity is not explored beyond that. However, for design
purposes of regular structures where the simultaneous yielding assumption is acceptable, FBD
still provides a straightforward procedure for reliably calculating design forces.
Regarding DBD, the ductility demand is generally also imposed by design codes,
according to the applicable performance limit states, but unlike in FBD, it is directly associated
with the response for each particular structure. Therefore, the yielding displacements associated
with the adopted plastic hinge dissipating mechanism must be calculated, because they are a key
component in the determination of the ductility capacity, and the difficulty of that assessment is
directly related with the complexity of the structure. That assessment is relatively
straightforward on regular structures such as buildings, since relations between cross-section,
structure geometry and the target displacement profiles are generally easy to determine by
analytical or numerical means. However, considering the geometry of the present bridge system
and its bent pier (and, particularly, the presence of the coupling beam) increased difficulties and
concerns are raised by the use of DBD.
4.2.2. IMPLICATIONS OF THE STRUCTURAL SYSTEM ON THE DESIGN STRATEGY
Figure 4.5 – a) presents illustrations of different height bent piers, roughly covering the
minimum and maximum values observed in Table 4.2. According to those values, the respective
width-to-height ratios vary between approximately 1.0 and 0.25, where the pier structure is
composed of two columns with varying heights along the full bridge length, and a constant size
coupling beam. In fact, the total span of the beam actually depends on the rail track
conditioning, because the columns are vertically aligned with a centered position relative to
each rail track, in order to eliminate eccentric moments from traffic loads and, for operating
speeds greater than 300 km/h, track centre distances are recommended to be adopted between
4.50m to 5.00m (UIC (2010)).
4.7
0.40
0.80
1.40
0.40
2.80 m
1.10 m
1.10 m
20 m
0.60 1.00 0.60 0.80
Column Cross-Section Beam Cross-Section
10 m
Column Beam
5m
A (m2) 3.040 1.120

4 * 1.021 0.183
I (m )
5m 5m 5m * for lateral displacements
b) Element cross-section and

a) Different pier heights
properties
Figure 4.5 – Double Column Bent-pier geometry
Consequently, the stiffness ratios between the columns and beam are not constant, i.e.,
they change according to pier height. In that regard, a simplistic evaluation can be made,
assuming no deformations take place in the beam-column nodes due to lateral displacement and
considering fixed base restraint conditions. For that purpose, the column and bending stiffness
ratios illustrated in Figure 4.6 can be represented by Equations 4.1 and 4.2, which are associated
with linear elastic properties calculated for the gross cross-sections shown illustrated in Figure
4.5 – b).
3EI
Kbeam = Cb
0.5 lbeam
4.085
𝐶𝑐 = 4.1
Cc 4.085 + 0.220 × 𝐻
4EI
Kcolumn =
H
H
0.220 × 𝐻
𝐶𝑏 = 4.2
4.085 + 0.220 × 𝐻
Figure 4.6 – Stiffness ratios for lateral displacements
Values of Cc and Cb for the pier height range (between 5.00m and 20.00m) were
determined for the following three scenarios of element stiffness and plotted in Figure 4.7 – a):
1. 100% of elastic stiffness on both column and beam (Cc-100C_100B and

Cb-100C_100B, respectively), corresponding to the reference structure;
2. 50% of the elastic column stiffness and 100% of the elastic beam stiffness
(Cc-50C_100B and Cb-50C_100B);
4.8
3. 100% of the elastic column stiffness and 50% of the elastic beam stiffness
(Cc-100C_50B and Cb-100C_50B);
In the referred plot, the upper bound value of 1.00 can be related to fully rigid behavior of the
respective element, while an intermediate value of 0.50 relates to equal stiffness between both.
The overtake threshold is represented by the colored triangle shape within each Beam-Column
result set. Concerning only the results of the reference system (100C_100B, represented by the
blue lines), it can be observed that the beam element provides higher stiffness to the bent-pier
structure for total heights greater than 15.50m (when the respective lines intersect each other),
whereas the opposite occurs for shorter piers.
When significant variations of the stiffness ratio between columns and beam are
considered, the overtake threshold shifts accordingly, as illustrated in Figure 4.7 – a) by the
additional plotlines (colored red and green). In addition, it can be argued that such change may
be representative of the structural impact caused by the occurrence of cracking or localized
damage (on the elements with the 50% reduced stiffness) during the structure’s lifecycle. The
red lines represent a situation where the contribution of the beam element to the total lateral
stiffness is larger, while the opposite is represented by the green lines, illustrating an amplitude
range of the overtake threshold between 7.00m in the case of the former (50C_100B), and a
value greater than the maximum height for this study of 20.00m, in the case of the latter
(100C_50B).
In both cases, the red and green plots are closer representations of the upper (fully rigid
cap beam behavior) and lower (non-existence of cap beam) bound scenarios under analysis,
respectively, but, nonetheless, a more direct comparison can also be made. Aiming at doing that,
the reference structure system (100C_100B) was assumed to provide the most accurate estimate
of the expected stiffness for the present structure (K100C_100B), enabling the evaluation of the
absolute variation of stiffness between itself and equivalent structures having…:
▪ Rigid: …fully rigid cap beam behavior, corresponding to a double column bent
𝐸𝐼
structure whose total lateral stiffness can be evaluated by 𝐾𝑅𝑖𝑔𝑖𝑑 = 24 × 𝐻3 ,
according to the concepts on Figure 4.6;

▪ No beam: …no cap beam (or a null stiffness beam), corresponding to two single
columns responding as vertical cantilevers, whose total lateral stiffness can be
𝐸𝐼
evaluated by 𝐾𝑛𝑜𝑏𝑒𝑎𝑚 = 6 × 𝐻3 , according to the concepts on Figure 4.6;
Furthermore, the stiffness relations Knobeam < K100C-100B and K100C-100B < Krigid are valid and,
consequently, the absolute variation of stiffness was evaluated differently for the Rigid and No
beam scenarios, according to equations 4.3 and 4.4, respectively, whose results are illustrated
Figure 4.7 – b):
4.9
𝑲𝒓𝒊𝒈𝒊𝒅
∆𝑲𝑹𝒊𝒈𝒊𝒅 = 4.3
𝑲𝟏𝟎𝟎𝑪−𝟏𝟎𝟎𝑩
𝑲𝑵𝒐𝒃𝒆𝒂𝒎
∆𝑲𝑵𝒐𝒃𝒆𝒂𝒎 = 𝟏 − 4.4
𝑲𝟏𝟎𝟎𝑪−𝟏𝟎𝟎𝑩
The black line that marks the 1.00 value illustrates the reference structure 100C-100B,
and its equivalent structural systems regarding total lateral stiffness. Likewise, values above and
below that threshold represent structures whose lateral stiffness is, respectively, greater than or
smaller than that of the reference case. Therefore, Figure 4.7 – b) shows that the difference
between the present structure and the boundary cases under analysis, regarding the lateral
stiffness, gradually decreases with total pier height. That also suggests that the only element
whose behavior varies between the three analyzed structural systems - the coupling beam -, has
negligible contribution to the lateral stiffness of very tall piers. Nonetheless, for the pier height
range under analysis differences are still greater than 30%, indicating that the contribution of the
coupling beam to the total lateral stiffness cannot be ignored, and that the resulting interactions
between the columns and the beam should be carefully evaluated.
1,00 3,00
Cc-100C_100B Cb-100C_100B ∆Krigid ∆KNobeam
0,90 Cc-50C_100B Cb-50C_100B 2,75
Cc-100C_50B Cb-100C_50B 2,50
0,80
Columns 2,25
0,70
2,00
0,60 1,75
Threshold shift
0,50 1,50
0,40 1,25
1,00
0,30
Beams 0,75
0,20
0,50
0,10 0,25
0,00 0,00
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
H (m) H (m)
a) Column vs. Beam stiffness ratios b) Rigid vs. No beam stiffness variation ratios
Figure 4.7 – Comparisons of structure stiffness relative element ratios and absolute variation for
pier heights in the assumed range and different analysis scenarios
A simple exercise based on moment-curvature analyses can be performed to provide

better understanding of the interaction between the columns and the beam. With that in mind,
2D fiber representations of the cross-sections illustrated in Figure 4.5 – b) were prepared and
subjected to a monotonic increasing curvature load history in the Cast3m FEM software
(http://www-cast3m.cea.fr/) to find the yielding curvature ranges for both the column and the
beam. For that purpose, an existing procedure on the referred software was used (labeled
MOCU), wherein the constitutive models adopted to simulate the stress-strain relationships of
concrete and steel fibers were a variation of Hognestad’s model (labeled BETON_UNI -
Hognestad et al. (1951)) in representation of a class C30/37 concrete, and Menegotto-Pinto
(labeled ACIER_UNI - Menegotto and Pinto (1973)) in representation of S500 reinforcing steel
4.10
(corresponding modeling parameters included in Annex A). In that regard, while the concrete
constitutive model can be considered fairly outdated (since other, more powerful, alternatives
can be found in the literature), it was still considered effective for the purpose of finding
yielding curvatures on the scope of this exercise, enabling the straightforward use of the above
referred procedure and avoiding resorting to more complex tools. For simplicity, a peripheral
distribution of longitudinal rebars was assumed for the column, while both top and bottom rebar
layers were considered in the beam. Two modeling assumptions were explored:
▪ Axial load ratio ν variation between 0.05, 0.10 and 0.20 for the column;
▪ Longitudinal reinforcement ratios ρ of 0.75%, 1.00% and 2.00% of the gross
concrete area for both column and beam;
The obtained results were compiled on Figure 4.8, where the color of the lines refers to
the axial load ratio (blue for 0.05, green for 0.10 and red for 0.20 on the columns, Figure
4.8 - a); blue only on the beam, Figure 4.8 - a)), while the line style refers to the reinforcement
ratio (dash lines for 0.75%, dash-dot lines for 1.00% and solid line for 2.00%). As observed, the
effective yielding curvatures for columns (φcy) are roughly between 0.0015rad/m and
0.0035rad/m and around 0.0025rad/m for the beam (φby).
50,00 10,00
ρ0.75%_ν0.05 ρ0.75%
9,00 ρ1.00%
ρ0.75%_ν0.1
ρ2.00%
40,00 ρ0.75%_ν0.2 8,00
ρ1.00%_ν0.05 7,00
30,00 ρ1.00%_ν0.1 6,00

Moment (MN.m)
Moment (MN.m)
ρ1.00%_ν0.2 5,00
ρ2.00%_ν0.05
20,00 4,00
ρ2.00%_ν0.1
3,00
ρ2.00%_ν0.2
10,00 2,00
1,00
0,00 0,00
0,000 φcy 0,005 0,010 0,015 0,020 0,025 0,000 φby 0,005 0,010 0,015 0,020 0,025
Curvature (rad/m) Curvature (rad/m)
a) Column cross-section b) Beam cross-section
Figure 4.8 – Moment curvature analyses
Comparison of these results is better achieved by merging with the influence of the
relative stiffness ratios presented in Figure 4.7 – a). Adopting a mean reference column yielding
curvature φcy of 0.0025rad/m, and also assuming that yielding occurs first through inelastic
incursion on the columns, enables determination of the expected beam curvature values
associated with column driven yielding. The corresponding results, which are illustrated in
Figure 4.9, are obtained by multiplying φcy by the ratio defined between the column and beam
stiffness ratios, Cc and Cb, respectively (for the sake of simplicity, it was also assumed that no
4.11
deformations occur inside the beam-column node during lateral displacement induced
rotations).
As it is possible to observe, the comparison of those values with the previously

estimated yielding curvature of the beam (φby), shows that rotations of the beam-column node
capable of inducing column yielding generally demand beam curvatures larger than the
reference φby value. These results suggest that the likelihood of the occurrence of significant
beam strains is relatively unavoidable for a large part of the pier systems within the assumed
height range, even if the demand decreases for the tallest structures. Likewise, it is also clear
that the assumption of cap beam rigid behavior, usually considered for multi-column bent pier
design, cannot be enforced in this structural system.
0.018
100C_100B 50C_100B
0.016
100C_50B φby
0.014
0.012
φb (rad/m)
0.01
0.008
0.006
0.004
0.002
0
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
H (m)
Figure 4.9 – Beam curvature φb demand variation (with column height) for node rotations
associated with column yielding for different column-beam elastic stiffness ratios.
The implications of that on general design strategy can be significant, since the plastic
hinge locations (Figure 4.4) and the preferred dissipating mechanisms should be defined
accounting for the most critical parts of the structure. Within that context, and considering the
beam free span with a fixed length of 2.80m and the section depth represented in Figure 4.5 - b),
the shear span-to-depth ratio αs is 1.0. If a plastic hinge length equal to the section depth is
considered, according to EC8 recommendations, then the full length of the beam would be
required for the formation of plastic hinges on opposing sides, which seems to be a conflicting
scenario.
Similarly to the present bent pier, this problem can also be found in coupling beams of
shear-wall systems (which are also characterized by high stiffness vertical elements), where
shear span-to-depth ratios of 1.0 are more common. Assuming that this system can display
4.12
similar behavior, additional design guidelines can be provided by analogy with context related
research. In that framework, beam deformations are usually characterized by the differential
displacements between the respective ends (Figure 4.10 – a)), caused by the motions of the
stiffer walls that they connect. The resulting system behavior can mainly relate to three different
collapse mechanisms (Figure 4.10 – b)):
1. Beam failure in bending;

2. Beam failure in shear;
3. Wall concrete crushing near the base;
1 2 3
a) Differential movement b) Different collapse mechanisms
between beam ends
Figure 4.10 – Coupled shear wall systems (Kumar Subedi (1991a), Kumar Subedi (1991b))
The third mechanism implies quasi-monolithic deformation of the two wall sets, which
can only be achieved through very rigid connections for displacement compatibility, with small
differential movement. The other two are related with less extreme cases, where the differential
movement is larger, causing large beam bending (1) and shear (2) demands. For the present bent
pier system, an intermediate scenario could be the most accurate representation, but that should
be carefully evaluated since the difference between mobilizing a mostly bending or a shear
dominated beam response is very relevant to design. In fact, the latter may lead to dangerous
strain concentrations at beam-wall interfaces, thus activating brittle failure modes such as
sliding shear or diagonal splitting (Figure 4.11), which can prevent the development of other,
more ductile, failure mechanisms.
4.13
c)
a) Start of the differential movement; b) Start of the diagonal cracking; c) Concrete crushing at beam ends;
Figure 4.11 – Beam shear dominated response. Diagonal splitting example (Kumar Subedi
(1991a))
All these aspects will be addressed further ahead, but nonetheless they indicate
potentially limiting factors to the behavior of the adopted bent pier, which may reflect on its
inelastic capacity evaluation. Therefore, rigorous numerical assessment of its behavior should
be required for the purpose of accuracy and reliability, and common empirical-based
assumptions of force-displacement relationships that are often used for more simple structures
could be considerably off target. Procedures such as pushover analyses, which are often used for
this purpose, are also difficult to apply to this case since only the geometrical properties of the
pier are known at this point (no reinforcement).
It is within this context that the simplicity of application of FBD methodologies can be
appreciated, where the definition of a single elastic force reduction factor encloses several of
these concerns. By doing so, it essentially enables disregarding the influence of many of these
subjects on a pre-analysis stage, contributing to fasten the preliminary design process. While
that may not always be a critical concern, this work was still subject to the timeframe defined
for the research project it was associated with, which essentially meant that a faster start of the
experimental campaign was desirable.
4.2.3. DESIGN STRATEGY
4.2.3.1. Introductory remarks
Based on the preliminary study and discussion addressed in the previous sub-section, it
was assumed that it can support the use of FBD procedures for the purpose of this work and, in
light of that decision, the Response Spectrum Method (prescribed in Eurocode 8 as the reference
methodology) was selected for the design. It should also be noted that the same procedure was,
in fact, used for the main seismic evaluation of the structures included in the original Poceirão-
Caia project, as it is possible to observe in its main document. Within that framework, the
definition of the elastic ground motion response spectra and the selection of an adequate elastic
4.14
force-reduction factor (or behavior factor q as stated in EC8) is paramount for performing the
analyses.
4.2.3.2. Seismic Action and Response Spectra
The general seismic design strategy for this work followed the main guidelines of EC8,
as discussed earlier in Chapter 2. With that in mind, combining the original layout of the
Poceirão-Caia presented in Figure 4.1 and the applicable seismic zoning maps shown in Figure
2.7 results in Figure 4.12, where the blue outline can be observed to cross seismic zones 1.3, 1.4
and 1.5 for Type 1 EQ, as well as 2.3 and 2.4 for Type 2 EQ. In addition, those zones were
identified with soil profiles ranging from categories A through C as presented in Table 4.3.
Figure 4.12 – Seismic zoning of the Poceirão-Caia HSRL line
For full definition of the seismic load, an importance factor must also be established. In
that regard, some argument could be made for an increased importance of viaducts included in a
HSRL line, considering that their full length amounts to hundreds of kilometers and is generally
associated with more expensive equipment/rolling stock than motorway structures, thus leading
to a high economic value at risk during a seismic event, even if human losses are not accounted
for. Despite that, the relevance of HSRL viaducts for post-earthquake communication in
Portugal isn’t as critical as, for example, in countries like France or Japan where it has
widespread use. Furthermore, the risk of human fatalities can likely be mitigated, to a certain
extent, by the systemic control over the traveling speed of trains that current monitoring
systems can provide (Boqueho (2002)), which may help with reducing the probability of
occurrence of seismic events capable of affecting structures where trains are still travelling close
to top speeds (even if they cannot fully prevent them). Within that context, it was decided that
importance class II was adequate for these analyses, leading to an importance factor γI = 1.0.
4.15
Table 4.3 – Poceirão-Caia Response Spectra Parameters

Seismic Soil Response Spectrum Parameters
ag (m/s2)
Zone Category S TB (s) TC (s) TD (s)
1.3 C 1.50 1.50 0.10 0.60 2.00
A 1.00 1.00 0.10 0.60 2.00
1.4
B 1.00 1.30 0.10 0.60 2.00
A 0.50 1.00 0.10 0.60 2.00
1.5
B 0.50 1.30 0.10 0.60 2.00
2.3 C 1.70 1.50 0.10 0.25 2.00
A 1.10 1.00 0.10 0.25 2.00
2.4
B 1.10 1.30 0.10 0.25 2.00
The numerical analyses were performed for the highest seismic intensities of each type
of earthquake; zone 1.3 for Type 1 and 2.3 for Type 2, which, combined with soil type C leads
to the most unfavorable conditions, as highlighted in Table 4.3 by the shaded rows. These
parameters result in the elastic ground motion response spectra illustrated in Figure 4.13.
0.70 0.70
1.3 zone 2.3 Zone
0.60 0.60
0.50 0.50
0.40 0.40
Sa (g) Sa (g)
0.30 0.30
0.20 0.20
0.10 0.10
0.00 0.00
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Period (s) Period (s)
Figure 4.13 – Elastic ground motion response spectra for soil type C
4.2.3.3. Behavior factor q
In the context of the Response Spectrum Method prescribed in EC8, behavior factors
are globally defined for the whole structure, enabling the definition of reduced intensity design
spectra relative to the elastic response spectra, reflecting the overall capacity for post-yielding
deformations; the corresponding maximum values for q factor are those represented in the EC8
table reproduced in Figure 4.14. In that regard, for RC piers the values presented reproduce
bending dominated failure, which can be adjusted according to the 𝜆(𝛼𝑠 ) parameter, to account
for the influence of shear in shorter elements.
4.16
Figure 4.14 – Maximum behavior factors for the Response Spectrum Method (EN1998-2
(2005))
Selection of q is determined from the structural elements designed to provide the main
energy dissipation on the overall system, taking into account the desired plastic hinge layout.
For usual bent pier structures, plastic hinges near the base should be mandatory, as well as
around the beam-column nodes, according to Figure 4.4 – c), and energy dissipation in the beam
should ideally be prevented with capacity design provisions and detailing. However, according
to previous discussion, it is possible that such strategy is not viable for the present structure.
Within that context, the behavior factor values presented in Figure 4.14 may not able to
reflect the expected level of ductility of the pier structure as they are related to straightforward
interpretations of the non-linear behavior usually reproduced by bridge columns. Additionally,
they don’t account for the possibility of the coupling beam being a limiting factor on the overall
ductile capacity. Therefore, two scenarios were considered for upper and lower bounds of the
behavior factor:
1. Maximum behavior factor value assuming ductility limited by the column: 𝑞1 =

3.00 ;
2. Minimum behavior factor value assuming ductility limited by the beam: 𝑞2 =
2.00;
The previous q factor values were calculated assuming null moment point at mid
element section. The first scenario corresponds to a mean column height (among all the pier
alignments between both abutments of a full length viaduct) value of 10.00m subjected to lateral
bending, leading to αs of around 2.30. The second scenario was determined from the free span
of the beam, which leads to αs of 1.00. Considering the range between the previous values,
4.17
adoption of an intermediate behavior factor value of 𝑞 = 2.50 was, instead, preferred. The
corresponding design spectra based on the elastic response spectra represented in Figure 4.13
were, therefore, defined for the Type 1 EQ and Type 2 EQ and 𝑞 = 2.50.
4.2.4. CASE STUDY VIADUCTS
Following the guidelines presented earlier, four viaducts were selected from the
Poceirão-Caia line as study cases, aiming to represent as different structural responses as
possible within the previously defined framework. A side view is presented for each of those
structures in Figure 4.15.
a) Macheda Viaduct
Figure 4.15 – Case Study Viaducts
b) Palheta Viaduct
c) Viaduct over Degebe River
4.18
d) Viaduct over Ribeira da Laje

Figure 4.15 (cont.) – Case Study Viaducts
As it can be observed, all the viaducts exhibit a regular layout, with constant span (except
near the abutments) and small pier height changes between alignments. For further detailed
information, Table 4.4 includes a summary of the main general geometric characteristics of each
of the selected viaducts.
Table 4.4 – Main characteristics of viaducts

# of Pier height (m)
Name Length (m)
spans Minimum Maximum Mean
1 - Macheda Viaduct 164 6 9.31 12.29 10.73
2 - Palheta Viaduct 192 7 8.48 9.38 8.64
3 - Viaduct over Degebe River 395 14 7.33 8.80 7.56
4 - Viaduct over Ribeira da Laje 520 18 9.45 19.00 14.14
Although transverse forces are transmitted to the piers throughout the bridge length,
longitudinal forces are anchored at selected points, as previously discussed. Figure 4.16
illustrates a representation of the support types expected on each pier alignment, as relevant for
bridge modeling and design, and for which the respective displacement constraints are further
detailed on Table 4.5.
4.19
( ) Pinned Connection
Translation Free
Connection
Monolithic Connection
Machede Viaduct
( ) ( ) STU / Damper
Palheta Viaduct
( )
Viaduct over Degebe River

( )
Viaduct over Ribeira da Laje
Figure 4.16 – Support layouts on the case study viaducts (illustrative scale).
Table 4.5 – Node Constraints between pier-head and deck nodes

Deck Constrained DOFs
connection type Translation Rotation
Pinned ux, uy, uz -
Free ux, uz -
Monolithic ux, uy, uz rx, ry, rz
ux - transverse direction
uy - longitudinal direction
uz - gravity direction
Furthermore, a summary of the expected dead-loads according to the non-structural

elements indicated in the Poceirão-Caia project is presented in Table 4.6, whereas structural
element dead-weight is considered using the typical RC unit weight of 25 kN/m3.
Table 4.6 – HSRL Bridge dead-loads

Dead Loads kN/m
Ballast 101.50
Waterproofing 8.70
Rails 5.60
Metallic guardrails and
10.00
concrete cornices
Ballast sleeper-walls 7.50
Gutter walls and cover 6.30
Total 139.6 ~ 140 kN/m
4.2.4.1. Numerical Models
The numerical models for pier design were developed in the SAP2000 structural software
(http://www.csiamerica.com/products/sap2000) using a combination of bar elements and
constraints for representation of both the piers and deck. Figure 4.17 – a) shows an example of
4.20
the numerical model for the Macheda Viaduct, wherein the element cross sections were defined
from the gross geometry of the structure as available in the Poceirão-Caia project. Therefore,
columns and beams follow the geometry illustrated in Figure 4.5 – b), while the deck follows
Figure 4.2 – a); the materials prescribed in the original project were also respected, leading to
the mechanical properties presented in Table 4.7. In addition, Figure 4.17 – b) presents a
frontview of a pier alignment, edited for description purposes.
Table 4.7 – Element properties

Column Beam Deck
A (m2) 3.040 1.120 6.694
4 *
I (m ) 1.021 0.183 61.649
J (m4) ** 1.372 0.154 7.081
C35/45 - Slab
Concrete class C30/37 C30/37
C50/60 - Precast "U" beams
* flexural moment of inertia for lateral displacements
** torsional moment of inertia
a) General Overview b) Pier alignment frontview

Figure 4.17 – Macheda Viaduct model in SAP2000
In order to accurately simulate the stiffness provided by the pier system, rigid connections
were introduced at the beam level, acting as joint offsets due to the large column cross-section.
The pier head nodes (red circles) were also connected to the deck element (green circle), at the
respective cross-section centroid, by associating the relevant DOFs using rigid body constraints
defined according to the respective bearing connection, as indicated in Table 4.5. No soil-
structure interaction was considered and fully fixed restraints were applied at the column bases.
Furthermore, following EC8 guidelines regarding member stiffness, the deck was modeled
using uncracked gross cross-section properties with 30% of the computed torsion stiffness.
Regarding the piers, however, EC8 points to the use of the effective secant-to-yielding stiffness
(𝐸𝐼𝑒𝑓𝑓 ) rather than the gross cross-section values (𝐸𝐼𝑔 ). It also suggests two methods for
evaluating the effective stiffness from the relation between the yielding moment and section
4.21
curvature; however, both of these methods require having previous knowledge of the
reinforcement ratios, which is one of the main objectives of this exercise. In order to address
that shortcoming, designers usually resort to iterations using the reinforcement ratios for
non-seismic loads, while taking into account the detailing rules and minimum ratios suggested
by EC8. Alternatively, some relationships, based on empirical knowledge, can also be found in
literature for predicting the effective stiffness using available information, such as pier geometry
and axial load, as indicated by Fardis et al. (2012). A common, but conservative, value to adopt
for the effective stiffness reduction is 50%. Another example is the proposal of Biskinis and
Fardis (2010), resulting from the numerical fitting of vast experimental data from columns, and
given by Equation 4.5:
𝐸𝐼𝑒𝑓𝑓 𝐿𝑠 𝑁 ∗
= 𝑎 × (0.8 + ln ) × (1 + 0.048 × ( ) ) 4.5
𝐸𝐼𝑔 ℎ 𝐴𝑐
where:
𝑎 – 0.081 for circular and rectangular cross-sections; 0.09 for hollow
rectangular piers;
𝐿𝑠 - Shear span;
ℎ - Cross section height;
𝑁 - Axial load;
𝐴𝑐 - Column cross section area;
* 𝑁
𝐴𝑐
in MPa;
Considering a main span of 30m and a mean column height of 10m (with null moment
assumed at mid-section), it is possible to calculate the overall axial load transmitted to a single
column from the dead loads, and determine the relation between effective secant-to-yielding
stiffness to the uncracked stiffness using equation 4.5:
𝑁𝑑𝑙,𝑐𝑜𝑙𝑢𝑚𝑛 = 0.50 × (6.70 × 25 + 140) × 30 = 4613 𝑘𝑁

𝐸𝐼𝑒𝑓𝑓 5.0 4613 × 10−3
≈ 0.081 × (0.8 + ln ) × (1 + 0.048 × ) = 0.141
𝐸𝐼𝑔 2.2 3.04
This low result should not be as accurate for the present bent pier case as for regular
rectangular columns, since cracking may develop differently on the columns and beam. In
addition, equation 4.5 was developed for a different application context, and mostly reflects
results based on bending-dominated column responses, disregarding the possible influence of
shear in the development of the moment-curvature behavior, which could be relevant for the
present structure, as previously discussed. In order to address that, a higher estimate of the
4.22
effective stiffness was assumed acceptable. For that purpose, a value of 30% of the uncracked
stiffness was globally adopted for all the bent-pier elements.
4.2.4.2. Analysis and Results
Aiming at using the Response Spectrum Method, a modal analysis was performed on all
the models. The resulting first vibration modes for all the studied viaduct decks are shown in
Figure 4.18, along with the associated period and Modal Participating Mass Ratio (MPMR).
T = 0.583s (MPMR= 81%) T = 0.452s (MPMR= 77%)

a) Macheda Viaduct b) Palheta Viaduct
T = 0.394s (MPMR= 50%) T = 0.998s (MPMR= 47%)

c) Viaduct over Degebe River d) Viaduct over Ribeira da Laje
Figure 4.18 – First vibration mode shapes, period and MPMR values
It can be observed that, apart from the Viaduct over Ribeira da Laje, the fundamental
period can be related to the higher intensity plateau of the design spectra for Type 1 EQ, and to
the descending slope (between TB and TC) for Type 2 EQ. Within this context, an elastic
spectral analysis was performed (accounting for the design spectra) in order to characterize the
seismic demand and establish a design threshold for the test models. For that purpose, stiffness
proportional viscous damping of 5% of the critical damping was also considered. Aiming for
accuracy, the minimum number of modes in consideration for each structure was defined to
guarantee a total Modal Participating Mass Ratio above 90%.
The obtained results are summarily presented in Figure 4.19, where the maximum base
shear value on each viaduct (calculated from all the respective transverse horizontal reactions),
is reported for both EQ types by the blue and red square markers. As observed, results for Type
2 EQ are always smaller than for Type 1 EQ, which could be expected considering that the first
vibration mode period of the analyzed structures was generally within the highest intensity
plateau for the latter, but on the descending slope in the case of the former. The overall
maximum was found to be around 2800 kN on the Palheta Viaduct. The numerical value of the
ratio (in percentage) between the maximum base shear on each structure and the base shear
determined at the respective tallest pier alignment is also plotted for each viaduct and EQ type,
representing a measure of the accuracy associated with just assuming peak demand at the tallest
pier. In this regard, differences were found ranging between 14% and 29%.
4.23
Within this context, a design base shear of 3000kN was assumed to provide a realistic
representation of the maximum seismic demand on the pier alignments for this type of structure,
enabling further supported decisions regarding reinforcement and detailing layouts for the
purpose of designing test models.
3000
Base Shear EQ1.3
2800 115%
114% Base Shear EQ2.3
2600 129%
2400
2200
121%
2000
kN
1800
115%
1600 117%
1400
1200 129%
1000 121%
800
Macheda Viaduct Palheta Viaduct Viaduct over River Viaduct over Ribeira
Degebe da Laje
Figure 4.19 – Spectral Analysis Results: Maximum base shear values among all the piers for
each viaduct
4.2.5. REINFORCEMENT AND DETAILING LAYOUTS
Energy dissipation mechanisms associated with seismic performance should, ideally,

reflect the natural ability of the structure to produce inelastic deformations, in addition to the
enhancements provided by design. Regarding the present structure, some discussion was
already provided in the 4.2.1 sub-chapter, mainly presenting the influence of the beam in the
overall deformation of the pier structure, and the impact it has on the usual bent-pier seismic
design strategy, reliant on column head and base plastic hinges (Figure 4.4 – c)). The
application of capacity design procedures, which are adopted to prevent significant energy
dissipation on unintended different locations, is conceptually challenged if the structure is more
likely to reach the yielding stage in the beam than in the columns.
Taking that into account, some level of damage and energy dissipation in the beam was
assumed to be unavoidable, increasing the ductility capacity of that element in order for it not to
be the limiting factor in the overall pier seismic performance. In addition, also according to the
previous discussion, some resemblance can be observed between this problem and that of
coupling beams of shear-walls, representing a framework that is better addressed in the relevant
literature. Within that context, the demand for ductility in coupling beams of shear walls has
been a common concern of several research works since the 1970s and, although some
proposals enabled some improvements, the general outlook on providing effective ductility
capacity is not straightforward.
4.24
The main shortcoming of coupling beams of shear walls is regarded to be the strong
influence of the shear driven strains and distortions in the overall deformation of those beams,
which are often responsible for the occurrence of considerable damage and even critical failure.
In structural terms, the vertical walls are usually considered as providing fixed restraints to the
beams. The latter’s behavior depends of the dominant deformation mode: when bending is
prevailing, double curvature occurs, leading to the alternate compressive and tensile strains
associated with the maximum bending moments near the walls (Figure 4.20 – a)); when shear is
dominant, the differential movement between beam ends tends to induce tensile strains
appearing on both sides of the element due to element elongation, while compressive forces are
transferred inside the element (Figure 4.20 – b)). The combined action of both bending and
shear deformation mechanisms, however, is considerably more complex, as the compressions
created by the double curvature in bending “conflict” with the tensile strains originated from
shear, potentially invalidating analysis through classic beam bending theory. In light of this,
correctly evaluating the relation between shear and bending deformations becomes critical for
accurate behavior assessment.
a) Bending b) Shear
Figure 4.20 – Beam strains during differential movement
As noticed by Paulay (1971), that relation is directly influenced by the shear span-to-depth
ratio αs, since taller beams are observed to have a larger prevalence of shear strains and
distortions, while slender beams are more likely to develop a deformation pattern similar to pure
bending. Paulay and Binney (1974) developed further into this subject, by testing several
conventionally reinforced beams, concluding that for elements with low αs, even when the shear
strength of the beam is large and capable of preventing diagonal splitting (Figure 4.11 – b)), the
tensile strains that occur due to heavy shear distortions lead to the formation of large cracks at
the interface between walls and beams (Figure 4.21), where aggregate interlock and rebar
dowell effect become the main forms of force transfer. During cyclic loading, the overall
ductility capacity of the beam is challenged by the continuous degradation of that section due to
friction. This is particularly noteworthy because that phenomena was revealed to be relatively
4.25
independent of the transverse reinforcement ratio, since it relates to a very localized occurrence
between the last stirrup and the wall interface.
Figure 4.21 – Sliding shear failure at beam-wall interface (Paulay and Binney (1974))
Those conclusions were also pointed by Tassios et al. (1996) or Galano and Vignoli
(2000), involving similar experimental tests, which showed that coupling beams exhibit low
energy dissipation with strong “pinching”, as evidenced in Figure 4.22.
a) Adapted from Tassios et al. (1996) b) Galano and Vignoli (2000)

Figure 4.22 – Cyclic loading results on conventionally reinforced coupling beams
In order to address this problem, Paulay and Binney (1974) proposed a layout that would
be the basis for the reference recommendation of both EC8 and ACI318 guidelines, according to
which rebars should be placed diagonally across the beam span, forming two vertically mirrored
struts, as presented in Figure 4.23, following the basic idea that short coupling beams are likely
to transfer shear forces through a single pair of strut and tie between the two walls.
4.26
a) NP EN1998-1 (2010)
b) ACI 318 (2014)

Figure 4.23 – Bi-diagonal Reinforcement Layouts
This system enables the diagonals to develop and carry the compressive or tensile stresses
according to the direction of the differential movements, and it is common to observe cracks
oriented along that geometry. The design capacity is determined from projecting the maximum
axial force carried by each strut in the transverse and longitudinal directions. Failure is then
expected due to rebar yielding, and concrete crushing may occur, leading to rebar buckling. For
this reason, codes suggest the adoption of individual transverse confinement for both diagonal
struts, in order to provide better control of the deformations occurring at the compressed sides,
contributing to an improvement of the overall ductility capacity. This layout provided better
results than conventional reinforcement, with larger energy dissipation and increased ductility,
as illustrated in Figure 4.24.
a) Adapted from Tassios et al. (1996) b) Paulay and Binney (1974)

Figure 4.24 – Cyclic loading results on diagonally reinforced coupling beams
4.27
In the context of technical design for coupling beams and short columns, Tegos and Penelis
(1988) presented a different layout, based on forming a rhombic truss with longitudinal rebars
(Figure 4.25). This configuration provides a peculiar combination of inclined and horizontal
reinforcement, increasing the flexibility for addressing both shear and bending demand, as
observed by other authors as well (e.g. Galano and Vignoli (2000)). A variant of this solution
was also tested by Tassios et al. (1996), where mainly the inclined parts of the rebars were
brought closer to the interface between beam and walls, essentially aiming to prevent the
occurrence of sliding shear. Other notable tested layouts include the adoption of horizontal
reinforcement across all the beam height, or localized dowell rebars at the beam-wall interfaces.
a) Original proposal by Tegos and b) Specimen tested by Galano and Vignoli

Penelis (1988) (2000)
Figure 4.25 – Rhombic truss reinforcement layout
General consensus was achieved on the fact that the diagonal reinforcement layout
provides the better performance for the shortest beams, and αs values smaller than 0.75 are
referred. In those cases, the slope of the diagonal struts is large enough, leading to high
effectiveness of the axial load carrying capacity of the struts that is assumed by design.
Simultaneously, conventional reinforcement is considered adequate for beams with shear span-
to-depth ratios αs larger than around 4⁄3, since the cyclic degradation of the beam shear capacity
can be observed to decrease considerably as the span increases. For intermediate values, the
rhombic truss was pointed as one of the most favorable layouts, because it provides similar
ductility and energy dissipation to diagonal rebar configurations and it is undeniably easier to
detail and supplement than the other, more complex, design proposals.
A common problem of these layouts is that they usually result in high reinforcement ratios
and large diameter bars (Figure 4.26), requiring careful evaluation of anchorage lengths and
confinement detailing, in addition to reducing the inherent constructive workability and
impacting productivity rates. Furthermore, casting concrete into a heavily reinforced cross-
section can pose challenges regarding concrete quality and integrity, due to difficulties in
4.28
obtaining a good aggregate spread and vibration. Some authors proposed alternative solutions
aiming at addressing this problem (Canbolat et al. (2005), Parra-Montesinos et al. (2010)),
based on the use of fiber reinforced concrete to reduce the overall rebar ratios (therefore
improving the constructability of the elements), for which positive results were obtained
regarding overall ductility and energy dissipation. This particular technical solution was out of
the present work scope, due to the industrial research partner interest, but it should still be
retained as a viable option.
Figure 4.26 – Rebar density in coupling beams of shear walls (Parra-Montesinos et al. (2010))
Within this context, the pier design strategy for this work is presented in Figure 4.27. A
column segment height of 15m was adopted (for reasons further on explained) and an equal
distribution of the base shear between columns was considered. Column end-section moments,
M1 and M2, are approximately equal (leading to null moment near the mid-height of the column)
that are possible to determine from the global equilibrium and the design base shear VEd. As for
the beam, either shear or moment can be assumed as the critical parameters for element design.
In the calculations, an effective span leff of 4.20m was considered in the definition of the design
forces for the beams, as defined by equation 4.6 adapted from EC2 (NP EN1992-1-1 (2010)),
𝑙𝑒𝑓𝑓 = 𝑙𝑛 + 2 × 𝑎 4.6
where:
𝑙𝑛 – free span of the beam;
𝑎 = min{0.5 × 𝑡; 0.5 × ℎ};
with: 𝑡 - column width; ℎ - beam height;
4.29
Global Layout: Columns: Beam:

2.80 m
Column axis
M2
Expected effects:
Rebar yielding
Energy dissipation
15 m
Damage accumulation V1
leff
VEd M1
M2
2
VEd
Figure 4.27 – Pier design strategy
An overview of all the relevant parameters is presented in Table 4.8. In this regard, it is
worth mentioning that, taking the base shear demands previously obtained in section 4.2.4.2, the
beam shear level determined with this strategy represents around 90% of the dead load
transmitted from the deck to the columns, which can lead to almost achieving decompression
during lateral displacements. Also, the strategy established to determine these reference
parameters was meant only to provide a realistic seismic demand expectation for guidance of
the test model design.
Table 4.8 – Reference design parameters for test models

Columns Beams
VEd (kN) 1500 4110
MEd (kN.m) 12210 8640
4.2.6. PRECAST CONCEPTION
Following the discussion of Chapter 3 and aiming to fulfill the interest and suggestions
established by the industry partner, the following set of requirements was defined for the precast
conception used in this work:
i). The precast solution should be adequate for road transportation from offsite precast
plants to construction sites;
ii). Mostly reinforced concrete elements should be used. Composite and steel solutions
were not desirable;
iii). The precast system should be aimed at fast construction, minimizing the amount of
work required on site;
4.30
Concerning requirement i), according to Portuguese law (ANSR (2013)), vehicles should
respect a certain set of geometrical dimensions in order to circulate freely in the national road
system:
▪ Maximum width: 2.60m;

▪ Maximum height: 4.00m;
▪ Maximum length: 18.75m depending on the type of vehicle;
Respecting the previous statements is not absolutely required, since the Portuguese regulation
allows individual permits to be issued for special cases, but it still indicates a range of
dimensions that are associated with increased transportation easiness and reduced costs.
Additionally, fast construction speed and minimal on site work requirements can strongly relate
to reducing the amount of precast connections, since those are generally the bottleneck of the
productivity rate, as previously discussed. Taking these guidelines into account mostly shows
that, on one hand, large elements are preferred by leading to a reduced number of precast
connections but, on the other hand, there is a limit to element size, for practical reasons.
That limit can also relate to precast units weight and the capacity for transportation and
lifting tasks. In fact, for most purposes vehicles in Portugal are limited to a maximum of 44 ton,
although that value is often surpassed through individual permits issued for special products
such as precast bridge girders. For the sake of reference, LASO - Transportes, S.A. (which is a
renowned and established service provider on the special cargo transportations Portuguese
market) has several tractors available with up to 250 ton maximum capacity. If required, the
weight of an individual precast unit can also be reduced by adopting smaller elements with
segmental construction procedures and, considering the desired framework of reinforced
concrete, maybe also adopting hollow cross-sections.
The influence of these operational constraints should be taken into account in addition to
more technical aspects. The mechanisms adopted for force transfer and/or displacement
compatibility between precast elements, such as rebar splicing, for example, come to mind as a
defining trait for the viability of most precast structures. In light of that, the two following sub-
sections aim to firstly discuss how the overall precast system is meant to function, focusing on
evaluating the pros and cons of a variety of different alternatives and leading to a decision
regarding the adoption of one of them. Afterwards, the details of precast connections are
addressed, again presenting different alternatives for the purpose of experimental study.
4.31
4.2.6.1. System Layouts
Before deciding on committing to a precast system, varying options were considered,

aiming to promote different layouts and functionalities. With that in mind, four systems were
conceptualized, and their pros and cons were discussed in order to reach a decision.
System A:
This layout corresponds to a straightforward application of the industry staple procedure of

resorting to descending precast cap beams onto a set of previously constructed columns (Figure
3.18). Associating the beam-column interface surfaces to suitable precast joint locations results
in three different precast elements required per pier alignment: one beam (B) and two columns
(C), as illustrated in Figure 4.28 – a). The concept for the connection mechanism is based on
some form of rebar continuity from the columns onto the beam and the footings. Therefore, this
system requires at least one foundation-to-element connection and one element-to-element
connection per column; an idealized procedure for assembling the complete structure is
illustrated in Figure 4.28 – b).
Temporary
B bracing may
be required
1 2
C C
3 4
a) Element division: b) Assembly procedure sequence: 1, 2, 3 and
B – Beam; C – Columns; 4
Figure 4.28 – Precast system: Option A
An important aspect of this proposal is the fact that both connections are placed on
locations expected to develop high strains and, as previously discussed, possibly rebar yielding.
That may present a challenge while accounting for anchorage forces for continuity rebars due to
the high stresses expected in those sections, especially if large diameters are used, considering
the limited space that is available. Taking into account that EC8 guidelines already lead to
densely reinforced beam-column nodes, increasing the amount of detailing to accommodate the
connection can create cumbersome design solutions, losing some of the appeal that a precast
alternative could present. However, since high local demand is expected and the precast joints
4.32
are prone to strain concentrations, it makes sense to adopt a design option aiming at providing
stability assuming that large deformations take place, instead of trying to prevent it from
occurring. The solution considered in this work aimed to enforce early yielding at that section
by reducing local moment capacity to about 50% (Figure 4.29 – a)) for the purpose of creating a
stable rotating mechanism similar to a hybrid rocking motion. In theory, comparing the
performance of equivalent monolithic and precast structures there would be expected larger
beam strains on the monolithic case and larger column strains on the precast system due to joint
opening, according to Figure 4.29 – b).
50% Asl
Moment: 100% Asl

VS.
Strain concentration Joint Opening
i) Monolithic Scenario ii) Idealized Precast Scenario

demand capacity
a) Moment demand and capacity along b) Beam-column node rotation illustration

the column and across the joint (i) monolithic; ii) precast)
(Asl – longitudinal reinforcement of the column)
Figure 4.29 – Beam-column precast connection mechanism

This layout assumes simultaneous and precise descending placement of the beam on top of
both columns, which presents a challenge which requires additional geometric clearance to
account for slight positioning errors. Dimension accuracy on precast elements is achievable, but
problems can arise to ensure precise verticality of the columns, for example. Moreover, the
connection is likely to be made at considerable heights while holding a precast unit weighing
several tons, thus adding to the operational complexity. On that topic, considering the pier
height range set for this work (maximum possible height of 20.00m) and the geometrical
properties defined earlier in Figure 4.5, the maximum weight per element can be calculated, and
the respective results are presented in Table 4.9. Considering these values, horizontal casting is
required for column segments, while vertical casting should be adequate for the beam element;
both elements are suitable for easy formwork.
Table 4.9 – Precast element geometric characteristics
Max. Dimensions Weight

Element Illustration
(m2, m) (ton)
H=18.30m
Columns (C) 3.04 x 18.30 141.77 A=3.04m2
2 x (3.04 x 1.70)
L=2.80m
Beam (B) + 37.33 H=1.70m
(1.40 x 1.10 x 2.80) A=3.04m2
4.33
This scenario indicates the possibility of using more conventional transportation for the
beam elements, although the columns are too heavy for that. Furthermore, considering
applications to long viaducts and fast construction purposes, the use of mobile or truck-mounted
cranes is adequate, enabling the assembly of one pier alignment and fast relocation to the next
one with minimal effort. It should also be taken into account that the capacity of the most
common mobile cranes in the market is generally smaller than lattice/boom cranes, which are
also less mobile due to the required extensive disassembly and reassembly procedures. For the
sake of context, Liebherr (http://www.liebherr.com/) is one of the main manufacturers of
construction machinery and their offer includes just one truck lattice/boom crane with 750 ton
of capacity; moreover while they also present just two mobile cranes for that weight range, there
are more than a dozen available cranes with capacity under 100 tons. Those values are a bit
misleading, however, as the full capacity of a crane is set for a short working range and low
heights, as enabled by the fully retracted boom. When considering lifting heights around 20m,
working with an extended boom becomes inevitable, and the leverage effectiveness decreases
considerably. Therefore, it takes a 350 ton capable crane such as the Liebherr LTM 1350
(illustrated in Figure 4.30) to be able to perform that operation.
a) Load chart b) Maximum lifting capacity (in ton)
Figure 4.30 – Liebherr LTM 1350 lifting capacity calculation procedure

(http://www.liebherr.com/)
Reducing the precast unit weight down to a range that is manageable using less capable
machinery would be desirable, as those are more common, require less operating constraints
(such as counter-weight establishment) and are generally less expensive, but that does not
prevent this from being a valid option even if it may not represent the most optimal one for
every situation. Nonetheless, a strategy to address this shortcoming could involve the use of
4.34
hollow sections, in order to reduce each individual unit’s weight. However, it was decided that
the columns’ inelastic behavior and potential plastic hinge zones should be associated with solid
cross-sections rather than hollow shapes, in order not to excessively increase the concrete
stresses and to further prevent local failure. Moreover, some of the precast connection
mechanisms designed for testing make use of the internal cross-section area, as will be
discussed on a later section.
Within this context, it is possible to perform a simple exercise to evaluate the gains from
adopting a hollow configuration on the remaining segment lengths. In order to obtain a rough
estimation of the plastic hinge length (Lh) on the columns, the general EC8 guidelines can be
taken into account, leading to the consideration of 𝐿ℎ = 2.20𝑚 (cross-section height along the
bending direction). If a maximum axial load ratio is also assumed, including the axial load
increase due to beam shear, it is possible to determine the minimum concrete area of the hollow
segment. Considering an axial load ratio of 0.30, the solid part of the cross-section is
approximately 1.00m2, representing 0.16m thickness walls if constant thickness is considered.
This design corresponds to a reduction of nearly two thirds of the original solid cross-section.
Consequently, the weight reduction of the hollow columns designed according to Figure 4.31 –
a), relative to equivalent solid columns, can be expressed as a function of the column height, as
presented in Figure 4.31 – b).
Solid 100
Weight Reduction Ratio
90
2.20m P.H 80
t = 0.16 m 70
Hollow 60
Var.
%
50
40
30
2.20m P.H 20
10
8723 0
𝐴𝑐 = = 0.96 𝑚2 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
0.3 × 30 × 106 H (m)
a) Hollow segment definition b) Weight reduction for varying height
Figure 4.31 – Hollow sections for precast column elements
As observed, for the tallest piers (where the column segment length is 18.30m) the weight
reduction for using a hollow section is around 50%, which is considerable. However, since most
of the bridge piers for a given viaduct should be under the established maximum, analyzing a
smaller size can be useful for understanding the implications of a hollow section on more
common occurrences. Considering a column length of 15.00m, the full-weight of a hollow
segment is around 60ton, while it is around 115ton using a solid cross-section. This also means
4.35
that the solid elements require around 200ton mobile cranes, while the hollow elements could
be assembled using more readily available 90ton mobile cranes.
On the other hand, the adoption of hollow sections also makes the detailing more difficult
and cumbersome, in addition to requiring the use of void forms that are not recoverable and that
may complicate the preparation of the reinforcement cages for horizontal casting. Furthermore,
the relative gain from using a hollow cross-section is not always as large as it may seem at a
first glance. For example, considering application for the mean pier heights indicated in Table
4.4 would result in a relative weight reduction of 20% or less, for two of the viaducts. With that
in mind, adopting a solid or hollow section should come down to a designer or contractor
decision based on the characteristics of each particular project and the actual gains that could be
achieved through it.
System B:
The concept for this system aimed at addressing two specific issues identified from system
A. As presented in Figure 4.32, instead of placing the connections in the locations with the
highest demand, the mid-height point was selected, as that is the vicinity of the expected null-
moment point for transverse seismic loading, leading to simpler connections with lower
stresses. Additionally, the bottom part of the columns (C) has significantly reduced height and
straightforward geometry, and CIP procedures can now be considered for faster construction.
The option of utilizing a precast element for that like in option A is still a viable possibility;
whether it could lead to considerable gains or not depends on the actual column heights. In
addition, the preparation of CIP columns can occur independently, as long as the contractor
teams have sufficient workers and resources.
BC
1 2
In situ
Construction
C C
3 4
a) Element division: b) Assembly procedure sequence: 1, 2, 3 and 4
BC – Beam with column segments;
C – Columns;
Figure 4.32 – Precast system: Option B
4.36
Therefore, this system includes only one precast unit (BC) required per pier alignment, but
the increased length of the column parts results in two to three times the weight of the B unit
from option A, depending on overall pier height. Some negative aspects can be associated with
that:
▪ The two large masses on the columns can easily cause damage in the beams due to
careless handling. Strong temporary bracing should be mandatory to prevent that
until the connecting operation;
▪ Column verticality is much more susceptible to accurate geometry assessment,
which may be harder to ensure since CIP column bases are adopted;
▪ The potential use of hollow sections is challenged by the need to execute the
connections at mid pier height. The additional detailing required for that design
can conflict with the existence of corrugated ducts or pocket holes, associated with
either rebar continuity or the splicing reinforcement required for connection
integrity and force transfers;
▪ Two cranes might be necessary for lifting the BC element in place for the
connecting operation due to the large column masses. That seems undesirable due
to the involved operating constraints, and guaranteeing the accessibility for two
cranes might also be a challenge on certain situations;
▪ The existence of a visible precast joint at the mid-section of different piers might
raise some concerns due to aesthetic reasons. Additional post-handling of the joint
in order to minimize this may have an undesirable impact on the overall duration
of those operations;
System C:
The concept for this system was aimed at finding an alternative layout to the double
column bent pier that could perform as adequately and be more precast friendly. Also, taking
into account the potential shortcomings related with the performance of the coupling beam
identified earlier, the idea was to use a wall-pier structure for stronger shear capacity. The
system involves a precast mechanism similar to that of the Sorell Causeway (Figure 3.17) pier-
sections, constituted by match-cast pieces (W) on top of each other, according to the illustration
of Figure 4.33. Flexible height (hel) on the precast elements is taken into account, enabling the
adoption of optimal sized segments for any given situation and preventing most of the size and
weight related problems of the two previous layouts, although it requires multiple precast
connections and assembly operations per pier. The geometry of each segment is easily suitable
for both manufacture and in situ assembly, and vertical direction casting is possible, which is
4.37
desirable for better concrete spread and rebar wrapping but was hardly achievable in the other
systems.
If no pier head enlargement is considered, the width of the wall-piers is relatively

predetermined as a result of the support required for the twin box girder deck. Therefore it is
safe to assume that a minimum of around 7.00m would be required, leading to considerably
higher stiffness than provided by the current bent pier. A minimum wall thickness (tw) of 1.00m
was also assumed, in order to be able to accommodate four (2 per girder) pot-bearing devices
with design shear strength of around 1900 kN each (an elastic horizontal shear force 2.5 times
greater than the design base shear calculated in 4.2.5 was considered for those calculations,
assuming that this wall-pier structure could be required to work under limited ductility
scenarios). For reference, the TE10i Mageba Reston Pot bearing (http://www.mageba.ch/)
fulfils the shear strength requirement (𝑉𝑅𝑑 = 1950 𝑘𝑁) and has a diameter of 930mm.
W hel = var.
1 2
lw =7.00m
3 4
a) Element division: b) Assembly procedure sequence: 1, 2, 3 and 4
W – Wall-pier segment
Figure 4.33 – Precast system: Option C
It is important to acknowledge that the structure change associated with this option can
result in considerable differences regarding overall seismic behavior if it leads to a significant
modification of structural periods. For the sake of context, three different lateral deformation
scenarios are illustrated in Figure 4.34, where Kpier represents the lateral stiffness of the bent
pier and Kbend and Kshear represent the bending and shear stiffness of the concept wall-pier,
determined using equations 4.7 and 4.8, respectively.
4.38
K pier K bend K shear

3𝐸𝐼
𝐾𝑏𝑒𝑛𝑑 = 4.7
ℎ𝑤 3
hw
5𝐺𝐴
𝐾𝑠ℎ𝑒𝑎𝑟 = 4.8
tw = 1.00m 6ℎ𝑤
lw =7.00m
Figure 4.34 – Pier vs. Wall stiffness comparison
Those can be compared as a function of pier height, by calculating Cbend and Cshear,
determined from dividing Kpier by Kbend, or by Kshear, respectively. According to section 4.2.1,
rigid beam-column nodes were assumed in the calculations, as well as a Poisson ratio equal to
0.20. The results are plotted in Figure 4.35, where it is possible to observe that the stiffness of
the double column structure is less than 20% of this wall pier for the height range under
analysis. However, the difference between bending and shear stiffness is still noticeable, and
illustrates the general seismic design guideline where a shear dominated response is expected
for squat walls and bending dominated response is expected for slender walls.
0.5
Cbending
Cshear
0.4
0.3
0.2
0.1
0
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
H (m)
Figure 4.35 – Bent pier to Wall pier stiffness ratio
This raises an important question when considering the precast option C for application.
According to Moehle et al. (2012), walls with very low aspect ratios (ℎ𝑤 ⁄𝑙𝑤 ≤ 0.5) tend to
resist lateral forces through a diagonal strut mechanism in which concrete and distributed
horizontal and vertical reinforcement resist shear. Conversely, slender walls (ℎ𝑤 ⁄𝑙𝑤 ≥ 2.0)
typically behave like vertical cantilevers, while a combination of the two mechanisms tends to
occur for intermediate geometric ratios. Considering the pier height range defined for this work
and taking into account the same concept wall illustrated in Figure 4.34, values for the ℎ𝑤 ⁄𝑙𝑤
ratio are between 0.70 and 2.85, meaning that the overall system design would have to address
4.39
both bending and shear critical failures, depending on the actual height of the pier under
analysis, making it considerably less optimal for large-scale applications. This characteristic has
particularly relevant impact in squat walls, as they exhibit limited deformation capacity under
shear loading. Due to that, yielding in shear is often considered unacceptable, although the shear
capacity of squat walls is often not a critical concern.
Within that context, two main challenges are associated with the precast system:
▪ Ensuring adequate shear stiffness for global element integrity and to enable the
formation of diagonal compressive struts;
▪ Providing sufficient reinforcement to enable inelastic deformations at the critical
sections, while accounting for anchorage of yielding forces along the plastic hinge
length;
When designing for shear, adopting match-cast shear-keys to complement the friction
forces along the joint length can be an adequate solution, granting interlock capability between
subsequent precast segments in order to establish shear force transfer. A similar disposition to
that of Figure 3.12 can be followed. However, providing adequate reinforcement for flexural
yielding might be more difficult, because rebar continuity between precast units is not easily
ensured with this system. Like in the Sorell Causeway, post-tension prestress is the most natural
solution for that problem, but as previously discussed that may lead to excessive compressive
stresses. Passive reinforcement could provide an interesting alternative, but the efficiency of a
bond mechanism along several precast units would be quite a challenge.
System D:
The concept idea for layout D focused on providing another alternative system that was
more capable of handling the flexural yielding requirements, while still providing the shear
performance of a wall pier. In order to do that, this structure concentrates the bending capacity
on the two wall column segments (WC) with continuous reinforcement, and provides shear
capacity due to the interlocking forces developed through the indentations of the wall panel
segments (WP). This can be related to the distribution of the coupling action (that is
concentrated in the beam of the original bent column) along the whole column height, to
emulate the behavior of coupled walls. An illustration of the different elements, as well as the
proposed assembly procedure is presented in Figure 4.36.
4.40
WC WP WC
1 2
twc = 1.00m twp = var.

2.20m 2.80m 2.20m
3 4
a) Element division: b) Assembly procedure sequence: 1, 2, 3 and
WC – Wall-pier column segment; 4
WP – Wall-pier panel segment;
Figure 4.36 – Precast system: Option D
A complete definition of the indentation mechanism was not fully explored, as this was
merely conceptual design. However, some guidelines can be introduced as there are basic
requirements this structure is challenged by. Design of the WP elements essentially takes into
account the shear transfer between indentations on opposing sides. The hwc distance between
two subsequent indentations is defined in order to ensure that a compression strut can form at an
angle of around 30º, according to the illustration on Figure 4.37 – a). However, that value can
easily be adjusted to accommodate varying pier heights, as long as it remains possible for
compressions to develop at an effective angle. Additionally, the geometry of the indentations
and the wall panel thickness twp are paramount to prevent concrete crushing. The twc thickness of
the WC elements was defined as 1.00m due to the same reasons as briefly discussed regarding
system C.
The core functionality of the WP element depends on the dominant deformation pattern
expected for each particular structure. For that purpose, two scenarios can be considered, as
illustrated in Figure 4.37 – b), although both may be relevant for intermediate situations:
▪ On taller piers, where bending is more relevant and longitudinal strains can be
expected;
▪ On smaller piers, where shear is dominant and significant distortions occur;
In bending, tensile strains reduce the friction forces between segments, and only the mechanical
interlock developed under indentation I1 is fully effective. In shear, that could be addressed by
4.41
the addition of shear keys along the horizontal joint, for improved force transfer between
adjacent panel elements and to enable the formation of a second concrete strut under I2.
VEd
shear
I1 distortion
tensile
strains
~30º hwc = 1.50m
I2
Shear Keys?
a) Shear transfer between columns b) Compressive struts formation in bending and shear
Figure 4.37 – WP element functionality
While this system could provide an easier solution to the problem of flexural yielding than
system C, as well as to incorporate the influence of shear in the overall pier response more
effectively than the coupling beam of systems A and B, it also has some unappealing
characteristics from a precast perspective. One of the main shortcomings is the complex and
dense detailing that should be required on the indentations. Both the protruding end on WP
elements and the receiving pocket on the columns need strong reinforcement and confinement,
in order to address the large compressions that are expected to develop there, which may also
further decrease the potential for hollow section solutions. Furthermore, while a methodical
assembly procedure can be established, it still represents several operations per pier, which can
lead to considerable time consumption, especially if dry joints and connections are not accepted
and grouting or local casting operations are also required. Even if that is not the case, some type
of bearing should possibly be required for the contact interfaces between WP and WC elements,
increasing the cost of the solution. Finally, while the other systems are roughly based on precast
solutions that have been successfully applied on other occasions, and therefore can be
envisioned within a reasonable level of previous industry experience, this system is a new and
unproven design idea. Considering the scope of the research programme associated with this
work, commitment to the full study of this layout was considered a big risk.
A general overview of the main characteristics of each system is presented in Table 4.10,
where related pros and cons are summarized. In light of this discussion, it was considered that
System A involved a set of characteristics making it the most suitable for this work and,
therefore, it was adopted for designing the test specimens. One of the main reasons for that
option is the ability to turn the potential weakness associated with precast joints located on high
demand sections into a strong design feature, by selecting those locations for provision of stable
energy dissipation. Furthermore, the main weaknesses of System A were also considered to be
more easily addressed than some of the innate shortcomings of the remaining options.
4.42
Table 4.10 – Precast systems overview
Precast
Main Characteristics Pros Cons
System
Potentially heavy columns;
Two precast columns Common assembly procedure;
Connection forces at critical
+ Suitable for easy manufacture;
A sections;
precast coupling beam; Precast joints for stable energy
Solution sensitive to
Four precast joints per pier; dissipation;
geometry;
Half CIP columns
Connections at low strain Difficult handling of the
+
sections; precast specimen;
B Half precast coupled
Single assembly operation per Solution very sensitive to
columns;
pier; geometry;
Two precast joints per pier;
Wall pier constituted by Difficult to provide reliable
Suitable for both easy
horizontal segments; bending capacity;
C Pier height dependent
manufacture and assembly;
Several connection
Strong shear capacity;
precast joints; operations per pier;
Precast side columns
Wall model for high shear Complex and cumbersome
+
capacity; design;
D Shear panels;
Continuous reinforcement on Unproven success of the
Four contact indentations
bending critical elements; system;
per panel segment;
4.3. TEST SPECIMENS
The framework for the experimental campaign, which will be further presented in Chapter
5, aimed to outline a three phase strategy:
▪ Phase 1: Analyzing different monolithic systems to develop an understanding of

the actual interaction between the columns and the beam, and to evaluate the
performance obtained from different sets of reinforcement layouts;
▪ Phase 2: Evaluating the performance of varied precast connections proposed for
the element-to-element connection between the columns (C) and the beam (B),
according to precast system A;
▪ Phase 3: Evaluating the performance of different precast connections proposed for
the foundation-to-element connection, according to precast system A;
Taking into account a maximum height constraint of 4.00m on the Laboratory of

Earthquake and Seismic Engineering (LESE), the test specimens had to be designed at a
reduced scale. Considering the upper bound of the pier height range set for this work (around
20.00m) and a reduced scale of 1:4, it would result in specimens around 5.00m tall, which is
still larger than intended. With that in mind, a decision was made to study only half structures,
taking advantage of the curvature inflexion point where null moments are observed under lateral
4.43
loading. Furthermore, design of the test specimens for an intermediate pier height within the
same range was considered appropriate. Therefore, the previously mentioned reference column
length of 15.00m was adopted (total pier height of 16.70m), leading to more manageable test
specimen heights around 2.30m on the 1:4 reduced scale.
Within that context, Phase 1 and 2 of this experimental campaign targeted the upper part of
the bent pier frame and the element-to-element connections (A section), while Phase 3 was
aimed at studying the bottom part and foundation-to-element connections (B section), as
illustrated in Figure 4.38.
Phase 1
A A Phase 2
M2
M2
M1 M1
B B Phase 3
Figure 4.38 – Experimental phases in correspondence with elastic moments distribution
In light of the adopted reduced scale, both element geometry and the design parameters
calculated earlier have to be reduced accordingly. Cauchy’s similitude relationships have been
used to determine the actual values for the test models (M) from those fit for a full scale
prototype (P), as represented in Table 4.11.
Table 4.11 – Cauchy’s similitude relationships for a 1:4 reduced scale

Length (m) 𝐿𝑃 = (4) × 𝐿𝑀
Area (m2) 𝐴𝑃 = (4)2 × 𝐴𝑀
Force (kN) 𝐹𝑃 = (4)2 × 𝐹𝑀
Moment (kN.m) 𝑀𝑃 = (4)3 × 𝑀𝑀
Strain (m/m) 𝜀𝑃 = 𝜀𝑀
Stress (MPa) 𝜎𝑃 = 𝜎𝑀
Furthermore, the same materials considered for the response spectrum analyses were also
adopted for test specimen design - C30/37 for concrete and A500 for steel rebars - although the
actual capacity of those used to construct the test specimens was generally significantly higher
than specified. In the relevant cases, the cement grout used for filling ducts and precast joints
4.44
was a proprietary product named Sika Grout (www.sika.pt), with compressive strength in the
range 55 − 65 𝑀𝑃𝑎 and tensile strength in the range 7.5 − 9.5 𝑀𝑃𝑎. An overview of the
general properties of the test specimens for the full experimental campaign, which are addressed
in further detail in the following sections, is presented in Table 4.12.
Table 4.12 – General properties for the full experimental campaign

Concrete properties 2 Column Beam
1 3
Specimen Type E fcc fct ρsl σ sly σslu ρst σsty σstu ρsl σsly σslu ρst 4 σsty σsty
GPa MPa MPa % MPa MPa % MPa MPa % MPa MPa % MPa MPa
Phase 1
SP_M01 M 40 48 4,6 0,74 504 590 0,64 504 590 1,63 601 702 0,26 504 590
SP_M02 M 36 45 3,8 0,74 504 590 0,64 504 590 1,96 601 702 0,50 504 590
SP_M03 M 36 41 4,4 0,74 504 590 0,64 504 590 1,80 601 702 0,79 504 590
SP_M04 M 40 48 3,6 0,74 504 590 0,64 504 590 1,96 601 702 0,79 504 590
Phase 2
SP_PC02A P 36 37 3,0 0,74 504 590 0,64 504 590 1,96 601 702 0,50 504 590
SP_PC02B P 37 45 3,4 0,74 504 590 0,64 504 590 1,96 601 702 0,50 504 590
SP_PC02C P 36 49 3,1 0,74 504 590 0,64 504 590 1,96 601 702 0,50 504 590
SP_M02C M 41 51 3,0 0,74 504 590 0,64 504 590 1,96 601 702 0,50 504 590
Phase 3
SP_F01 M 40 47 2,8 0,74 504 590 0,64 504 590
SP_F02 P 45 57 3,5 0,74 504 590 0,64 504 590 N/A
SP_F03 P 37 47 3,7 0,74 504 590 0,64 504 590
1 - P for Precast; M for Monolithic
2 - Values observed on the test date, or on the closest possible day
3 - Contribution of the diagonal bars calculated by taking into account the slope angle
4 - Not including the contribution of diagonal bars
4.3.1. PHASE 1 – MONOLITHIC SPECIMENS
The geometry outline for all the phase 1 and 2 specimens follows Figure 4.39, respecting
the original double column design of Figure 4.2.
0.55 0.70 0.55
0.41 0.35
2.30
1.88
[m]
Figure 4.39 – 1:4 Reduced scale model geometry
4.45
In addition, a constant enlargement width of 0.02m was considered on the beam-column

nodes, aiming to provide more space for reinforcement placement (since high densities are
expected) and for connection detailing (further discussed on 0). Furthermore, the same column
reinforcement was adopted in all the tests, designed accounting for both maximum and
minimum axial load due to beam shear transfers (corresponding to axial load ratio range
between approximately 0.00 and 0.10), and is illustrated in Figure 4.40.
18 10
6 // 0.10
0,20
0,40
0,015
6 // 0.10
6 // 0.10
6 // 0.10
0,25
0,55
[m]
Figure 4.40 – General column reinforcement for all specimens.

According to the design strategy presented in Figure 4.27, some level of damage and
energy dissipation was also expected in the coupling beam. Aiming to evaluate different
reinforcement layouts to address that problem, four test specimens were designed, where only
the beam reinforcement layouts were distinct. Those specimens were labeled according to the
SP_M# tag, where SP stands for the SIPAV project, M relates to the monolithic aspect, and # is
a number according to the following:
▪ 01: EC8 bi-diagonal layout (Figure 4.23);
▪ 02: Rhombic truss layout (Figure 4.25);
▪ 03: Conventional layout, with horizontal main reinforcement and vertical stirrups;
▪ 04: Rhombic truss layout with adjustment of the diagonal position;
4.3.1.1. Specimen SP_M01
This specimen was designed according to EC8 provisions for coupling beams of shear
walls. In that regard, the design capacity of the beam is dependent of the vertical projection of
the axial force developed in each diagonal, according to equation 4.9:
𝑉𝐸𝑑 ≤ 2 × 𝐴𝑠𝑖 × 𝜎𝑠𝑦𝑑 × sin 𝛼 4.9

Where:
𝐴𝑠𝑖 – Total reinforcement area of one diagonal;
𝜎𝑠𝑦𝑑 - Reinforcement yield strength;
𝛼 – Diagonal slope angle;
4.46
In addition, each diagonal must be detailed as if it were a column element, accounting for a
minimum cross-section size equal to 50% of the beam width (leading to diagonal slope angles
under 25º) and individual confining reinforcement. Additional horizontal and vertical
reinforcement is also included for crack control and to ensure concrete integrity for large
deformations, although it is not expected to contribute to the overall capacity. The result is
presented in Figure 4.41, corresponding to a very dense layout that was a challenge to build,
which is most evident on the beam-column nodes, as illustrated in Figure 4.41 – d). For further
reference, due to an error of the formwork, the actual value of the beam-column node
enlargement cap width on this specimen was 0.05m.
A B
A B
a) Beam detail drawings
b) Beam detail picture

A-A B-B
6
6//0.10 6//0.07
(810) x 2
c) Beam cross sections d) Beam-column node reinforcement
Figure 4.41 – Specimen SP_M01 reinforcement layout
4.47
This specimen was designed according to the model proposed by Tegos and Penelis (1988),
where shear transfer is expected to occur through a complex interaction between different
mechanisms. Design equation 4.10 can be used to calculate the shear strength of the beam,
where Vsd is the shear component carried by the axial load of the main reinforcement (can be
determined from equation 4.9 by using different diagonal slope values), Vst is the shear capacity
associated with the design truss model, and Vc is related with a parallelogram of compressions
carried through the concrete.
𝑉𝑅𝑑 = 𝑉𝑠𝑑 + 𝑉𝑠𝑡 + 𝑉𝑐 4.10
As it stands, the diagonal bars carry most of the load, and the critical aspect for calculation
of the transverse reinforcement is ensuring sufficient capacity to prevent failure at the mid-
section of the beam, where diagonal rebar folding occurs. The resulting layout is presented in
Figure 4.42, where a higher slope of nearly 40º for the diagonal parts of the main reinforcement
is duly noted. The confining reinforcement used at the beam-column node was also adjusted
from the previous specien, accounting for both easier assembly and reduction of reinforcement
ratio, as presented in Figure 4.42 – d).
4.48
A B
A B

A-A 416 B-B
6 // 0.10
8 // 0.10
416
6 // 0.10
6 // 0.10
6 // 0.10
c) Beam cross sections d) Beam-column node cross-section

SP_M03 corresponds to a standard beam design, using horizontal reinforcement on both

sides of the beam, as well as strong transverse reinforcement in the form of vertical stirrups; the
shear capacity of this beam is determined using the regular truss model of Eurocode 2, indicated
in equation 4.11. The confining reinforcement on the beam-column node is the same of
SP_M02.
𝐴𝑠𝑡
𝑉𝑅𝑑 = × 𝑧 × 𝜎𝑠𝑦𝑑 × cot 𝜃 4.11
𝑠
4.49

A-A 212
216
8 // 0.10
6 // 0.10
216
212
c) Beam cross sections

This test specimen is a variant of SP_M02, where the relative position of the diagonal parts
of the reinforcement was adjusted to be centered with the vertical interface between the beam
and the adjacent nodes. The main objective behind such change was to seek an increase in the
local capacity of that location, regarding the large shear transfers that occur therein, aiming to
prevent heavy degradation of the interlock mechanism that leads to failure, by having the
diagonals acting as dowel bars. However, this option has the consequence of creating a zone in
4.50
the mid-section of the beam where only horizontal bars exist, therefore locally reducing the
shear capacity of the beam. In light of that, the transverse reinforcement had to be globally
increased to the level of SP_M03, while the calculation model for the diagonals was the same of
SP_M02 (equation 4.10).
A B
A B

A-A B-B
16
6 // 0.10
8 // 0.10
16
c) Beam cross sections

4.51
4.3.2. PHASE 2 – PRECAST SPECIMENS
In phase 2 tests, all the specimens aimed to evaluate different connection mechanisms and
detailing, therefore all the models used globally the same reinforcement layout. In particular, the
column reinforcement illustrated in Figure 4.40 and the rhombic truss layout of SP_M02
(Figure 4.42) were adopted for that purpose; the only relevant modifications included therein
were designed for the beam-column nodes and the inherent connection mechanisms.
Considering the previously described precast system A, the introduction of rebars through
existing corrugated ducts in the precast beam for moment continuity was deemed the best
option. In light of that, and taking into account that yielding is expected due to the joint
location, some concerns can be raised regarding the efficiency of the connection for seismic
loading. To expand on that, Figure 4.45 presents an illustrative representation of beam-column
nodes subjected to negative moment. Under this load condition, the tensile forces developed
through the longitudinal reinforcement balance the compressions that occur due to element
bending, and moment continuity is achieved by ensuring adequate transfer of those forces
between beam and column rebars. This can be provided through splicing, or by shear anchorage
of the rebars on the surrounding concrete, especially when hooks are included, as illustrated in
Figure 4.45 – a), although on most cases a minimum splicing or anchorage length is required.
When considering a precast element, a vertical bar must be considered for ease of assembly,
preventing the possibility for the mechanical anchorage of hooks. However, direct interaction
between rugged indentations is hindered by the presence of the corrugated ducts, and interface
friction forces develop to engage the rebar inside the duct, as illustrated in Figure 4.45 – b).
0.135
lbd
a) Monolithic node b) Precast node

Figure 4.45 – Beam-column node rebar force transfers
The usual bond length (lbd) code formulas do not directly consider the influence of that
mechanism, and while high quality grouts can enable yielding development on short distances,
the accuracy of lbd calculations for these cases may be uncertain. The implications thereof for
the current study case may be evaluated with a simple example, for discussion sake.
Considering the application of Sika Grout, whose bond strength with deformed rebars is around
15 MPa, the resulting lbd in tension would be around 10ϕ, depending on the actual conditions of
4.52
the application. If ϕ10 bars are used, the anchorage length would be 0.10m, which is
dangerously close to the maximum available length between the top of the column and the
bottom side of the hook.
Aiming to provide improved bond behavior to the connection, three different strategies
were followed in this work, each related to an individual test. The respective specimens were
labeled SP_PC02#, where SP stands for the SIPAV project, PC relates to the precast aspect, 02
characterizes the beam reinforcement layout based on SP_M02, and # is a letter according to the
connection type tested in each:
▪ A: Protruding rebars from the column with post-connection bolting against a steel
plate placed over the node;
▪ B: Splicing rebars inserted into the column, with post-connection bolting against a
steel plate;
▪ C: Splicing “U-shaped” stirrups inserted from the top of the node into the column;
Furthermore, a fourth specimen was also designed, labeled SP_M02C. This last model is a
monolithic replica of SP_PC02C, which was aimed at evaluating the effects of the precast
technology itself, in comparison with the effects of the 50% continuity reinforcement reduction
that was considered for every specimen of phase 2, using precast system A.
4.3.2.1. Specimen SP_PC02A
This specimen accounts for columns produced with protruding rebars, as illustrated in
Figure 4.46. Specifically, in this case, out of the total 18ϕ10 bars of the column, 8ϕ10 bars are
extended onto the beam, and introduced through existing ducts during its descending placement.
Subsequently, 10mm steel plates are put in place on the top of the node, and the joint sections
are externally sealed, after which gravity-flow grouting is performed. The final step of the
connection requires the rebars to be mechanically bolted against the steel plates for increased
anchorage.
4.53
A
A A
1
2 B
C
A
A
Beam cross-section
1
1 B 1
B 1 1
Descending beam into column Assembled structure Column top view
1 - 10 rebars; 2 - M20 rods; A - 30 ducts; B - 50 ducts; C - Steel plates
Figure 4.46 – Specimen SP_PC02A overview
Two M20 rods are also introduced into an inwards position in the columns, aimed at acting
similarly to dowel bars or shear keys and to improve the shear capacity of the precast joints. The
diameter for the ducts was defined as 2 to 3 times the diameter of the bars, in order to provide
tolerance for misalignments and inaccurate geometry. Figure 4.47 briefly illustrates the
sequence of operations required for full assembly of the system.
a) Descending placement of the beam b) Sealing of the joint
Ducts for M20 rods
Steel Plates Bolted ϕ10 bars
c) Node with the steel plates d) Grouted ducts with bolted connections
Figure 4.47 – Specimen SP_PC02A assembly operations
4.54
4.3.2.2. Specimen SP_PC02B
Specimen SP_PC02B mainly aimed to test the efficiency of a connection that could be
performed without the protruding rebars, as those tend to complicate manufacture, handling and
assembly. For that purpose, 100% of the longitudinal reinforcement is cut at the joint section,
and continuity is instead provided by lap splicing. Therefore, additional rebars are introduced
downwards through the beam, into existing ducts extending 0.30m inside the column, placed at
locations that result from a direct offset of the original rebars towards the centre of the cross-
section, according to Figure 4.48. This strategy leads to a reduction of the maximum lever arm
between rebars subjected to tensile and compressive forces at the joint section (0.435m
compared to 0.50m of the original cross-section), which is to be compensated by the adoption of
a larger rebar diameter of ϕ12, in order to avoid further moment capacity reductions.
B B
C
A
3 B
B
Second phase Beam cross-section

of grouting
A
1 B
1
First phase
of grouting
Descending beam into column Assembled structure Column top view
1 - 10 Column rebars; 2 - 12 Splice rebars; 3 - M20 rods; A - 30 ducts; B - 50 ducts; C - 15mm Steel plates;
Figure 4.48 – Specimen SP_PC02B overview
In this case, the gravity-flow grouting process is phased, occurring firstly on the duct
sections that are located in the column, with the upper portion of the ducts grouted only when
the first lower portion is hardened. That leaves the option of performing the beam descending
operation before or after the first grouting phase. While it is believed that the former should be
simpler to execute in a real full-scale application, the latter was adopted in this test specimen
due to time constraints. An overview of different stages required for full assembly of this
specimen is presented in Figure 4.49.
4.55
a) First phase grouting b) Beam descent
c) Sealing of the joint d) Second phase grouting

Figure 4.49 – Specimen SP_PC02B assembly operations
4.3.2.3. Specimen SP_PC02C
The precast connection of specimen SP_PC02C is based on the same concept tested in
SP_PC02B, with 100% of column reinforcement interrupted at the joint section and continuity
ensured by additional splicing rebars. Consequently, the same general layout of Figure 4.48 is
also adopted for the current specimen, whereas the only difference is introduced in the
anchorage system for the splicing rebars. As illustrated in Figure 4.50, SP_PC02C contemplates
the use of “U-shaped” folded rebars (similar to open stirrups) that are introduced through
existing ducts in the node and into the column. This layout enables more flexibility for design
because, while rebar embedment length is limited in the beam, a larger depth is available in the
columns for duct placement (although the same 0.30m length of SP_PC02B was also used in
this case).
4.56
B
Second phase
of grouting Beam top view
1 A
1 B
First phase
of grouting
Column top view

Descending beam into column Introducing the splicing rebars
1 - 10 Column rebars; 2 - 12 "U-shaped" rebars; A - 30 ducts; B - Canals for the stirrups;
Figure 4.50 – Specimen SP_PC02C overview
The same grouting process used in SP_PC02B was also considered for SP_PC02C, with
phased grouting, first in the column sections and later in the upper part of the node. However, in
SP_PC02C, four canals were included on the top of the node, providing connection between
vertical stirrup branches and enabling a horizontal surface. An overview of different stages
required for full assembly of this specimen is presented in Figure 4.49.
“U-shaped” stirrup
a) Beam descent b) Stirrup placement
c) Grouting d) Fully assembled specimen

Figure 4.51 – Specimen SP_PC02C assembly operations
4.57
4.3.2.4. Specimen SP_M02C
As mentioned before, SP_M02C is a monolithic replica of the previous specimen

SP_PC02C. With that in mind, it follows the original rhombic truss reinforcement layout of the
SP_M02 specimen (Figure 4.42), while also including all the alterations introduced to perform
the precast connection, namely the reduction of continuity reinforcement (Figure 4.48) and the
“U-shaped” stirrups (Figure 4.50).
4.3.3. PHASE 3 – FOUNDATION-TO-ELEMENT CONNECTIONS
In phase 3, three different specimens were designed with the main objective of assessing
the behavior of precast foundation-to-element connections. Therefore, the models were labeled
SP_F#, where SP stands for the SIPAV project, F represents the focus on foundation-to-element
connections, and # is a number according to the following:
▪ 01: Fully monolithic specimen;

▪ 02: Precast column with protruding rebars and CIP footing;
▪ 03: Precast column with protruding rebars introduced in corrugated ducts in a
precast or previously constructed CIP footing;
4.3.3.1. Specimen SP_F01
The first specimen of phase 3 tests is a monolithic system, including a column and the
respective foundation, designed for providing a benchmark for the precast models. The
reinforcement layout of the column is the same used for all the phase 1 and phase 2 specimens,
consisting of 18ϕ10 bars placed around the outer perimeter of the cross-section, according to
Figure 4.40. The column element is 1.85m tall and is placed on the top of a square footing with
1.30m width and 0.70m height, which was designed to perform as a rigid element while fixed to
the lab floor through prestressed connections placed near its corners, (see Figure 4.52).
4.58
# 12//0.10
a) Side view b) Footing top view

Figure 4.52 – Specimen SP_F01 overview
Specimen SP_F02 incorporates a similar strategy to that presented earlier in Figure 3.6.
Essentially, that involves precasting the column segment in a previous stage, accounting for the
protruding reinforcement required to provide continuity between the footing and the element.
Afterwards, the element is put in place, and both the reinforcement cage and formwork of the
footing are prepared around it for posterior casting. This process is briefly illustrated in Figure
4.53. The reinforcement layouts for both the column and the footing are the same used in the
monolithic specimen SP_F01.
4.59
Stage 1 Stage 2 Stage 3
Production of the precast segment Preparation of the reinforcement cage CIP production of the footing
and formwork for the footing
Figure 4.53 – Specimen SP_F02 construction process
The main design focus for SP_F03 aimed to enable independent preparation of the precast
column and its respective footing, trying to further improve the appeal of a precast solution as a
powerful tool for flexibility in construction management. For that purpose, the precast system is
required to be fairly cheap and easy to assemble, avoiding complex connections, expensive gear
and disproportionately time-consuming operations. Therefore, a solution using corrugated ducts
was chosen for the present case, in light of the same strategy adopted at the beam-column
precast joints for phase 2 tests. As illustrated in Figure 4.54 – a), the precast column is
manufactured with 0.65m long protruding extensions, which are introduced through ϕ50 ducts
extending 0.50m into the footing (Figure 4.54 – b) and c)).
This strategy involves leaving an empty band of 0.15m on the top of the footing
(cross-section cut 1-1, Figure 4.54 – c)), which is meant for second stage casting after the
associated transverse reinforcement is prepared. Furthermore, while the current column
cross-section (2-2, see Figure 4.54 – a)) follows the original layout, this section is temporarily
supported on top of a 0.15m width concrete core cube (previously cast from the same concrete
mix used for the precast segment), to avoid resorting to more elaborate bracing solutions,
according to Figure 4.54 – c). Normal or micro-concrete should be provided for the second
stage casting of the empty band, although in the present work normal grout was used, and the
concrete core cube used for temporary supporting of the precast column segment is not
removed, remaining within the section as an integral part of the structure.
4.60
# 12//0.10
b) Footing top view
a) Precast column side view

Second stage casting
Concrete cube
c) 1-1 Cross-section cut d) Footing side view
Figure 4.54 – SP_F03 overview
The complete assembly process of SP_F03 is the following, for which steps 3 to 6 are
illustrated in Figure 4.55:
1. Manufacture of the precast column;

2. Construction of the footing with embedded corrugated ducts;
3. Placement of the column on the top of the footing by introducing the protruding
4.61
rebars through the relevant ducts. The system is temporarily supported by a 0.15m
concrete cube;
4. Gravity-flow grout is used to fill the ducts, sealing the bottom part of the
longitudinal reinforcement;
5. Preparation of the transverse reinforcement in the empty band;
6. Second stage casting of the bottom section of the column. The concrete core cube
serving as temporary support of the precast segment is left within;
4-5 6
Figure 4.55 – SP_F03 assembly process
4.4. FINAL REMARKS
This chapter was aimed at describing the strategy and thought process associated with
conceptual design of the test specimens. For that purpose, the HSRL Poceirão-Caia design
proposal was presented as a core element of this study, from which it was possible to evaluate
the structure layouts and design options that could benefit from large-scale application. In that
regard, a decision was made to focus on medium to long viaducts, (100m to 600m)
characterized by double box-girder decks, and supported by double-column piers between
5.00m and 20.00m tall, where the transverse performance was the main concern.
Within that context, the proposed double-column pier structure is defined by geometrical
characteristics that may pose a challenge for adequate seismic design, due to low ductility
capacity and proneness to shear-critical performance. This is mainly caused by the short span of
the coupling beam which, as observed, has non-negligible impact on the stiffness relation
between the columns and the beam and, therefore, also on the overall pier structure behavior for
pier heights within the range relevant to this work. Nonetheless, the previously described
chosen structural layout enabled the selection of a few case study viaducts for numerical
4.62
modeling, aiming to obtain reference data using the Response Spectrum Method to help with
test specimen design.
In that regard, the outline of the experimental campaign and its test specimens were also
presented, characterized by a three phase process and a total of 11 reduced scale (1:4)
specimens tested under lateral cyclic loading conditions. The first phase involves the study of
monolithic half-pier structures, resorting to different beam reinforcement layouts designed for
provision of improved ductility under strong shear demand, using similar strategies to those
developed for reinforcement of coupling beams on shear walls. Its main purpose is aimed at
observing and analyzing the actual interaction between the columns and the beam, as well as
evaluating the ductility capacity and failure characteristics of the current double column pier.
Additionally, the first phase enables choosing one of the beam reinforcement layouts (according
to the observed performance on the monolithic structures) to use in the design of phase two
specimens.
Phase two is entirely focused on evaluating precast solutions for the double column pier,
which is achieved by testing specimens designed with equivalent reinforcement and structural
layouts, but using different precast mechanisms developed with attention to some of the main
industry concerns. Finally, phase three is comprised of only three specimens, which were aimed
at evaluating the performance of precast connections established at the footing-to-element
interface.
4.63
Experimental Campaign
5
EXPERIMENTAL CAMPAIGN
As previously discussed and highlighted in Table 4.12, the experimental campaign devised
for this work involved three different stages, each related with different objectives. Therefore,
the first section of this chapter will focus on presenting the main test setup details according to
the clear distinction between bent pier and single column tests.
The later sections focus on presenting the main experimental observations, and on
providing comparison between the performances of different systems according to generalized
response parameters, which is also performed separately for testing stages 1 and 2 and testing
stage 3.
5.1. TEST SETUP
Two test setups were developed at LESE to support this work, based on uniaxial cyclic
loading of the test specimens according to the different circumstances associated with each
testing stage. As previously mentioned, all the tests follow a reduced scale of 1:4.
5.1.1. PHASE 1 AND PHASE 2 – BENT PIER TESTS
The test setup assembled for this part of the work aimed to make the most use of the
already existing infrastructures and equipment of the LESE laboratory. With that in mind, the
main features associated with it were the following, also highlighted in Figure 5.1 for
illustration purposes:
▪ 1 horizontal actuator (+/- 500 kN capacity; +/- 150mm stroke);

▪ 2 vertical jacks (500 kN capacity);
5.1
▪ 4 prestress threadbars (Dywidag 26WR);

▪ Mechanical hinged system;
▪ Reaction frame;
Vertical Jacks Reaction Frame
N S
Horizontal Actuator
Prestress Threadbars
Hinge System
a) Construction drawings b) Highlighted picture view

Figure 5.1 – Phase 1 and Phase 2 test setup
For the purpose of accuracy, the development of this experimental layout for assessing the
cyclic performance of reduced scale bridge bent piers requires realistic simulation of the forces’
transfer that occur during the seismic motion of an equivalent real scale structure. With that in
mind, three main aspects were defined as the core guidelines for test setup design:
▪ Constant axial loading on the columns;

▪ Free rotation ability for simulation of curvature inflexion points due to moment
symmetry simplification;
▪ Application of the lateral loading with respect to the expected interaction between
bridge decks, bearings and the piers;
In order to address specific issues related to each of these topics, separate discussion is
promoted in the following sections.
5.1.1.1. Axial Loading
The applied type to a bridge pier is mostly caused by the dead weight of the bridge deck
and associated permanent loads, which remain constant for the duration of any seismic event.
With that in mind, keeping constant axial load in the test specimens is paramount, as it is well
known that axial load variations affect the behavior of cyclically loaded columns. On single
column structures, it is common to use vertically placed hydraulic actuators to compress the
element body, possibly including some type of rotation or sliding enabling device designed to
accommodate the pier head displacements (e.g. Delgado et al. (2009), Rodrigues et al. (2013)).
More importantly, when that issue is accounted for, the vertical deformations on the columns
are mainly due to the initial compression, while during lateral deflection the overall variation of
5.2
axial deformation is negligible. Therefore, significant axial load variations are also not
expected.
On laterally loaded bent piers, the structural system may require different handling of that
problem, since individual element deformations can be critical for keeping constant axial load.
As seen in Chapter 2, the most common configuration associated with bent piers relates to
girder bridges, or equivalent deck layouts that provide a wide support length for good transverse
load distribution. The bent systems also usually represent a strong beam and weak column
design, enabling nearly rigid beam behavior. In that case, many authors opt to study individual
beam column joints (“T” or “knee” joints), as opposed to studying complete bents, since
capacity design reinforced beams ensure similar behavior between all of the vertical elements.
Nonetheless, when the complete pier is considered, dead-load application on test specimens can
be performed in different ways, depending on the laboratory features, budget and purpose of the
study:
▪ Including deck masses in the test setup: This approach can be adopted by
including correctly placed and/or scaled external masses on top of the bent (Figure
5.2 – a)), or by constructing equivalent deck structures (Figure 5.2 – b)). It has the
clear advantage of enabling the actual interaction between superstructure and
substructure, with the column axial load variations determined by the geometric
and material properties of the bent, as well as the loading pattern. It is the most
accurate experimental simulation of axial loading conditions, especially suited for
dynamic shaking table tests. On the other hand, the resources and budget required
to establish this type of setup are considerable, and often impractical, for the
purpose of most experimental studies.
a) Mostafa et al. (2004) b) Pantelides and Gergely (2008)

Figure 5.2 – Test setup examples including deck masses
▪ External prestress application: This approach is generally more frequent in

quasi-static cyclic loading tests. One option is to use hydraulic jacks in
5.3
conjunction with prestress rods to continually apply pressure on each column

(Figure 5.3 – a)). Another one is to globally prestress the whole structure, using a
large beam or similar structure on top of the bent frame to enable a wider
distribution of the vertical load (Figure 5.3– b)). It is a testing layout that occupies
a larger space, but generally requires fewer resources (such as jacks or prestress
rods), while also being faster to install regarding multiple column structures.
a) Mclean et al. (1998) b) Hose and Seible (1999)

Figure 5.3 – Test setup examples including vertical prestress for axial load application
The bent system studied in this work is constituted by strong columns, which are enhanced
by the coupling action provided by the short beam. In this case (as discussed in Chapter 4) the
shear forces transferred through the coupling beam lead to the occurrence of significant
variations of the total axial load that is mobilized through the columns, which are dependent of
the lateral displacement of the structure and, thus, should be taken into account.
Considering the resources available at the LESE laboratory and the scope of quasi-static
cyclic loading tests, an external prestress solution was the most suited for this problem.
Therefore, a system similar to that of Figure 5.3 – a) was designed. Two Dywidag 26WR
threadbars and a 500 kN jack (Enerpac CLRG-502) enabled the individual application of
targeted prestress loads of 300 kN per column (which is around the dead-weight per column of
4613 kN calculated earlier, scaled down to a 1:4 reduced value). Furthermore, to keep the
overall external forces as constant as possible, while accounting for the beam shear induced
axial load variations, the jacks were kept under force-controlled hydraulic pressure, involving
real time adjustments for the purpose of maintaining the intended threadbar stress level.
In addition, a load distribution beam, made of reinforced UNP200 steel shapes was also
adopted (Figure 5.4 – a)), including two hinged devices on each edge to enable the connection
and simultaneous tensioning of both threadbars by the same jack, according to the red highlight
in Figure 5.4 – b). A similar hinged device was also installed on the other end of the threadbars,
5.4
enabling them to follow the rotation of the columns to maintain the direction of their respective
axial load forces.
a) Load distribution reinforced beam b) Hinged device at the beam edge

Figure 5.4 – Axial loading system details
Finally, it must be acknowledged that the hydraulic flow capacity of the jacks and the
corresponding power rig is naturally limited, which also limits the rate at which the strain
increments due to shear can be compensated. Because of that, even though this system aims to
provide real-time adjustments to the threadbars stress level, small variations can still occur if the
rate at which those vary differs from the rate at which the system is capable of adjusting. With
that in mind, strain gauges were installed on all prestress bars for monitoring purposes.
5.1.1.2. Free Rotation
According to the design concepts previously discussed on Chapter 4, a decision was made
to only study half structures, taking advantage of the null moment point expected from the
lateral loading of the bent pier. In order to fulfill that requirement, curvature inflexions should
be simulated with free rotation constraints at the column bases, while also accommodating
increasing intensity pier-head lateral displacements and the resulting high shear and axial load
levels. For that purpose, column bases were bolted against free rotation steel plates, placed over
mechanical hinges (Figure 5.5) designed for multi-directional (vertical and horizontal) shear
stress states within the expected axial and shear loading variation thresholds.
5.5
a) Top view b) Side view

Figure 5.5 – Mechanical hinge and free rotation steel plates
This system provides free rotation ability to each of the columns, enabling the rotation of
their vertical axes, which is also accounted for in the axial loading system. In that regard,
aiming to prevent undesired eccentric moments and/or second order effects, the threadbar
rotation centre should be the same as the column’s, enabling equivalent movement and
concentric axial loads. That was not possible to implement, however, since the mechanical
hinge was not designed to include a connection for anchoring the threadbars. Therefore, a
compromise solution was adopted, involving the attachment of the bottom hinged devices in the
shortest possible vicinity, which was just below the mechanical hinges, as illustrated in Figure
5.6 – a). As a consequence, the column and threadbar axial loads become increasingly
misaligned as pier-head displacements increase, as illustrated in Figure 5.6 – b). Nonetheless,
for the horizontal displacement level associated with the actuator stroke (150mm), the
misalignment would be around 2cm on the base of the columns, leading to a maximum
eccentric moment of 6 kN.m. That value corresponds to less than 2% of the maximum expected
moment, and therefore was considered negligible for the purpose of this study.
5.6
Threadbar Hinge device
Mechanical
Hinge
2,47
Hinge
device
Column axial load
Mechanical Hinge Center
0,47
Threadbar axial load
Threadbar Hinge Device Center
a) Setup detail b) Eccentric moment from lateral loading

Figure 5.6 – Mechanical hinge and threadbar cap relative positions
5.1.1.3. Lateral Loading
The application of a lateral force aims to represent the action of the inertia forces
developed during seismic events (namely from bridge deck masses), which are transmitted to
the piers through the associated bearing devices. According to Marioni (2006), in HSRL
structures elastomeric devices are generally not considered, with the adopted bearing schemes
usually revolving around a combination of pot-bearing devices and shear keys in order to
provide fixed connections between decks and piers. Therefore, since seismic forces are applied
at the pier head level, aiming for accurate experimental simulation involves establishing a
similar shear interaction between the hydraulic actuator and the test specimens.
In single column piers, the seismic force is fully transferred at the same location, and
experimental simulations only require that actuator heads are firmly attached to the targeted
loading point. Typical applications generally require the use of steel pieces bolted around the
column for enabling the pier head movement, as highlighted in the examples of Figure 5.7.
a) Detail from Delgado et al. (2011) b) Detail from Pang et al. (2010)
Figure 5.7 – Lateral load application examples for single columns
5.7
Considering the geometry of this double column bent, however, replicating the previously
described procedure is of increased difficulty, since the full seismic force developed at the deck
level is transferred to two columns instead, meaning that the experimental setup should aim to
provide simultaneous loading of both vertical elements. Using two in sync actuators (one for
each column) under displacement control conditions could be possible if their respective
loading rates were kept low enough, but that option was not eligible due to the availability of
only one reaction frame for a single actuator. With that in mind, the only suitable solution
considered application of the lateral load on a single column, while ensuring displacement
compatibility over the column heads. For that purpose, the mechanical device presented in
Figure 5.8 was designed for attachment to both columns, maintaining constant displacement
between them for the expected load levels.
Coupled Loaded
Column Column
a) 3D view b) Overall placement in the setup

Figure 5.8 – Displacement compatibility device
Despite that, preliminary numerical simulations revealed that forcefully coupling the
column heads actually increases the overall bent pier lateral stiffness, since the columns rotate
around the base mechanical hinges, and the coupled points on the pier heads are vertically
displaced while keeping constant (but not always horizontal) distance between them, as
illustrated in Figure 5.9 - a). This means that a constant length device cannot actually ensure
equal displacements between the two columns without being subjected to non-negligible axial
strains, which may be a source of additional stiffness due to creating an artificial displacement
constraint to the natural rotation of the column heads. This is further illustrated in Figure 5.9 -
b), where reaction force results from a monotonic plane stress analysis (using the same
modeling strategy that will be presented in Chapter 6) while accounting for constant distance
between column heads are plotted against the imposed displacements at the loaded column. It
can be observed the occurrence of a negative reaction force at the coupled column head,
meaning that the constant width constraint was actually limiting the natural movement of the
latter and, consequently, increasing the overall load applied at the other column. Eventually, on
a real setting this issue might be less noticeable due to the unavoidable flexibility of mechanical
5.8
connections (gaps and slip between bolts and screws), contrasting with the rigid limitations
imposed by a numerical constraint. Nonetheless, the possibility of affecting the experimental
results and undermining the respective conclusions led to the decision of completely avoiding
the use of the displacement compatibility device.
800
600
400
Force (kN)
200
dv
0
dh dh dv
-200
-400
-600
Loaded column
Coupled column
-800
0.00 5.00 10.00 15.00 20.00 25.00 30.00
FCoupled FLoaded Displacement (mm)
a) Movement during lateral displacements b) Horizontal reaction forces

Figure 5.9 – Numerical simulation including constant displacement between coupled points
With that in mind, despite this seeming to be the most suitable solution to avoid disturbing
the natural behavior of the bent piers, it inevitably leads to differences between the
displacements recorded at each column. The magnitude of those differences is dependent of the
damage accumulated in the beam, since that is the element through which the lateral load is
transmitted from the loaded column to the coupled column, and also dependent of the load
application point. Assuming a sideways loading procedure, two options were established:
▪ Application of the lateral load at the beam horizontal axis level;

▪ Application of the lateral load through shear interaction at the column head;
In a real bridge setting, horizontal forces are transferred through the bearings mainly by
shear, at the pier head surface. If the beam-column nodes are small, it is generally safe to
consider the load applied at the beam horizontal axis level because the distance between that
point and the pier head surface is also small, therefore no significant changes in the overall force
distribution are expected. This would be the preferable option for experimental implementation
(similarly to single column procedures (Figure 5.7)), since it enables easier displacement
monitoring and load application, by attaching the actuator head directly to the column.
However, that simplification may not be reasonable for the present case, considering the
large dimensions of the test specimens and, particularly, of the beam-column nodes, since there
is a non-negligible difference between the horizontal beam axis and the column head surfaces,
which may result in considerably different distributions of internal force. Figure 5.10 illustrates
a possible strut-and-tie representation of the force distributions inside the beam-column node for
5.9
positive (left) and negative (right) moments where it is clear that applying the load at beam axis
level could disturb the internal force distribution and result in a different stress state.
Considering the importance of the beam-column node behavior for the overall performance of
these structures, a decision was made to establish shear induced loading, in order to represent
the bridge pier seismic loading scenario as realistically as possible.
Shear Loading Shear Loading
Beam Axis Beam Axis

Loading Loading
Figure 5.10 – Strut-and-tie representation of beam-column nodes under moment loading
With that in mind, a system had to be designed for appropriate shear loading of the
specimens. The adopted solution was constituted by sets of two HEB200 steel shapes welded
together and reinforced with additional ribbing plates, which were placed on the top of each
column (Figure 5.11 – a)). That steel grid structure also included eight holes for the same
number of M20 rods (designed for working at 50% of their design shear strength and
considering an embedment length of 0.20m into the column) for fixation and shear loading
purposes, according to the illustration presented in Figure 5.11 – b).
a) Picture view before a test b) Plant view

Figure 5.11 – Lateral shear loading system
Finally, the lateral loading system required the establishment of two external connections
for force transfer to the reaction frame: a fairly self-explanatory one for the actuator, and an
5.10
additional one for a force retention mechanism on the base of the experimental setup. In that
regard, while the application of the lateral load on the test specimens has already been
thoroughly discussed, appropriate anchoring for the whole structural system is also a concern.
In particular, since free rotation in the column bases was expected by design, footings could not
be considered to receive and transmit the loads to the strong slab of the laboratory.
Instead, a base supporting system for the bent pier specimens had to be developed to
provide appropriate reaction against both vertical and horizontal loads, including connections to
the vertical load cells and the prestressed threadbar connections, according to Figure 5.6 – a).
Furthermore, accurate readings on the vertical load cells require that mostly axial displacements
are recorded on their sensitive contact surfaces, but that is difficult to enforce when reaction
forces are expected to rotate in correspondence with the pier drift. In order to address that issue,
the force retention system installed on the base of the experimental setup (illustrated in Figure
5.12) consisted of an assembly of several steel beams connecting the column bases to the
reaction frame just above the load cells. The purpose of this mechanism was to provide high
stiffness to lateral movements, aiming to redirect the shear force from each column back to the
reaction frame, avoiding significant rotations at the contact surface on the load cells underneath
and, therefore, enabling more accurate readings.
Load Cells
a) Threadbar connections through the

b) Picture during assembly
slab
Figure 5.12 – Base supporting system with force retention beams
In addition, the first test (SP_M01) was performed using a rigid connection between the
different steel beams along the force retention system (highlighted with the red shade within the
red rectangle in Figure 5.12 – a)). That solution proved to be inefficient, causing the occurrence
of shear/bending in addition to the expected axial load. In order to address that issue on the
subsequent tests, the rigid connection was replaced by two mechanical hinge connections,
illustrated in Figure 5.13 – a) and b). Preliminary laboratory tests showed that, with this
improved setup version, no shear forces were transmitted through the force retention system to
the reaction frame, and further data recorded during the actual tests showed that peak forces at
5.11
this level were always above 75% of the horizontal load applied at the top of the piers,
evidencing the role of this force retention system as the main horizontal load reaction element.
a) Hinged connection to the supporting

b) Hinged connection to the reaction wall
system during preliminary test
Figure 5.13 – Hinged force retention system
5.1.1.4. Out-of-Plane Bracing
As described, this experimental setup was developed to analyze the behavior of the bent
pier under lateral seismic loading, representing the corresponding “in-plane” motion. The
adoption of a testing layout especially suited to provide uniaxial demand was presented, but a
bracing system was also prepared to accommodate eventual out-of-plane deformations. That
possibility is a concern because it is impossible to achieve perfect symmetry in the test
specimens along the loading plane. On that regard, not only the internal issues that may arise
from a difficult concrete casting procedure (e.g. bad vibration, unequal concrete cover depths
and/or rebar layouts) are worthy of mention, but the assembly of each test specimen on the
experimental setup can also relate with slightly unsymmetrical loading (mainly due to unequal
connection gaps) of both columns.
The impact of these issues on the overall behavior of a single pier testing system is
generally perceived to be negligible and easily controlled by the testing setup. However, in the
presented setup only the displacement of one of the bent pier columns is controlled by an
external device, as the coupled column is dependent of the forces transferred through the beam.
Within that context, it is understandable that with increasing progress of structural damage, an
eventual small initial asymmetry can lead to larger differences along the loading plane and,
therefore, increased possibility of out-of-plane deformations.
With that in mind, a bracing system was prepared around the column heads and coupling
beam, aiming to provide increased stiffness against such movement. It was constituted by a set
of several steel elements connected to a reaction wall, including a rolling system to enable
5.12
uniaxial longitudinal displacements while reacting against transverse out-of-plane

displacements. This feature is further illustrated in Figure 5.14, highlighting the rolling system
inside the red rectangle shape.
Figure 5.14 – Out-of-plane bracing system
5.1.1.5. Loading Protocol
The tests involved the application of uniaxial cyclic loading under displacement controlled
conditions. Two loading displacement histories were considered: LH1 for the first test,
SP_M01, while LH2 was applied for all the remaining tests, both characterized in Table 5.1 in
terms of drift, and plotted in Figure 5.15 – a) and Figure 5.15 – b), respectively.
Table 5.1 – Loading displacement histories characterization – Phase 1 and 2

LH1 LH2
Actuator Actuator
Drift Load rate Drift Load rate
Displacement Cycles Displacement Cycles
(%) (mm/s) (%) (mm/s)
(mm) (mm)
0.05 1.15 1 0.2 0.05 1.20 1 0.2
0.10 2.30 3 0.2 0.10 2.40 1 0.2
0.19 4.37 1 0.5 0.20 4.80 3 0.5
0.29 6.67 3 0.5 0.30 7.20 1 0.5
0.38 8.74 1 0.5 0.40 9.60 3 0.5
0.48 11.04 3 0.5 0.50 12.00 1 0.5
0.72 16.56 3 1.0 0.75 18.00 3 1.0
0.96 22.08 3 1.0 1.00 24.00 3 1.0
1.44 33.12 3 1.0 1.50 36.00 3 1.0
1.92 44.16 3 1.0 2.00 48.00 3 1.0
2.40 55.20 3 1.0 2.50 60.00 3 1.0
2.88 66.24 3 1.0 3.00 72.00 3 1.0
3.35 77.05 3 2.0 3.50 84.00 3 1.0
3.83 88.09 3 2.0 4.00 96.00 3 1.0
4.31 99.19 3 2.0 4.50 108.00 3 1.0
4.79 110.21 3 2.0 5.00 120.00 3 1.0
5.13
5.00% 120 5,00% 120
96 96
Displacement (mm)
Displacement (mm)
Horizontal Drift
Horizontal Drift 3.00% 72 3,00% 72
48 48
1.00% 24 1,00% 24
0 0
-1.00% -24 -1,00% -24
-48 -48
-3.00% -72 -3,00% -72
-96 -96
-5.00% -120 -5,00% -120
a) LH1 b) LH2
Figure 5.15 – Loading displacement histories
The differences between the two were not significant, and mainly relate to some minor
adjustments performed in light of observations made during the first test (using LH1), namely:
▪ Reducing the loading rate on the larger displacement cycles;

▪ Adjusting the occurrence of repeated loading cycles only after cracking begins;
▪ Readjustment of the target drift values;
With respect to the first of the previous adjustments, it resulted from the observation of
axial load variations of ±50 kN on the last few cycles, caused by insufficient jack hydraulic flow
for real-time compensation of the axial strains caused by the cyclic displacement loading rate of
2.0 mm/s. The second adjustment, as mentioned, was aimed at enabling the occurrence of the
repeated loading cycles to the 0.20% to 0.40% drift levels, since degradation is only expected to
occur with the onset of cracking (which was observed to occur at that stage). The last
adjustment mainly intended to adapt the target displacement levels for provision of drift values
more suited for presentation.
Taking into account all of the latter, it must be acknowledged that a direct comparison
between the SP_M01 test and all the other tests is innately hindered by the differences in their
respective loading protocols and test setup. Despite that, a conscientious analysis and
interpretation of experimental results should still enable comparison of the overall behavior and
main parameters.
5.1.1.6. Monitoring Layout
A set of Linear Displacement Voltage Transducers (LVDTs), complemented with a few

Draw Wire Transducers (DWT, potentiometer based) were used to monitor the displacements of
several points in the structure, and two tiltmeters were placed on the top of the column heads to
record beam-column node rotations. In addition, strain gauges were also installed on the
threadbars, as previously discussed, as well as on the main beam of the force retention system,
the latter aiming to evaluate bending or shear deformations. Furthermore, imposed loads and
reaction forces were recorded by the following:
5.14
▪ Internal load cell (LC) of the horizontal actuator;

▪ Pressure transducer of the hydraulic jack lifting system;
▪ External Load Cell 1 (LC1) placed below the southside column;
▪ External Load Cell 2 (LC2) placed below the northside column;
▪ External Load Cell 3 (LC3) placed in the horizontal force retention system;
Figure 5.16 presents a full overview of the layout used in this work including all the
previously mentioned features, where the digit based numbering scheme refers to LVDTs while
LF## refers to DWTs:
North Side (+) South Side (-)
35 36 37 39 Internal LC and LVDT

0,100
60a 50a
0,338
8 3 4 14 25
15 23
26 11 20
24
16 19
18
34 22 33
0,225
27 6 1 2 7 21
46a 47a 48a 49a
32 29
0,05
0,479
1 2
LF 03 LF 04
0,491
5 9
LF 01 LF 02
Tiltmeters
0,868
Strain Gauges
LC3
LC2 LC1
Figure 5.16 – Schematic overview of Phase 1 and 2 monitoring features
5.15
5.1.2. PHASE 3 - SINGLE COLUMN TESTS
The test setup assembled for the third phase of this work was adapted from other existing
and proven layouts used on previous works at LESE, to the specific conditions of these
columns. A general overview of the adopted system is presented in Figure 5.17, including the
following:
▪ 1 horizontal actuator (+/- 500 kN capacity; +/- 150mm stroke);

▪ 1 vertical jack (500 kN capacity);
▪ 2 prestress threadbars (Dywidag 26WR);
▪ Reaction frame;
Horizontal Actuator
Axial Loading System
Hinged anchorage and

Footing fixation
Figure 5.17 – Phase 3 test setup
The lateral loading system of phase 1 and 2, constituted by a horizontal 500 kN capacity
actuator and the respective shear load transfer system (Figure 5.11) was installed for this
application, as well as the previously used axial loading system (due to its capability to rotate
and adjust to large displacements of the pier head) prepared for the target load of 300kN.
5.16
However, due to this testing phase focusing on the bottom part of the bent system in study, the
elements of the corresponding threadbar hinge device had to be modified. As observed in Figure
5.17, the adopted solution took advantage of the prestress connection used to attach the footings
against the laboratory slab, to also provide fixation for steel beams placed parallel to the
column, where the hinged anchorages were attached.
5.1.2.1. Loading Protocol
Phase 3 tests were performed under displacement control conditions, involving the
application of a displacement history with the same characteristics of LH2 (Table 5.1 and Figure
5.15 – b)), calculated for the displacement levels of the present horizontal actuator level,
according to Table 5.2.
Table 5.2 – Loading time-history characterization – Phase 3

LH3
Actuator
Drift Load rate
Displacement Cycles
(%) (mm/s)
(mm)
0.05 0.98 1 0.2
0.10 1.97 1 0.2
0.20 3.93 3 0.5
0.30 5.90 1 0.5
0.40 7.86 3 0.5
0.50 9.83 1 0.5
0.75 14.74 3 1.0
1.00 19.65 3 1.0
1.50 29.48 3 1.0
2.00 39.30 3 1.0
2.50 49.13 3 1.0
3.00 58.95 3 1.0
3.50 68.78 3 1.0
4.00 78.60 3 1.0
4.50 88.43 3 1.0
5.00 98.25 3 1.0
5.1.2.2. Monitoring Layout
This testing stage was mostly focused on recording overall lateral displacement profiles for
the single columns, and analyzing the behavior of their respective foundation-to-element
connections. With that in mind, a simple layout of LVDTs was adopted along both sides of the
whole column height, involving a concentration of recordings near the expected plastic hinge
region for the two following specific output results:
▪ Lateral displacements – horizontal LVDTs and DWTs;

▪ Longitudinal deformations (crack openings) – vertical LVDTs;
Figure 5.18 illustrates the recording layout just for the North side of the columns, although
the South side includes equivalent instrumentation.
5.17
Actuator Axis Actuator Axis

60
34 + LF
15
LF09
LF01
LF05
12
LF03
4
35
a) Front view (horizontal displacement b) Side view (vertical displacement

transducers) transducers)
Figure 5.18 – Schematic overview of Phase 3 monitoring features and overall dimensions
5.2. EXPERIMENTAL RESULTS – BENT PIER TESTS
5.2.1. PHASE 1 OBSERVATIONS
The focus of this section regarding the results of Phase 1 is to present and discuss the
experimental evidence related to the damage progression and failure of each specimen. The
force – drift curves that are addressed were defined for the internal displacement of the actuator,
according to Figure 5.16. Furthermore, a collapse threshold corresponding to a reduction of
20% of the peak force was also defined (Park and Ang (1985)) for the purpose of analysis and
discussion, although tests continued until complete failure of the specimens, determined through
visual inspection. In addition, the values recorded for lateral force and beam shear are also
presented and compared with predictive estimates calculated using the real material properties
presented in Chapter 4. For this purpose, the equilibrium model represented in Figure 5.19 can
be used to determine the lateral peak force, assuming overall capacity governed by column
bending or by beam shear. In the present case the latter was considered (critical beam shear),
according to experimental observations, enabling the calculation of the lateral force for two
scenarios:
▪ Yielding: Peak strength assumed at yielding of the main beam reinforcement;

▪ Ultimate: Peak strength assumed after hardening of the beam reinforcement;
5.18
1.250m
Flateral
For critical beam shear:
Vbeam
Vbeam 𝑉𝑏𝑒𝑎𝑚 × 1.250
𝐹𝑙𝑎𝑡𝑒𝑟𝑎𝑙 =
2.565m 2.565
2.050m
For critical column bending:

𝑀𝑐𝑜𝑙𝑢𝑚𝑛
Vcolumn Vcolumn 𝐹𝑙𝑎𝑡𝑒𝑟𝑎𝑙 = ×2
2.050
Rv,north
Rv,south
Figure 5.19 – Equilibrium model
Due to the way the lateral load is applied to the structure, some level of response
asymmetry is expected. In fact, the axial load that is transmitted from the loaded column to the
𝐹𝑙𝑎𝑡𝑒𝑟𝑎𝑙
coupled column is approximately 2
. However, for the positive (pushing) loading direction
it is carried through compression of the coupling beam, while for the negative (pulling)
direction it is carried through tension. The effective loading of the structure is then dependent of
the level of damage incurred in the beam, and it may impact the stiffness of the coupling action
differently depending on whether the element is compressed or under tension. For that reason,
predictive estimates according to the equilibrium model are performed only for the positive
loading direction.
5.2.1.1. SP_M01 specimen results
The cyclic force-drift curve for this specimen is illustrated in Figure 5.20, where the
collapse threshold is also identified, occurring at the 2.50% drift level on the positive loading
direction. In addition, there is a clear distinction between positive and negative loading
directions, which causes a difference in the peak force level recorded on both, of around 16%.
Regarding this subject, the lateral force estimations are also illustrated in Figure 5.20:
5.19
225
Collapse
175 Ultimate
Yielding
125
75
Force (kN)
25
-25
-75
-125
-175
Fp+ = 196.41 kN
Fp- = 169.70 kN
-225
-4.00 -3.50 -3.00 -2.50 -2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Lateral Drift (%)
Figure 5.20 – SP_M01 force-drift results
As it is possible to observe, the experimental peak force overtakes the values calculated using
the equilibrium model presented earlier by almost 30%, considering the shear capacity of the
beam as defined by EC8 (equation 4.9). This can be further analyzed in Figure 5.21, which
illustrates a comparison between the theoretical shear capacity of the beam calculated for the
Yielding and Ultimate scenarios and the peak value recorded in the experimental test,
determined by subtracting the axial load carried by the threadbars from the vertical reactions.
Again, the values calculated using the bi-diagonal design model are both smaller than the
experimental value, although with smaller difference, around 23% for the Yielding scenario and
4% for the Ultimate scenario.
In light of these results, there is the possibility that the additional vertical reinforcement
provided for crack control is also contributing to the shear capacity of the beam, which is not
accounted for in the coupling beam design model nor in these calculations, and could help
explain the larger capacity of the experimental model. Furthermore, the shear capacity of the
EC8 design model is governed by the tensile force on the steel rebars, and no additional strength
provision is considered through concrete alone, which may be a conservative evaluation.
It should also be noted that the experimental capacity may, in fact, have also been
artificially overestimated by non-monitored shear forces transferred through the retention
system, since this test was performed with the rigid connection (Figure 5.12 – a)).
Unfortunately, the first version of the test setup was not adequately prepared to deal with that
issue, therefore it was not possible to understand how much did it influence the results.
5.20
350
Yielding
Ultimate 326.79
300 Experimental 313.34
250 265.54
Beam Shear (kN)

200
150
100
50
Figure 5.21 – SP_M01 Beam shear: comparison of experimental and theoretical estimates
The observed failure mode of the specimen can be attributed to the occurrence of “sliding
shear” on the interface between the beam and the north side column, and a relatively
undamaged structure can be observed on other locations, as illustrated in Figure 5.22.
Figure 5.22 – SP_M01 after testing
The cracking pattern associated with this specimen also showed significant differences
between the beam and the column. The first cracks appeared in the beam around 0.20% drift,
with inclined orientation typical of shear, and new cracks continued to form up until 1.00%
drift. Afterwards, that process stabilized and only crack width increases could be detected. This
can be observed in Figure 5.23, for a sequence of 0.20%, 0.50%, 1.00% and 2.50% drift.
0.20% 0.50%
S N S N
Figure 5.23 – SP_M01 Beam cracking pattern
5.21
1.00% 2.50%
S N S N
Figure 5.23 (cont.) – SP_M01 Beam cracking pattern
This contrasts with what was observed in the columns, which is illustrated in Figure 5.24,
where cracks developed only after 0.75% drift and stabilized at 1.00% drift, displaying a regular
distribution that seemed to closely follow the distance between transverse reinforcement stirrups
(0.10m) until collapse.
0.50% 0.75%
N S N S
2.50%
N S
~ 0.10m ~ Half column height
Figure 5.24 – SP_M01 Column cracking patterns
5.22
Furthermore, the damage taken by the structure was mostly localized in the beam and,
more specifically, at the interface sections with the beam-column nodes. In fact, the first signs
of spalling and concrete crushing occurred at those locations for 2.0% drift, on the enlargement
cover cap of the nodes, while the rest of the mid-section of the beam still showed a relatively
low damage progression. As far as it could be observed, the degradation of that interface caused
a sudden localized loss of shear strength that ultimately led to failure by sliding of the beam
over the column in the subsequent displacement cycles, as is illustrated in Figure 5.25.
2.00% 2.50%
N S N S
Figure 5.25 – SP_M01 Beam-Column node interface
This is also attributable to the low slope angle (~25º) of the diagonals, resulting in reduced
efficiency of the axial force carried through the rebars, which increases the reliance of the
structure in achieving stable compressive struts within the concrete. Since the development of a
large crack (with opening widths of around 1.50 mm at 1.00% drift, 3.70mm at 2.00% and
9.95mm at 2.50% were recorded) at the interface sections removes that ability, failure took
place thereafter.
The cyclic force-drift plot for this specimen is illustrated in Figure 5.26, exhibiting a
slightly pinched response, as well as the collapse threshold, which occurs at the 3.00% drift
level. In that regard, a large force asymmetry of nearly 25% was observed. Lateral peak force
estimates are also included in Figure 5.26, calculated for the Yielding and Ultimate scenarios
using the rhombic truss design model of equation 4.10 but disregarding the influence of the
compressions parallelogram (Vc term on the same equation) since, for beam shear critical cases,
the total compression carried through the concrete is influenced by the force applied by the
actuator and, therefore, is not known beforehand. Thus, although both are lower than the
experimental value, the Ultimate scenario provided a closer approximation, within 8%
difference.
5.23
225
Collapse
175 Ultimate
Yielding
125
75
Force (kN)
25
-25
-75
-125
-175
Fp+ = 218.03 kN
Fp- = 174.99 kN
-225
-4.00 -3.50 -3.00 -2.50 -2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Lateral Drift (%)
Similar differences can be observed when theoretical shear capacity of the beam
(dependent of reinforcement) is compared with the peak experimentally recorded shear value, as
seen in Figure 5.27, where a relative difference of around 6% for the Ultimate scenario can be
found. This seems to indicate that, just like in the SP_M01 case, the influence of the
compressions (the previously mentioned Vc term) carried through the concrete is non-
negligible.
500
Yielding
450 Ultimate
Experimental 440.11
400 415.82
350
Beam Shear (kN)
347.35
300
250
200
150
100
50
0
Failure of this specimen can be attributed to the occurrence of “sliding shear” on the
interface between the beam and the north side column, but with increased damage spread. A
picture of the final stage of the test is included in Figure 5.28.
5.24
The evolution of cracking observed during the test was similar to the previous specimen, as
shear cracks appeared in the beam very early on, at around 0.20% drift. Formation of new
cracks mostly stabilized after 1.00% drift, as it is possible to observe identical cracking patterns
at 1.00% and 2.00% drift, in Figure 5.29 (color edited for visibility at the lower drifts).
0.20% 0.50%
N S N S
1.00% 2.00%
S N S N
The cracking pattern of the columns showed horizontal cracks, first forming at around
0.75% drift. Crack depth and width increased with the drift, and formation of new cracks
progressed downwards to around half the column height, similarly to SP_M01. This is
5.25
0.50% 0.75%
S N S N
3.00%
N S
~ 0.10m
~ Half column height
As it is possible to observe in Figure 5.28 and in the last panel of Figure 5.30, relevant
damage was observed in the beam, occurring at the collapse level of 3.00% drift, while the
columns displayed only mild cracking. Furthermore, spalling first occurred at the enlargement
cover cap of the beam-column nodes at 2.00% drift, as a result of the progression of the vertical
cracks in those sections, whose damage evolution is illustrated in Figure 5.31 for the north
beam-column interface. The local crushing of concrete hinders the formation of a stable
compression strut able to carry the shear load, causing the sliding movement to occur.
5.26
1.50% 2.00%
N S N S
2.50% 3.00%
S N S N
Figure 5.31 – SP_M02 Beam-column interface
The cyclic force-drift curve for this specimen is illustrated in Figure 5.32, displaying very
clear “pinching”, with collapse determined at the 2.50% drift level. Peak force asymmetry was
also observed, representing a difference of nearly 20%. In this case, predictive estimates for the
lateral peak force using the shear capacity of the beam (equation 4.11) for the Yielding and
Ultimate scenarios can lead to very different results depending on the value adopted for cot 𝜃,
according to the truss model for shear in beams. With that in mind, a single estimation is
presented, which was found to provide the best prediction of the lateral peak force on the
positive loading direction, and was calculated using rebar yielding properties and cot 𝜃 = 1.80
5.27
(corresponding to 𝜃 ≈ 30°, which is around the slope of the geometrical diagonals of the
rectangular shape of the beam).
225
Collapse
Yielding_cot1.8
175
125
75
Force (kN)
25
-25
-75
-125
-175
Fp+ = 205.48 kN
Fp- = 171.44 kN
-225
-4.00 -3.50 -3.00 -2.50 -2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Lateral Drift (%)
The value cot 𝜃 = 1.80 also leads to a small difference under 4% between the theoretical
shear capacity of the beam and the maximum value recorded in the experimental test, as
550
Yielding_cot1.8
500
Experimental
450
400 424.82 440.01
Beam Shear (kN)
350
300
250
200
150
100
50
0
Specimen SP_M03 exhibited a very clear case of “diagonal splitting” due to shear, which is
illustrated in Figure 5.34, representative of the damage state of the beam at the end of the test.
This occurred as a result of transverse reinforcement failure at the beam mid-section,
immediately leading to the almost complete loss of overall structure strength. With respect to
visual evidence alone, this specimen showed the most extensive damage from all the Phase 1
tests, but that can mostly be associated with the brittle occurrence at 2.50% drift.
5.28
a) East Side
Stirrup failure
a) West side
Figure 5.34 – SP_M03 Beam after testing
A typical shear cracking pattern in the beam was again evident throughout the test, with its
evolution starting at 0.20% drift, and the development of new cracks until 0.75% drift.
Afterwards, crack width and depth progression was observed, and by 2.00% drift,
spalling/detachment of the top surface of the beam was starting to occur, as highlighted in the
last panel of Figure 5.35. Eventually, when that occurred, failure of the stirrups ensued, leading
to global collapse.
5.29
0.20% 0.50%
S N S N
1.00% 2.00%
S N S N
Horizontal cracking in the columns developed essentially between 0.75% drift and 1.00%
drift, but the overall cracking pattern was overall less pronounced than in the previous tests. The
depth progression of the cracks was smaller, and shorter development into the column could
also be observed, which did not reach half height, as represented in Figure 5.36.
0.50% 0.75%
N S N S
5.30
3.00%
N S
Figure 5.36 (cont.) – SP_M03 Column cracking patterns
The cyclic force-drift curve for this specimen is illustrated in Figure 5.32, also exhibiting
visible “pinching”, with collapse determined at 3.00% drift for both loading directions, and a
force asymmetry of nearly 18% was also observed. Peak force estimations calculated from the
rhombic truss design model of equation 4.8 (disregarding the effect of the axial load) are also
included in the same figure and, in this case, the maximum value calculated under the Yielding
scenario compares better with the experimental force, with a relative difference of around 8%.
5.31
225
Collapse
175 Ultimate
Yielding
125
75
Force (kN)
25
-25
-75
-125
-175
Fp+ = 184.28 kN
Fp- = 156.25 kN
-225
-4.00 -3.50 -3.00 -2.50 -2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Lateral Drift (%)
That is also evidenced when comparing the experimental shear with its theoretical shear
capacity (dependent of reinforcement) under the Yielding and Ultimate cases. As illustrated in
Figure 5.38, the difference between Yielding and Experimental values is minimal, and this
seems to indicate that the behavior of SP_M04 is more heavily associated with the capacity of
the reinforcement than the compressions carried through the beam. In fact, this may be plausible
since the original rhombic truss compressions parallelogram is disrupted with the shifted
position of the diagonal rebars.
500
Yielding
450 Ultimate
Experimental 449.18
400
350 375.63 373.67
Beam Shear (kN)
300
250
200
150
100
50
0
Failure of SP_M04 occurred mainly due to the influence of shear, since the diagonal
cracking under cyclic displacements deteriorated concrete integrity within the beam, causing
progressive strength loss. This process was further increased when transverse reinforcement
failure also occurred at the beam mid-section, leading to considerable degradation of the
structural response and, eventually, failure. A picture of the final stage of the test is shown in
Figure 5.39, illustrating the damage incurred by the beam.
5.32
Stirrup failure
The same shear cracking pattern observed in other cases also occurred in SP_M04, with
crack formation between 0.20% drift and 1.00% drift. Afterwards, stable progression of crack
depth and width could be observed and, at 2.00% drift, the concrete cover of the upper surface
of the beam was spalling, as illustrated in Figure 5.40.
0.20% 0.50%
N S N S
1.00% 2.00%
N S N S
As for the column cracking pattern, similar evidence to the other tests was observed in this
case, with horizontal crack formation occurring between 0.75% drift and 1.00% drift, and
progression through the column down to around half height. In general, the north column
exhibited stronger cracking than the south one, as evidenced in Figure 5.41.
5.33
0.50% 0.75%
N S N S
3.00%
N S

~ 0.10m
5.2.1.5. Summary of main results
The main observations taken from Phase 1 of the experimental campaign were the following:
▪ Beam shear dominated failure occurred in all the tests, typically involving considerable
damage recorded in the beam, and only mild cracking in the columns;
▪ Crack formation occurred between 0.20% and 1.00% drift for the beams, and between
0.75% and 1.00% drift for the columns.
▪ The peak force asymmetry between loading directions ranged between 16% and 25%;
▪ The collapse threshold was calculated at 2.50% drift for SP_M01 and SP_M03, while
the two rhombic truss variants SP_M02 and SP_M04 were able to achieve the 3.00%
drift level;
▪ SP_M01 and SP_M02 experienced beam shear values greater than the calculated
capacity for both Yielding and Ultimate scenarios, suggesting that those bent pier
5.34
structures have an increased shear capacity than that calculated using the design models,
possibly due to larger influence of the interlock mechanism or the concrete
compressions;
▪ SP_M03 and SP_M04 experienced beam shear values similar to the theoretical capacity
for the Yielding scenario, and exhibited stronger “pinching” than the other specimens
(particularly SP_M03);
▪ All the specimens were subjected to beam shear values larger than the axial load level
previously installed on each column, leading to decompression on at least one of them;
The experimental evidence regarding Phase 2 tests follows a similar presentation layout to
that used for Phase 1. Force – drift curves will be presented according to the internal
displacement recordings of the horizontal actuator (Figure 5.16), and the 20% peak force
reduction will be used as the collapse threshold defining parameter. In this case, the predictive
estimates of the peak lateral force were calculated using the real material properties using the
equilibrium model of Figure 5.19 for the following two scenarios:
▪ Beam_shear: Assuming peak force occurs at shear yielding on the beam;

▪ Joint_bending: Assuming failure occurs at moment yielding of the longitudinal
reinforcement crossing the joint section;
The beam shear capacity associated with the reinforcement layouts used in the Phase 2
specimens is the same of SP_M02, calculated according to the rhombic truss model. In order to
determine the moment capacity of the joint section for the second case, a moment - curvature
analysis was performed in Cast3m for each specimen of Phase 2, assuming that it can be
represented by a regular concrete section with reduced reinforcement (this also assumes that
compressions can be directly transferred between the precast beam and column elements
through contact, and that full bond is achieved at both ends of the continuity rebars). The
constitutive relationships considered for simulation of the uniaxial behavior of concrete and
steel fibers were the previously mentioned models of Hognestad and Menegotto-Pinto,
respectively, calibrated for the material properties presented in Table 4.12 (the corresponding
modeling parameters are included in Annex A). Figure 5.42 - a) presents the results obtained for
each specimen, calculated considering the application of the axial dead-load of 300 kN. In
addition, Figure 5.42 - b) shows the peak force values calculated for all the specimens on both
described scenarios.
5.35
200
180
160
140
Beam_Shear Joint_Bending
Moment (kN.m)
120 (kN) (kN)

100 SP_PC02A 169,27 146,93
80
SP_PC02B 169,27 173,96
60
SP_PC02A SP_PC02C 169,27 177,25
40 SP_PC02B
SP_M02C 169,27 177,79
20 SP_PC02C
SP_M02C
0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0.045 0.050
Curvature (rad/m)
a) Moment – curvature analyses b) Peak force estimates

Figure 5.42 – Strength results for Phase 2 specimens
5.2.2.1. SP_PC02A specimen results
The cyclic force-drift curve for SP_PC02A is illustrated in Figure 5.43, where the collapse
threshold is identified at the 3.50% drift level for the two loading directions. A peak force
asymmetry of around 7% was observed between them. Furthermore, the relative difference
between the lateral peak force recorded in the test and the values estimated under Beam_shear
and Joint_bending scenarios is around the 8% for both.
200
Collapse
150 Joint_bending
Beam_shear
100
50
Force (kN)
-50
-100
-150
Fp+ = 158.85 kN
Fp- = 147.99 kN
-200
-5,50 -4,50 -3,50 -2,50 -1,50 -0,50 0,50 1,50 2,50 3,50 4,50 5,50
Lateral Drift (%)
Figure 5.43 – SP_PC02A force-drift results
The experimentally recorded beam shear seems to be in line with these results, as
illustrated in Figure 5.44, where the difference between the actual beam shear and the
theoretically determined values is around 6% and 9% for the Beam_shear and Joint_bending
scenarios, respectively.
5.36
450
Beam_shear
400 Joint_bending
Experimental
350
347.35
Beam Shear (kN)

300 328.63
301.49
250
200
150
100
50
0
Figure 5.44 – SP_PC02A Beam shear: comparison of experimental and theoretical estimates
Failure of SP_PC02A was mainly due to the influence of shear in the beam, since a strong
degradation of the beam-column interface led to sliding of the beam over the column element as
a typical “sliding shear” occurrence. Figure 5.45 presents a picture illustrating the described
mechanism during the final stage of the test (after the collapse threshold), corresponding to the
4.00% drift level.
Figure 5.45 – SP_PC02A at the final stage of the test
Beam cracks first formed between 0.30% drift and 0.40% drift, having continued forming
until 1.50% drift, where the bulk of cracking stabilized. In general the pattern was typical of
shear, with diagonal orientation on both loading directions and concentration of vertical cracks
near the beam-column interfaces. Figure 5.46 illustrates its evolution along four increasing drift
levels.
5.37
0.50% 1.00%
N S N S
2.00% 3.00%
N S N S
Figure 5.46 – SP_PC02A Beam cracking pattern
Column cracking is illustrated in Figure 5.47, and mainly developed between the 1.00%
and 1.50% drift levels, displaying horizontal orientation and progressively increasing depth. At
1.50% drift, the cracking pattern was mostly stabilized.
0.75% 1.50%
S N S N
Figure 5.47 – SP_PC02A Column cracking patterns
5.38
3.50%
N S
~ 0.10m
Figure 5.47 (cont.) – SP_PC02A Column cracking patterns
Despite that, most of the deformation observed in the column occurred at the joint sections,
where visible signs of cracking were apparent from 0.75% drift, related to joint opening widths
that had exceeded the maximum reading limit (5mm) of the associated LVDT when the 3.00%
drift was reached (joint opening evolution illustrated in Figure 5.48).
0.75% 3.00%
Figure 5.48 – SP_PC02A North precast joint opening progression
In fact, that was the moment when spalling first occurred, and roughly also when the
vertical cracks at the south beam-column interface started showing signs of sliding failure,
related to local crushing of the concrete. This is further illustrated in Figure 5.49, where the
magnitude of the damage incurred due to progression of that mechanism is presented until the
end of the test (at 4.00% drift), and the contribution of the diagonal rebars in providing capacity
to a heavy shear loaded interface is also understandable.
5.39
3.00%
S N
Large width opening
Spalling
4.00%
S N
Visible main
rebars
Figure 5.49 – SP_PC02A progression into failure
5.2.2.2. SP_PC02B specimen results
The cyclic force-drift curve for SP_PC02B is presented in Figure 5.50, illustrating a
response to the full loading history (up to 5.00% drift). It is important to note that, for
unexpected technical reasons, in the negative loading direction, the actuator went to the full
extent of its stroke, which is why the last few cycles do not show displacement increase. Even
so, the collapse threshold is identified at the loading step corresponding to the 4.00% and 3.50%
drift level for positive and negative directions, respectively. The peak force asymmetry on this
test was of around 28%, but the relative difference between the experimental value and those
determined according to the Beam_shear and Joint_bending scenarios is negligible (less than
3%).
5.40
200
Collapse
150 Joint_bending
Beam_shear
100
50
Force (kN)
0
-50
-100
-150
Fp+ = 175.62 kN
Fp- = 137.60 kN
-200
-5,50 -4,50 -3,50 -2,50 -1,50 -0,50 0,50 1,50 2,50 3,50 4,50 5,50
Lateral Drift (%)
Figure 5.50 – SP_PC02B force-drift results
That difference slightly increases regarding the comparison between the experimentally
recorded beam shear and the theoretical values. Specifically, both Beam_shear and
Joint_bending scenarios indicate larger force than the experimental value, and the respective
differences are around 6% and 9%.
450
Beam_shear
400 Joint_bending
Experimental
350
347.35 356.96
Beam Shear (kN)
300 327.43
250
200
150
100
50
0
Figure 5.51 – SP_PC02B Beam shear: comparison of experimental and theoretical estimates
The failure mode of this specimen can be attributed to heavy degradation of both beam-
column interfaces due to shear, although considerable sliding motions of the beam over the
columns were not apparent like in the previous specimen. In addition, failure of transverse
reinforcement due to spalling of the top surface of the beam (at 4.00% drift only), and
continuity rebars’ fracture could also be detected. Figure 5.52 shows the final state of the beam
after loading at 5.00% drift, and is believed to provide a fairly good picture of the failure
mechanism of SP_PC02B by itself.
5.41
Figure 5.52 – SP_PC02B after testing
Beam cracks first formed at 0.40% drift, and that process stabilized after 1.50% drift. The
same shear typical cracking pattern was also observed in this beam, with heavy concentration of
cracks near the beam-column interfaces, as illustrated in Figure 5.53.
0.50% 1.00%
N S N S
2.00% 3.00%
N S N S
Figure 5.53 – SP_PC02B Beam cracking pattern
Column cracking in SP_PC02B only occurred between 1.00% drift and 1.50% drift,
stabilizing after 2.00% drift, but the general appearance of the cracking pattern was the same of
all the previously presented cases, with horizontal cracks until roughly half the column height,
5.42
as evidenced in Figure 5.54. However, the cracking on the first 0.30m below the precast joint
was scarce, and marked by vertical cracks indicative of heavy straining of the longitudinal
reinforcement and, possibly, of rebar slip.
1.00% 1.50%
S N S N
4.00%
N S
~ 0.30m
Vertical cracks
~ 0.10m
Figure 5.54 – SP_PC02B Column cracking patterns
The level of deformation recorded at the joint sections was also large, as the opening
widths exceeded the maximum limit of the LVDT devices by the 2.50% drift level. In this case,
however, the evolution of the joint opening led to fracture of one rebar of the longitudinal
reinforcement that was providing continuity over the joint, as shown in Figure 5.55.
5.43
1.50% 3.00%
Rebar fracture
Figure 5.55 – SP_PC02B: Progression of north precast joint opening
5.2.2.3. SP_PC02C specimen results
The cyclic force-drift curve for SP_PC02C is shown in Figure 5.56, where collapse is
identified at the 4.00% drift level for the positive loading direction and near 3.00% for the
negative loading direction. In addition, the peak force asymmetry was 23%, but the strength
degradation is clearly more noticeable in the negative than in the positive loading direction.
Beam_shear and Joint_bending predictive estimations are fairly on point, with minimal
difference to the experimental peak force (again less than 3%).
200
Collapse
150 Joint_bending
Beam_shear
100
50
Force (kN)
-50
-100
-150
Fp+ = 174.44 kN
Fp- = 142.22 kN
-200
-5,50 -4,50 -3,50 -2,50 -1,50 -0,50 0,50 1,50 2,50 3,50 4,50 5,50
Lateral Drift (%)
Figure 5.56 – SP_PC02C force-drift results
Similar maximum relative difference of 3% is observed when comparing the shear

determined from experimental records with the theoretical shear values computed according to
the two scenarios, as evidenced in Figure 5.57.
5.44
450
Beam_shear
400 Joint_bending
Experimental
350 363.71
347.35 352.21
Beam Shear (kN)

300
250
200
150
100
50
0
Figure 5.57 – SP_PC02C Beam shear: comparison of experimental and theoretical estimates
Failure of SP_PC02C was related to issues caused by both beam shear and joint bending.
The degradation of the concrete due to crack propagation and concrete crushing was
considerable at the beam-column interfaces. In addition, the deterioration and progressive
increase of opening width at the precast joint sections also led to fracture of continuity rebars.
Figure 5.58 presents the final state of the specimen, after loading up to 5.00% drift.
Figure 5.58 – SP_PC02C after testing
The first cracks in the beam formed at 0.30% drift, and progression of crack depth, width,
and formation of new cracks stabilized after 1.50% drift. The cracking pattern of SP_PC02C
(Figure 5.59) was also indicative of heavy shear loading, and the vertical cracks in this
specimen had a strong development from 1.00% drift, having shown to be the dominant factor
in beam deformation from an early stage of the test.
5.45
0.50% 1.00%
N S N S
2.00% 3.00%
N S N S
Figure 5.59 – SP_PC02C Beam cracking pattern
The outlook of the column cracking pattern of SP_PC02C was very similar to the previous
specimen, with mainly horizontal cracks occurring at 1.00% drift and with a stable development
from 1.50% drift, as presented in Figure 5.60. However, and just as in the previous specimen, in
the zone below the precast joint, vertical cracks were observed instead of horizontal ones,
aligned with the longitudinal reinforcement that crossed the precast joint.
0.75% 1.00%
S N S N
Figure 5.60 – SP_PC02C Column cracking patterns
5.46
4.00%
N S
~ 0.30m
Vertical cracks
~ 0.10m
Figure 5.60 (cont.) – SP_PC02C Column cracking patterns
The joint deformations (Figure 5.61) were also significant in this specimen, as the opening
widths exceeded the 5mm maximum by the 2.00% drift level. In addition, as the progression of
the vertical cracks increased in the concrete below the precast joint, the deterioration of the
vertical beam-column cracked interface (due to concrete crushing and spalling of the specimen)
also occurred, exposing the reinforcement and, eventually, leading to rebar fracture in the south
joint, as illustrated in Figure 5.61 – a).
1.50% 5.00%
Rebar fracture
a) South joint (at 1.50% and 5.00% drifts)
Figure 5.61 – SP_PC02C precast joint opening progression
5.47
1.50% 5.00%
b) North joint
Figure 5.61 (cont.) – SP_PC02C precast joint opening progression
5.2.2.4. SP_M02C specimen results
The cyclic force-drift curve for SP_M02C (the monolithic replica of SP_PC02C) is
presented in Figure 5.62, for which early collapse was determined at the 2.50% drift level for
both loading directions. The peak force asymmetry between them was around 26%. Like in the
previous cases, the values estimated using the theoretical scenarios Beam_shear and
Joint_bending are very close to the experimental peak force for the positive loading direction
(less than 3% difference).
200
Collapse
150 Joint_bending
Beam_shear
100
50
Force (kN)
-50
-100
-150
Fp+ = 174.20 kN
Fp- = 137.95 kN
-200
-5,50 -4,50 -3,50 -2,50 -1,50 -0,50 0,50 1,50 2,50 3,50 4,50 5,50
Lateral Drift (%)
Figure 5.62 – SP_M02C force-drift results
Negligible relative difference is also observed when comparing the experimental beam
shear with the theoretically determined values, as illustrated in Figure 5.63.
5.48
450
Beam_shear
400 Joint_bending
Experimental
350 364.82
347.35 345.11
Beam Shear (kN)

300
250
200
150
100
50
0
Figure 5.63 – SP_M02C Beam shear: comparison of experimental and theoretical estimates
The failure mode on SP_M02C occurred due to fracture of the longitudinal reinforcement
on the beam-column node. The specimen, in general, experienced low damage. But early on, the
cracking of the monolithic concrete in the zone corresponding to the precast joint on the
SP_PC## cases led to very large opening widths and fracture of the crossing reinforcement, as
well as subsequent sudden strength loss. Figure 5.64 presents a picture of the specimen after
testing, where this is very noticeable by the large crack openings at the beam-column
intersections, in contrast with the minimal damage on the rest of the structure.
Figure 5.64 – SP_M02C after testing
The beam cracking pattern was typical of a shear dominated response, with diagonal
oriented cracks and vertical cracks at the beam-column vertical interfaces that first formed
around the 0.30% drift level (Figure 5.65). Formation of new cracks nearly stopped after 0.75%
drift, when mostly crack depth and width were observed to be increasing. The influence of the
strong shear mechanism that was apparent in the precast specimens could, nonetheless, also be
identified in this case, and the vertical cracks saw a large development, as clearly illustrated by
the last picture of Figure 5.65.
5.49
0.50% 1.00%
S N S N
1.50% 2.50%
S N S N
Figure 5.65 – SP_M02C Beam cracking pattern
The columns observed virtually no cracking outside of the previously identified zone. The
formation of the main crack first occurred at 0.50% drift. Eventually, crack depth and width
increased significantly, as shown in Figure 5.66, to the point where rebar fracture was detected.
0.50% 1.50% 2.50%
Figure 5.66 – SP_M02C South beam-column joint crack width evolution
5.2.2.5. Summary of main results
▪ The failure mode of the precast specimens was related to shear failure at the vertical
beam-column interfaces, involving large damage in the beam and low damage in the
columns;
▪ Before failure, the precast specimens evidenced large deformations at the horizontal
precast joint;
5.50
▪ By contrast, the reference monolithic specimen (SP_M02C) exhibited a brittle failure

due to fracture of beam-column node reinforcement caused by concentration of column
deformations on a single crack below the beam;
▪ Beam cracking started forming between 0.30% to 0.40% drift, and developed until
between 0.75% and 1.50% drift;
▪ Column cracking generally developed between 1.00% drift and 1.50% drift, except on
the monolithic specimen where it occurred earlier at 0.50% drift;
▪ The peak force asymmetry between loading directions ranged between 7% and 28%;
▪ The collapse threshold was determined at 2.50% drift in the monolithic SP_M02C
specimen, while in the precast specimens it was between 3.50% drift and 4.00% drift;
▪ The predictive estimates of the peak force according to the Beam_shear and
Joint_bending scenarios were in good agreement with the experimental values, showing
differences smaller than 10% in every case.
▪ The experimental beam shear was in good agreement with the values determined
according to the Beam_shear and Joint_bending scenarios, since the relative differences
obtained were under 3%;
▪ All the specimens were subjected to beam shear values that exceeded the axial load
level previously installed on each column, leading to decompression on at least one of
them.
5.2.3. RESULTS DISCUSSION
Result analysis and discussion for the present work will mostly be provided in two
segments, according to the different phases of experimental testing and their respective
objectives. Nonetheless and for the purpose of global comparison, it is important that the
analyses reflect the performance of each specimen over generalized demand parameters. With
that in mind and taking into account that the collapse threshold and failure for all tests were
determined at different drift levels, it seems appropriate to introduce the concept of cumulative
ductility. Essentially, it enables characterization of the desired performance parameters over a
standardized representation of inelastic incursion that inherently reflects the actual loading
history and the mechanical properties of each structural system (e.g.: stiffness). By doing so, it
also provides improved flexibility for comparing results of tests where different conditions were
explored.
Within that context, in this work the concept is associated with displacement ductility,
considered in cumulative terms for values recorded in the horizontal actuator’s internal LVDT,
calculated according to equation 5.1 where i represents each half-cycle on post yielding phase,
aiming at evaluating the inelastic incursion of specimens under cyclic loading. For that purpose,
5.51
it is necessary to define the yielding displacement, which can be done by adopting an idealized
equivalent bilinear response; in this work it is characterized by an elastic branch determined
using the ¾ rule proposed in Park (1989) and a hardening branch established for the peak
force. The yielding displacement and the hardening branch slope for the idealized system
are calculated to display the same area under the curve up to the peak force displacement as
the actual monotonic envelope of the cyclic tests. That procedure is further highlighted in
Figure 5.67, in which the ductility of the first half-cycle after yielding occurrence is calculated
according to equation 5.2, where dy represents the yielding displacement. For each subsequent
half-cycle the values of ∆µi (calculated according to equation 5.3) are determined for the
maximum absolute displacement |𝑑𝑖 | > |𝑑𝑦 | with ∆dy defined by equation 5.4. This
methodology was adapted from a similar proposal by Galano and Vignoli (2000) established for
ductility values in rotation, in order to use displacement based recordings and to consider the
possibility for a hardening branch. The described methodology has already been successfully
used by the author for the purpose of experimental result analysis, according to the work
published in Monteiro et al. (2017b) (and also in Monteiro et al. (2017a), currently under
review).
𝜇𝑖𝑐𝑢𝑚 = 𝜇𝑖−1
𝑐𝑢𝑚
+ ∆𝜇𝑖 5.1
𝑑1 5.2
𝜇1 =
𝑑𝑦
∆𝑑𝑖 5.3
∆𝜇𝑖 =
𝑑𝑦
∆𝑑𝑖 = 𝑑𝑖 − 𝑑𝑦 5.4
Figure 5.67 – Procedure for definition of equivalent bilinear systems
5.52
5.2.3.1. Global Behavior and Damage Incursion
Comparison of the global behavior exhibited by each specimen is performed by analyzing

the drift levels at which the experimental evidence shows increased damage. For that purpose,
four performance levels were defined, according to the observations made during the tests:
▪ Performance Level 1 (PL1): Detection of beam cracking;

▪ Performance Level 2 (PL2): Detection of column cracking;
▪ Performance Level 3 (PL3): Detection of concrete crushing/spalling;
▪ Performance Level 4 (PL4): Collapse of the specimen;
PL1 PL2 PL3 PL4

4,50%
Phase 1 Phase 2
4,00%
3,50%
3,00%
2,50%
2,00%
1,50%
1,00%
0,50%
0,00%
SP_M01 SP_M02 SP_M03 SP_M04 SP_PC02A SP_PC02B SP_PC02C SP_M02C
Figure 5.68 – Performance level comparison between Phase 1 and Phase 2 specimens
Figure 5.68 summarizes many of the points that were previously discussed, and presents
them in a more intuitive and visual form, in which the red arrow serves as a reminder that all
Phase 2 specimens were designed according to the same rhombic truss beam layout of SP_M02.
In light of those results, it is clear that the precast specimens of Phase 2 displayed better
performance than all of the monolithic specimens, even SP_M02C, since nearly all the
performance levels occurred at higher drift levels. This is especially more relevant for the
performance levels PL3 and PL4, which are the most revealing of considerable damage
experienced by the specimens, and are generally delayed by around 1.00% drift in all the precast
tests.
Regarding Phase 1 specimens, cumulative ductility and respective yielding drifts are
illustrated in Figure 5.69. According to those results SP_M04 and SP_M01 achieved yielding
significantly earlier than SP_M02 and SP_M03 (especially in the positive loading direction).
This is particularly important because it also relates with the cumulative ductility experienced
5.53
by the specimens, since they were subjected to inelastic deformations from an earlier stage,
potentially leading to overall increased energy dissipation. Despite that, only SP_M04 seemed
to benefit from this, showing the highest cumulative ductility value of all the Phase 1 test
specimens while SP_M01 mobilized around half that value, which may be explained by the
earlier failure it also experienced. On the other hand, SP_M02 achieved yielding at a later stage,
but also mobilized greater peak strength than both SP_M01 and SP_M04, and later failure than
the former of the two, enabling the second largest cumulative ductility of all the test specimens,
which are some of the reasons that supported the decision of designing Phase 2 specimens based
on the reinforcement layout of SP_M02 (although results seem to indicate better overall
performance on SP_M04). Finally, the conventionally reinforced specimen SP_M03 showed the
overall lowest cumulative ductility values, which is not surprising considering its performance
also showed a large “pinching” effect and the brittle failure that ensued.
45
40
35
30
25
μcum
20
15
10
0
SP_M01 SP_M02 SP_M03 SP_M04

dy+ (%) 0,86 1,29 1,34 0,82
dy- (%) 0,74 0,84 0,94 0,81
Figure 5.69 – Cumulative ductility at the collapse threshold (for the positive loading direction)
and yielding drifts (for both directions) for Phase 1 specimens.
Regarding Phase 2 tests, the cumulative ductility was shown to be significantly larger in
the precast than in the monolithic specimens, as observed in Figure 5.70, where results from the
reference monolithic SP_M02 specimen (from Phase 1) were also included (in grey) for
comparison. In that regard, SP_PC02A shows the lowest cumulative ductility value among the
three precast models, but it still almost doubles the value of SP_M02C (highest of the
monolithic specimens under analysis). That difference increases for the other two specimens,
particularly for SP_PC02C, which, considering that similar yielding drifts were achieved in
those three cases (SP_PC02B, SP_PC02C and SP_M02C), further evidences the difference in
5.54
performance between these precast and monolithic specimens. By contrast, yielding drifts are
higher for SP_PC02A and SP_M02, which helps to explain the lower cumulative ductility
values they exhibit, although, in the case of the precast specimen, the peak force was also lower
than observed in the cases of SP_PC02B and SP_PC02C. Nonetheless, the overall performance
of the precast specimens in comparison with the monolithic models seemed to show significant
improvements.
120
100
80
μcum
60
40
20
0
SP_PC02A SP_PC02B SP_PC02C SP_M02C SP_M02

dy+ (%) 1,07 0,88 0,88 0,89 1,29
dy- (%) 0,85 0,89 0,78 0,87 0,84
Figure 5.70 – Cumulative ductility at the collapse threshold (for the positive loading direction)
and yielding drifts (for both directions) for Phase 2 specimens.
5.2.3.2. Beam Shear Degradation
As discussed before, shear forces mobilized in the beam are paramount for the cyclic
response of this frame type because considerable lateral stiffness is added by the shear
interaction between columns. The tests performed in this work allowed obtaining the actual
internal forces involved in that structural mechanism, by relating the vertical reactions recorded
in load cells at the columns’ base sections with the axial loads imposed by the prestressed rods.
Thus, regarding Phase 1 specimens the peak beam shear forces (Vp) obtained in the four tests
are presented in Table 5.3 for both loading directions, while Figure 5.71 shows the plots for
each half-cycle beam shear forces’ ratios to their peak values (V/Vp) versus the cumulative
ductility (μcum), which illustrates its influence on the inelastic incursion experienced by each
specimen. Both positive and negative loading results are presented, where the former is uses a
vivid blue color and the latter a light blue color. Furthermore, the drift levels recorded at 20%
peak horizontal force reduction (which is equivalent to the base shear in these tests) are also
5.55
shown, where the beam shear values they represent correspond to the ultimate beam shear
(Vu/Vp) recorded at the ultimate drift du.
Table 5.3 – Peak Beam Shear results for Phase 1 specimens
Specimen Vp+ (kN) Vp- (kN)
SP_M01 326.79 287.41
SP_M02 440.11 320.72
SP_M03 440.01 287.77
SP_M04 376.67 290.43
1.00
SP_M01
0.90
0.80 SP_M02
0.70 SP_M03
0.60
SP_M04
V/Vp
0.50
20% Red.SP_M01
0.40
0.30 20% Red.SP_M02
0.20 20% Red.SP_M03
0.10
20% Red.SP_M04
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
μcum
a) Positive loading direction

1.00
SP_M01
0.90
0.80 SP_M02
0.70 SP_M03
0.60
SP_M04
V/Vp
0.50
20% Red.SP_M01
0.40
0.30 20% Red.SP_M02
0.20 20% Red.SP_M03
0.10
20% Red.SP_M04
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
μcum
b) Negative loading direction
Figure 5.71 – Beam shear vs. Cumulative ductility for Phase 1 specimens
5.56
Despite significant differences in Vp values between specimens, particularly for the

positive direction, it can be seen that the 20% reduction threshold was reached for beam shear
forces Vu around 90% of their Vp values on all specimens except SP_M04, for which a bit lower
ratio of 80% was found. Such findings agree well with experimental observations, which
evidenced shear dominated brittle failure in SP_M01 and SP_M03, while in the “rhombic truss”
variants (SP_M02 and SP_M04) it was observed after gradual strength degradation. In addition,
20% peak force reduction is achieved for beam shear reductions between 10% and 20%, thus
indicating that the overall strength reduction is strongly related with the strength degradation in
the beam. After the 20% peak force reduction threshold, it can be seen that the conventionally
reinforced specimen SP_M03 was the least resilient one, keeping beam shear above 0.8V u just
up to cumulative ductility values about 20. All the other specimens performed similarly between
them, being able to sustain the same shear load (0.8Vu) nearly until around twice the μcum value.
A relative difference around 35% is observed between overall peak beam shear values,
with the smaller loads obtained again for SP_M01 and SP_M04, which is strongly related to the
smaller base shear values also recorded in those tests, as opposed to SP_M02 and SP_M03.
Nonetheless, the shear forces mobilized through the coupling beam in all the specimens are
greater than the level of applied dead load (300 kN), highlighting the contribution of the
coupling beam in this system.
Regarding Phase 2 specimens, Table 5.4 presents the maximum beam shear values
recorded on the respective tests (including results from SP_M02 for comparison, in the grey
shaded row), where, again, a considerable difference can be observed between loading
directions. In these results, the important contribution of the coupling beam is again evidenced,
since values for positive loading again exceeded the level of the applied dead load, although
slightly smaller forces were observed for the negative loading direction.
Table 5.4 – Peak Beam Shear results for Phase 2 specimens
Precast Vp+ (kN) Vp- (kN) Monolithic Vp+ (kN) Vp- (kN)
SP_PC02A 328,63 277,53

SP_M02C 345,11 263,40
SP_PC02B 327,43 275,13
SP_M02 440,11 320,72
SP_PC02C 352,21 281,06
In addition, Figure 5.72 presents the plot of the normalized maximum beam shear (V/Vp) of
each half-cycle against cumulative ductility for all the specimens, illustrating its evolution
according to the inelastic incursion. Evolutions are again separated for positive and negative
direction half-cycles’ curves; the former are plotted in vivid red and the latter in light red, and in
5.57
both plots the values for the reference monolithic specimen SP_M02 are also presented in grey
color for comparison. Within this context, it is possible to observe that a cumulative ductility
value between 20 and 30 is the threshold upon which shear degradation starts to notably
increase.
1.00
SP_PC02A
0.90
SP_PC02B
0.80
SP_PC02C
0.70
SP_M02C
0.60
SP_M02
V/Vp
0.50
20% Red.SP_PC02A
0.40
20% Red.SP_PC02B
0.30
20% Red.SP_PC02C
0.20
20% Red.SP_M02C
0.10
20% Red.SP_M02
0.00
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320
μcum
1.00
SP_PC02A
0.90
SP_PC02B
0.80
SP_PC02C
0.70
SP_M02C
0.60
SP_M02
V/Vp
0.50
20%Red.SP_PC02A
0.40
20%Red.SP_PC02B
0.30
20%Red.SP_PC02C
0.20
20%Red.SP_M02C
0.10
20%Red.SP_M02
0.00
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320
μcum
Figure 5.72 – Beam shear vs. Cumulative ductility for Phase 2 specimens
The results of the monolithic and the precast specimens present some noteworthy
differences. Referring only to the positive direction, the former experienced beam shear drops to
around 70% of the corresponding maximum at cumulative ductility values between 50 and 60,
while the others were able to keep the same performance level up to nearly twice the cumulative
ductility values (over 116). Additionally, the overall evolution of all precast specimens presents
similar shear loss, although with slightly different decrease rates.
5.58
Regarding the previously defined collapse threshold of 20% reduction of the lateral peak
force, it is possible to observe that specimens SP_PC02A and SP_PC02B reached that stage
with beam shear loss around 20%, but in the SP_PC02C case, significant differences were
recorded in both directions. For positive loading a value slightly above 80% can be observed,
while for negative loading the shear loss was almost 40%. By contrast, the beam shear loss at
the collapse threshold on the monolithic models was less than 20%, particularly for positive
loading (around 90%). These findings support claiming that the precast specimens benefitted
the most of exploring column capacity, since the 20% peak force reduction could be achieved
with larger beam shear loss, in comparison with the other tests. Conversely, structural response
in monolithic models shows more dependency of the beam, since force degradation is
associated with smaller losses in beam shear.
5.2.3.3. Stiffness Degradation
Stiffness degradation can be an important aspect to take into account regarding the seismic
performance of bridge piers. In fact, common design strategies require adequate structural
ductility, which is often related with the secant stiffness for the last half-cycle before collapse is
reached. Thus, lower stiffness degradation means that post-peak strength is kept at less reduced
levels for larger deformation and, consequently, higher ductility is achieved.
Figure 5.73 shows stiffness degradation plotted against cumulative ductility for Phase 1
specimens, in terms of the ratio (K/Ky) of secant stiffness (K) to yielding stiffness (K y), where
the both loading directions are represented separately (vivid blue for positive loading and light
blue for negative loading). The behavior was globally similar between all specimens, but around
cumulative ductility values of 35 the conventionally reinforced model (SP_M03) showed
increased rate of stiffness degradation, since the corresponding secant stiffness was found to be
around 10% of the yielding stiffness at cumulative ductility of 45 on negative loading. On the
other cases, cumulative ductility values around 70 could be achieved before similar degradation
of the secant stiffness was achieved.
When looking at the 20% peak force reduction markers that are also illustrated in Figure
5.73, SP_M03 was found at more than 50% of the corresponding yielding stiffness, while the
two rhombic truss variants SP_M04 and SP_M02 showed increased degradation up to nearly
30% of the yielding stiffness (for positive loading); SP_M01 ended up in between those two
values.
5.59
1.00
SP_M01
0.90
0.80 SP_M02
0.70 SP_M03
0.60
SP_M04
K/Ky
0.50
20% Red.SP_M01
0.40
0.30 20% Red.SP_M02
0.20 20% Red.SP_M03
0.10
20% Red.SP_M04
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
μcum
1.00
SP_M01
0.90
0.80 SP_M02
0.70 SP_M03
0.60
SP_M04
K/Ky
0.50
20% Red.SP_M01
0.40
0.30 20% Red.SP_M02
0.20 20% Red.SP_M03
0.10
20% Red.SP_M04
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
μcum
Figure 5.73 – Stiffness degradation vs. Cumulative ductility for Phase 1 specimens
Figure 5.74 illustrates the values of the secant stiffness (K) normalized by the yielding
stiffness (Ky) calculated for the Phase 2 tests, which are plotted against the corresponding
cumulative ductility. The values for positive loading are reported in vivid red, while negative
loading values are shown in light red; the reference SP_M02 model results are also included in
grey lines for comparison purposes.
It can be observed that the general performance of all the specimens was globally similar,
according to the general development of the plotted curves. It can be observed that all the
specimens reach 50% of the yielding stiffness very early on, for cumulative ductility values
between 10 and 20. Regarding the 20% collapse threshold, in precast specimens it was achieved
5.60
with secant stiffness values between 15% and 30% of the yielding stiffness, while the
monolithic specimens were found at upwards of 30% of the yielding stiffness. Furthermore, all
the tests of Phase 2 showed relatively stabilized curves for values of K/Ky less than 20%.
1.00
SP_PC02A
0.90
SP_PC02B
0.80
SP_PC02C
0.70
SP_M02C
0.60
SP_M02
K/Ky
0.50
20% Red.SP_PC02A
0.40
20% Red.SP_PC02B
0.30
20% Red.SP_PC02C
0.20
20% Red.SP_M02C
0.10
20% Red.SP_M02
0.00
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320
μcum
1.00
SP_PC02A
0.90
SP_PC02B
0.80
SP_PC02C
0.70
SP_M02C
0.60
SP_M02
K/Ky
0.50
20% Red.SP_PC02A
0.40
20% Red.SP_PC02B
0.30
20% Red.SP_PC02C
0.20
20% Red.SP_M02C
0.10
20% Red.SP_M02
0.00
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320
μcum
Figure 5.74 – Stiffness degradation vs. Cumulative ductility for Phase 2 specimens
5.2.3.4. Energy Dissipation
Energy dissipation mechanisms in RC structures are mobilized mostly resorting to the

inelastic capacity of steel to accommodate large deformations without significant force increase
in the post-yielding range, which, for regular structures, occurs in the “plastic hinge” zones
5.61
usually associated with maximum bending moment locations such as column bases. In this
structure, significant damage occurs in the beam region, but it is mostly originated by shear
distortions and not flexural deformations of the reinforcement, which are mostly caused by the
small span-to-depth geometric ratio. In this regard, although beam reinforcement is able to
achieve yielding, it is believed that the same concept of plastic hinge should not be considered
for interpretation of the energy dissipation of these structures, particularly because while
concrete integrity holds, the load can be transferred between columns through a single
compressive strut, which helps increase the total shear capacity above that which would be
possible by reinforcement alone.
Regarding comparison of all Phase 1 specimens, Figure 5.75 shows the plot of cumulative
energy dissipation against the number of loading half-cycles. The overall maximum dissipated
energy recorded at the displacement corresponding to 20% peak force reduction was greatest on
the SP_M02 specimen (about 58 kN.m), which was another reason supporting the adoption of
that layout for Phase 2 tests. By contrast, SP_M03, with the conventional reinforcement layout,
was notably the lowest dissipative specimen with 30 kN.m at that stage. Such results are not
surprising, taking into account that the SP_M03 force vs. drift plot (Figure 5.32) exhibited
severe “pinching” effect, considerably more pronounced than in SP_M01, SP_M02 or SP_M04
specimens. This is further confirmed by its Energy Dissipation curve, which shows considerably
lower values than the other three specimens. Concerning the SP_M01 test, it showed similar
energy dissipation evolution to the rhombic truss variants but lower value at 20% peak force
reduction than SP_M02, suggesting that a potentially better performance was hindered by
sudden occurrence of sliding shear failure, as previously discussed.
100.00
SP_M01
90.00
80.00 SP_M02
Dissipated Energy
70.00 SP_M03
60.00
(kN x m)
SP_M04
50.00
20% Red.SP_M01
40.00
30.00 20% Red.SP_M02
20.00 20% Red.SP_M03

10.00
20% Red.SP_M04
0.00
0 10 20 30 40 50 60 70 80 90 100
Half-cycles
Figure 5.75 – Energy dissipation vs. Loading half-cycles for Phase 1 specimens
5.62
Results for cumulative dissipated energy for Phase 2 tests are presented in Figure 5.76,
where related specimens’ values are represented by the red lines and the grey lines link to
results for the reference SP_M02 model, included for comparison purposes. As it is possible to
observe, the rate of increase of energy dissipation is similar in all the specimens, since they
generally follow the same outline. However, when taking into account the half-cycle at which
the 20% force reduction threshold is achieved, it is possible to understand that the two
monolithic models performed poorly relative to the precast specimens, since they provided
considerably less energy dissipation until that stage. This is particularly relevant in the case of
the SP_M02C specimen, confirming that the deformation mechanism provided by the
opening/closing of the large crack formed at the beam-column intersection (Figure 5.66, which
is naturally less ductile than the behavior evidenced by the precast specimens) had reasonable
impact in the overall performance of that pier. Finally, the precast specimens SP_PC02B and
SP_PC02C (where splicing ϕ12 rebars were used to provide moment continuity) enabled the
best results in this regard, closely followed by SP_PC02A.
160.00
SP_PC02A
140.00
SP_PC02B
120.00 SP_PC02C
Dissipated Energy
100.00 SP_M02C
(kN x m)
SP_M02
80.00
20% Red.SP_PC02A
60.00 20% Red.SP_PC02B
40.00 20% Red.SP_PC02C
20% Red.SP_M02C
20.00
20% Red.SP_M02
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Half-cycles
Figure 5.76 – Energy dissipation vs. Loading half-cycles for Phase 2 specimens
5.2.3.5. Beam-column deformations
Two different ratios were considered for evaluating the relative deformations at the beam
and column elements. The concepts indicated in Figure 5.77 were calculated according to the
available data recorded during the tests, at the south column (S, see Figure 5.1 – a)), determined
as follows:
1) 𝑅1 = 𝜑𝑐 ⁄(𝜑𝑐 + 𝜑𝑏 ): ratio between the column and beam curvatures,

calculated from mean rotations recorded on positive bending moments by
local LVDTs, at the immediate vicinity of both interfaces of the beam-
column node;
5.63
2) 𝑅2 = 𝛼 ⁄𝛾: ratio between the beam rotation (α), directly recorded by

tiltmeters on the top of the column and the chord rotation (γ) of the column,
obtained from horizontal LVDTs at its top section;
b

c
a) Ratio R1 b) Ratio R2
Figure 5.77 – Illustration of the parameters used for calculating the ratios used in this work for
comparison of beam-column deformations
The above defined ratios were calculated only at the following four drift levels: 0.50%,
1.00%, 1.50% and 2.00%, since some LVDT recordings were not available for the full extent of
the applied loading histories. Values of R1 closer to 100% indicate that the local deformations
recorded at the column side of the node are considerably more relevant than those at the beam,
while the contrary is associated with values closer to 0%. Regarding ratio R2, values of 100%
can only be ensured if fully rigid motions of the column body are observed, since that would
essentially be equivalent to have the same rotations of column axis and column head.
Conversely, if the beam element exhibited fully rigid behavior, then the column head surface
rotation would be negligible, and all deformations would occur at and below the beam-column
intersections. For reference, the described methodology was also presented in Monteiro et al.
(2017a).
Ratio R1 values for Phase 1 (Figure 5.78) specimens were generally found between around
20% at 0.50% drift and 60% at 2.00% drift. In that regard, an evolution of R1 values between
the referred lower and upper bound could be observed for specimens SP_M01 and SP_M02,
whereas less variation could be found in the case of SP_M03 and SP_M04. In general, these
results tend to show that moment demands in the beam-column node mobilize greater
curvatures in the beam than the column for a significant part of the applied loading history,
which seems consistent with experimental observations of greater damage occurrences in the
beams.
5.64

100%
80%
60%
40%
20%
drift
0%
0.50% 0.75% 1.00% 1.25% 1.50% 1.75% 2.00%
Figure 5.78 – R1 deformation ratio vs. drift evolution for Phase 1 specimens
Figure 5.79 illustrates the values of R1 values calculated for Phase 2 specimens (and the
reference monolithic SP_M02 model), which can provide a good measurement of the local
demand expected for this type of structure and, most importantly, of the contribution of the
precast connection to the overall behavior and performance. This is particularly true because the
LVDTs, from which column curvature data was calculated, were recording deformations within
a short 0.05m distance from the joint, making it safe to assume that the calculated 𝜑c values
were mostly due to joint opening/closing, rather than concrete strains.
Within that context, it is not surprising to see how R1 values for the precast specimens are
generally larger than for the monolithic models, almost achieving 100% in all three cases while
the latter only achieve around 70% at 2.00% drift. Likewise, all precast specimens showed
values greater than 60% from nearly 0.75% drift, while monolithic specimens only go above
60% well after 1.50% drift. It is also noteworthy that values for SP_PC02A and SP_02B show a
significant decrease to the 70%-80% range at 2.00% drift, after having reached almost 100% a
few cycles earlier; this is likely caused by the increase of beam deformations that occurs due to
shear degradation of the vertical beam-column interfaces.

100%
80%
60%
40%
20%
drift
0%
0.50% 0.75% 1.00% 1.25% 1.50% 1.75% 2.00%
5.65
Figure 5.80 illustrates R2 values for Phase 1 specimens, and nearly all of them show low
variation throughout the test. In general, values between around 60% - 70% at 0.50% drift and
80% - 90% at 2.00% drift were recorded, showing that deformations along the column body
have a relatively low impact on the overall rotation of the column head. The case of SP_M01 is
the only one where significant variation occurs, since R2 values go from 40% at 0.50% drift to
nearly 90% at 2.00% drift, but this finding could also be related to gradual increases of the
sliding motions observed during the test, since it accommodates most of the deformations.

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
drift
0%
0.50% 0.75% 1.00% 1.25% 1.50% 1.75% 2.00%
Regarding Phase 2 specimens, corresponding R2 values are illustrated in Figure 5.81 (also
including the reference monolithic SP_M02 model). Observed values show, in general, very
different scenario from that observed in Phase 1 tests. In this case, all specimens exhibit a
decrease of R2 values throughout the tests, going from between 70% - 90% at 0.50% drift to
40% - 60% at 2.00% drift. A likely explanation is that the precast columns (and SP_M02C due
to the formation of its large concrete crack at the joint location) accumulate large deformations
at the joint section, leading to a shift in the beam-column stiffness ratio in favor of the beam.
Although the decrease rate of R2 seems to be similar on all the specimens, SP_PC02A (precast)
and SP_M02C (monolithic) reach the 2.00% drift level with the smaller values of R2, around
and below 40%, respectively.
5.66

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
drift
0%
0.50% 0.75% 1.00% 1.25% 1.50% 1.75% 2.00%
5.3. EXPERIMENTAL RESULTS – SINGLE COLUMN TESTS
Experimental observations regarding Phase 3 tests are based on force – drift curves
established for the internal displacement of the horizontal actuator (Figure 5.18), and the same
strategy is also considered for defining the collapse threshold. In addition, a single predictive
estimation scenario is presented for the peak lateral force, related to the bending capacity
calculated in the base of the columns using the real material properties (Table 4.12) of each
specimen, in order to calibrate the constitutive relationships for moment – curvature analyses in
Cast3m (using the same constitutive models as before). For that purpose, the precast
connections were assumed to enable similar performance to a monolithic cross-section under
the same axial load (300 kN), leading to the results presented in Figure 5.82.
300
250
200
Moment (kN.m)
150
100
50
SP_F01
SP_F02
SP_F03
0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0.045 0.050
Curvature (rad/m)
Figure 5.82 – Moment – curvature analysis results for Phase 3 specimens
5.67
5.3.1.1. SP_F01 specimen results
The cyclic force-drift curve for SP_F01 is illustrated in Figure 5.43, where the collapse
threshold is identified at the 3.50% drift level for the positive loading direction only. A minimal
peak force asymmetry of around 5% was observed between them, and the difference between
the bending capacity determined from the results of Figure 5.82 (taking into account force
application at 1.95m from the column base), and the experimental maximum values is also
negligible. The failure mode on SP_F01 occurred due to buckling and consequent fracture of
the longitudinal reinforcement.
200
Collapse
Bending Capacity
150
100
50
Force (kN)
-50
-100
-150
Fp+ = 138.41 kN
Fp- = 131.61 kN
-200
-4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00
Lateral Drift (%)
Figure 5.83 – SP_F01 force-drift results
Column cracking (Figure 5.83) was first detected at the 0.50% drift level, with a few thin
cracks near the footing. That pattern developed between the 0.75% and 1.00% drift levels,
where the crack width and depth increased and spread over a larger column height. In general, it
could be observed that by the 1.00% drift level, the bottom crakcs from both the North and
South side of the column transformed into a single crack spanning the whole cross-section
depth. Furthermore, cracking was regularly distributed, since crack distance was similar to that
of the transverse reinforcement.
5.68
0.50% 0.75% 1.00%

~ 0.10m
Figure 5.84 – SP_F01: Evolution of column cracking pattern
Afterwards, that process stabilized and only crack width increases could be observed. At
the 2.50% drift level, mild spalling was detected, which evolved until full concrete cover
detachment started occurring at 3.00% drift, as well as main rebar buckling that would
eventually lead to rebar fracture at 3.50% drift, as illustrated in Figure 5.85. The main spread of
concrete crushing and spalling damage was located in both sides of the column, and
concentrated in roughly the first 0.20m near its base.
2.50% 3.00% 3.50%
Rebar fracture
Figure 5.85 – SP_F01: South side onset of spalling, buckling and fracture
The cyclic force-drift curve for SP_F02 is illustrated in Figure 5.86, and the collapse
threshold was found at the 3.00% drift level for both positive and negative loading directions.
The peak force asymmetry was negligible, less than 3%. In addition, the bending capacity
determined from the moment curvature analyses agrees well with the results obtained, since the
difference between the numerical and experimental values is less than 4%. Failure of SP_F02
occurred through fracture of several longitudinal rebars, as a consequence of bending induced
buckling.
5.69
200
Collapse
Bending Capacity
150
100
50
Force (kN)
0
-50
-100
-150
Fp+ = 131.04 kN
Fp- = 134.24 kN
-200
-4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00
Lateral Drift (%)
Column cracking was first detected at the 0.30% drift level, on the north side of the column
only. From 0.40% drift onwards, those cracks developed in the same way as SP_F01, with
increases in crack width, depth, and progression over a larger column height. The observed
cracks followed a horizontal orientation with regular distribution similar to the distance between
stirrups (0.10m), as shown in Figure 5.87. Around 0.75% drift, the cracks formed at opposing
sides of the columns nearest to the column base were sufficiently developed that they converged
into a single crack.
0.50% 0.75% 1.00%
~ 0.10m
Figure 5.87 – SP_F02: Evolution of column cracking pattern
The bulk of cracking occurred until 1.00% drift, and a stabilized cracking pattern could be
observed afterwards. Eventually, concrete crushing was detected at 2.00% drift, leading to
spalling and buckling of the main reinforcement at 2.50% drift and subsequent fracture, first
detected at 3.00% drift, according to the pictures shown in Figure 5.88. As it is possible to
5.70
observe, the main damage spread (particularly concrete crushing and spalling) was located quite
near the footing, roughly in the first 0.10m to 0.15m length, on both sides of the column.
2.00% 2.50% 3.00%
Rebar fracture
Figure 5.88 – SP_F02: North side onset of spalling, buckling and fracture
Figure 5.89 shows the force-drift results for SP_F03, where the collapse threshold can be
found at the 2.50% drift level for both loading directions. However, the observed response
shows unusual and unexpected behavior, where an effect similar to stiffening, rather than
softening, can be observed for the post-peak displacement cycles (particularly for loading at the
2.50% and 3.00% drift levels). The reason for that is not clear, but one explanation may be the
occurrence of sliding of the column along the precast joint section, leading to the progressive
adjustment of the compressive struts, seeking stability after initial grout crushing and enabling
further exploration of the reinforcement capacity. On the other hand, it may more likely be
caused by a progressive failure of rebar anchorages, forcing the structure to find the required
rebar strength at an increasingly lower depth, according to respective increases in drift demand,
and gradually causing the degradation of surrounding grout. With the present monitoring layout
it was not possible to assess this problem with certainty, however both explanations seem
plausible.
The peak force asymmetry was around 6%, although the difference between experimental
values and the theoretical bending capacity was of 35% and 26% for positive and negative
loading directions, respectively. This relative difference is larger than that observed for either
SP_F01 or SP_F02, but it can be explained by the general behavior observed in the specimen. In
fact, despite the failure of SP_F03 having also occurred through rebar fracture at 3.00% drift,
there are distinct characteristics of its performance pattern.
5.71
200
Collapse
Bending Capacity
150
100
50
Force (kN)
0
-50
-100
-150
Fp+ = 100.55 kN
Fp- = 106.85 kN
-200
-4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00
Lateral Drift (%)
Column cracking was first detected at the 0.40% drift level, as a clear crack formed exactly
between the grout and concrete parts of the precast connection. That crack fully developed over
the total cross-section depth and, while a few more cracks formed over the column height until
1.00% drift, the larger width of the first one was noticeable at naked eye. Nonetheless,
horizontal cracks with a regular distribution could still be observed, as illustrated in Figure 5.90.
0.50% 0.75% 1.00%
Figure 5.90 – SP_F03: Evolution column cracking pattern
After 1.00% drift, crushing and spalling of the grout below the main horizontal crack was
increasingly more relevant and, at 2.00% drift, the longitudinal reinforcement was fully exposed
and buckled. At this point, there was a clear gap between the grout and the column concrete,
and, with the bulk of buckling occurring exactly at this section, it could be observed to be
working similar to a hybrid rocking joint. Eventually, the large strains of the main
reinforcement caused fracture of multiple rebars, as evidenced in Figure 5.91, and total collapse
of the transition section constituting the first 0.20m of column length.
5.72
1.50% 2.00% 3.00%
Rebar fracture
Figure 5.91 – SP_F03: North side spalling, buckling and fracture
5.3.1.4. Summary of results
▪ Cracking first occurred between 0.30% and 0.50% drift levels, and developed until
1.00% drift;
▪ Collapse was determined at 3.50% drift level for SP_F01 for the positive loading
direction only, at 3.00% drift levels for both loading directions on SP_F02, and at the
2.50% drift level on SP_F03;
▪ All specimens’ failure mode is associated with fracture of longitudinal reinforcement;
▪ Specimens SP_F01 and SP_F02 exhibited roughly equivalent behavior, with low peak
force asymmetry between loading directions and similar peak forces in close agreement
with the values estimated from moment-curvature analyses;
▪ Specimen SP_F03 exhibited considerably smaller peak force than the estimated value,
related with early crushing of the grout part of the connection and overdevelopment of a
gap between the grout and the column concrete;
▪ The large gap on SP_F03 could be observed to be performing similarly to a hybrid
rocking joint;
▪ Specimen F03 exhibited a response in which the rebars were still able to mobilize
greater strength for increasing displacement cycles, even after peak lateral force was
recorded.
5.3.2. RESULT DISCUSSION
5.3.2.1. Global Behavior and Damage Incursion
Comparison of the global behavior exhibited by each specimen is performed by analyzing

the drift levels corresponding to the occurrence of the following performance levels, defined
according to experimental observations:
5.73
▪ Performance Level 1 (PL1): Detection of cracking;

▪ Performance Level 2 (PL2): Detection of concrete crushing/spalling;
▪ Performance Level 3 (PL3): Detection of longitudinal rebar buckling;
▪ Performance Level 4 (PL4): Fracture of longitudinal reinforcement;
PL1 PL2 PL3 PL4

4.00%
3.50%
3.00%
2.50%
2.00%
1.50%
1.00%
0.50%
0.00%
SP_F01 SP_F02 SP_F03
Figure 5.92 – Performance level comparison between Phase 3 specimens
Figure 5.92 presents the results associated with the performance of Phase 3 specimens
regarding each of the previously defined levels. Taking into account that SP_F01 was a fully
monolithic specimen, while SP_F02 and SP_F03 involved at least partial precast, it is clear that
neither of the proposed footing-to-element connections could perform as good as intended. In
that regard, both the precast models experienced the same performance as the monolithic
specimen but at earlier drifts; considering that no significant material property or constructive
detail, other than the precast mechanism, can be pointed out as a potential cause, it is firmly
believed that the latter is the main reason for the worse performance of SP_F02 and SP_F03.
Nonetheless, there are still considerable differences between the precast tests themselves, since
SP_F03 incurred considerable damage earlier than SP_F02. In fact, the relative difference
between the drift cycle when the performance levels are achieved in SP_F01 and SP_F02 is
very similar to the difference between SP_F02 and SP_F03, which corresponds to a shift of
around 0.50% drift.
The methodology for definition of an equivalent bilinear system previously presented in

Figure 5.67 was also applied to these tests. In that regard, the obtained cumulative ductility
values achieved by each specimen (until collapse is determined in the positive loading direction)
are presented in Figure 5.93; relative diferences can be observed between the precast specimens,
and both of them also exhibit smaller values than the monolithic model. The latter achieves a
cumulative ductility of 97.4, which is around 30% larger than the value reached by SP_F02
5.74
(75.6), but almost three times larger than the value obtained in SP_F03 (35.1). The yielding
drifts calculated for both loading directions according to the previously mentioned methodology
are also given in Figure 5.93, showing similar values between them, with a slightly smaller one
determined for SP_F02.
120
100
80
μcum
60
40
20
0

dy+ (%) 0.73 0.66 0.71
dy- (%) 0.71 0.66 0.72
Figure 5.93 – Cumulative ductility at the collapse threshold for the positive loading direction
5.3.2.2. Stiffness Degradation
The stiffness degradation according to the cyclic loading of the Phase 3 columns was
calculated for each post-yielding half-cycle and is plotted against cumulative ductility in Figure
5.94. For comparison purposes, the values for each specimen are normalized by the
corresponding yielding stiffness Ky, and are represented by different line styles: solid, dash-dot
and dashed, for SP_F01, SP_F02 and SP_F03 respectively. The cumulative ductility achieved
as the columns reach the collapse threshold is also identified in the same figure by the different
markers placed on the corresponding curves. In this regard, the general evolution of the secant
stiffness is similar on all the tests, with degradation of around 50% of the yielding stiffness
observed for cumulative ductility values about 10. Furthermore, while the collapse threshold is
achieved within just above 30% of the yielding stiffness for SP_F03, both SP_F02 and SP_F01
are able to perform for longer, by reaching that mark within less than 25% of the yielding
stiffness. The secant stiffness at the final stage (corresponding to a post-collapse threshold drift
cycle) of the SP_F03 test (at about cumulative ductility of 110) was at nearly 15% of the
corresponding yielding stiffness. By contrast, on the other two tests similar secant-to-yielding
5.75
stiffness ratio was achieved at cumulative ductility values of 135 and 140 (18%) for SP_F02
and SP_F01, respectively.
1.00
SP_F01
0.90
0.80 SP_F02
0.70 SP_F03
0.60 20% Red.SP_F01

K/Ky
0.50 20% Red.SP_F02

0.40
20% Red.SP_F03
0.30
0.20
0.10
0.00
0 20 40 60 80 100 120 140 160 180 200
μcum
Figure 5.94 – Stiffness Degradation vs. Cumulative Ductility
5.3.2.3. Energy Dissipation
Cumulative energy dissipation is plotted against the number of loading half-cycles in

Figure 5.95, and different specimens are again represented by distinct line styles: solid, dash-dot
and dashed, for SP_F01, SP_F02 and SP_F03 respectively. Furthermore, the half-cycles, during
which the collapse threshold was achieved, are also represented by different markers, placed
over the corresponding curves.
According to experimental evidence, the most significant column damage due to increasing
cyclic drifts was associated with concrete crushing/spalling, as well as rebar buckling and
eventual fracture, which occurred in every test at comparable drift levels. Therefore, it is not
surprising that the overall rate of energy dissipation is relatively similar between all the
specimens, even if the shape of the force-drift response is not, since it is heavily dependent of
the deformation incurred by the main longitudinal reinforcement, which is also similar in all of
them.
Despite that, a closer look shows that, for the half-cycle (45) after the collapse threshold is
achieved in SP_F03, the increase rate for that specimen slows down when compared with the
other tests, which is arguably expected due to the more fragile behavior exhibited during the last
drift stages. Furthermore, SP_F02 is the specimen that dissipates the most energy until the 57 th
half-cycle, which can be attributed to the fact that it also achieved yielding earlier (Figure 5.93).
Still, the monolithic specimen SP_F01 shows the highest energy dissipation before collapse,
which is also around 35% larger than for SP_F02, and almost three times as large as SP_F03.
5.76
100.00
SP_F01
90.00
80.00 SP_F02
Dissipated Energy 70.00 SP_F03
60.00
(kN x m)
20% Red.SP_F01
50.00
20% Red.SP_F02
40.00
20% Red.SP_F03
30.00
20.00
10.00
0.00
0 10 20 30 40 50 60
Half-cycles
Figure 5.95 – Energy Dissipation vs. Loading Half-Cycles
5.3.2.4. Lateral Displacement and Curvature Profiles
The lateral displacement and vertical strains for the columns were recorded over several
points along their height, according to the monitoring layout presented earlier in Figure 5.18. In
order to characterize the relative differences between the deformation patterns for each of the
columns, the following methodology was adopted:
▪ Comparison of the lateral displacement profile, defined for a mean of the

horizontal displacement values obtained on both sides of the column, recorded at
the height wise column levels of 0.275m, 0.550m, 1.200m and 1.950m;
▪ Comparison of the mean curvature profile calculated from the North and South
side vertical deformation values, considering the mid cross-section between the
application points of each vertical LVDT, corresponding to column height values
of 0.069m, 0.206m, 0.413m, 0.875m and 1.525m;
For clarity, only three different drift levels were considered for calculations, namely: 0.50%
drift (which corresponds to a pre-yield stage), 1.00% drift (corresponding to a post-yield stage)
and 2.00% drift (which is a peak or post-peak stage, depending on the actual specimen).
Figure 5.96 illustrates the lateral displacement profiles obtained for the monolithic
specimen SP_F01 for the three drift levels, represented by different colors and markers,
according to the respective legend. It indicates a relatively linear shape, with slightly larger
relative displacement increases occurring near the column head, as expected.
5.77
2.00
1.80
1.60
Height wise column levels (m)

1.40
1.20
1.00
0.80
0.60
0.40
0.50%
0.20 1.00%
2.00%
0.00
0 10 20 30 40
Disp. (mm)
Figure 5.96 – SP_F01 lateral displacement profiles
Equivalent displacement profile shapes are obtained on the other specimens, with slight
differences that are, however, difficult to evaluate at the scale of Figure 5.96. In order to provide
better comparison, the ratio of the precast specimens’ horizontal displacements to the equivalent
displacement on SP_F01 is plotted in Figure 5.97 instead, and the respective results are
represented as a percentage relative to the latter. In this regard, displacement profile differences
between SP_F01 and both precast specimens at 2.00% drift were relatively small, as a
maximum ratio of around 13% was found in the cross-section closest to the footing. However, a
different scenario is observed at earlier drifts, where ratios above 20% could be found in several
cases. Most of those differences are not relevant, because they result from relative variations of
quite small absolute displacements. Still, at 1.00% drift ratios for SP_F02 were found all below
10%, but larger differences of up to 55% were observed in the case of SP_F03. Taking into
account that this was also the same drift level where the precast connection section first showed
signs of grout crushing, it may possibly indicate that some sliding might have occurred along
the grout-concrete interface, which is located below the first reading point.
5.78

1.95
+0.47 1.95
-1.21 1.95
+0.31
+1.47 +2.55 +1.15
Column height (m)

Column height (m)
Column height (m)

1.20 -5.98 +13.11 1.20 +11.62 1.20 +5.65
+2.90 +4.66
+31.14 +5.25
0.55 -19.44 +26.79 0.55
+0.10
0.55
+9.34
0.28 -35.63 +37.85 0.28 -8.84 +55.46 0.28 -5.58 +13.22
-60.00 -30.00 0.00 30.00 60.00 -60.00 -30.00 0.00 30.00 60.00 -60.00 -30.00 0.00 30.00 60.00
% % %
a) 0.50% drift b) 1.00% drift c) 2.00% drift

Figure 5.97 – SP_F01 to precast specimens lateral displacements ratio
The mean curvature profile obtained for each drift level is presented in Figure 5.98, where
SP_F01, SP_F02 and SP_F03 are represented by the red, blue and green lines and markers,
respectively, and an indicative yielding threshold is also included (Figure 5.82). Regarding
these results, SP_F01 and SP_F02 display very similar progression for all the calculated drift
levels, but SP_F03 presents an interesting detail. In fact, while the highest curvature on the
former two specimens is always located at the cross-section closest to the footing, as expected
on a bending-dominated reinforced concrete column response, the second closest cross-section
always shows the largest value in the case of the SP_F03. However, despite the uncommon
nature of that particular observation, it seems to be in line with the experimental evidence, since
both rebar buckling and fracture occurred next to the grout-concrete interface, as illustrated in
the last picture of Figure 5.91, and not closer to the footing as in the SP_F01 and SP_F02
specimens (Figure 5.85 and Figure 5.88, respectively).
The analysis of the curvature profiles shown in Figure 5.98 also seems to confirm the
occurrence of yielding after the 0.50% drift cycle, because the largest curvatures are lower than
the yielding threshold on all the tests. Furthermore, at 2.00% drift, the largest maximum
curvature is recorded for SP_F03, followed by SP_F02 and SP_F01 as the second largest and
the smallest maximum curvatures, respectively. This presents further evidence to understand the
differences in the ultimate capacity of each specimen, which follows the same relative order.
5.79
2.00 2.00 2.00

1.80 SP_F03
1.80 SP_F03 1.80 SP_F03
Yield Yield Yield
1.60 1.60 1.60
Column height (m)
Column height (m)

Column height (m)
1.40 1.40 1.40
1.20 1.20 1.20
1.00 1.00 1.00
0.80 0.80 0.80
0.60 0.60 0.60
0.40 0.40 0.40
0.20 0.20 0.20
0.00 0.00 0.00

0.000 0.025 0.050 0.075 0.100 0.000 0.025 0.050 0.075 0.100 0.000 0.025 0.050 0.075 0.100
Curv. (rad/m) Curv. (rad/m) Curv. (rad/m)
a) 0.50% drift b) 1.00% drift c) 2.00% drift
Figure 5.98 – Mean curvature profile
5.4. FINAL REMARKS
The present chapter involved presenting the experimental campaign performed on the
scope of the present work in full detail. According to the methodology adopted for the referred
campaign, which was presented earlier in Chapter 4, a three phase approach was required. In
that regard, two different experimental setups were developed for reduced scale (1:4)
applications in order to adequately address the different challenges posed by each of them,
where the same setup was used for Phase 1 and 2 and the second setup was used only on Phase
3. The first setup was aimed at handling the cyclical loading of the earlier defined monolithic
(Phase 1) and precast (Phase 2) bent pier specimens, where some of the main challenges were
the constant axial loading of the columns, free rotation at the column bases simulating null
moment curvature inflexion points and lateral load application by shear at the expected interface
between bearings and structure. The second experimental setup was a typical layout adequate to
cyclical in-plane loading of single column structures.
The experimental tests performed on the Phase 1 specimens, which consisted of only
monolithic models, showed structural responses largely dominated by shear at the coupling
beam (exceeding the level of applied dead-load), wherein brittle failure modes could also be
observed at relatively low drift levels (under 3.00%). In all cases, structural damage in the
columns was moderate, and well distributed horizontal cracking patterns could be found.
Analysis of experimental evidence and predictive peak force estimates suggested that the
influence of shear could not be very well evaluated by the design models used in this work,
5.80
possibly due to the formation of a single strut-and-tie mechanism between columns, and
increased reliance on the shear interlock mechanism at beam-column interfaces.
Regarding Phase 2 tests, performed on mostly precast specimens, significant improvements

could be found. Although beam shear was still observed to be a highly relevant factor to the
overall performance of the bent pier structures (the beam shear loads still exceeded the level of
applied dead load), an increased reliance on rebar deformations at precast joint sections could
also be observed. Due to that, all the precast specimens were able to show a more ductile
response, where loading up to 4.00% drift could be performed within reductions of the peak
force of 20%. Structural damage in the columns was also kept moderate, except on the
SP_M02C case (monolithic structure with a reinforcement layout designed for a precast
specimen) where a large horizontal crack could be found at the beam-column intersection.
Finally, Phase 3 specimens addressed foundation-to-element connection tests, but, in the

tested cases, the structural performance of the precast specimens was always worse than the
reference monolithic model. Nonetheless, precast specimen SP_F02 was able to show similar
behavior patterns to those of the monolithic model, particularly regarding the occurrence of
spalling, buckling and failure, suggesting that improvements on that regard may be easier to
achieve than using the layout of SP_F03, which exhibited significantly worse performance.
5.81
Numerical Simulations
6
NUMERICAL SIMULATIONS
6.1. INTRODUCTION
In order to extend conclusions from the experimental observations discussed in the

previous chapter to the wider range of piers associated with the previously defined conceptual
framework (double column piers with short span coupling beams with total height in between
5m to 20m for application in medium to long viaducts), those results should be independent of
testing conditions and, namely, of the total height of the specimens. Within that context and due
to the strong association of the structural response with the variable stiffness ratios, between the
beam and the columns, that were previously shown to depend of the total height, doubts can be
raised regarding the extrapolation of testing conclusions to other scenarios. Taking this into
account and intending to evaluate the problem in further detail, a strategy was established for
the work developed in this chapter, which was based on accomplishing two distinct objectives:
▪ Providing additional insight on the relative differences between monolithic and

precast specimens aiming at improving the knowledge gathered by the previous
experimental conclusions;
▪ Developing a modeling strategy capable of exploring different geometrical
conditions and constraints than those tested at the laboratory, in order to assess the
validity of extrapolating test results to a wider range of applications;
6.2. REFINED MODELING
In order to explore different applications of the bent pier systems presented in Chapter 4
within a purely numerical setting, a modeling strategy capable of addressing the main
phenomena reported during the experimental tests is paramount. Additionally, considering the
6.1
availability of experimental data, a two-staged approach was adopted, first focused on providing
a comparison of the numerical results obtained with the modeling approach presented in the
current work, with those recorded during the experimental tests, for validation purposes. In this
context, the following set of specimens was selected for numerical study:
▪ Monolithic specimens: SP_M02; SP_M03; SP_M04;

▪ Precast specimens: SP_PC02A; SP_PC02B;
▪ Single column specimens: SP_F01;
Afterwards, considering the experimentally validated numerical applications and the

corresponding calibration parameters, the second stage focused on exploring other scenarios not
experimentally tested; for the most part, this was based on introducing targeted structure
changes, such as different total height, or support restraints.
For that purpose, the finite element method (FEM) was used with a refined approach,
including thorough detailing of the inherent geometric and material properties of each specimen.
Appropriate constitutive models were also chosen and calibrated against the available test data.
Furthermore, since the numerical simulations in question generally require extensive and long
analyses that are often very prone to numerical convergence problems, the modeling strategy
was also developed with attention for computation time requirements and result precision. In
this regard, all the analyses were performed with the structural analysis software Cast3m
(http://www-cast3m.cea.fr/, Fichoux (2011)), considering the author’s previous experience with
the program, the support for both 3D and 2D applications, as well as the wide spectrum of
different constitutive models that is available within its software environment.
6.2.1. MODELING STRATEGY
6.2.1.1. Phase 1 and Phase 2 tests
Using a FEM based methodology for the numerical applications of this work requires a
very early decision regarding the associated assumptions. In particular, it is important to decide
beforehand between a three or two dimensional approach. Taking into account the scope of the
experimental tests, focused on recording the cyclic behavior of the bent pier specimens under
uniaxial loading, as well as the setup features installed to deal with out-of-plane displacements,
it seems reasonable to assume that the structural problem in question can mostly be reduced to
the loading plane. With that in mind, a valid strategy can consist on elaborating a 2D FEM mesh
for analyses under the plane stress assumption.
Since the bent piers in question are essentially symmetric along the loading plane, reducing
their full geometry to a two-dimensional representation can be done by adopting the mid-section
6.2
plane cut as the basis for mesh geometry, according to Figure 6.1. Therefore, the 2D model can
be developed by adopting constant thickness in the beam elements and varying thickness on the
column elements, aiming for an equivalent overall pier lateral stiffness. The solution adopted for
this work involved the consideration of three different thickness zones, A, B and C.
A
Column Cross Section
a) 3D geometry: mid-section plane b) 2D Plane stress mesh
A B A
Column Cross Section A Top view
Beam B A
Column Cross Section
c) Column cross-section d) Beam section
C
A
Beam B
Top view A
C
Beam Top view
e) Column thickness layers f) Beam thickness layer

Figure 6.1 – Geometric definition for numerical modeling
In addition to theC finite elements representative of the bent piers, the numerical model also
considered the inclusion of 4 linear elastic element bands above and below the two columns
(two per column) to enable lateral load application and column rotation occurring at the same
relative positions (Figure 6.2) of the test setups.
6.3
All the analyses were performed considering at least two types of finite elements, defined
for the representation of the concrete continuum (eight-node rectangular elements) and the steel
reinforcement (two-node bar elements). Within that context, the adopted mesh density was the
same for all the analyses, and considered 0.050m sized element divisions. The previous mesh
density value was defined as the result of a parametric study regarding element size, which
essentially intended to evaluate the computational efficiency associated with different meshes
regarding calculation times (while taking into account result quality and detail), and is included
in Annex B. An overview of the adopted FEM mesh layout is presented in Figure 6.2 as defined
for the monolithic specimens of Phase 1, including an illustration of the associated support
restraints and applied loading conditions, as well as the element thickness for the different
zones.
Figure 6.2 – Adopted generic layout for the FEM mesh
Accurate characterization of the concrete cover would be incompatible with the previously
presented plane mesh due to its constant mesh size. Changing the mesh to accommodate that
feature was possible but, ultimately, was not deemed a good option. The main reason supporting
that decision was that due to the 2D modeling approach, only a part of the cover concrete can
actually be represented, located in the external finite element layers according to the 2D
representation. In addition, Figure 6.3 illustrates a hypothetical layer of finite elements for
simulation of cover concrete, where the green and red colors represent elements with expected
low and high ductility demand in compression (the color transitions represent the uncertainty
associated with pinpointing the actual zones where that occurs). Elements in green (the
majority) should have a low impact on the overall performance of the models; elements in red,
however, are crucial for the development of the internal concrete struts that carry the load
transferred through the beam and, therefore, are also critical to the accuracy and numerical
6.4
stability of the structural response. Within that context, it can be argued that the ductility
capacity of those zones should be defined by careful assessment through calibration against
experimental results, since it may be influenced by surrounding columns providing a local
confinement effect, therefore making it less important to simulate the actual behavior of the
cover concrete.
Figure 6.3 – Outer finite element layer and ductility demand
Regarding the simulation of the precast models of Phase 2, the finite element meshes need
to include an additional feature, in order to enable the representation of the grouted joints.
Within that framework, joint or interface elements are generally used to represent small/thin
layers that incorporate considerably distinct constitutive characteristics over another continuum
element (often with much larger dimensions), for example to model the thin mortar layer
located between masonry blocks. In a way, that is believed to be similar to the current case,
where the vertical columns are separated from the beam element by a passively reinforced
grouted joint with negligible thickness. Therefore, the adoption of a joint element to simulate
the force transfers occurring in that interface becomes a natural choice.
For that purpose, 6 node joint elements were used to establish a binding connection
between the nodes on the beam segment and the vertical columns (by making use of the RACC
operator available in Cast3m). Those joint elements follow the proposal of Bfer (1985), as
implemented in Cast3m, and require the definition of adequate constitutive laws for the stress
(normal σn, and tangential τs) – strain evolutions along their normal (δj) and tangential (γj)
degrees of freedom, according to Figure 6.4. The structural analysis software includes several
constitutive models developed for simulation of joint behavior, which will be discussed further
ahead.
6.5
a) Joint element 2D representation
n s
j
j
n s
b) Joint degrees of freedom

Figure 6.4 – Joint element in Cast3m (Pegon), Costa (2009))
With this approach, the nodes of the rebar elements and the concrete continuum mesh
cannot be connected directly, since the constant mesh size prevents coinciding points for both.
In order to address that shortcoming, a linear kinematic constraint (resorting to the RELA
‘ACCRO’ operator in Cast3m) was established to bind together the displacements of concrete
and steel finite element nodes.
The finite element layouts for the reinforcement were defined considering a similar
approach to that presented in Figure 6.1 for the concrete continuum, by projecting all existing
rebars onto the mid-plane section cut, involving the consideration of equivalent diameters for
the various rebar elements of each structure. Figure 6.5 shows an overview of the upper part of
the bar element meshes used for all the specimens of Phase 1 and 2, wherein the colored lines
represent equivalent rebars with different diameters or mechanical properties tuned so as to
match the set of original ones. It should be noted that according to the original specimen
layouts, the same column mesh is used in all the analyses of Phase 1 specimens, since only the
beams involve different detailing. By contrast, the beams respect the same mesh layout on
Phase 2 specimens, and detailing differences are introduced on the vertical reinforcement
around the precast joint section. Finally, the transverse reinforcement of the beam on SP_M03
was defined with shorter spacing than the other tests because calculation of this specimen
presented particularly difficult numerical convergence issues due to high local distortions
caused by shear. Shorter spacing on the stirrups enabled better handling of that difficulty, while
accounting for reduced cross-sections for maintaining overall equivalent transverse
reinforcement contents.
6.6
a) SP_M02 b) SP_M03
c) SP_M04
d) SP_PC02A e) SP_PC02B
Figure 6.5 – FEM rebar meshes for Phase 1 and Phase 2 analyses
The interaction of the bar elements at the joint section on precast specimens required
special attention, because, in addition to the disturbance of regular force transfers that it
introduces, the discontinuity leads to weaker bond behavior between steel and concrete in the
local vicinity. The implications of that can be related to the considerable differences of
performance observed between specimens SP_PC02C and SP_M02C, which can be attributed
to that specific issue, since both included the same reinforcement layouts whereas the only
significant difference was the existence of the precast joint. In the present application, this was
addressed with a careful selection of the nodes and elements to be bound with the previously
mentioned kinematic constraint, since binding all the rebar nodes in the precast joint vicinity
can easily lead to an artificially increased stiffness because this creates an indirect connection
between superposed nodes at the joint (which then become linked through the kinematic
relation in addition to the intended joint stiffness). Therefore, the binding strategy adopted for
continuity rebars skipped the joint nodes, while the rebars that were cut at that level were bound
only to the column part, as illustrated in Figure 6.6 for one of the beam-column joints.
6.7
Figure 6.6 – Binding strategy at the precast joints
6.2.1.2. Phase 3 tests
Modeling single columns under bending dominant loading is generally a more

straightforward task than what is required for simulation of phase 1 and 2 tests, and good results
can be obtained by involving experimental calibration and using accurate mesh geometry with
adequately assessed constitutive relationships. Within that context, and owing to the reduced
complexity of that numerical application, 3D modeling was considered, aiming to improve the
overall accuracy of the results in relation to comparable 2D tests (for example, by fully
modeling the cover concrete). It is important to note, however, that this is a reasonable approach
because the same constitutive models and general behavior relationships used for the 2D tests
can also be applied to a 3D application.
Within this context, taking advantage of the essentially symmetric nature of the column
structure, only half of its body continuum was considered, and the volumetric envelope for the
FEM mesh was defined according to the representation of Figure 6.7.
6.8
a) Total specimen b) Half specimen
Figure 6.7 – Geometric 3D envelope definition for columns.
The FEM mesh elaborated for this purpose focused on the use of eight-node cubic elements
for representation of the concrete continuum and two-node bar elements for the steel
reinforcement. Different mesh densities were used in the column elements (0.025m and
0.050m), to provide a more refined mesh on the column base region (defined for roughly half
cross-section height), where the concentrated inelastic phenomena occurrences are mostly
expected. Linear elastic behavior was adopted for the footing elements, as well as the upper
finite element band (red elements) included for load application, supporting the use of a more
coarse mesh. Figure 6.8 illustrates an overview of the described mesh.
Mesh density: 0.050m
Mesh density: 0.025m
b) Mid-section cut view

a) 3D view
(with boundary conditions)
Figure 6.8 – Phase 3 FEM mesh
6.9
Concerning steel reinforcement (Figure 6.9), longitudinal (blue elements) and transverse
reinforcement (red elements) were also defined individually, accounting for every rebar placed
on its accurate position inside the column. A similar strategy to that presented for the bent pier
specimens was also adopted to bind the displacements of steel reinforcement nodes to the
concrete mesh.
Figure 6.9 – Steel reinforcement finite element mesh
6.2.2. CONSTITUTIVE MODELS
Selection of adequate constitutive relations for simulation of concrete and steel is of the
utmost importance on the context of non-linear modeling of reinforced concrete structures.
Moreover, increasing the quality and detail of the numerical responses given by those
constitutive models has always been a relevant research topic, with ever more robust and
efficient models being developed continuously. In that regard, one of the main advantages of
using the Cast3m software for the current exercise is the amount of non-linear models available
for use, for which several different formulations can be considered, from plasticity based
models to others associated with continuum damage mechanics. A choice between different
alternatives should, then, inevitably be related to each model’s ability to reproduce (efficiently
and accurately) the required local phenomena. Regarding this work, that choice is evaluated by
promoting separate discussion over concrete, steel and joint constitutive models. All the adopted
model parameters are included in Annex A.
6.10
6.2.2.1. Concrete Model
There are several models available in Cast3m for simulation of concrete, from which the
following were selected for consideration, because they are more suitable for plane stress cyclic
applications:
▪ BETON_INSA: This model was developed at INSA (Institute national des

sciences aplicquées de Lyon) within the framework of plasticity theory for
uncracked concrete in plane stress, following the Ottosen criterion Ottosen (1977)
for compression and tension.
A smeared fixed crack approach is also considered, and when the ultimate failure
surface is reached in tension, the model considers the opening of a crack, leading to
the uncoupling of compressive and tensile behaviors, which are afterwards
modeled by an orthotropic law (Ile and Reynouard (2000)).
▪ DAMAGE_TC: A constitutive model developed on the framework of continuum

damage mechanics (Faria and Oliver (1993), Faria (1994)), using the concept of
effective stresses defined on the principal directions space, to characterize the
non-linearity according to the energy dissipation produced by the accumulation of
damage. Two independent and scalar damage variables are introduced to describe
the strength degradation under tensile and compressive stresses separately,
according to individually calibrated evolution laws. Furthermore, fracture
mechanics is taken into account, as a smeared crack approach is adopted for the
tensile behavior. In addition, plastic deformations in compression are also
considered in the implemented version in Cast3m (but not in tension).
▪ DRUCKER_PRAGER_2: A full plasticity based model (Jason and Durand (2007))

developed with two yield surfaces, following different criteria according to the
loading state. In compression, a Drucker-Prager criterion is adopted, including a
hardening-softening law calibrated to enable an equivalent response to
experimental uniaxial tests. By contrast, a Rankine type surface governs the tensile
behavior, considering an exponential softening law defined in accordance with
fracture mechanics. The parameters required to calibrate this model are
straightforward and easy to obtain from standard material tests.
RICRAG: This model was developed in the framework of thermodynamics of

irreversible processes (Richard et al. (2010)), intended to address inelastic
phenomena by introducing an isotropic variable that is linked to the
6.11
thermodynamic state of the material volume under consideration. The loading state
is then characterized by a single damage variable (whether compression or tension
occurs) and a single flow rule. It also aims to accurately reproduce reinforced
concrete monotonic and cyclic behavior by decoupling the elastic free energy in
order to address inelastic effects separately (e.g. crack opening/closure and
frictional sliding).
Understanding the options and the differences between is deemed a quite valuable help to
choose an adequate model for use in the subsequent numerical analyses. Nonetheless, seeking a
deep and attentive review of all of these models is out of the scope for this work. With that in
mind, a very simple analysis was performed instead, focusing on modeling a single 1m2 element
with all the four concrete constitutive model alternatives, considering monotonic (in tension and
compression, Figure 6.10 – b)) and cyclic (Figure 6.11) loading conditions. Its main purpose
was to obtain the numerical responses at the Gauss point level, and to analyze potential
outcomes of application in the full model. The characterizing parameters of each model were
calibrated for reasonable fit of a class C25/30 concrete, defined for the corresponding mean
values of fcm and Ecm according to Eurocode 2, and considering similar tension-stiffening effect
for the tensile behavior in all the models.
The loading was made as imposed displacements, with appropriate boundary conditions as
shown in Figure 6.10 - a). Figure 6.10 – b) highlights the differences exhibited between the four
models for the same monotonic loading. As observed, there are some clear differences on the
softening behavior of the curves, although the overall envelope shapes stays relatively
unchanged in both compression and tension. It is only when cyclic loading results are observed
that larger differences become apparent.
6.12
5.00
Imposed Displacements ε (1x10 -3)
0.00
-6.00 -5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00
-5.00
-10.00
-15.00
-20.00
-25.00
-30.00
-35.00
σ (MPa)
-40.00
BETON_INSA DAMA_TC DRUCKER_PRAGER2 RICRAG
a) Single element model b) Compression and tension envelopes

Figure 6.10 – Concrete models Gauss point monotonic response
Figure 6.11 displays the cyclic response obtained with all the tested models. There are
striking differences between them, especially in the RICRAG and DRUCKER_PRAGER2
results, in light of key aspects inherent to the respective formulations. For example, the
RICRAG model takes into account a single damage variable to characterize the concrete
degradation, for both tension and compression. When cyclic loading is applied, a part of the
damage accumulated in compression or tension goes through for the other loading direction,
since the model just includes a partial unilateral effect. When large strains are expected for both
loading directions, this effect is more prevalent, affecting the overall structure stiffness and
energy dissipation substantially. On that regard, Cast3m also includes an upgraded version of
this model, called RICBET (Richard and Ragueneau (2013)), which accounts for a full
unilateral effect but, unfortunately, is only implemented for 3D elements. As for the
DRUCKER_PRAGER2 model, it does not include any unloading/reloading stiffness
degradation, thus always adopting elastic values instead. That results in larger permanent
deformations than expected and, potentially, higher energy dissipation as well.
6.13
5.00 5.00
ε (1x10 -3) ε (1x10
-3)
0.00 0.00
-6.00 -5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00 -6.00 -5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00
-5.00 -5.00
-10.00 -10.00
-15.00 -15.00
-20.00 -20.00
-25.00 -25.00
-30.00 -30.00
σ (MPa) σ (MPa)
-35.00 -35.00
a) BETON_INSA model b) DAMAGE_TC model

5.00 5.00
ε (1x10
-3) ε (1x10
-3)
0.00 0.00
-6.00 -5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00 -6.00 -5.00 -4.00 -3.00 -2.00 -1.00 0.00 1.00
-5.00 -5.00
-10.00 -10.00
-15.00 -15.00
-20.00 -20.00
-25.00 -25.00
-30.00 -30.00
σ (MPa) σ (MPa)
-35.00 -35.00
c) DRUCKER_PRAGER2 model d) RICRAG model

Figure 6.11 – Concrete models Gauss point cyclic response
With that in mind, the results obtained from the BETON_INSA and DAMAGE_TC models
can be expected to provide more accurate cyclic behavior simulation than the other two, as both
include stiffness degradation, full unilateral effect, permanent deformations and display
comparable envelope shapes. Their main difference (which isn’t apparent in Figure 6.11) is that
the BETON_INSA model takes into account permanent deformations for both compression and
tension, while the DAMAGE_TC version implemented in Cast3m only accounts for permanent
deformations in compression. Despite that, the former revealed to be harder to calibrate, since
testing different combinations of characterizing parameters led to substantially more numerical
problems than the latter. In addition, DAMAGE_TC can provide a very valuable result
assessment tool for post-processing, due to its two independent damage variables.
Within this context, an argument could be made for the viability of using the
BETON_INSA model, as permanent deformations in tension are important for accurate
simulation of unloading/reloading stiffness and overall energy dissipation, especially in shear
dominant situations where the tensile behavior of concrete is likely more relevant. Nonetheless,
DAMAGE_TC has already been used for applications similar to those in the current work with
considerable success by Vila Pouca (2001), Monteiro (2009) or Delgado et al. (2011) even
without plasticity in tension, and the advantage of enabling straightforward result interpretation
6.14
due to the damage variables should not be underestimated. These reasons and the author’s
previous experience with it supported a choice for the DAMAGE_TC model, which was used to
simulate the concrete continuum for all the numerical analyses of this work. A more detailed
overview of the uniaxial cyclic behavior computed by this model is presented in Figure 6.12,
including a numbered path (from 1 to 7) related to the associated loading variations.
 
 
 
 
Figure 6.12 – Example of DAMAGE_TC uniaxial cyclic behavior
According to Faria (1994), the complete characterization of this model requires the
definition of the independent evolution laws of damage variables for compressive and tensile
behavior (equations 6.1 and 6.2, respectively). With that in mind, the parameters required for
that purpose, like the elastic or cracking thresholds, can mostly be obtained from uniaxial
sample tests. Additionally, one evolution law is adopted for compressive behavior and another
for tensile behavior (although two are available in the current Cast3m implementation). In
compression, two fitting points must be provided to compute the evolution of the 𝑑− damage
variable (related to the A- and B- parameters of equation 6.1). Those points can readily be
obtained from experimental uniaxial stress-strain curves, if available. Otherwise, the
DAMAGE_TC model can be calibrated against theoretical concrete curves, such as Mander et
al. (1988) or Kent and Park (1971), among others available in the literature.
−
𝑟0− 𝜏̅
𝐵− ×(1− − )
𝑑 = 1 − − × (1 − 𝐴− ) − 𝐴− × 𝑒
− 𝑟0
6.1
𝜏̅
Where:
𝜏̅ − : Effective stress; 𝐴− : Fitting point related parameter;
𝑟0− : Elastic threshold; 𝐵− : Fitting point related parameter;
6.15
𝑟0+ 𝑟𝑢+
1− × ( − 1) , 𝑖𝑓 𝑟0+ ≤ 𝜏̅ + ≤ 𝑟𝑢+
𝑑+ = { 𝑟𝑢+ − 𝑟0+ 𝜏̅ + 6.2
1 , 𝑖𝑓 𝜏̅ + > 𝑟𝑢+
where:
𝜏̅ + : Effective stress; 𝒓+𝒖 : Effective stress corresponding to the ultimate
strain in tension;
𝑟0+ : Cracking threshold;
In tension, two evolution laws are available for the d+ damage variable, one with
exponential softening decay, and another with linear decay, which was used in this work. With
this approach, the 𝑟𝑢+ effective stress is related with depletion of the tensile capacity on the
concrete fracture width, considering a smeared crack based formulation. In reinforced concrete
problems, fully depleted tensile capacity is usually considered to occur close to reinforcement
yielding, due to steel-concrete bonding. The overall Gf fracture energy involved in the formation
of a discrete crack (Figure 6.13 – a)), is equivalent to the total fracture energy involved in the
smeared crack approach (Figure 6.13 – b)), which can then be determined from equation 6.3,
taking into account the l1 characteristic length of the FEM mesh. This was the approach
followed in the current applications, as linear tension softening was considered, as described in
Figure 6.13.
wu wu l1 l1
   
ft ft ft ft
g1 g1
gf
Gf Gf
wu w wu w u  u 
a) Discrete crack b) Equivalent smeared crack
Figure 6.13 – Fracture energy in smeared crack based models
𝜀𝑢 × 𝑓𝑐𝑡 × 𝑙1
𝐺𝑓 = = 𝑔𝑓 × 𝑙1 6.3
2
where:
𝑔𝑓 , 𝐺𝑓 : Fracture energy; 𝑙1 : Characteristic mesh length;
𝑓𝑐𝑡 : Cracking stress; 𝜺𝒖 : Ultimate tensile strain (yielding strain of the
main reinforcement);
6.16
For 2D and 3D applications, the definition of l1 requires taking into consideration different
possibilities for the cracking direction, thus it is usual to adopt simplified relations based on the
square root of the Gauss Point represented area or volume, respectively, as indicated by
equations 6.4 and 6.5.
√𝐴𝐺𝑎𝑢𝑠𝑠 , 𝑓𝑜𝑟 2𝐷 𝑝𝑟𝑜𝑏𝑙𝑒𝑚𝑠 6.4

𝑙1 = { 3
√𝑉𝐺𝑎𝑢𝑠𝑠 , 𝑓𝑜𝑟 3𝐷 𝑝𝑟𝑜𝑏𝑙𝑒𝑚𝑠 6.5
Regarding the numerical simulations, each analysis was performed using two domains for
concrete characterization, wherein distinct behavior curves were used. On Phase 1 and Phase 2
tests, a high ductility concrete was defined in addition to the regular concrete, aiming to provide
increased ductility on beam edges (as previously discussed, and based on Figure 6.3), as
illustrated in Figure 6.14 – a). For Phase 3 analyses, unconfined cover concrete and core
concrete domains were defined, according to the representation shown in Figure 6.14 – b).
a) Phase 1 and Phase 2 analyses b) Phase 3 analyses

Figure 6.14 – Different domains for concrete behavior laws
Within that context, model parameters were determined individually for each model by
aiming to reproduce the material properties identified in Table 4.12, by considering an
amplification of peak strength and ultimate strain based on Mander’s stress-strain theoretical
model for confined concrete (except for the SP_F01’s cover concrete, which has no
confinement). The main results obtained for the corresponding confined-to-unconfined ratios
are presented in Table 6.1, which were calculated considering confinement enabled by not only
the vertical stirrups, but also the diagonal parts of the main rebars crossing the beam web. In
fact, in the rhombic truss model, for each pair of compressed rebars there is another set of rebars
in tension. Due to their position and the fact that they fully cross the web length, it can be
argued that they may apply a similar effect as the inwards confining pressure of enveloping
6.17
stirrup corners, although likely not as efficiently. Still, assuming that the effectively confined
concrete area may be larger than the area calculated accounting only for stirrups’ contribution
(e.g. the area represented in the right side of Figure 6.15 versus the area on its left side), then the
contribution of the diagonal parts of the main beam rebars were also required, which is why it
was included in the calculations, although at 50% reduced effectiveness.
Table 6.1 – Confinement factors
𝑓𝑐𝑐 ⁄𝑓𝑢𝑐
Specimen Beam Column
SP_M02 1.13 1.17
SP_M03 1.05 1.21 vs.
SP_M04 1.13 1.17
SP_PC02A 1.15 1.22
SP_PC02B 1.14 1.19
SP_F01 N/A 1.25
Figure 6.15 – Effectively confined concrete assumption
In addition, trial and error calibration of the ultimate strain of the high ductility concrete
was made, aiming to have the numerical tests indicating similar concrete compressive damage
and failure as reported on the experimental results, particularly because it is heavily associated
with most of the phenomena reported for collapse characterization (concrete crushing onto
sliding failure, or onto spalling and subsequent tensile failure, like diagonal splitting).
6.2.2.2. Steel and steel-concrete bond models
Simulating the behavior of steel rebars is often performed with simple bilinear models,
depending on the purpose of the application. Generally, only the elastic stiffness and hardening
stiffness are required in addition to yielding and ultimate strengths for full characterization of
the envelope for a given cyclic loading. This is a well-known pattern, for which several models
have been developed over the years. On that regard, the Menegotto-Pinto model (Menegotto and
Pinto (1973)) has been one of the most widely accepted proposals, including the addition of the
softening aspect of the curves related to the Bauschinger effect, and is also implemented in
Cast3m (designated by the ACIER_UNI alias). An overview of the corresponding uniaxial
cyclic behavior enabled by this model is presented in Figure 6.16.
6.18

E h
E E

E h
Figure 6.16 – Menegotto-Pinto uniaxial cyclic behavior
In addition to the steel rebar constitutive model, it is important to discuss the steel-concrete
bond behavior. A rebar embedded in concrete and subjected to tensile forces accumulates
strains over the embedment length that cause an extension of the bar, which can be related with
the overall crack opening width for tensile strained concrete. However, rebar slip relative to the
embedment concrete can also occur, causing increased displacements/rotations for no increase
of force. This issue is also very relevant to the present bent pier system due to the large cracks
observed at the beam-column interfaces, and also due to the large opening width experienced at
the precast joints, as identified in Chapter 5, wherein some vertical cracks were actually
observed, and suggesting rebar slip.
The steel-concrete bond behavior is generally described by bond stress (τb) defined as a
function of the displacement of rebars, relative to the surrounding concrete and along the
anchorage length (Lanch, in Figure 6.17), which is commonly referred to as slip (s). The overall
relation between them depends on the actual stage of the bond connection, with a total of three
different stages usually referred as the most relevant:
▪ The first stage, controlled by chemical adherence, occurs up to very small slip
values and depends of the molecular connections established between the steel and
the cement paste.
▪ The second stage, governed by mechanical adherence, initiates when the chemical
adherence is broken and friction forces start developing (up to s1 in Figure 6.17),
causing internal micro cracks to form as well. This mechanism gradually increases,
in proportion with the friction and mechanical interlock, especially in the presence
of rugged indentations of deformed bars (with peak bond force developed between
the s1 and s2 slip thresholds).
▪ The third and last stage, ruled by residual adherence (up to and after the s3
threshold), occurs after the peak bond stress (τb,peak) is achieved and degradation
6.19
occurs, since a small residual strength (τb,res) can generally be maintained through
friction.
b
b,peak b Legend:
b,peak b - average bond stress over Lanch;
b,peak - peak bond stress;
b,res b,res
b,res - residual bond stress;
s1, s2 and s3 - slip displacement thresholds
for the different stages of the bond-model;
s1 s2 s3
s1 s2 s3 ss
Figure 6.17 – Eligehausen bond model as implemented with ACIER_ANCRAGE in Cast3m
Within this context, several theoretical models were proposed to enable taking into account the
effects of slip between concrete and rebars. Works presented by Tassios (1979), Eligehausen et
al. (1982), Harajli et al. (1995) or Huang and Engstrom (1996), for example, were well-received
by the community and some were incorporated in design codes. In Cast3m, the consideration of
a constitutive behavior for the bond simulation can be performed by explicitly modeling the
steel-concrete interface using specific finite elements for that purpose, or by adopting a
constitutive model for the steel bar finite elements that implicitly takes into account the
occurrence of both slip and rebar deformations. The latter approach was adopted in this work,
using the constitutive model labeled ACIER_ANCRAGE, which considers the previously
described Menegotto-Pinto constitutive behavior for the rebars as well as the model proposed by
Eligehausen (Eligehausen et al. (1982)) for the bond-slip behavior. Its formulation requires the
calculation of the effective rebar strain and slip from the total strain εtotal established at the bar
finite element level, according to Equation 6.6:
𝑠
𝜀𝑡𝑜𝑡𝑎𝑙 = 𝜀𝑠 + 6.6
𝐿𝑎𝑛𝑐ℎ
where:
𝜀𝑡𝑜𝑡𝑎𝑙 : is the total strain calculated at the finite element level;
𝜀𝑠 : is the strain associated with rebar deformations;
𝑠: is the mean slip displacement value over the 𝐿𝑎𝑛𝑐ℎ length;
𝐿𝑎𝑛𝑐ℎ : is the embedment length of the anchored rebar;
The overall behavior enabled by this model relies on the separation of slip and rebar strains
in order to, simultaneously, develop both bond and rebar stresses, according to the
corresponding constitutive laws. During cyclic loading, the unload and reload stages of
bond-slip behavior are managed with the corresponding elastic stiffness, while the rebars’
behavior is subjected to the hysteresis rules of the aforementioned Menegotto-Pinto model. The
6.20
main parameters required for calibration of the bond-slip model are those represented in Figure
6.17, where s1, s2 and s3 are the slip values corresponding to the start and end of peak
mechanical adherence stage and start of residual adherence stage, respectively, while 𝜏b,peak
and 𝜏b,res represent the peak and residual bond stresses.
Calibration of the Menegotto-Pinto model for use in each test was performed by
considering the material properties identified on Table 4.12 and adopting the default values for
the cyclic behavior controlling parameters. Regarding the bond model, it was considered for
simulation of all the heavily strained rebar elements located at the beam and precast joints,
whereas it was disregarded for general modeling of the columns. Therefore, it was used on all
but the single column test SP_F01, and the corresponding values were evaluated according to
literature references (CEB (1993), Santos (2012)) while taking into account adjustment for the
concrete strength values relative to each test. In this regard, the s 3 value, which is usually taken
as the distance between rebar ribbings, was evaluated for the highest diameter bar of each test,
considering A500 NR steel.
6.2.2.3. Joint Model
There are several constitutive models available in Cast3m for joint modeling. The most
relevant selected for this application are the following:
▪ JOINT_DILATANT: This model follows the proposal of Snyman et al. (1991),

using a Mohr-Coulomb criterion characterized by cohesion (c0, corresponding to
null normal joint stress) and friction angle (ϕj) (See Figure 6.18 – a)). Normal
displacements in the joint (δj – see Figure 6.4 – b)) are managed by the elastic
stiffness Kn, originating normal stresses σn which are limited in tension (σn0+ -
Figure 6.18 – b)). Shear forces are managed by an elasto-plastic model according
to the elastic stiffness Ks, originating from shear displacements along the joint
transverse direction Figure 6.18 – c)).
▪ JOINT_SOFT: This model also relates to a Mohr-Coulomb criterion. There are

several implementations available, according to the actual Cast3m version, since
some variants were developed from the original version of the model. However, all
of them essentially aimed to improve the accuracy of results by enabling the
consideration of hardening/softening rules for both normal and tangential joint
behaviors, and improving the associated hysteresis rules. Further details can be
reviewed in Costa (2009).
6.21
s n s
j
n0+ Ks
s = c0 + n . tan j c0 Kn
n0+ n j j
a) Yielding surface b) Normal behavior c) Shear behavior

Figure 6.18 – JOINT_DILATANT model in Cast3m (Costa (2009))
Within this context, the adoption of the more complex formulations can broaden the scope
of structural aspects to take into account, although generally at the cost of added computational
effort and, potentially, of undesirable increased numerical problems. Considering that the
present work FEM tests involve large and concentrated inelastic strains at the joint sections, it
was considered important to keep with simple approaches, aiming to reduce the difficulty of
achieving numerical convergence.
In that regard, the importance of the shear behavior in the joint elements can be argued. In
fact, in addition to usual friction forces (under the prestress axial load), an important dowel
action should be expected from the many rebars that cross the connection interfaces, leading to
a high shear stiffness and low displacements. Furthermore, since considerable sliding is not
expected, then softening and capacity depletion in shear should also not be relevant features for
these applications. Therefore, only the normal degree of freedom seems to be critical for
accurate simulations, as it is directly associated with the opening and closure of the joint during
cyclic lateral motions, thus regulating the contact forces that are transferred between beam and
columns.
Taking that into account, the choice between JOINT_SOFT or JOINT_DILATANT should
be related to the importance of softening in the normal tensile behavior of the models, since it is
considered on the former but not in the latter. In this regard, considering post-yielding loading
levels, where the joint motions were observed to lead to large opening widths that can be related
with full depletion of joint tensile capacity, its contribution to the overall energy dissipation
should be negligible. Therefore, considering the modeling tools presented above, while the
JOINT_SOFT model could be an option, it does not seem like it would provide considerable
benefits for this specific application. The alternative of using the simpler JOINT_DILATANT
model also addresses the main concerns regarding accurate simulation of the joint
opening/closure motions with less calibration effort and less subject to numerical convergence
6.22
problems. Nonetheless, it also requires that the cracking threshold is defined as a low or null
value, otherwise the elasto-plastic behavior in tension would lead to unrealistic energy
dissipation in the joints at large openings. With this approach, the early behavior of the
numerical models can exhibit slightly less stiffness than the experimental counterparts before
joint cracking occurs, but accurate joint performance is expected to be better addressed for the
later stages, namely for the onset of yielding and collapse. With that in mind, the option was
made to use JOINT_DILATANT as the constitutive model for precast joints. For that purpose,
the shear and normal stiffnesses Ks and Kn, respectively, are required for full characterization of
the elastic behavior of this model, while the friction angle and cracking threshold σn0 define the
yielding surface. When applicable, a dilatancy angle can also be defined to enable plastic flow.
The values for the Ks and Kn parameters were estimated from actual test data, due to the
unavailability of adequate sampling results to evaluate from. Furthermore, the monitoring
layout used in the experimental campaign was not developed accounting for this requirement,
leading to considerable difficulty in obtaining adequate values. Nonetheless, a methodology was
developed with that purpose which, for the normal stiffness Kn, involved the following
assumptions and procedure, also illustrated in Figure 6.19:
▪ Evaluation of the peak displacement instant for one of the first few cycles of each
test (until a maximum of 0.40% drift), gathering data for analysis from those load-
steps only;
▪ Assuming a linear strain distribution over the connection section, for determination
of the neutral axis and the compressive strain εn- from the recorded displacements,
δj- and δj+;
▪ Calculation of σn- by equilibrium, in order to balance the total vertical reaction Rv-
recorded at the load cell underneath the corresponding columns; at the local joint
section, all the compressive force is transferred through the zone under
compression, which is assumed to be adequately represented by the strain diagram
from Figure 6.19;
▪ Determination of an adequate value for Kn from the experimental σn- - εn-
relationship;
6.23
 𝜀𝑛− 
𝜎𝑛-
𝜀𝑛−
-
-
𝛿𝑗++ −
+  -𝛿𝑗
𝜀𝑛+
-
Rv
Figure 6.19 – Methodology for determination of Kn
The calibration of shear stiffness Ks was performed against experimental data, by adopting
a procedure based on comparison of the plots between the experimental σn - τs relationship, and
a calculated version of σn - τs using the previously obtained Kn and assumed values for Ks
instead:
▪ Evaluation of the peak displacement instant for one of the first few cycles of each
test (until a maximum of 0.40% drift), gathering data for analysis from those load-
steps only;
▪ Assuming equal distribution of the applied horizontal force between the two
precast joints (as shear) for calculating the experimental shear stress τs;
▪ Using the experimental values for σn and the previous shear stress τs, determination
of the experimental σn - τs relationship;
▪ Calculation of a numerical σn - τs relationship assuming σn governed by the
previously defined normal stiffness Kn, and τs by an initial Ks shear stiffness;
▪ Iterative adjustment of the initial Ks parameter until both σn - τs relationships are
characterized by similar evolution. In this procedure, the experimental σn - τs curves
and the normal stiffness Kn are kept constant, and only Ks is a variable input;
A friction angle ϕ of 60º and a residual cracking threshold (σn0+) of 25 KPa were also
considered for full definition of the joint model.
6.3. EXPERIMENTAL RESULTS VS. NUMERICAL SIMULATIONS
The results presented in this section were obtained from numerical simulations with the
models elaborated according to the methodology described in 6.1. The loading conditions
considered for the analyses respected a constant axial loading of 300 kN per column and aimed
to reproduce the same displacements recorded for each tested specimen. However, considering
6.24
the modeling strategy presented earlier and the characteristics of the constitutive models, strong
degradation is not expected between the repeating cycles for each individual drift level.
Therefore, the horizontal drift loading history applied in the numerical analyses only included
one full cycle per drift level.
6.3.1. MONOLITHIC SPECIMENS: SP_M02, SP_M03 AND SP_M04
Figure 6.20 presents a comparison of the applied force results obtained in both the
numerical and experimental tests, plotted against the actual displacements recorded on top of
the piers in terms of drift percentage, only the first full cycle for each drift.
225 225
Experimental Experimental
175 175
Numerical Numerical
125 125
75 75
Force (kN)
Force (kN)
25 25
-25 -25
-75 -75
-125 -125
-175 -175
-225 -225
-3,50 -2,50 -1,50 -0,50 0,50 1,50 2,50 3,50 -3,50 -2,50 -1,50 -0,50 0,50 1,50 2,50 3,50
Lateral Drift (%) Lateral Drift (%)
a) SP_M02 b) SP_M03
225
Experimental
175
Numerical
125
75
Force (kN)
25
-25
-75
-125
-175
-225
-3,50 -2,50 -1,50 -0,50 0,50 1,50 2,50 3,50
Lateral Drift (%)
c) SP_M04
Figure 6.20 – Numerical vs. Experimental Force – Drift results for monolithic specimens
As it is possible to observe, the numerical and experimental results are in good agreement,
presenting similar peak forces (between 1.50% and 2.00% drifts for SP_M04 and around 2.50%
drift for both SP_M02 and SP_M03) and general envelope; the numerical models were also able
to characterize the force asymmetry between the two loading directions. In fact, the asymmetry
for the negative loading direction was generally even slightly larger in the numerical simulation
than in the experimental test, particularly on the negative unloading stiffness on SP_M02 and
SP_M03. The lack of numerical simulation of plasticity in tension comes to mind as a possible
6.25
explanation of these results. Nonetheless, this result may also be explained by a larger demand
applied to the shear interlocking mechanism on that direction, which would help increase the
corresponding beam stiffness, although that feature is not accounted for by the adopted
modeling strategy. Recalling what was previously discussed about the coupling action at the
beam (section 5.2.1), on the negative loading direction it is mobilized with a tensile pulling
force, leading to overall reduced beam stiffness due to less efficient crack closing. However, the
interlocking mechanism mostly depends on the differential shear displacements occurring at
existing cracks, rather than the overall lateral movement, and therefore it should be relatively
the same for both loading directions (except when the crack widths are too large to enable
efficient shear transfers). Taking that into account, it is possible that for pushing coupling action
(positive direction) the numerical model is able to mobilize the beam shear force transfer mainly
through concrete compression struts, without significant contribution of interlocking, but for
pulling coupling action (negative direction) the lack of interlocking leads to smaller forces.
With this in mind, results will be further presented addressing the positive loading direction
only, as for instance, the maps of compressive damage variable 𝑑− , shown in Figure 6.21 for the
monolithic specimens.
a) SP_M02 at peak force b) SP_M02 at failure
c) SP_M03 at peak force d) SP_M03 at failure
(scale)
e) SP_M04 at peak force f) SP_M04 at failure

Figure 6.21 – Maps of compressive damage variable 𝒅− maps on monolithic specimens
6.26
At the 3.00% drift loading level, the numerical models experienced a strength loss on all
the specimens, leading to the occurrence of severe numerical convergence difficulties that
caused the simulation to stop on all but SP_M03. An in depth analysis of the problem showed
that it was mostly caused by failure of the beam in compression, because the respective damage
variable 𝒅− developed values around 1.0 (which is representative of fully damaged section with
no further capacity) near the edges close to the columns. Figure 6.21 evidences the previous
description, where it is possible to observe the incursion of damage in compression at the
loading levels corresponding to peak and failure, showing that increasing damage progresses
upwards along the beam-column interface.
The failure of the concrete in compression at the beam-column interface goes in line with
experimental observations, where some level of concrete crushing and/or spalling was detected
in those locations for the largest drifts (a clear example is Figure 5.25, for SP_M02). In this
regard, the importance of the shear force transfer between columns is also noted, considering
the strength drops recorded in accordance with disruption of the compression struts mobilized
through the beam. In fact, based on the plots of principal compressions σ22 shown in Figure
6.22 for the monolithic specimens, the main shear transfer mechanism until reaching the peak
force drift level (Figure 6.22 – a), c) and e)) is related with the formation of a single
compression strut; it is only after the concrete failure that multiple struts are formed (Figure
6.22 – b), d) and f)).
Figure 6.22 – Principal compressions σ22 on monolithic specimens
6.27

Figure 6.22 (cont.) – Principal compressions σ22 on monolithic specimens
Regarding the level of shear that is transferred through the beams, the numerical models
were able to simulate with reasonable accuracy the values recorded during the experimental
tests. Figure 6.23 illustrates a comparison of maximum beam shear values from the numerical
analyses (red bars) with the respective experimental results (blue bars), where the presented
values were evaluated from the amplitude variation of the vertical reactions at the base of each
column. Furthermore, numerical and experimental result values are also indicated within their
respective columns, showing minor differences as highlighted by the colored labels above the
numerical one.
550
Experimental
500
+2% Numerical
450 - 4%
+6%
400 440,11 447,32 440,01
422,81
397,01
Beam Shear (kN)
350 373,67
300
250
200
150
100
50
0
SP_M02 SP_M03 SP_M04
Figure 6.23 – Beam shear results for monolithic specimens: numerical vs. experimental
The damage pattern observed in the pier system throughout the tests indicated early cracks
in the beam at the first few drift levels, progressing to considerable beam damaged close to the
collapse level. However, the cracking pattern was generally observed to be mostly stabilized
after 1.00% drift, with just mild cracking on the columns recorded around 0.75% drift. Figure
6.24 presents the results for the tensile damage variable 𝑑+ for 0.20% drift, 0.50% drift and
6.28
1.00% drift, where a similar progression can also be observed. The regular distribution of
tensile damage on the columns is also noted.
a) SP_M02 - 0.20% drift b) SP_M02 - 0.50% drift c) SP_M02 - 1.00% drift
d) SP_M03 - 0.20% drift e) SP_M03 - 0.50% drift f) SP_M03 - 1.00% drift
(scale)
g) SP_M04 - 0.20% drift h) SP_M04 - 0.50% drift i) SP_M04 - 1.00% drift

Figure 6.24 – Maps of tensile damage variable 𝒅+ maps on monolithic specimens
The previous results show that the demand is large at the beams, where damage is mostly
found near the nodes, similarly to experimental evidence. In addition, the virtual loss of
concrete tensile capacity indicative of generalized cracking that is observed in the previous
figure for the 1.00% drift level, suggests that crack width increases are likely to occur along
with progression of main reinforcement yielding. This is confirmed by Figure 6.25, which
provides illustrations of the principal positive (tensile) strains ε11over the deformed shapes (with
5.0 amplification factor) for loading levels corresponding to the first yielding of the main
reinforcement and to failure, thus clearly reflecting what was previously described.
a) SP_M02 at first yielding b) SP_M02 at failure
Figure 6.25 - Principal strain ε11 maps over the deformed shape on monolithic specimens
6.29
c) SP_M03 at first yielding d) SP_M03 at failure
(scale)
e) SP_M04 at first yielding f) SP_M04 at failure

Figure 6.25 (cont.) – Principal strain ε11 maps over the deformed shape on monolithic specimens
Concerning the main reinforcement yielding, it was observed to always occur firstly at the
beam reinforcement, according to the axial stress maps shown in Figure 6.26, determined for
the first yielding of the main column or beam reinforcement. Despite that, column
reinforcement also develops high stresses and, at peak force, several of its rebars have achieved
yielding. For ease of analysis, results are presented in normalized terms, with main rebar
stresses σsl divided by the corresponding individual yielding stress σsly.
a) SP_M02 at first yielding b) SP_M02 at peak force
(scale)
c) SP_M03 at first yielding d) SP_M03 at peak force
Figure 6.26 - Maps of rebar stress ratios (𝜎𝑠𝑙 ⁄𝜎𝑠𝑙𝑦 ) on monolithic specimens
6.30
e) SP_M04 at first yielding SP_M04 at peak force

Figure 6.26 (cont.) – Maps of longitudinal reinforcement stress ratios (𝝈𝒔𝒍 ⁄𝝈𝒔𝒍𝒚 ) on monolithic
specimens
The drift values corresponding to the first occurrence of yielding represented in the
previous figures can also be compared with the values determined through the bilinear
equivalent system procedure adopted in Chapter 5 (Figure 5.67). Figure 6.27 illustrates that
comparison, where relatively small differences under 15% are observed between numerical and
experimental results of SP_M02 and SP_M04.
2
Experimental
1,8
Numerical
1,6
+9%
1,4
Yielding Drift (%)
1,41 - 12%
1,2 1,29 1,34
1,18 +14%
1
0,8 0,94
0,82
0,6
0,4
0,2
0
SP_M02 SP_M03 SP_M04
Figure 6.27 – Yielding drift for monolithic specimens: numerical vs. experimental
Lastly, for the drift range simulated in the analyses, the numerical models did not show
significant demand on the transverse reinforcement on either the columns or, more importantly,
the beam, particularly for the peak force level. Considering the level of shear that is transferred
through that element, as previously discussed, those results are interesting to note. Figure 6.28
illustrates the axial stress σst map of beam transverse reinforcement bars for each specimen,
normalized by their respective yielding stress σsty, for loading drifts corresponding to the peak
force level and failure.
This means that most of the vertical component of the shear load is handled by the
concrete struts and the diagonal parts of the reinforcement (when applicable), instead of the
6.31
vertical stirrups, actually as intended by design of all specimens except SP_M03. Moreover,
it is believed to be also a consequence of the modeling strategy not being able to adequately
explore the post-peak stage, since the compressive failure and the occurrence of numerical
convergence problems prevented longer simulations. This is particularly relevant because,
as seen in Chapter 5, experimental evidence showed that significant tensile degradation or
failure (e.g: Figure 5.34, Figure 5.39) were detected after compression driven occurrences.
In light of these results, it is believed that the demand for transverse reinforcement
capacity is mostly relevant at the larger drifts, as the result of the concrete degradation
associated with the shear load transfer on the beam. In fact, the stirrups tend to prevent loss
of integrity of the damaged core concrete, thus it is only upon failure of the compressive
strut mechanism that originally enables shear loading through the beam, that their role as
the main shear resisting element becomes apparent. This is the reason why these results do
not show that effect, because they refer to the behavior mostly before failure of the
compressive strut.
(scale)

Figure 6.28 – Maps of transverse reinforcement stress ratios (𝝈𝒔𝒕 ⁄𝝈𝒔𝒕𝒚 ) on monolithic
specimens
6.3.2. PRECAST SPECIMENS: SP_PC02A AND SP_PC02B
Results obtained for the precast specimens subjected to numerical simulation are presented
in Figure 6.29, where a comparison of the applied force plotted against the percentage of drift
6.32
(associated with the displacements at the top of the piers) is shown for both the simulations and
experimental tests. Regarding the latter, only results for the first full loading cycle at each drift
level are included, similarly to the strategy adopted for presenting the monolithic simulations.
In the precast specimen analyses, calculations were only performed until the 3.50% drift
loading level, since evaluation of those results for larger drifts was deemed unnecessary, and
because it is believed the obtained responses already characterize performance for an adequate
demand range capable of detailing the most important phenomena. In addition, it is also noted
that due to the small loading increment used for all the analysis (for the purpose of numerical
convergence), the output file size for the later cycles becomes quite large, leading to impractical
post-processing.
In general, the responses obtained with the numerical analyses provide satisfactory results,
showing similar peak forces (at near the 2.50% and 3.00% drift loading levels for SP_PC02A
and SP_PC02B, respectively) and unloading behavior. Despite that, some differences can be
observed between the numerical and experimental output on both specimens. In SP_PC02A and
roughly between drifts 0.50% to 1.50% (corresponding to the yielding range) the lateral force is
greater than the experimentally obtained, although for larger drifts the difference reduces to a
negligible margin. In addition, both specimens show higher stiffness in the reloading stages
(wider force-drift loops, particularly in the positive loading direction) that lead to increased
energy dissipation relative to the experimental tests. Although this was not evaluated in depth, it
is believed that it can be related with differences between the numerical and experimental
bond-slip behavior, enabling stronger bond on the former (which could also help explain the
earlier yielding range observed on SP_PC02A), or with the joint stiffness parameters.
Regardless, this was not further adjusted because, by contrast with the reloading stiffness, the
unloading stiffness is believed to be closely captured on both cases and, despite the energy
dissipation may be slightly higher than intended, the overall performance is deemed to be
adequately similar to experimental results. The loading direction asymmetry was also well
captured, particularly for SP_PC0B, where peak values for both the positive and negative
directions are quite close to the experimental records.
6.33
225 225
175 175
Numerical Numerical
125 125
75 75
Force (kN)
Force (kN)
25 25
-25 -25
-75 -75
-125 -125
-175 -175
-225 -225
-4,50 -3,50 -2,50 -1,50 -0,50 0,50 1,50 2,50 3,50 4,50 -4,50 -3,50 -2,50 -1,50 -0,50 0,50 1,50 2,50 3,50 4,50
d) SP_PC02A e) SP_PC02B
Figure 6.29 – Numerical vs. Experimental Force – Drift results for precast specimens
In general, the performance in both cases was dependent on both the joint and beam
behavior, contrasting with the monolithic tests where clear prevalence of beam failure was
observed. This can be seen in Figure 6.30, which illustrates the values recorded for the
compressive damage variable 𝑑− at the loading displacement associated with peak force for the
positive direction, and the 3.50% drift cycle. As it is possible to observe, accumulation of
concrete damage in compression near the beam edges still occurs. However, crushing of the
beam-column node concrete along the precast joints is also evident around 3.50% drift in both
of the precast specimens, which indicates the importance of the balance between the
compressive forces transferred from the beam to the column. Nonetheless, full degradation of
the associated concrete capacity in compression along the precast joint was not evidenced by the
experimental tests as much, in the outer edges of the beam-column nodes, although the inner
edge experienced significant crushing (e.g. Figure 5.49 and Figure 5.55). In light of this, it is
possible that the concrete compressive ductility calibrated for these numerical applications was
too low for those particular zones. Additionally, incorrectly calibrated joint model stiffness
could also lead to increased local stresses (which may be another reason supporting the
evidence above described).
6.34
a) SP_PC02A at peak force b) SP_PC02A at 3.50% drift
(scale)
c) SP_PC02B at peak force d) SP_PC02B at 3.50% drift

Figure 6.30 – Maps of compressive damage variable 𝒅− maps on precast specimens
Despite the concrete failure in compression (at the largest drifts) that is detected in the
numerical analyses, the general distributions of compressive forces is very similar to those of
the monolithic specimens, as illustrated by the σ22 principal compressions’ maps shown in
Figure 6.31. In fact, at the peak force level, a single compression strut in the beam is responsible
for shear transfer between columns in both SP_PC02A and SP_PC02B. Furthermore, increasing
damage causes modifications on the principal stresses, leading to the formation of multiple
struts from the original one, when approaching the ultimate drift stages.
Figure 6.31- Principal compressions σ22 on precast specimens
6.35

Figure 6.31 (cont.) – Principal compressions σ22 on precast specimens
The difference between the maximum beam shear evaluated from experimental tests and
the corresponding value determined from numerical analyses was small in both cases. Figure
6.32 illustrates the comparison of those values, where experimental and numerical data are
represented by the blue and red columns, respectively. The accuracy of the numerical
simulations is confirmed by the relative minor differences observed and expressed in percentage
over the numerical columns.
500,00
Experimental
450,00
Numerical
400,00
+9%
+1%
350,00
Beam Shear (kN)
356,81
300,00 328,63 330,14 327,43
250,00
200,00
150,00
100,00
50,00
0,00
SP_PC02A SP_PC02B
Figure 6.32 – Beam shear results for precast specimens: numerical vs. experimental
Concerning the tensile damage on the precast specimens, it was found concentrated on the
coupling beam, first observed around the 0.20% drift level near the beam-column node vertical
interfaces, progressing to complete tensile capacity degradation over the total length of the
element and also into the node. Column cracking occurred after the 0.75% drift level, but not to
the same extent as in the beam.
6.36
a) SP_PC02A b) SP_PC02A c) SP_PC02A

0.20% drift 0.50% drift 1.00% drift
(scale
)
d) SP_ PC02B e) SP_ PC02B f) SP_ PC02B

0.20% drift 0.50% drift 1.00% drift
Figure 6.33 – Maps of tensile damage variable 𝒅+ maps on precast specimens
The previous results suggest that the performance of both precast specimens shows similar
demand on the beam but reduced demand on the columns, when compared with the previous
monolithic cases (Figure 6.24). In that regard, concentrated rotations do occur at the interface
connection section, enabling reduced strains in the beam comparatively with the monolithic
specimens, especially on SP_PC02A, as evidenced by the principal tensile strains ε11 maps
plotted over the deformed shapes (with amplification factor of 5.0) represented in Figure 6.34.
Nonetheless, the deformations that take place in the beam are still large enough so that
considerable damage cannot be prevented, as reported on the experimental tests and adequately
simulated by the numerical models. Furthermore, as it can be observed in Figure 6.34, at 3.50%
drift, concrete crushing in the edges of the precast joints is apparent, as well as the large strains
in the beams.
a) SP_PC02A at first yielding b) SP_PC02A at 3.50% drift
(scale)
c) SP_PC02B at first yielding d) SP_PC02B at 3.50% drift
Figure 6.34 – Principal strain ε11 maps over the deformed shape on precast specimens
6.37
It is interesting to note that SP_PC02B seems to develop larger strains in the beam at an
earlier stage relative to SP_PC02A, and that, at larger drifts, it also shows increased column
damage. As presented in Chapter 5, the moment capacity of SP_PC02B at the joint section is
slightly larger than SP_PC02A’s, leading to increased connection stiffness and larger
deformations on the beams and columns, at the expense of more reduced joint motions. In this
regard, Figure 6.35 illustrates a comparison of the maximum precast joint opening widths
recorded in the experimental tests and those determined from the numerical analysis, for the
0.50% drift, 1.00% drift and 1.50% drift levels on both specimens. Values for SP_PC02A and
SP_PC02B are not very different, especially regarding the experimental data. Nonetheless,
numerical results seem in line with the previous, even though lower values are generally
reported by the numerical models (particularly on SP_PC02B), which could also help explain
the larger reloading stiffness on the numerical force-drift curves, relative to the experimental
values.
3,00
SP_PC02A_exp
SP_PC02B_exp
2,50
SP_PC02A_num
Joint Opening Width (mm)
SP_PC02B_num
2,00
1,50
1,00
0,50
0,00
0,50% 1,00% 1,50%
Lateral Drift (%)
Figure 6.35 – Precast joint maximum opening width results: numerical vs. experimental
A major difference between numerical results for monolithic and precast specimens is that
while first yielding was always detected in the beam reinforcement in the former, joint
reinforcement yielding occurs first in the precast specimens. Figure 6.36 illustrates the axial
stress σsl maps for main reinforcement bars, normalized by the corresponding yielding stresses
σsly, where this can be observed in further detail. Moreover, despite the occurrence of rebar
yielding first at the joint section, as intended by design, the beam reinforcement still shows
relevant stresses, particularly at the peak force stage for each specimen, thus again
confirming that the behavior pattern of the precast models is associated with complex
interactions between the beams and columns.
6.38
a) SP_PC02A at first yielding b) SP_PC02A at peak force
(scale)
c) SP_PC02B at first yielding d) SP_PC02B at peak force

Figure 6.36 – Maps of longitudinal reinforcement stress ratios (𝝈𝒔𝒍 ⁄𝝈𝒔𝒍𝒚 ) on precast specimens
Still concerning reinforcement yielding, the comparison of values determined for the
yielding drift from experimental and numerical results shows larger differences than those
observed in the monolithic cases. Specifically, the yielding drift determined from the bilinear
equivalent system for experimental results on SP_PC02A is around 26% lower than the lateral
drift at which first yielding was detected in the numerical analysis. By contrast, an even larger
difference of around 38% in favor of the numerical value was identified regarding SP_PC02B.
These results can be observed in Figure 6.37, and are relatively unsurprising, considering all
that was already discussed regarding the calibration of joint stiffness and bond behavior.
2
Experimental
1,8
Numerical
1,6
1,4
+ 38%
Yielding Drift (%)
1,2
1,22
1 1,07 -26%
0,8 0,88
0,79
0,6
0,4
0,2
0
SP_PC02A SP_PC02B
Figure 6.37 – Yielding drift for precast specimens: numerical vs. experimental
6.39
Figure 6.38 illustrates the axial stress σst map of beam transverse reinforcement bars for
each specimen, normalized by their respective yielding stress σsty, for lateral drifts
correspondent to the peak force level and 3.50% drift. As observed for the monolithic tests,
the numerical analyses for the precast specimens also do not show significant demand on
the transverse reinforcement for the drift range applied in the simulations, which is likely
related to the same reasons already mentioned before.
(scale)
Figure 6.38 – Maps of transverse reinforcement stress ratios (𝝈𝒔𝒕 ⁄𝝈𝒔𝒕𝒚 ) on precast specimens
6.3.3. SINGLE COLUMN SPECIMENS
Numerical simulation of single columns was only performed for the SP_F01 specimen.
Nonetheless, the corresponding analysis provided results that compare quite well against the
experimental data. Figure 6.39 illustrates the results for lateral force plotted against the
displacements recorded at the top of the column in terms of drift percentage. As it is possible to
observe, the numerical and experimental curves are in good agreement, displaying similar peak
force and general envelope. The unloading and reloading behavior is also well captured,
showing a similar slight “pinching” effect between both responses. Despite that, however, the
numerical results show increased reloading stiffness relative to the experimental findings,
particularly for the positive loading direction. This can be explained by the fact that no bond-
slip relationship was used to model the interaction between the rebars and the concrete, resulting
in increased energy dissipation in the numerical case. In addition, the experimental results
exhibit slight strength degradation between subsequent cycles due to the damage accumulated in
the column, but although the global force envelope is well simulated by the numerical model,
this effect is not as apparent in its results.
6.40
200
Experimental
150
Numerical
100
Force (kN)
50
-50
-100
-150
-200
-3,50 -2,50 -1,50 -0,50 0,50 1,50 2,50 3,50
Lateral Drift (%)
Figure 6.39 – Numerical vs. Experimental Force – Drift results for SP_F01
The general behavior pattern exhibited by the specimen in the numerical analysis was
typical of a standard reinforced column test subjected to bending, with the response limited by
yielding of the longitudinal reinforcement at its base, and corresponding increase of
compressive damage around the edges of the cover concrete. Furthermore, the displacement at
which yielding is first identified in the numerical analysis corresponds to 0.56% drift, compared
to the value of 0.73% reported from the experimental tests, resulting in a difference of around
22%, which could likely also be better simulated if bond-slip had been considered.
Results for normalized rebar stresses and the compressive damage in concrete are presented
on a front-view of the symmetry plane in Figure 6.40 for the last displacement cycle,
corresponding to the 3.50% drift level. Although both illustrations that are included represent
different measurements, the same scale applies to both of them, where the blue color indicates
values close to or less than zero, and the red color shows values equal or greater than 1.0.
6.41
a) Map of longitudinal rebar stress b) Map of compressive damage (scale)
ratios (𝜎𝑠𝑙 ⁄𝜎𝑠𝑙𝑦 ) variable 𝑑−

Figure 6.40 – Numerical results representing typical column bending behavior at 3.50% drift
Moreover, most of the deformations occur on the base of the column, as expected,
specifically in the first 0.275m layer of smaller elements, according to Figure 6.41 - a)
(deformed shape with an amplification factor of 5.0), comparing reasonably well to the value of
0.20m reported in Chapter 5 for the main damage location. Despite that, cracking extends
further along the column, as it can be observed from Figure 6.41 - b).
(scale) (scale)
a) Principal strain ε11 map over the

b) Map of tensile damage variable 𝑑 +
deformed shape
Figure 6.41 – Numerical results for the tensile behavior of concrete at 3.50% drift
6.42
6.4. NUMERICAL SIMULATIONS OF UNTESTED PIER SYSTEMS
The main objective for this section is to provide application context to the results obtained
during the experimental campaign, and to evaluate the possibility of using them for the
calibration of global bridge modeling tools for seismic performance assessment, which will be
addressed in Chapter 7. For that purpose, as previously discussed in the introductory section of
the present chapter, two main scenarios were established:
▪ Comparison of results obtained from the experimental setup based on half-pier

structures, with those determined by using full height columns;
▪ Assessment of possible behavior pattern differences evidenced in shorter piers,
taking into account that the tested structures are already close to the maximum
value of the height range proposed for study in Chapter 4 (5.00m to 20.00m);
Within this context, the previous modeling assumptions and strategy were kept mostly
unchanged; only very specific modifications were introduced to create the required differences
for simulation of these untested pier scenarios.
6.4.1. SIMULATIONS WITH FULL HEIGHT PIERS
The test setup used throughout the experimental campaign was developed taking advantage
of the expected null moment point in the distribution of moments of the pier structure for lateral
loading. Taking that into consideration, numerical simulation of full height piers essentially
involves extending the column length to account for the full structure, and adjustment of
supporting constraints to characterize a fully fixed footing. With this in mind, the original finite
element mesh used for the pier tests (Figure 6.2) was modified according to Figure 6.42. The
full height of the modified piers then becomes 4.693m, which is determined from doubling the
height of the previous mesh up to the horizontal axis of the beam (including the 0.165m length
of the linear elastic elements connecting the pier elements to the rotation points, for coherence
relative to curvature inflexions).
6.43
Figure 6.42 – Finite element mesh layout for full height pier numerical simulations.
Within this framework, two simulations were performed, aiming to numerically describe
the behavior for the full structure in representation of one monolithic case and one precast case
as follows:
▪ SP_M02_full – Full height pier simulation of the SP_M02 test, using the same
model calibration parameters;
▪ SP_PC02A_full – Full height pier simulation of the SP_PC02A test, using the
same model calibration parameters;
Figure 6.43 presents a comparison of the applied force results obtained in both tests, plotted
against lateral displacements in terms of drift percentage. The figures also include the original
results obtained with the half pier simulation, as well as the corresponding experimental
records. As it is possible to observe, results for the full pier simulations are in good agreement
with both the experimental records and the previous half pier analyses. For positive loading,
similar peak forces and unloading stiffness are recorded on both the numerical analyses. The
increased reloading stiffness issue that was previously discussed for precast specimens is also
apparent. In the case of SP_M02_full, numerical difficulties were observed near 3.00% drift due
to concrete failure, similarly to what was reported on the original half pier analysis.
6.44
225 225
175 Numerical_half 175 Numerical_half
Numerical_full Numerical_full
125 125
75 75
Force (kN)
Force (kN)
25 25
-25 -25
-75 -75
-125 -125
-175 -175
-225 -225
-4.50 -3.50 -2.50 -1.50 -0.50 0.50 1.50 2.50 3.50 4.50 -4.50 -3.50 -2.50 -1.50 -0.50 0.50 1.50 2.50 3.50 4.50
a) SP_M02_full b) SP_PC02A_full
Figure 6.43 – Force – drift results for full pier tests
Regardless, the responses for the full pier analyses seem to display less loading direction
asymmetry than the corresponding half pier simulations, and that is particularly evident on the
precast specimen SP_PC02A. As far as it was possible to assess, this might be related with the
fact that the test setup used for the half pier analyses essentially forces the inflexion point of
column curvature to be at the same column section (base rotation hinges), while the curvatures
develop freely in the full structure. Therefore, the inflexion point in the latter can change
between loading directions, leading to unsymmetrical bending moments between the two
columns, which can cause the longitudinal reinforcement that exists therein to perform
according to different demands.
A characterization of what was described is presented on Figure 6.44, with a rough

illustration (according to the green color bands) of curvature inflexion zones for positive and
negative loading at 2.50% drift on SP_PC02A_full, established by visually determining the
column locations where the compression forces are equally oriented (essentially vertical) along
the cross-section (indicative of approximately null moment). Despite resulting from
approximate strategy, Figure 6.44 does seem to suggest that the inflexion zones change between
loading directions, thus agreeing with the previous argument.
This is particularly relevant because, while on the tested half pier structures the demand is
mostly directed to the beam and beam-column nodes (where structural damage was found hard
to prevent or mitigate), on full piers the column bases are also expected to develop strong
moments. Therefore, full structures have the ability to adapt to the unsymmetrical nature of
damage and loading, without the limitations that the present half pier testing conditions impose
in that regard.
6.45
a) at +2.50% drift b) at -2.50% drift

Figure 6.44 – σ22 compressions on SP_PC02A_full
Both structures showed shear transfer mechanisms similar to their half-pier counterparts,
where the force is transferred mostly through a single compressive strut (which can also be
observed in Figure 6.44), later degrading into several struts on larger drifts. For comparison
purposes, the data from Figure 6.23 and Figure 6.32 was reworked to include results from
SP_M02_full and SP_PC02A_full into Figure 6.45, where quite close values can be readily
identified for the different scenarios of both cases.
500
Experimental
450
440.11 447.32 444.73 Numerical_half
400
Numerical_full
350
Beam Shear (kN)
300 328.63 330.14 331.68
250
200
150
100
50
0
SP_M02 SP_PC02A
Figure 6.45 – Maximum beam shear value comparisons for full structure analyses
6.46
Accumulation of damage in compression is also observed on beam edges in both cases, as a

result of the high compressions caused by the transferred shear forces, as well as generalized
depletion of the tensile capacity of the concrete on the beam element. Moreover, according to
the maps represented in Figure 6.46, determined for failure in the case of SP_M02_full or
3.50% drift in the case of SP_PC02_full, it is possible to observe the corresponding damage
incurred in the bottom sections of the columns, indicating generalized cracking of concrete and
some crushing.
(scale)
a) SP_M02_full b) SP_M02_full c) SP_PC02A_full d) SP_PC02A_full

𝑑− map 𝑑+ map 𝑑− map 𝑑+ map
Figure 6.46 – Maps of damage variables for full structure analyses (at failure for
SP_M02_full and at 3.50% drift for SP_PC02A_full)
The two models showed some differences regarding reinforcement yielding, as illustrated
in Figure 6.47, where normalized values for rebar stresses are provided for the displacements
corresponding to first reinforcement yielding and peak lateral force. While on SP_PC02A_full it
was first detected at the precast joint reinforcement just like in the half pier simulation, on
SP_M02_full that occurred first on column bases. Nonetheless, progression of each pier’s
behavior into the peak force level eventually led to the same patterns observed before, including
generalized yielding in the beam and beam-column node on the monolithic model
SP_M02_full, and localized yielding at the precast joint of the SP_PC02_full simulation, with
complementary beam contribution.
6.47
(scale)
a) SP_M02_full b) SP_M02_full c) SP_PC02A_full d) SP_PC02A_full

first yielding peak force first yielding peak force
Figure 6.47 – Map of longitudinal rebar stress ratios (𝜎𝑠𝑙 ⁄𝜎𝑠𝑙𝑦 ) from full height pier
analyses
6.4.2. SHORTER PIER SCENARIOS
The full scale height of the prototypes corresponding to the tested piers is around 18.80m
(four times 4.693m, according to the originally reduced scale of 1:4 and the pier height of
Figure 6.42). In that regard, considering the height range established for study in Chapter 4,
which focused on structures between 5.00 to 20.00m, and also taking into account already
addressed findings concerning the complex height-dependent relationship between column and
beam stiffness ratios, it is in the interest of this work to evaluate whether or not the observations
made for the tested structures are representative of the complete pier range. Within this context,
the tested pier represents nearly its maximum value, so shorter piers should be studied instead.
However, reducing pier height naturally shifts the stiffness ratio between beams and columns in
favor of the latter, which become increasingly more rigid in shorter structures, causing increased
beam strains relative to columns’. Taking that into account, the usefulness of studying shorter
monolithic piers can be argued, as they already revealed to be quite limited by the performance
of the beam, both on the experimental as well as the numerical case studies. On the other hand,
the precast specimens revealed larger concentrated column deformations at the precast joint
section, caused by the localized yielding. In that framework, it may be useful to evaluate the
impact of reducing column height only on precast specimens on the global pier performance.
6.48
For that purpose, the full height mesh for SP_PC02A_full was modified according to the
two following scenarios:
▪ SP_PC02A_50: Reduction of the column length to account for a shorter pier with
50% of the original height (4.693m to 2.347m);
▪ SP_PC02A_75: Reduction of the column length to account for a shorter pier with
75% of the original height (4.693m to 3.520m);
Furthermore, the same modeling assumptions and calibration parameters were used for
these simulations as for SP_PC02A_full, and the loading history applied therein was scaled to
target the same drift values in each simulation. For comparison purposes, the lateral peak forces
from SP_PC02A_full were also scaled to the reduced heights, according to the representation of
Figure 6.48 and equation 6.7. Essentially, the scaled results of SP_PC02A_full represent a
scenario where the interaction between beam, columns, and precast joint of that particular
analysis is kept unchanged for different height piers. It also assumes no relevant influence of
stiffness ratio shifts and, therefore, characterizes a global behavior that is influenced by yielding
of the column reinforcement at the column base and precast joint, as well as by the beam
degradation due to shear. Furthermore, this strategy also assumes that the peak moment from
the lateral load (determined from the peak lateral force) is the same despite the column height
reductions, which can only occur if the available capacity is explored to a considerable extent
on both the beam and the columns, since yielding was achieved on both elements in the original
SP_PC02A_full simulation.
𝐻𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙
𝐹𝑠𝑐𝑎𝑙𝑒𝑑 = 𝐹𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙 × 6.7
𝐻𝑠𝑐𝑎𝑙𝑒𝑑
Foriginal
Fscaled
Horiginal
Hscaled
SP_PC02A_full scaled SP_PC02A_full
Figure 6.48 – Illustration of the procedure for obtaining scaled results from SP_PC02A
Lateral force results obtained with the described approach are presented in Figure 6.49,
plotted against displacement in terms of drifts, for both the simulations, including the
6.49
corresponding peak force scaling. It is possible to observe that both the positive and negative
loading direction responses compare well against the scaled peak forces, since a maximum
absolute difference of 5.50% was recorded amongst all results. According to what was assumed
earlier, that seems to indicate that despite a reduction in column length and overall pier height,
demand at local level is still quite close to the available capacity in all the relevant locations
(column bases, precast joint and beam).
450 450
SP_PC02A_full_scaled SP_PC02A_full_scaled
350 350
SP_PC02A_50 SP_PC02A_75
250 250
150 150
Force (kN)
Force (kN)
50 50
-50 -50
-150 -150
-250 -250
-350 -350
-450 -450
-4.50 -3.50 -2.50 -1.50 -0.50 0.50 1.50 2.50 3.50 4.50 -4.50 -3.50 -2.50 -1.50 -0.50 0.50 1.50 2.50 3.50 4.50
a) SP_PC02A_50 b) SP_PC02A_75
Figure 6.49 – Force – drift results for reduced height pier scenarios
Rebar stress results for first yielding and peak force levels agree with the previous
conclusion, because generalized yielding is identified on column bases, at the precast joint and
at the beam. This can be observed in Figure 6.50, where rebar axial stresses σsl are normalized
by the respective yielding stresses σsly. Reinforcement yielding is first detected at the column
bases in both cases, being achieved in the precast joint and beam at a later stage.
(scale)
a) SP_PC02A_50 b) SP_PC02A_50 c) SP_PC02A_75 d) SP_PC02A_75

first yielding peak force first yielding peak force
Figure 6.50 – Map of longitudinal rebar stress ratios (𝜎𝑠𝑙 ⁄𝜎𝑠𝑙𝑦 ) from reduced height
pier analyses
6.50
Accordingly, results for the two concrete damage variables indicate similar damage
progression to that calculated for SP_PC02A_full (Figure 6.46 – c) and d)) on both cases.
Particularly, full degradation of the tensile capacity in the beam and at column bases, as well as
accumulation of compressive damage at the beam edges and column bases can be identified in
Figure 6.51, which is indicative of the large local demands and in agreement with what was
observed previously.
(scale)
a) SP_PC02A_50 b) SP_PC02A_50 c) SP_PC02A_75 d) SP_PC02A_75

𝑑− map 𝑑+ map 𝑑− map 𝑑+ map
Figure 6.51 – Maps of damage variables for reduced height pier analyses
Considering these results, it seems reasonable to assume that: the experimental

observations reported on Chapter 5 can represent a wider range of different pier applications,
namely for different pier heights; and direct scaling of total lateral moment according to pier
height provides a fairly good estimation of the overall capacity.
6.5. FINAL REMARKS
The work developed on this chapter aimed to address two specific issues with the present
work: a strong dependency between the conclusions extracted from the experimental work and
the specific testing conditions of the adopted setup; the fixed height of the test specimens, which
raised doubts over the accuracy of the experimental results for representing the range of
structures set for study (piers between 5.00m to 20.00m tall). Therefore, the strategy adopted for
dealing with these issues was supported by the numerical simulation of structural scenarios
different than those subjected to experimental testing, using a finite element based methodology
with individual constitutive characterization of concrete, steel reinforcement, bond-slip behavior
and joint behavior, when applicable.
6.51
The first set of numerical applications performed under this strategy targeted the simulation
of some of the experimental tests presented in Chapter 5, focusing on evaluating the quality and
precision of the obtained results in comparison with experimental evidence. In that regard, the
numerical simulations provided results which were generally in good agreement with the
experimentally observed values, particularly concerning the evaluation of peak forces, overall
force-displacement envelopes and damage spread. Furthermore, they enabled confirmation that
beam shear was a critical factor for the behavior and failure of the pier structures (whether
through concrete crushing or sliding shear), and that the precast mechanism devised for Phase 2
specimens helped improve the overall ductility of the specimens, by mobilizing increased
deformations on the rebars crossing the respective precast joints.
Within that context, the adopted modeling strategy and associated parameters were
assumed to provide fairly good results, enabling further testing of other structural scenarios for
which experimental calibration was not available. The first of those was focused on evaluating
significant result differences when considering equivalent cyclic loading of a full height pier,
with clamped support column base connections, instead of half-pier structures with free rotation
base connections. These analyses showed that full structures are able to adapt better to the
occurrence of damage or unsymmetrical loading, by developing different force patterns within
the columns under those conditions. Nonetheless, no significant differences were observed
between half and full piers regarding the main response defining parameters, such as peak force,
which further emphasizes the precision of the experimental results on a global sense.
Finally, the last simulations presented in this chapter focused on evaluating the validity of
the general conclusions, extracted from the work presented in both Chapter 5 and 6, for
addressing pier structures of different heights. In that regard, only shorter piers were analyzed,
since the tested structures already represent close to the upper bound of the height range set for
the present study (5.00m to 20.00m), and the obtained results were generally in good agreement
with most of the previous observations. Therefore, it was considered reasonable to assume that,
for that pier height range, the critical factor for pier behavior and failure is still the shear load
transferred through the coupling beam, and that the possibility of scaling the experimental
results according to the moment mobilized through lateral loading at variable height is a valid
strategy.
6.52
Seismic Performance Assessment
7
SEISMIC PERFORMANCE
ASSESSMENT
7.1. OBJECTIVES AND MAIN METHODOLOGY
The current chapter presents the last part of this work, which focuses on considering the
application of the pier systems studied in previous chapters in real railway bridge structures, for
the purpose of evaluating the resulting seismic performance. With that in mind, full structure
models respecting the structural layout defined for study (Chapter 4) are calibrated against the
results of precast and monolithic specimens (Chapter 5), which are scaled for generalized
application using a pier height dependent strategy that was validated earlier (Chapter 6).
Furthermore, the viaducts presented in Table 4.4 were selected for this case study.
Within that context, it was explicitly assumed that the main focus of the present chapter
should essentially target a comparison between the two types of pier system - monolithic and
precast - rather than a full and comprehensive performance assessment report. The main reason
for that decision is that doing so would require extensive study, which would be out of scope for
the present work to address thoroughly, because performance based engineering is a wide and
complex framework, for which other sources can also provide more in depth discussion
(Vision2000 (1995), Moehle (2003)). This also means that the main scope of this work is the
application of well-established methodologies to the case studies, rather than further
developments. With that in mind, the PEER’s Performance-Based Earthquake Engineering
proposal (Krawinkler and Miranda (2004)) was considered as the present study reference, which
is supported on a probability based approach. Essentially, that proposal addresses the
characterization of system performance involving four different stages (Günav and Mosalam
(2012)), taking into account the variability and uncertainties related with each of them:
7.1
▪ Hazard analysis: Characterization of the earthquake hazard, considering the

existence of nearby faults, magnitude recurrence rates, fault mechanisms,
source-site distance, site conditions, etc... It mostly targets the adequate selection
of a number of compatible ground motions for representation of the variability of
the earthquake action, enabling the definition of the Intensity Measure (IM) for
seismic assessment of the intended structure;
▪ Structural analysis: Prediction of structure behavior while accounting for the
seismic loading with variable characteristics, which generally requires elaboration
of a suitable computational model to perform adequately under dynamic loading
conditions and reasonable modeling assumptions. It aims to determine the values
of a given selection of Engineering Demand Parameters (EDPs), which are deemed
capable of characterizing the response of the structure for the intended
performance levels;
▪ Damage analysis: Definition of the Damage Measures (DM) used for evaluation of
the required performance levels, which are generally related with the occurrence of
physical damage or loss of functionality at element or system levels. The
probability of the EDPs exceeding the corresponding DMs is then calculated for
the varying range of IM, and is often characterized through the adoption of
Fragility curves;
▪ Loss analysis: Evaluation of the probability of exceedence for the Decision making
Variables (DV) regarding losses on each of the pre-defined damageable groups.
Loss analysis often involves the determination of loss curves;
In the present work, the adopted methodology essentially aims to determine the
probability of exceedence of certain DMs for both monolithic and precast structures, to be
expressed in the form of fragility curves. Within that context, the work presented herein mostly
focuses on the structure analysis and damage analysis parts of the PBEE methodology. For
further discussion on probabilistic seismic assessment methods, Romão (2012), Monteiro
(2011), and Marques (2011), among other works, can provide more in depth information.
7.2. GROUND MOTION CHARACTERIZATION
One of the main requirements of seismic performance assessment is having ground motion
records capable of adequately representing code compatible mean seismic demand. In that
regard, it is widely known how the characteristics of different earthquake events may have
distinct effects on the response of structures, as for instance the type of record (artificial, real or
simulated (Iervolino and Manfredi (2008), Katsanos et al. (2010))), the scaling methods used
7.2
for providing correspondence with a given target spectrum (Cantagallo et al. (2014), Grant and
Diaferia (2013)), the number of records required for assurance of acceptable representativeness
(Monteiro (2011)), etc…
Admittedly, the framework of ground motion input addresses one of the most complex
issues of seismic engineering, due to the large number of variables capable of influencing the
outcome of seismic demand characterization. Additionally, some of them are also difficult to
evaluate thoroughly, thus it is not of interest for the present work to tackle this subject in depth.
Within that context, ground motion record scaling and selection was instead performed using
the SelEQ program (Macedo et al. (2013), Macedo and Castro (2016)), a tool developed at
FEUP, which searches known earthquake record databases and incorporates several meta-
heuristic algorithms and filtering criteria, enabling users to find compatible record sets for any
type of target response spectrum.
For the current application, Type 1 EQ was previously observed to be the most relevant
within the test specimen design stage (Chapter 4). Following those results, the ground motion
spectra represented in Figure 4.13 – a) was considered as the target spectrum for record
selection using the SelEQ tool. In that regard, the program was configured to find a suitable set
of ground motions within the range defined by 0.20T1 and 2.00T1, according to EC8 guidelines,
where T1 represents the fundamental period of the structure under analysis. However, since this
study aims to compare the seismic performance on four different structures, it was deemed
necessary to have the same records used for all the analyses, which essentially limits the search
to a single period range that should, ideally, be representative for all of them. Therefore, the
largest fundamental period T1 of all the structures was considered, corresponding to that of
Viaduct over Ribeira da Laje, as will be presented in further ahead.
Twenty ground motion records were then obtained from the Harmony Search algorithm
incorporated in SelEQ, involving the optimization of the differences between the mean
spectrum of the record set and the target spectrum, within a given tolerance and the pre-
established period range. Results of that procedure are illustrated in Figure 7.1, for the ground
motion records listed in Table 7.1 and further characterized in Annex C. Regarding the scaling
of these records for the purpose of determining fragility curves, the adopted IM characterization
parameter was the spectral acceleration Sa1 (corresponding to the first vibration mode of each
structure), whose values are also listed in Table 7.1.
7.3
Figure 7.1 – SelEQ ground motion selection output
Table 7.1 – Scaling factors for the Ground Motion records for all viaducts
11-SuperstitionHills-02
3-TaiwanSMART1(40)
7-TaiwanSMART1(40)
12-Chi-ChiTaiwan-04
14-Chi-ChiTaiwan-06
19-ImperialValley-06
8-Chi-ChiTaiwan-05
2-ChalfantValley-02
1-ImperialValley-06
9-ImperialValley-06
20-VictoriaMexico
10-Chi-ChiTaiwan
13-Chi-ChiTaiwan
5-Chi-ChiTaiwan
17-Northridge-01
18-Northridge-01
6-Northridge-01
16-BigBear-01
15-LomaPrieta
4-LomaPrieta
Earthquakes
SelEQ scaling
0,60 3,40 1,84 0,79 3,61 0,33 1,13 2,90 3,17 0,89 2,50 3,56 1,10 3,70 3,77 2,70 2,69 2,86 1,21 3,44
parameter
Sa1 (g) - A 0,65 0,42 0,76 0,76 0,56 0,55 0,53 0,45 0,69 0,50 0,49 0,57 0,34 0,34 0,57 0,45 0,56 0,59 0,62 0,54
Sa1 (g) - B 0,67 0,49 0,68 0,59 0,62 0,49 0,35 0,49 0,74 0,41 0,61 0,45 0,49 0,41 0,79 0,40 0,46 0,76 0,83 0,40
Sa1 (g) - C 0,76 0,39 0,64 0,64 0,69 0,37 0,38 0,61 0,68 0,43 0,75 0,38 0,51 0,45 0,69 0,34 0,56 0,79 0,75 0,60
Sa1 (g) - D 0,28 0,24 0,41 0,35 0,26 0,40 0,37 0,29 0,28 0,37 0,37 0,33 0,35 0,34 0,45 0,31 0,27 0,35 0,24 0,27
A - Macheda Viaduct
B - Palheta Viaduct
C - Viaduct over Degebe River
D - Viaduct over Ribeira da Laje
7.4
7.3. STRUCTURE CHARACTERIZATION
A computational model for seismic response evaluation was elaborated in OpenSees

(McKenna et al. (2010)) for the four viaducts under analysis. For the sake of simplicity and
faster dynamic analyses, a 2D approach was adopted, following the same strategy presented in
Delgado (2009), which is illustrated on Figure 7.2 – a) for a generic viaduct case. Essentially,
this methodology focuses on characterizing the transverse response of the piers, which are
constrained to the lateral behavior of the deck by means of axially rigid bars (to impose
coupling of ux displacements). This approach disregards deck torsion and longitudinal effects,
which is an acceptable assumption for the present chapter scope.
Therefore, for consistency in calibrating the lateral stiffness of each viaduct while having
piers with different heights (Hpier), the values for Hpier take into account the total length between
the fixed restraint at the base of the piers and the deck’s centroid, where inertia forces are
considered. The double column bent pier system is modeled by a single vertical element with
linear elastic properties, adjacent to a zero-length plastic hinge at the base, according to Figure
7.2 – b). Deck masses were lumped at several locations along the total length of the structure,
separated by 5 segments per span, while roughly half the total mass of the piers (including the
beam segment) was lumped at the beams’ centroid level.
For the characterization of the dynamic properties of each viaduct, a Rayleigh-type

damping approach proportional to the mass and initial stiffness was adopted, assuming 5% of
the critical damping at the first and second transverse vibration modes of each structure.
F
F
Lspan Linear Elastic
Element
Hpier
Concentrated
Plastic Hinge
uy Double Column Numerical Pier

ux Prototype Representation
a) 2D representation b) Pier model

Figure 7.2 – Structural modeling scheme for a generic viaduct
7.5
With this methodology, the inelastic behavior is globally considered only at the plastic
hinge level, which is calibrated for performing in the degree of freedom corresponding to
rotations and according to a uniaxial material model in OpenSees. For that purpose, the software
includes a large variety of user selection models, from which the Modified
Ibarra-Medina-Krawinkler (IMK, Ibarra et al. (2005)) deterioration model with “pinched”
hysteretic response was adopted, having already been successfully used in other performance
based earthquake engineering studies (e.g. Lignos (2008), Lignos and Krawinkler (2012)).
Calibrating this model assumes definition of a backbone curve, from which cyclic degradation
is determined according to several stiffness and strength related parameters.
7.3.1. PIER BEHAVIOR CALIBRATION
Realistic assessment of the seismic performance on the present case studies, involves
ensuring that the computational models are capable of returning dynamic responses compatible
with the experimental evidence gathered during the cyclic tests. For that purpose, the numerical
model chosen for characterization of the inelastic behavior of the modeled viaducts should
incorporate similar damage incursion patterns and collapse features as those observed in the
experimental tests. In other words, it should be characterized by equivalent (but necessarily
scaled) force-displacement or moment-rotation results to those presented in Chapter 5. With that
in mind, the following two scenarios were considered:
▪ Monolithic: Plastic hinge uniaxial model parameters calibrated to simulate the

monolithic specimen SP_M02;
▪ Precast: Plastic hinge uniaxial model parameters calibrated to simulate the precast
specimen SP_PC02C;
The selection of SP_M02 and SP_PC02C as the calibration sources is related to their
performance, which, as presented in Chapter 5, is among the best in each respective category
(monolithic and precast). Furthermore, only the positive loading direction was taken into
account in this process, since it is believed that the observed loading direction asymmetry would
not be so evident during a real earthquake event, considering that seismic loading is
simultaneously applied to both columns, instead of the one-sided loading as adopted on the
experimental tests.
The actual calibration procedure was based on the uniaxial cyclic loading of a fixed base
sample pier, aiming to reproduce the same behavior of the corresponding experimental models
(monolithic or precast, as mentioned above), and by taking into account the applicable scale
factors to peak force (due to both the reduced scale of the test models and the half-pier testing
constraint), which is illustrated in Figure 7.3. Furthermore, IMK model parameters related with
7.6
unloading/reloading stiffness, energy dissipation and cyclic degradation were adjusted within
the set of recommended values through trial and error, aiming to minimize differences to the
experimental values as much as possible.
F
2x4xd 42xF
d
Scale factor for length
H and displacements:
x4
2xH 2 x4 x H
Scale factor
for forces:
x 42
Experimental Model Sample Pier
Figure 7.3 – Sample pier calibration against experimental model results
The stiffness of the linear elastic element is also an important part of the calibration
procedure, since it directly influences both static and dynamic results, considering that the
moment mobilized at the base plastic hinge depends on the rotation conveyed through the
vertical element. In turn, this depends on the stiffness relationship between both elements of the
pier assembly (with global stiffness denoted by Kassembly), namely the plastic hinge and vertical
element (whose stiffnesses are designated by Khinge and Kelastic, respectively), which is unknown
beforehand. Furthermore, considering the adoption of stiffness proportional Rayleigh-type
damping for dynamic analysis purposes, this issue extends to the overall energy dissipation that
is obtained for a given displacement history, particularly because the inelastic behavior is
concentrated only in the plastic hinge.
In order to address this, the approach presented by Zareian and Medina (2010) was
adopted, where a set relationship between the hinge and vertical element stiffness components is
enforced for the sample pier, governed by the parameter n, as described in equations 7.1 and
7.2. In this application, according to the conclusions of that same work where a large value of n
is recommended, it was assumed 𝑛 = 20. Within this context, the value of Kassembly represents
the global lateral stiffness of the pier system (governed by equation 7.3), which should be
equivalent to the experimental values of the applicable scenarios. Therefore, for a pre-yielding
stage, the Kassembly stiffness can be determined from the yielding displacements of Figure 5.69
7.7
and Figure 5.70 (and the corresponding yielding forces), taking also into account the force and
displacement scaling procedure presented in Figure 7.3. After calculation of Kassembly, the values
of Khinge and Kelastic can be determined using the n parameter and equations 7.1 and 7.2,
respectively. Furthermore, this strategy considers initial stiffness based Rayleigh damping
applied only to the elastic elements, since the IMK uniaxial model controls the overall value of
Khinge, where energy dissipation due to inelastic deformations is expected to be dominant.
𝐾ℎ𝑖𝑛𝑔𝑒 = (𝑛 + 1) × 𝐾𝑎𝑠𝑠𝑒𝑚𝑏𝑙𝑦 7.1

𝑛+1
𝐾𝑒𝑙𝑎𝑠𝑡𝑖𝑐 = ( ) × 𝐾𝑎𝑠𝑠𝑒𝑚𝑏𝑙𝑦 7.2
𝑛
1 1 1
= + 7.3
𝐾𝑎𝑠𝑠𝑒𝑚𝑏𝑙𝑦 𝐾𝑒𝑙𝑎𝑠𝑡𝑖𝑐 𝐾ℎ𝑖𝑛𝑔𝑒
The remaining part of the calibration process is related to the estimation of the backbone
curve, from which cyclic degradation is established. Ideally, the backbone curve characteristics
are determined from monotonic loading results, which were, unfortunately, unavailable. As an
alternative, a trial and error estimation of the backbone curves was attempted, by taking into
account in-cycle and cyclic strength and stiffness degradation according to recommendations by
Haselton et al. (2009), in order to adapt the model parameters that control those effects to the
present case studies. The resulting moment – drift relationships obtained with this procedure are
presented in Figure 7.4 for both Monolithic and Precast cases (whose corresponding model
parameters are provided in Annex A). As it is possible to observe, there is a good fit between
the experimental and numerical curves, particularly regarding moment values for the positive
loading direction and regarding the cyclic degradation for the repeating cycles. A residual
strength ratio of 20% of the peak value was adopted in both models (which, however, is not
apparent for the drift values shown in the plots).
70.00 70.00
Monolithic_experimental Precast_experimental
IMK_Pinched_Monolithic IMK_Pinched_Precast
Backbone Backbone
50.00 50.00
30.00 30.00
Moment (103 kN.m)
Moment (103 kN.m)
10.00 10.00
-10.00 -10.00
-30.00 -30.00
-50.00 -50.00
-70.00 -70.00
-5.50 -4.50 -3.50 -2.50 -1.50 -0.50 0.50 1.50 2.50 3.50 4.50 5.50 -5.50 -4.50 -3.50 -2.50 -1.50 -0.50 0.50 1.50 2.50 3.50 4.50 5.50
Drift (%) Drift (%)
a) Monolithic b) Precast
Figure 7.4 – Comparison of the moment – drift relationship calibration results
7.8
Despite providing satisfactory moment-drift comparisons against the scaled experimental

records, the IMK model’s precision is less noticeable for simulating of the energy dissipation
per half-cycle, tending to lead to higher values relative to the experimental results, mostly due to
the larger unloading stiffness branch. It is noteworthy that, the above presented moment – drift
relationships already reflect the best result of several parameter adjustment attempts, performed
to address that particular issue, but the energy dissipation difference between the experimental
results and those produced by the IMK models is still noticeable. This can be identified in
Figure 7.5, where post-yielding energy dissipation per positive loading half-cycle is presented
for both Monolithic and Precast cases, considering representation on normalized terms, obtained
by dividing each dissipated energy value by the energy dissipation of the first experimental
cycle after yielding, according to equation 7.4:
𝐸𝑖
𝐸𝑛𝑜𝑟𝑚 = 7.4
𝐸1,𝑒𝑥𝑝
9.00 10.00
Monolithic_experimental Precast_experimental
8.00 9.00
Normalized Energy Dissipation
Normalized Energy Dissipation
IMK_Pinched_Monolithic IMK_Pinched_Precast
7.00 8.00
7.00
6.00
6.00
(Ei/E1,exp)
(Ei/E1,exp)
5.00
5.00
4.00
4.00
3.00
3.00
2.00 2.00
1.00 1.00
0.00 0.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Post-yielding Half Cycles Post-yielding Half Cycles
(positive loading) (positive loading)
a) Monolithic b) Precast
Figure 7.5 – Comparison of the Energy dissipation per cycle results
Nonetheless, while energy dissipation is larger on the numerical models than observed on
the test specimens, they still exhibit similar increase rate. In addition, it may be reasonable to
say that this strategy can provide sufficiently adequate simulation, considering that it mainly
targeted the comparison between the global behavior of monolithic and precast structures.
General application of a methodology where the behavior of all the piers is determined
from a single sample pier calibration can be arguable, when considering the accuracy of
stiffness scaling for different pier heights. In that regard, the adopted strategy involved the same
hinge component on all the piers, and the elastic component is derived from an equivalent
inertia calculated for the sample pier stiffness, hereby labeled method A, which means that
stiffness scaling according to different pier heights is managed by the elastic component alone.
This also means that the relative contribution of hinge and stiffness components determined for
7.9
the sample pier actually changes according to pier height. A preferable solution would be to
enforce the same 𝑛 = 20 relationship on all the piers, hereby labeled method B, but that would
involve significant work, requiring individual calibration of different hinge models for all the
different pier heights. Regardless, the actual difference between these two approaches is not that
large, as seen in Figure 7.6, where the ratio between the total lateral pier stiffness calculated
with the two approaches (method B values divided by method A values) is illustrated for the
pier height range under analysis. The overall difference between the two approaches is never
greater than 15%, and only for pier structures shorter than 9.50m it is greater than 5%. In light
of that, it is believed that method A still provides reasonable accuracy while being significantly
easier to perform, which is why it was adopted for the current work.
1.15
1.10
1.05
1.00
0.95
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Pier Height (m)
Figure 7.6 – Lateral stiffness ratio between method B and A
7.3.2. NON-LINEAR STATIC PUSHOVER ANALYSES
In order to develop an understanding of the demand distribution over all the piers of the
selected viaducts, in order to help with defining the control piers for further dynamic evaluation,
non-linear static pushover analyses were performed. For that purpose, force-based conventional
pushover with first-mode proportional load pattern was considered (EN1998-1 (2004)). In that
regard, the first vibration mode of each viaduct is represented for the corresponding deck nodes
in Figure 7.7, considering normalized modal displacements for unitary maximum value.
Comparing the associated periods (which are essentially the same within the Monolithic or
Precast models) with the values previously evaluated in SAP2000 earlier in Chapter 4, small
differences are found, which can be attributable not only to the fact that the current structure
models were developed in a 2D framework, but also to the calibration that was made herein
against actual experimental data.
7.10
180 225 450

T=0.462s
550 T=1.038s
T=0.547s T=0.418s
160 200 400 500
140 175 350 450
400
120 150 300
Viaduct Length (m)

Viaduct Length (m)
Viaduct Length (m)

Viaduct Length (m)
350
100 125 250
300
80 100 200 250
75 200
60 150
150
40 50 100
100
20 25 50
50
0 0 0 0
0.0 0.5 1.0 0.0 0.5 1.0 0.0 0.5 1.0 0.0 0.5 1.0
c) Viaduct over d) Viaduct over
Degebe River Ribeira da Laje
Figure 7.7 – First lateral vibration mode for the case study viaducts
The pushover analyses were performed for a target drift of 10% on the pier closest to the
mid section of the viaducts, at 82.00m, 100.00m, 200.00m and 256.70m for Figure 7.7 - a) to
Figure 7.7 – d) viaducts, respectively. Results for ductility demands were normalized by the
ductility demand of those piers and are presented in Figure 7.8 for monolithic and precast
behavior models. Unsurprisingly, the ductility demand is largest on the piers close to the mid
section of the viaducts, considering that first-mode proportional load pattern was adopted and
the fact that all the viaducts are quite regular, with low spread between the median pier height
and the corresponding minimum and maximum values (Table 4.4). Taking into account these
results, the following piers were chosen for control sections regarding dynamic performance
evaluation:
▪ Macheda Viaduct: P3 at 82.00m of total length;

▪ Palheta Viaduct: P4 at 112.00m of total length;
▪ Viaduct over Degebe River: P7 at 200.00m of total length;
▪ Viaduct over Ribeira da Laje: P10 at 286.70m of total length;
7.11
1.50 1.50
Monolithic Monolithic
Precast Precast
1.25 1.25
Normalized Ductility Demand

1.00 1.00
(μi/μcontrol)
(μi/μcontrol)
0.75 0.75
0.50 0.50
0.25 0.25
0.00 0.00
P1 P2 P3 P4 P5 P1 P2 P3 P4 P5 P6
Pier Pier

1.50 1.50
Monolithic Monolithic
Precast Precast

1.25 1.25
1.00 1.00
(μi/μcontrol)
(μi/μcontrol)
0.75 0.75
0.50 0.50
0.25 0.25
0.00 0.00
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17
Pier Pier

Figure 7.8 – Ductility demand on the piers of the case study viaducts
7.4. DAMAGE CHARACTERIZATION
This stage of performance based earthquake engineering ultimately aims to be able to

establish relations between the intensity of earthquakes (IM) and the damage caused by them.
For that purpose, structural response is numerically characterized by estimates of EDPs, and
their values are compared against the established performance targets (DMs). In that regard, the
three following DMs were chosen for this study:
▪ DM #1: Structural performance (sp)

DMsp1 – Yielding of the main pier reinforcement;
DMsp2 – Conventional collapse of the pier structure due to strength loss;
▪ DM #2: Lateral deflection (ld)
DMld1 – Maximum angular rotation of the deck for running safety of trains;
DMld2 – Maximum angular rotation of the deck for restorability of the track;
▪ DM #3: Derailment conditions (dr)
DMdr1 – Maximum lateral vibration of the deck for running safety of trains;
7.12
7.4.1. DM #1 STRUCTURAL PERFORMANCE
In performance based earthquake engineering studies, structural performance is usually

evaluated from a selection of common damage states involving yielding, spalling, buckling,
rebar fracture and collapse (Kunnath et al. (2006), Mander et al. (2007)). In the present study,
yielding and collapse were adopted as the performance targets, and the corresponding EDP
associated with DM #1 was the rotation recorded at the simplified plastic hinge model on the
control pier for each viaduct. The associated performance limits (included in Table 7.2 for
reference) were calculated as the rotation values (at the hinge level) that characterize the same
load state of the equivalent experimental results. In the case of yielding, for example, (which is
characterized by the displacements from Figures 5.69 and 5.70), the value of 0.88% found in the
experimental results from SP_PC02C addresses the same load state (yielding) as 0.80% on
Figure 7.4 – b). Accordingly, at the local plastic hinge level the same yielding stage is
associated with a rotation of 0.037 x 10-2 rad, which can then be defined as the corresponding
EDP limit.
Table 7.2 – EDP limits for DM#1

EDP - Hinge Rotation (10-2 rad)
DM#1 targets Monolithic Precast
DMsp1 - Yielding 0.054 0.037
DMsp2 - Collapse 1.587 2.859
It is important to discuss, however, that the strategy for collapse definition adopted in the
present work may not be reasonable for every such application. For example, Vamvatsikos and
Cornell (2002) state that “If the model is realistic enough it ought to explicitly contain such
information, i.e. show a collapse by non-convergence instead of by a definite DM output”,
leading to the adoption of so-called flatlines, in correspondence with the IM values for which
the capacity points occur.
Additionally, Haselton et al. (2009) recommend that non-convergence should only be

identified as collapse when it occurs for large demands, representative of structural failure. This
is, in fact, because non-convergence should ideally be caused by dynamic instability at the onset
of structure collapse, and not just due to failure of the solution algorithm. However, this requires
that the model is adequately prepared for representing the structure behavior up to that moment
when dynamic instability occurs, which may not be the case with the present modeling strategy,
particularly because the plastic hinge models are only experimentally validated up to the drift
threshold of the test specimens that served as calibration source. Within this context, the
adoption of a plastic hinge rotation limit for definition of the capacity point may just represent a
conservative check of the no collapse limit state, but nonetheless, that is believed to be the most
reasonable approach considering the present application.
7.13
On another note, is should also be said that, under normal circumstances, EDP
characterization for performance based earthquake engineering involves the evaluation of
uncertainties related with identification of the material properties and/or the modeling strategy
employed for structural behavior assessment. For example, on Monso and Miranda (2013)
fragility functions developed by Berry and Eberhard (2003) accounting for a statistical
distribution of column strength parameters are used to assess the mean DM limits and the
corresponding dispersion, while in Delgado (2009), Latin Hypercube simulations are performed
to obtain the probability distribution that characterizes structural strength. In the present work,
the small number of experimental tests for each scenario prevents application of common
statistical procedures to determine the dispersion of the adopted EDP limits. Likewise, since the
previously presented modeling strategy focuses on the local reproduction of the recorded
moment-drift histories and not constitutive characterization based on material properties,
extensive numerical simulations using statistical sampling methods would require the
calibration of other computational models more suited for such task. In this regard, it is worth
mentioning that, while the refined 2D FEM models presented in Chapter 6 are constitutive
based and provided generally good results, the amount of time required for a single full analysis
is quite large, rendering them very inefficient for this particular purpose.
Even so, it may be argued that the comparison between the two established scenarios is still
well served by assessing the probability of exceedence of the limit states in consideration while
disregarding the uncertainties related with DM#1 performance limits, because it can be assumed
that they affect both scenarios equally. Consequently, this acknowledges the values indicated in
Table 7.2 as deterministic measurements of those structural systems, leading to performance
assessment representing only the response variability that is introduced by considering different
ground-motions, which seems to be a reasonable compromise.
7.4.2. DM #2 LATERAL DEFLECTION
The lateral deflection of the viaducts was evaluated according to the maximum angular
rotation observed between subsequent deck spans (which is mostly relevant to simply supported
span structures), notwithstanding the fact that the present modeling strategy reflects the
displacements recorded at the deck centroid level, and not at the upper deck surface, as would
be preferable. However, this difference is deemed not to be relevant so as to compromise the
conclusions of the present study. The corresponding value for angular rotation, calculated at the
end of each span, is graphically illustrated in Figure 7.9.
7.14
di-1
Deck displacements
Lspani
di
Lspani+1 Angular rotation
di+1
Figure 7.9 – Lateral deflection evaluation procedure.
EDP limits for the two selected damage states on this subject are presented in Table 7.3,
corresponding to the values proposed by the RTRI guidelines (Table 2.2). The choice to adopt
RTRI values instead of European Eurocode angular limits was made because the former
explicitly considers differences between performance under ordinary travel conditions and
during a seismic event, according to previous discussion on Chapter 2. Furthermore, it also
provides more consistent methodology since it relies on a performance based approach. In that
regard, this work focused on the performance objectives related with running safety of high
speed trains and the assurance of restorability conditions when significant track damage occurs,
for DMld1 and DMld2, respectively. Additionally, for the DMld2 damage state, the most severe
track type conditions were considered, which in this case refer to a slab track. It is worth
mentioning that Table 7.3 shows identical limits for monolithic and precast specimens, since
this DM focuses on a global track related assessment.
Table 7.3 – EDP limits for DM#2 (according to RTRI (2007) )

EDP – Angular Rotation (rad)
DMld1 – Running Safety 0.002 0.002
DMld2 – Track Damage 0.006 0.006
7.4.3. DM #3 DERAILMENT CONDITIONS
European codes do not provide clear guidance regarding the evaluation of derailment
during earthquake, and designers are often required to perform track-structure interaction
studies to address that in further detail, such as in Montenegro (2015), for example, which
proposes a methodology for the assessment of running safety of trains on HSRL, based on
7.15
wheel-track contact criteria. Other authors like Luo (2005), Luo and Miyamoto (2008),
proposed an approach based on the concept of Spectral Intensity (SI, equation 7.5, where
𝑆𝑣 (𝜉, 𝑇) is the pseudo-velocity spectrum).
2.50
𝑆𝐼 = ∫ 𝑆𝑣 (𝜉, 𝑇) 𝑑𝑇 7.5
0.10
This concept is contemplated in the RTRI guidelines, essentially aiming to reflect the
relationship between the energy content of the seismic input and the structural response, which
is simpler to account for with the present modeling strategy. Within this context, SI has been
selected as the EDP for assessment of derailment, and the corresponding limits for safety can be
identified in Figure 2.6. It should be noted that, while the values presented therein were
determined for a wide variety of soil conditions and derailment mechanisms still, they reflect
the dynamic properties of Japanese rather European trains. However, owing to the lack of a
better alternative for the European case and taking into account the structural vibration periods
for the case study viaducts, the limit value of 4100mm included in Figure 2.6 was adopted for
SI, according to Table 7.4. Similarly to DM#2, these damage measure limits are equal for both
scenarios, since they also report to a global track assessment.
Table 7.4 – EDP limits for DM#3

EDP – Spectral Intensity (mm)
DMdr1 – Running Safety 4100 4100
The original procedure for SI assessment involves the characterization of the full structure
as a SDOF model with equivalent properties, and computation of the corresponding
displacements at the track level when subjected to seismic loading. Afterwards, the velocity
response spectrum of the response may be determined, from which the SI is calculated. This
methodology is illustrated in Figure 7.10. However, in this study, the full length of all the
viaducts was explicitly considered instead of using equivalent SDOF structures, since the
response data at all the deck nodes was already available from the full set of dynamic analyses
ran for the other DMs. For that purpose, calculation of the velocity spectrum of the response
was made at the control pier for each viaduct, using only the corresponding first vibration mode
periods, which may constitute a conservative approach for the evaluation of SI.
7.16
Figure 7.10 – Procedure for running safety assessment using SI (Luo (2005)).
7.5. SEISMIC PERFORMANCE OUTPUT
For the purpose of seismic performance assessment, non-linear dynamic analyses were
performed on all the four case study viaducts according to an Incremental Dynamic Analysis
(IDA) procedure (Vamvatsikos and Cornell (2002)), for which the earthquake intensity scaling
of the previously presented records was considered. For that purpose, the Newmark integration
method (Newmark (1959)) was used (with typical parameters 𝛾 = 0.50 and 𝛽 = 0.25) and an
IM scaling factor λscaling was individually defined for each ground motion, in proportion to the
spectral acceleration Sa1 corresponding to the first vibration mode of each structure.
Furthermore, this means that, for IDA purposes, a different scaling factor was determined for
each structure and ground motion i, according to equation 7.6, where ΔIDA and nIDA represent the
scaling increment and the increment number, respectively. In that regard, a ΔIDA of 0.20g was
used for the full analysis range of [0.0 ; 6.0] g.
1
𝜆𝑠𝑐𝑎𝑙𝑖𝑛𝑔 = × (∆𝐼𝐷𝐴 × 𝑛𝐼𝐷𝐴 ) 7.6
𝑆𝑎1,𝑖
The output of the IDA analyses can be organized in the form of IDA curves, representing
the distribution of the maximum EDP values, determined individually at each IM level (Figure
7.11 – a)) for all the analyzed ground motions. An example of such a curve is illustrated in
Figure 7.11 – b), relating to data obtained for the Macheda Viaduct with precast behavior and
DM#1, where the 16th, 50th and 84th fractiles of the data distribution are also represented. Due to
7.17
the large amount of data produced by the IDA methodology, all the IDA curves are not
presented in the main body of this document, but were instead included in Annex C.
6.00 0.050
Maximum value 16th fractile
5.00 0.045
Maximum Hinge Rotation (rad)

for the IM 50th fractile
4.00 0.040
84th fractile
Hinge Rotation (10-3 rad)
3.00 0.035
2.00 0.030
1.00 0.025
0.00 0.020
-1.00 0.015
-2.00 0.010
0.005
-3.00
0.000
-4.00
0.0 1.0 2.0 3.0 4.0 5.0 6.0
-5.00 Intensity Measure, Sa 1 (g)
a) EDP response at 𝑆𝑎1 = 1.0𝑔 and 5th GM b) IDA curve for DM#1
Figure 7.11 – Example results obtained for the Macheda Viaduct with Precast behavior.
On account of the IDA output, fragility curves were determined for each performance level
of the three selected DMs. For this purpose, the probability of exceedence of the corresponding
damage states i was determined for each IM level, by selecting the performance points from
EDP j values regarding the lowest IM occurrence (DM based approach). Furthermore, this
information was gathered in the form of a cumulative distribution function (CDF) according to
equation 7.7, to which lognormal curves were fitted using procedures developed by Baker
(2015).
+∞
𝐶𝐷𝐹(𝐷𝑀𝑖 |𝐼𝑀) = ∫ 𝑃[𝐸𝐷𝑃𝑗 > 𝐷𝑀𝑖 |𝐼𝑀 = 𝑥] 𝑑𝑥 7.7
0
7.5.1. DM#1: STRUCTURAL PERFORMANCE RESULTS
The fragility curves obtained for the structural performance DMs are presented in Figure
7.12 for all the viaducts, including the comparison between monolithic and precast systems’
results. In addition, the first mode spectral acceleration values were evaluated from the elastic
ground motion acceleration response spectra and for each structure, considering the reference
PGA defined for the damage limitation and no collapse requirements. Those values are also
indicated by the green and purple vertical lines (for damage limitation and no collapse,
respectively).
7.18
1.00 1.00
0.90 0.90
0.80 0.80
0.70 0.70
CDF (SP | IM)
CDF (SP | IM)

0.60 0.60
0.50 0.50
0.40 0.40
0.30 0.30
DMsp1 - Mono DMsp1 - Mono
0.20 DMsp2 - Mono 0.20 DMsp2 - Mono
0.10 DMsp1 - Precast DMsp1 - Precast
0.10
DMsp2 - Precast DMsp2 - Precast
0.00 0.00
0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 4.0
Intensity Measure, Sa 1 (g) Intensity Measure, Sa 1 (g)

1.00 1.00
0.90 0.90
0.80 0.80
0.70 0.70
CDF (SP | IM)

CDF (SP | IM)
0.60 0.60
0.50 0.50
0.40 0.40
0.30 0.30
DMsp1 - Mono DMsp1 - Mono
0.20 DMsp2 - Mono 0.20 DMsp2 - Mono
DMsp1 - Precast 0.10 DMsp1 - Precast
0.10
DMsp2 - Precast DMsp2 - Precast
0.00 0.00
0.0 1.0 2.0 3.0 4.0 0.0 1.0 2.0 3.0 4.0

Figure 7.12 – Fragility curves for DM#1 – Structural Performance results
The structural performance inherent to these results shows that the precast model is able to
maintain stable behavior up to larger ground motion intensities, as the respective DMsp2
fragility curves generally exhibit lower probability of exceedence than those relative to
monolithic behavior. The contrary is observed regarding yielding, where precast DMsp1 occur
earlier than for the monolithic model. That is an expected result, and can be understood as the
natural outcome of the precast piers having lower yielding threshold, whereas the monolithic
cases lead to slightly larger yielding displacements. Regardless, this agrees with the conclusions
obtained through reduced scale testing, where the precast model was able to carry larger
ductility capacity than the monolithic counterpart.
With respect to the performance observed on different structures, a general trend for earlier
collapse can be observed on the longer viaducts (Viaduct over Degebe River and Viaduct over
Ribeira da Laje, Figure 7.12 – c) and d), respectively). That can be attributed to the fact that
longer structures have less overall contribution of their abutments to the transverse load carrying
capacity, leading to significant demand to be observed on a relatively larger part of the piers. In
these cases, the difference between monolithic and precast systems is less pronounced, and both
7.19
exhibit very similar exceedence probability values, particularly within the CDF range between
0.00 and 0.30.
Representation of the first mode spectral acceleration provides an indication of the

performance that is expected relative to EC8 requirements. In that regard, the damage limitation
requirement could be at risk on the Macheda viaduct with precast behavior, since the
corresponding probability of exceedence for yielding is significant. In addition, a low
probability of exceedence is observed on every other relevant scenario, although further detailed
analyses should be performed, particularly for Viaduct over Degebe River and Viaduct over
Ribeira da Laje, if increased precision is required.
7.5.2. DM#2: LATERAL DEFLECTION RESULTS
Lateral deflection DM#2 fragility curves are illustrated in Figure 7.13 for all the viaducts
and both behavior models.
1.00 1.00
0.90 0.90
0.80 0.80
0.70 0.70
CDF (LD | IM)
CDF (LD | IM)
0.60 0.60
0.50 0.50
0.40 0.40
0.30 0.30
DMld1 - Mono DMld1 - Mono
0.20 DMld2 - Mono 0.20 DMld2 - Mono
DMld1 - Precast 0.10 DMld1 - Precast
0.10
DMld2 - Precast DMld2 - Precast
0.00 0.00
0.0 1.0 2.0 3.0 4.0 5.0 6.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0

1.00 1.00
0.90 0.90
0.80 0.80
0.70 0.70
CDF (LD | IM)
CDF (LD | IM)
0.60 0.60
0.50 0.50
0.40 0.40
0.30 0.30
DMld1 - Mono DMld1 - Mono
0.20 DMld2 - Mono 0.20 DMld2 - Mono
0.10 DMld1 - Precast 0.10 DMld2 - Precast
DMld2 - Precast DMld2 - Precast
0.00 0.00
0.0 1.0 2.0 3.0 4.0 5.0 6.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0

Figure 7.13 – Fragility curves for DM#2 – Lateral Deflection results
In general, these results illustrate two different trends regarding differences between the
monolithic and precast scenarios. On the shorter viaducts (Macheda Viaduct and Palheta
Viaduct) those scenarios show quite close fragility curves, suggesting that the transverse
7.20
stiffness of the deck is more relevant to the control of corresponding displacements than pier
stiffness, which seems reasonable. On the longer viaducts that is no longer the case, and a clear
distinction between precast and monolithic fragility curves can be identified, particularly on
Viaduct over Degebe River. As far as it was possible to assess, that occurs because that is a long
structure (395.00m of length) with relatively short piers (mean pier height of 7.56m), which
benefits from the earlier yielding of precast behavior to see a larger ductility demand on a
greater number of piers, in comparison with the monolithic model, and therefore less overall
angular variation between subsequent deck spans. It is also worth mentioning that running
safety and track damage related fragility curves (DMld1 and DMld2, respectively) are closer
together in the longest viaducts than in the shortest viaducts.
7.5.3. DM#3: DERAILMENT CONDITIONS RESULTS
Figure 7.14 illustrates the fragility curves relative to derailment conditions, evaluated
through DM#3 for all the viaducts and behavior models.
1.00 1.00
0.90 0.90
0.80 0.80
0.70 0.70
CDF (Dr | IM)
CDF (Dr | IM)
0.60 0.60
0.50 0.50
0.40 0.40
0.30 0.30
0.20 0.20
DMdr1 - Mono DMdr1 - Mono
0.10 0.10
DMdr1 - Precast DMrd1 - Precast
0.00 0.00
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
e) Macheda Viaduct f) Palheta Viaduct

1.00 1.00
0.90 0.90
0.80 0.80
0.70 0.70
CDF (Dr | IM)
CDF (Dr | IM)
0.60 0.60
0.50 0.50
0.40 0.40
0.30 0.30
0.20 0.20
DMdr1 - Mono DMld1 - Mono
0.10 0.10
DMdr1 - Precast DMld2 - Precast
0.00 0.00
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
g) Viaduct over Degebe River h) Viaduct over Ribeira da Laje

Figure 7.14 – Fragility curves for DM#3 – Derailment Conditions results
Small differences can generally be observed between the precast and monolithic results for
all the viaducts. Considering that Spectral Intensity values are determined from pseudo-velocity
7.21
which, in correspondence, can be related with the dynamic lateral displacement records at the
control pier, this evidences that the differences between the precast and monolithic models deck
displacements are mostly negligible. Unsurprisingly, this is particularly true for the shortest
length viaducts (Macheda Viaduct and Palheta Viaduct), whereas slightly larger differences can
be identified on the other two study cases.
7.6. FINAL REMARKS
The present chapter encloses the seismic performance assessment study that was developed
as the final part of this thesis. Its main objective was to provide a comparative overview of
performance differences that can be expected from using the proposed precast systems (which
were previously tested and analyzed on earlier chapters) regarding more common monolithic
applications.
For that purpose, a methodology based on the calculation of fragility curves for different
damage measures was adopted, wherein ground motion characterization was made using twenty
seismic records obtained from the SelEQ tool for the applicable EC8 target spectrum, and
structural characterization was evaluated within a computational model elaborated in OpenSees.
In that regard, 2D models were defined, in which modeling the transverse behavior of the
structure was the main focus. Inelastic behavior was governed by a uniaxial model mobilized at
the piers, which was calibrated according to a force-displacement scaling procedure and
experimental results corresponding to two scenarios: one monolithic and one precast.
Within that context, incremental dynamic analysis were performed on four different
viaducts for a seismic intensity range of [0.0 ; 6.0] g, and the corresponding results were
evaluated regarding three damage measures, where the first was related with structural
performance damage state, while the other two consisted of track related assessments.
Interestingly, results generally showed small differences between the precast and monolithic
scenarios on these track related DMs, suggesting that they are more sensitive to deck stiffness
variations than pier behavior. Nonetheless, in the cases where larger differences were observed,
the precast scenario tended to show lower probability of exceedence than the monolithic case
for the most severe damage state under evaluation (Non-Collapse for structural performance and
Track Damage for lateral deflection). Even though the present study was developed from a
simplified approach, those observations arguably support an improved seismic performance of
the precast solutions relative to monolithic specimens.
7.22
Conclusions
8
CONCLUSIONS
8.1. MAIN REMARKS
The main purpose of the work presented in the current document was to provide additional
insight into the framework of the seismic behavior of bridge piers designed for HSRL viaducts,
therefore involving work under two wide and important research fields: the seismic
performance of bridge structures and the strict requirements of HSRL transportation.
Furthermore, it also encloses several activities that were carried out on account of the SIPAV
project, which is a previous research program carried out at the Faculty of Engineering of the
University of Porto, from which the aim to study precast applications was originally derived. In
that regard, while widespread application of the precast technology can be observed for the
construction of bridge deck structures, the same cannot be said regarding the construction of
bridge piers, where cast-in-place construction is most often considered.
Within that context, the underlying concept of precast bridge piers for HSRL is believed to
be innovative, in nature, since it is inherent to structure designs that are not only uncommon, but
also virtually nonexistent in the context of HSRL. This was also the reason why most of the
work was focused on assessing the different features and outcomes of a precast application in
contrast with a monolithic CIP construction, since that has been the historically preferred
option. Additionally, according to research performed during this study, widespread application
of precast for bridge piers is limited by several factors, not only manufacture and operational
issues that influence the cost-effectiveness comparison against typical monolithic solutions, but
also critical technical aspects that lead to weaker or unreliable designs. And while on this topic,
seismic loading related weaknesses are undeniably some of the most difficult to address, since
8.1
Conclusions
they stem from having to account for ductile deformations whilst HSRL SLS criteria demand
high stiffness and low displacements.
This may lead to the design of limited ductility structures, such as wall-pier solutions
where shear yielding tends to be avoided, or generally more ductile single column structures,
where large deformations are more easily accounted for. In that case, if the Portuguese venture
project into HSRL can serve as good representation of the general outlook for HSRL viaducts,
the vast majority of structures face long lengths and short/medium pier heights. Therefore,
according to previous discussion, both of the aforementioned designs were deemed not
particularly well suited for large scale precast applications. As an alternative, the double column
bent pier structure designed for the same project can represent an acceptable solution, providing
considerable stiffness on the bridges’ transverse direction, while being easier to accommodate
for a precast application due to the inherent division between columns and beam, which can be
tackled using relatively common construction industry techniques, such as the top-down cap
beam assembly methodology.
Nonetheless, the seismic performance of the double column structure can still be a concern,
since it is able to provide high lateral stiffness, in a large part, due to the robust cross-section of
the columns, resulting in a short free span length on the coupling beam. This raises some doubts
regarding the efficiency of the usual seismic design methodologies for bridge bents in the
current structure, since cap beams are often targeted for rigid behavior, aiming to mostly serve
as elements capable of enforcing similar displacement demand on all columns, and general
deformation compatibility between them. According to what was possible to assess, the
provision of the large column stiffness comes at the cost of not being able to design the beam
for that rigid behavior, since its geometrical limitations cannot be easily compensated by
reinforcement alone. As a consequence, some level of damage in the beam can be expected,
related with a significant shear demand that is difficult to address within an acceptable level of
ductility. Moreover, a very similar problem is also experienced by coupling beams of shear
walls with low shear span-to-depth ratios, having been a strong research field in the past few
decades, from which alternative design strategies can be evaluated for the current application.
Therefore, the complex interactions between beam and columns that could be understood
from the preliminary study of the double column bent pier structure constituted a design
challenge of particularly difficult nature, which was deemed better addressed by experimental
studies. For that purpose, a first stage 1:4 reduced scale testing was performed at LESE on four
distinct monolithic pier models, which had the same reinforcement in the columns, but different
in the beams. In that regard, all the specimens revealed shear critical behavior on the beams,
where the conventional design layout based on horizontal reinforcement and vertical stirrups
8.2
Conclusions
showed particularly brittle failure. As far as it was possible to assess, that is a consequence of
the large strains that occur at the beam, and of the cyclic degradation of the cracked concrete
surfaces. In addition, the beam-column interfaces were observed to concentrate most of the
deformations, and the reported collapse modes could be linked with concrete failure in those
same locations. In fact, the high level of damage experienced by all specimens was largely tied
to those occurrences, because once the compression failure threshold is achieved, damage
progression is significantly more pronounced. Likewise, it can be argued that up until that point,
the level of damage experienced by the piers could be repaired easily, but it would require a
significantly bigger effort if further loading was considered, particularly for the specimens that
showed the most sudden failures. Furthermore, the rhombic truss based reinforcement layouts
were those that provided arguably the best results. However, it should also be stated that, in
order to deal with the large demand in the coupling beam, all the resulting reinforcement layouts
are required to be quite strong, particularly those that were inspired by EC8 or ACI318, which
can lead to significant difficulties considering real-scale construction. In fact, this issue would
most likely cause any precast application based on such a design to not be economically
competitive.
A conclusion that could be assessed from the previous results was that the large
deformation demand at the beam level is related to the general inability of forming stable energy
dissipating mechanisms located on the columns rather than the beams (since significant damage
occurs earlier on the latter), therefore preventing solid mobilization of the flexural capacity of
the columns. In light of those results, the precast specimens were designed aiming to forcefully
induce earlier column yielding, by reducing moment continuity reinforcement on the beam-
column section by nearly 50%, seeking a reduction of the overall beam deformations by
providing a more desirable alternative energy dissipation mechanism. This was performed along
with the adoption of a precast system based on a top-down assembly of an integral beam
element similar to a cap beam, descending onto two previously placed columns. Thus, a second
stage of experimental assessment was devised, focused on evaluating the effects of three
different connection detailing layouts, designed for the purpose of decreasing construction
complexity.
All the precast double column pier specimens performed reasonably well, reaching higher
drifts, with larger energy dissipation, and the intended design goal of promoting increased
deformations at the precast joint for earlier column yielding was clearly achieved. In fact,
comparison of two equivalent and identically reinforced specimens, where the only variation
was the adoption of precast construction on one of them (requiring the respective joints),
showed noticeable differences and, in particular, considerably improved behavior with the
8.3
Conclusions
precast specimen. Nonetheless, even if that strategy enabled larger ductility capacity, the
collapse mode of all the specimens was still related to shear deformations occurring at the beam,
particularly at the beam-column vertical interfaces, where most of the structural damage was
observed. Taking that into consideration and, if similar performance can be ensured with any of
the tested systems, then the one that is associated with less operational constraints or increased
flexibility in the construction procedures can be sustained as the preferred alternative.
Regardless, despite the precast specimens having showed globally improved cyclic behavior,
the level of damage observed at the last few drift cycles was still significant and, as mentioned
before, related with the shear demands on the beam.
The experimental study also included a third testing stage, focused on evaluating precast
mechanisms for the foundation-to-column connections that are required for column bases. This
part of the study was not developed to the same extent as for the upper part of the frame but,
nonetheless, two precast connection systems were designed and tested. One of those was based
on a precast construction of the column element with protruding rebars and second phase
casting of the footing while holding the column above it. The other was also based on
constructing the column with protruding rebars, but considering their introduction into existing
foundations through duct holes. In this regard, the former was able to exhibit acceptable
performance, even if slightly worse in nearly every way (ductility, energy dissipation and failure
mode), when compared to a monolithic specimen. By contrast, the latter showed significant
weaknesses and undesirable failure modes that could be linked to the precast design itself,
particularly the grout crushing that resulted in early rebar buckling and fracture. As far as it was
possible to assess, that was mostly due to a design misconception, since grout was observed to
be inadequate for the intended purpose. Unfortunately, it was not possible to repeat this test due
to the project constraints, but it is believed that better performance was achievable if an
adequate micro-concrete mix was used instead.
The information obtained during the experimental campaign is undoubtedly one of the
strongest features of the present work, and it is the author’s belief that it provides an interesting
perspective on the framework of HSRL bridge piers subjected to seismic loading. In that regard,
in order to further understand the experimental observations and to extend conclusions of that
initiative to a wider range of similar applications, several numerical simulations were made,
based on 2D FEM applications. A two stage approach was considered, wherein a first set of
numerical analyses were performed for the purpose of experimental validating the modeling
strategy, and a second set of analyses for evaluating other scenarios not experimentally studied,
namely full length columns with clamped base supports, and shorter piers. In that regard, the
modeling strategy was successful in simulating several of the experimental tests, enabling
8.4
Conclusions
confirmation of several assumptions regarding experimental observations, such as the overall

importance of beam deformations on the global behavior of the double column pier structure or
the role of the precast joint in the overall ductility increase of the structural system.
Furthermore, it enabled validation of the test setup by providing similar results between the half
pier and full pier analyses. Likewise, it was also possible to confirm that most of the phenomena
observed on the experimental campaign (e.g. large beam deformations, mild cracking at the
columns) are also as relevant on other untested scenarios, since similar response features can be
identified therein for equivalent capacity, from which it was possible to conclude that the
overall double column behavior can be scaled according to the associated lateral seismic
moment.
The final part of the work aimed at using the previously analyzed test data to calibrate
numerical applications, for evaluation of the global behavior of idealized HSRL viaducts
involving some of the tested pier systems. Essentially, this required characterizing the moment
capacity of piers within a numerical viaduct simulation through use of lumped plasticity 2D
models, where the inelastic behavior was defined according to the experimental lateral forces,
scaled for different application heights, and set only at the base of each pier. In this regard, only
two scenarios were considered: one for a monolithic construction and one for a precast
construction, calibrated from the two assumed best test results for each category. The
methodology used for the assessment of the viaducts’ performance was based on well
established performance based methodologies, but due to the inherent limitations of the present
study, only the response variability was taken into account by considering different ground
motion records. Furthermore, three Damage Measures were evaluated, one related with the
structural performance of the viaducts and the other two with service criteria relative to
performance on the framework of HSRL bridges, namely lateral deflection and derailment
conditions. In this regard, the results were evaluated by means of IDA analyses, from which
fragility curves were determined, which essentially showed that within this scope there is a large
probability of the required safety and service conditions to be met for the analyzed structures.
Nonetheless, it is acknowledged that the adopted methodology is not sufficiently refined to
extract those conclusions in a more definitive manner, namely due to lack of, for example,
soil-structure interaction or DM limits dispersion data, although that was also never its main
purpose.
On the scope of all that was presented, it is a firm belief of the author that the main
objectives of this work were fairly accomplished, and that some of the results achieved present
an interesting alternative for precast applications on these structures, with respect to the rapid
construction of bridges.
8.5
Conclusions
8.2. FUTURE DEVELOPMENTS
Concerning future developments, it is not possible to address this subject without first
reinforcing the importance of the SIPAV research project in the development of the present
work. In fact, that project aimed to study new opportunities for precast solutions while
envisioning application on the first part of the foreseen Portuguese HSRL network: the
Poceirão-Caia project; the political and economical junctures at that time were highly favorable
to the establishment of HSRL in Portugal, which naturally provided a demand for improved
knowledge that the engineering community sought to address. That mindset has changed,
unfortunately, throughout the time period between the beginning and conclusion of this work,
since the deep economic crisis that the country had been facing in the past few years radically
modified public and institutional perceptions of the relative worth of HSRL on the global
context of the national economy.
Despite the above described, the data that was obtained during this work certainly still
holds valuable scientific information that is relevant on its own, particularly since the seismic
behavior of HSRL bridges is not a very frequently addressed topic by the community.
Furthermore, while a Portuguese HSRL network may be out of consideration for the foreseeable
future, the discussion on precast design and overall performance of these structures may be
applicable for other endeavors.
Therefore, some ideas that could further improve the work presented in this document are
enlisted as follows:
▪ Economic assessment of the viability of the tested systems would be a very

valuable feature, enabling the extraction of matured conclusions regarding the
practical outcome of seeking precast applications in comparison with the common
CIP construction alternatives;
▪ For budget reasons and time constraints, the third stage of the experimental
campaign was quite limited in both its scope and number of tested precast systems.
An extension of this stage to involve additional design options and construction
methods would likely help refining the knowledge on that particular subject and,
possibly, improving the outlook over the foundation-to-element connections. At
the very least, considering the weaknesses that were identified on the model most
suited for pre-fabrication, repeating that test using a different precast connection
material would be desirable;
▪ Considering the high level of damage experienced by all the double column tested
specimens, it would be fairly important to evaluate alternative construction features
8.6
Conclusions
aiming to improve this condition, such as, for example, using fiber reinforced
concrete; Additionally, definition of adequate performance thresholds for
repairable and non-repairable damage levels could help with evaluating the
effectiveness of repair measures;
▪ The global behavior of the double column structures was observed to be highly
commanded by either the shear level in the beam or the bending moments in the
columns/precast joint. Due to that, the capacity prediction models used throughout
this work were somewhat simple, and in some cases have not been able to provide
satisfactory results. A stronger understanding of the relative importance of beam
shear and column bending on the overall pier capacity could certainly help on
improving capacity prediction and, ultimately, providing better informed design
decisions;
▪ The performance assessment numerical study that is included in Chapter 7 of the
current work was mainly developed aiming for the comparison between the two
monolithic and precast scenarios, which is why the conclusions extracted from that
section may hold valuable information despite the simplicity of the approach.
Nonetheless, considering further applications and an eventual need to develop
more in depth knowledge on that subject, there are several aspects of the adopted
methodology that could be improved. For example, future works could seek to
include soil-structure interaction, or at least different scenarios for soil flexibility
due to how it may affect both the forces and the displacements recorded during an
earthquake event. Additionally, if a refined performance assessment study is
sought, then the performance levels for the selected damage measures would likely
need to be better reviewed, including their EDP limits and the associated
dispersion data.
8.7
REFERENCES
References
A
ACI 318 (2014). "Building Code Requirements for Structural Concrete (ACI 318-14)"
Farmington Hills, MI 48331,American Concrete Institute (ACI)
Alinaitwe, H. M., Mwakali, J. A. and Hansson, B. (2006). "Assessing the degree of

industrialisation in construction." Journal of Civil Engineering and Management 12(3): 221-
229
Altavia (2009). "BAFO ". Concessão RAV Poceirão - Caia. Linha Ferroviária de Alta
Velocidade. Eixo Lisboa - Madrid. Troço Poceirão - Caia
ANSR (2013). "Código da Estrada". Autoridade Nacional de Segurança Rodoviária. Decreto-

Lei nº 72/2013
AREMA (2003). "Railway Track Design"
B
Baker, J. W. (2015). "Efficient Analytical Fragility Function Fitting Using Dynamic Structural
Analysis." Earthquake Spectra 31(1): 579-599
Berry, M. P. and Eberhard, M. (2003). "Performance Models for Flexural Damage in

Reinforced Concrete Columns". Pacific Earthquake Engineering Research Center Report.Pacific
Earthquake Engineering Research Center
Bfer, G. (1985). "An isoparametric joint/interface element for finite element analysis."
International Journal for Numerical Methods in Engineering 21(4): 585-600
Billington, S. L., Barnes, R. W. and Breen, J. E. (1999a). "A precast segmental substructure
system for standard bridges." PCI Journal 44(4): 56-73.
Billington, S. L., Barnes, R. W. and Breen, J. E. (1999b). "A precast substructure design for
standard bridge systems". Research Report 1410-2.Center for Transportation Research. Austin,
Texas.
Billington, S. L., Barnes, R. W. and Breen, J. E. (2001). "Alternate substructure systems for
standard highway bridges." Journal of Bridge Engineering 6(2): 87-94.
Billington, S. L. and Yoon, J. K. (2004). "Cyclic response of unbonded posttensioned precast

columns with ductile fiber-reinforced concrete." Journal of Bridge Engineering 9(4): 353-363.
Biskinis, D. and Fardis, M. N. (2010). "Deformations at flexural yielding of members with

continuous or lap-spliced bars." Structural concrete 11(3): 127-138.
Boqueho, G. (2002). "SNCF's TGV Safety System". International Railway Safety Council,
Tokyo.
R.1
References
C
California Department of Transportation (2000). “Bridge Design Specifications Manual:7 -
Substructures"
Calçada, R., Delgado, P. and Matos, A. C. (2008). "Bridges for High-Speed Railways". Revised
Papers from the Workshop, Porto, Portugal, 3 - 4 June 2004. R. Calçada, P. Delgado and A. C.
Matos, CRC Press.
Canbolat, B. A., Parra-Montesinos, G. J. and Wight, J. K. (2005). "Experimental study on

seismic behavior of high-performance fiber-reinforced cement composite coupling beams." ACI
Structural Journal 102(1): 159-166
Cantagallo, C., Camata, G. and Spacone, E. (2014). "Seismic Demand Sensitivity of Reinforced
Concrete Structures to Ground Motion Selection and Modification Methods." Earthquake
Spectra 30(4): 1449-1465
Carmichael, A. and Desrosiers, N. (2008). "Comparative Highway Bridge Design".Faculty of

the Worcester Polytechnic Institute.
Castilho, V. C. and Lima, M. C. V. (2012). "Comparative Costs of the Production, Transport

and Assembly Stages of Prestressed Precast Slabs using Genetic Algorithms." International
Journal of Optimization in Civil Engineering 2(3): 407-422.
CEB (1993). "CEB Bulletin 217". Selected Justification Notes - Bond and Detailing CEB.
Lausanne, Switzerland
Couchard, I. and Detandt, H. (2003). "Viaducts for high-speed trains consisting of a series of
skew portal frames and intermediate simply supported girders". IABSE Symposium, Antwerp
Costa, C. (2009). "Análise Numérica e Experimental do Comportamento Estrutural de Pontes

em Arco de Alvenaria de Pedra". Porto, Tese de Doutoramento, Faculdade de Engenharia da
Universidade do Porto
Cruz Lesbros, L. F., Carranza Aubry, R. and Reinoso Angulo, E. (2003). "Design and
construction of the Ayuntamiento 2000 Bridge, Cuernavaca, Mexico." PCI Journal 48(4): 58-65.
Cutillas, A. (2007). "Track-bridge interaction problems in bridge design". Track-bridge

interaction on High-Speed railways. FEUP. Porto, Portugal.
D
Davis, P., Janes, T., Eberhard, M. and Stanton, J. (2012). "Unbonded Pre-Tensioned Columns
for Bridges in Seismic Regions". PEER Report.Pacific Earthquake Engineering Research
Center.Department of Civil and Environmental Engineering of University of Washington
Delgado, P. (2009). "Avaliação da Segurança Sísmica de Pontes". Porto, Tese de

Doutoramento, Faculdade de Engenharia da Universidade do Porto
R.2
References
Delgado, P., Monteiro, A., Arêde, A., Vila Pouca, N., Delgado, R. and Costa, A. (2011).
"Numerical simulations of RC hollow piers under horizontal cyclic loading." Journal of
Earthquake Engineering 15(6): 833-849
Delgado, P., Monteiro, A., Arêde, A., Vila Pouca, N., Delgado, R. and Costa, A. (2011).
"Numerical simulations of RC hollow piers under horizontal cyclic loading." Journal of
Earthquake Engineering 15(6): 833-849
Delgado, R., Delgado, P., Vila Pouca, N., Arêde, A., Rocha, P. and Costa, A. (2009). "Shear
effects on hollow section piers under seismic actions: Experimental and numerical analysis."
Bulletin of Earthquake Engineering 7(2): 377-389
Design Standards for Railway Structures and Commentary (2007a). "Seismic Design" Railway
Bureau of the Ministry of Land, Infrastructure and Transport Government of Japan
Design Standards for Railway Structures and Commentary (2007b). "Displacement Limits"
Railway Bureau of the Ministry of Land, Infrastructure and Transport Government of Japan
DGF. (2007). "Instrucción de acciones a considerar en puentes de ferrocarril (IAPF)".

Gobierno de España Ministerio de Fomento, Secretaria General Técnica Ministerio de Fomento
Dutoit, D., Wouts, I. and Martin, D. (2004). "Seismic Design of Structures in the French
Mediterranean and Asian High Speed Railway Lines". Bridges for High-Speed Railways, Porto,
Portugal.
Dutoit, D. (2007). "New Evolutions for High Speed Rail Line Bridge Design Criteria and
Corresponding Design Procedures". Track-bridge interaction on High-Speed railways. FEUP.
Porto, Portugal
E
Elgawady, M. and Sha'lan, A. (2011). "Seismic Behavior of Self-Centering Precast Segmental
Bridge Bents." Journal of Bridge Engineering 16(3): 328–339.
Eligehausen, R., Popov, E. and Bertero, V. (1982). "Local bond stress-slip relationships of
deformed bars under generalized excitations".Universitätsbibliothek der Universität Stuttgart.
Stuttgart
EN1990 (2002). "Eurocode 0: Basis of structural design" Brussels,European Committee for

Standardization (CEN)
EN1990-1, +A1:2005 (2002). "Eurocode - Basis of structural design" Brussels,European

Committee for Standardization (CEN)
EN1991-2 (2003). "Eurocode 1: Actions on Structures - Part 2: Traffic Loads on Bridges"

Brussels,European Committee for Standardization (CEN)
R.3
References
EN1998-1 (2004) "Eurocode 8: Design of structures for earthquake resistance. Part 1: General
rules, seismic actions and rules for buildings" Brussels,European Committee for
Standardization (CEN)
EN1998-2 (2005). "Eurocode 8: Design of structures for earthquake resistance. Part 2:

Bridges" Brussels,European Committee for Standardization (CEN)
F
Fardis, M. N., Kolias, V., Panagiotakos, T. B., Katsaras, C. and Psychogios, T. (2012). "Guide
for bridge design with emphasis on seismic aspects"
Faria, R. (1994). "Avaliação do Comportamento Sísmico de Barragens de Betão Através de um

Modelo de Dano Contínuo". Porto, Tese de Doutoramento. Faculdade de Engenharia da
Universidade do Porto
Faria, R. and Oliver, J. (1993). "A rate dependent plastic-damage constitutive model for large
scale computations in concrete structures.". CIMNE Monograph, Barcelona
fib (2008). "fib Bulletin 43 Structural connections for precast concrete buildings”.fib
(fédération international du béton). Switzerland
Fichoux, E. (2011). "Présentation et Utilisation de Cast3M." from http://www-

cast3m.cea.fr/html/Documentation_Cast3M/Presentation_Cast3M.pdf
Figueiredo, H., Calçada, R. and Delgado, R. (2009a). "Aspectos Regulamentares no Cálculo de

Pontes Ferroviárias em Linhas de Alta Velocidade"
Figueiredo, H., Calçada, R. and Delgado, R. (2009b). "Soluções Estruturais para Pontes
Ferroviárias em Linhas de Alta Velocidade"
Fouad, F. H., Rizk, T. and Stafford, E. L. (2006). "A Prefabricated Precast Concrete Bridge
System for the State of Alabama". UTCA Report 05215.University Transportation Center for
Alabama (UTAC)
Freeby, G., Medlock, R. and Slage, S. (2003). "Prefabricated Bridge Innovations". Second New
York City Bridge Conference, New York
G
Galano, L. and Vignoli, A. (2000). "Seismic behavior of short coupling beams with different
reinforcement layouts." ACI Structural Journal 97(6): 876-885
Gibbens, B. and Smith, P. S. (2004). "Design-Construction of Sorell Causeway Channel Bridge,

Hobart, Tasmania." PCI Journal 10(4): 56-66
R.4
References
Goicolea, J. M. (2007). "Service Limit States for Railway Bridges in New Design Codes IAPF
and Eurocodes". Track-Bridge Interaction on High-Speed Railways
Grant, D. N. and Diaferia, R. (2013). "Assessing adequacy of spectrum-matched ground motions

for response history analysis." Earthquake Engineering & Structural Dynamics 42(9): 1265-
1280
Günav, M. and Mosalam, K. (2012). "PEER Performance Based Earthquake Engineering

Methodology, Revisited". 15th World Conference on Earthquake Engineering, Lisboa
H
Haber, Z. B., Saiidi, M. S. and Sanders, D. H. (2013). "Precast Column-Footing Connections
for Accelerated Bridge Construction in Seismic Zones".Reno University of Nevada.Center for
Civil Engineering Earthquake Research
Haraldsson, O., Janes, T., Eberhard, M. and Stanton, J. (2013). "Seismic Resistance of Socket
Connection between Footing and Precast Column." Journal of Bridge Engineering 18(9): 910-
919
Harajli, M., Hamad, B. and Rteil, A. (1995). "Effect of Confinement on Bond Strength between
Steel Bars and Concrete." Structural Journal 101(5)
Haselton, C. B., Liel, A. B. and Deierlein, G. G. (2009). "Simulating structural collapse due to
earthquakes: Model idealization, model calibration and numerical solution algorithms".
COMPDYN - Computational Methods in Structural Dynamics and Earthquake Engineering,
Rhodes, Greece
Hewes, J. T. and Priestley, M. J. N. (2002). "Seismic Design and Performance of Precast

Concrete Segmental Bridge Columns". Technical Report UCSD/SSRP-2001/25.State of
California Department of Transportation
Hieber, D. G., Wacker, J. M., Eberhard, M. O. and Stanton, J. F. (2005). "Precast Concrete Pier
Systems for Rapid Construction of Bridges in Seismic Regions".Washington State
Transportation Center (TRAC).Washington State Transportation Commission
Hognestad, E., University of, I., Engineering Experiment, S., Reinforced Concrete Research, C.
and Engineering, F. (1951). "A study of combined bending and axial load in reinforced concrete
members". Urbana, University of Illinois
Hose, Y. and Seible, F. (1999). "Performance Evaluation Database for Concrete Bridge
Components and Systems under Simulated Seismic Loads". PEER Report.Pacific Earthquake
Engineering Research Center.Department of Structural Engineering of University of California,
San Diego
Hoshikuma, J.-i., Unjoh, S. and Sakai, J. (2009). "Seismic Performance and Structural Details
of Precast Segmental Concrete Bridge Columns". 25th US - Japan Bridge Engineering
Workshop, Tsukuba, Japan
R.5
References
Huang, Z. and Engstrom, B. (1996). "Experimental and analytical studies of the bond behaviour
of deformed bars in high strength concrete.". 4th international symposium on utilization of
high-strength/high-performance concrete, Paris, France
I
Ibarra, L. F., Medina, R. A. and Krawinkler, H. (2005). "Hysteretic models that incorporate
strength and stiffness deterioration." Earthquake Engineering and Structural Dynamics 34(12):
1489-1511
Iervolino, I. and Manfredi, G. (2008). "A Review of Ground Motion Record Selection Strategies
for Dynamic Structural Analysis". Modern Testing Techniques for Structural Systems:
Dynamics and Control. Oreste S. Bursi and David Wagg. Vienna, Springer Vienna: 131-163
Ile, N. and Reynouard, J. (2000). "Nonlinear analysis of reinforced concrete shear wall under
earthquake loading." Journal of Earthquake Engineering 4(2): 183-213
International Building Code (2009). "International Building Code" .U.S.A., International Code
Council
J
Jason, L. and Durand, S. (2007). "A two-surface plastic model for concrete behaviour." Revue
Européenne de Génie Civil 11(5): 579-602
K
Kent, D. C. and Park, R. (1971). "Flexural members with confined concrete." Journal of the
Structural Division 97(7): 1969-1990
Khaleghi, B. (2005). "Use of precast concrete members for accelerated bridge construction in
Washington State". 6th International Bridge Engineering Conference: Reliability, Security, and
Sustainability in Bridge Engineering, Boston, MA
Konishi, J. (2012). "Breakthrough in Japanese Railways 10 - Railways and Bridges 2." Japan
Railway and Transport Review(59): 48 - 55
Koyama, Y. (1997). "Railway Technology Today 1: Railway Construction in Japan." Japan

Railway & Transport Review(14): 36-41
Kumar Subedi, N. (1991a). "RC-coupled shear wall structures. I. Analysis of coupling beams."
Journal of structural engineering New York, N.Y. 117(3): 667-680
R.6
References
Kumar Subedi, N. (1991b). "RC coupled shear wall structures. II. Ultimate strength
calculations." Journal of structural engineering New York, N.Y. 117(3): 681-698
Kunnath, S. K., Larson, L. and Miranda, E. (2006). "Modelling considerations in probabilistic

performance-based seismic evaluation: case study of the I-880 viaduct." Earthquake
Engineering & Structural Dynamics 35(1): 57-75
L
Lignos, D. G. (2008). "Sidesway Collapse of Deteriorating Structural Systems Under Seismic
Excitations". Department of Civil and Environmental Engineering, Stanford University
Lignos, D. G. and Krawinkler, H. (2012). "Development and Utilization of Structural

Component Databases for Performance-Based Earthquake Engineering." Journal of Structural
Engineering: 1382-1394
Luo, X. (2005). "Study on methodology for running safety assessment of trains in seismic design
of railway structures." Soil Dynamics and Earthquake Engineering(25): 79-91
Luo, X. and Miyamoto, T. (2008). "Examining the Adequacy of the Spectral Intensity Index for
Running Safety Assessment of Railway Vehicles during Earthquakes". 14th World Conference
on Earthquake Engineering, Beijing, China
M
Macedo, L., Araújo, M. and Castro, J. M. (2013). "Assessment and calibration of the harmony
search algorithm for earthquake record selection". Vienna Congress on Recent Advances in
Earthquake Engineering and Structural Dynamics, Vienna, Austria
Macedo, L. and Castro, J. M. (2016). "SelEQ: An advanced ground motion selection and
scaling tool." Computer-Aided Civil and Infrastructure Engineering (under submission)
Mander, J. B., Priestley, M. J. N. and Park, R. (1988). "Theoretical Stress‐Strain Model for
Confined Concrete." Journal of Structural Engineering 114(8): 1804-1826
Mander, J. B., Dhakal, R. P., Mashiko, N. and Solberg, K. M. (2007). "Incremental dynamic
analysis applied to seismic financial risk assessment of bridges." Engineering Structures 29(10):
2662-2672
Manrique, J. D., Hussein, M. A., Telyas, A. and Funston, G. (2007). "Case Study-Based
Challenges of Quality Concrete Finishing for Architecturally Complex Structures." Journal of
Construction Engineering and Management 133(3): 208-216
R.7
References
Manterola, J. and Cutillas, A. M. (2004). "Prestressed Concrete Railway Bridges". Bridges for
High Speed Railways, Porto
Marioni, A. (2006). "Bearing Systems for High Speed Railway Bridges". 6th Joints, Bearings
and Seismic Systems for Concrete Structures, Halifax (Canada)
Marques, M. (2011). "Avaliação Probabilística da Segurança Sísmica de Edifícios". Porto, Tese

de Doutoramento, Faculdade de Engenharia da Universidade do Porto
Marsh, L. (2011). "Application of Accelerated Bridge Construction Connections in Moderate-

to-High Seismic Regions", Transportation Research Board
Marsh, M., Stanton, J. and Eberhard, M. (2010). "A precast bridge bent system for seismic
regions". Fully Precast Bridge Bents for Use in Seismic Regions. FHWA
Matsumoto, E. E., Waggoner, M. C., Kreger, M. E., Vogel, J. and Wolf, L. (2008).
"Development of a precast concrete bent-cap system." PCI Journal 53(3): 74-99
Mclean, D., Kuebler, S. and Mealy, T. (1998). "Seismic Performance and Retrofit of Multi-
Column Bridge Bents". Research Project T9902 - Multi-Column Retrofit - Experiments.
Washington State Department of Transportation
McKenna, F., Fenves, G. and Fillipou, F. (2010). "OpenSees." from http://www-

cast3m.cea.fr/html/Pegon/elemjoint.pdf
Menegotto, M. and Pinto, P. (1973). "Method of analysis for cyclically loaded reinforced
concrete plane frames including changes in geometry and non-elastic behaviour of elements
under combined normal force and bending.". Resist. Ultimate Deform. of Struct. Acted on by
Well-Defined Repeated Loads, Final Report.IABSE Symposium. Lisbon
Millanes Mato, F. (2004). "Composite and Prestressed Concrete Solutions for Very Long
Viaducts: Analysis of Different Structural Designs for the Spanish High Speed Lines". Bridges
for High Speed Railways, Porto
Millanes Mato, F. and Ortega Cornejo, M. (2007). "Track-Structure Interaction and Seismic
Design of the Bearings System for Some Viaducts of Ankara-Istanbul HSRL Project". Track-
Bridge Interaction on High-Speed Railways, FEUP, Porto
Moehle, J. P. (2003). "A framework for performance-based earthquake engineering". ATC-15-9

Workshop on the Improvement of Building Structural Design and Construction Practices, Maui,
HI
Moehle, J. P., Ghodsi, T., Hooper, J. D., Fields, D. C. and Gedhada, R. (2012). "Seismic Design
of Cast-in-Place Concrete Special Structural Walls and Coupling Beams: A guide for
Practicing Engineers". NEHRP Seismic Design Technical Brief No. 6.NEHRP Consultants
Joint Venture.U.S. Department of Commerce
Monso, G. M. and Miranda, E. (2013). "Performance-Based Seismic Assessment of a Recently

Built High-Speed Rail Viaduct in Spain. The Archidona Viaduct". 7th National Seismic
Conference on Bridges and Highways, Oakland, CA
R.8
References
Monteiro, A., Arêde, A. and Vila Pouca, N. (2017a). "Comparative Analysis of Monolithic vs.
Precast Solutions for Bent-Type Reinforced Concrete Bridge Piers Under Cyclic Loading:
Experimental Assessment (Under review)." Bulletin of Earthquake Engineering.
Monteiro, A., Arêde, A. and Vila Pouca, N. (2017b). "Seismic behavior of coupled column
bridge RC piers: Experimental campaign." Engineering Structures 132: 399-412
Monteiro, A. (2009). "Simulação Numérica e Validação Experimental do Comportamento

Cíclico de Pilares Ocos de Pontes de Betão Armado". Porto, Tese de Mestrado, Faculdade de
Engenharia da Universidade do Porto
Monteiro, R. (2011). "Probabilistic Seismic Assessment of Bridges". Porto, Tese de

Doutoramento, Faculdade de Engenharia da Universidade do Porto.
Montenegro, P. (2015). "A Methodology for the Assessment of the Train Running Safety on
Bridges", Tese de Doutoramento, Faculdade de Engenharia da Universidade do Porto.
Mostafa, K., Sanders, D. H. and Saiidi, M. S. (2004). "Impact of Aspect Ratio on Two-Column
Bent Seismic Performance".California Department of Transportation.Center for Earthquake
Engineering Research, Department of Civil and Environmental Engineering, University of
Nevada
N
Newmark, N. (1959). "A Method of Computation for Structural Dynamics." ASCE Journal of
Engineering Mechanics Division 85(EM3)
NP EN1992-1-1 (2010). "Eurocódigo 2: Projecto de estruturas de betão Parte 1-1: Regras

gerais e regras para edifícios." Bruxelas,Comité Europeu de Normalização (CEN)
NP EN1998-1 (2010). "Eurocódigo 8: Projecto para estruturas para resistência aos sismos
Parte 1: Regras gerais, acções sísmicas e regras para edifícios." Bruxelas, Comité Europeu de
Normalização (CEN)
NCHRP (2003). "Prefabricated Bridge Elements and Systems to Limit Traffic Disruption
During Construction - A Synthesis of Highway Practice". NCHRP Synthesis 324.Transportation
Research Board.National Cooperative Highway Research Program. Washington, D.C
O
Ou, Y., Lee, G. C., Wang, P., Tsai, M. and Chang, K. (2008). "Large scale cyclic tests of
precast segmental concrete bridge columns with unbonded post-tensioning tendons". Bridge
Maintenance, Safety Management, Health Monitoring and Informatics. H.M. Koh and D. M.
Frangopol, Taylor & Francis
R.9
References
Ou, Y., Tsai, M., Chang, K. and Lee, G. C. (2010). "Cyclic behavior of precast segmental
concrete bridge columns with high performance or conventional steel reinforcing bars as
energy dissipation bars." Earthquake Engineering and Structural Dynamics 39(11): 1181-1198
Ottosen, N. (1977). "A Failure Criterion for Concrete"
P
Palermo, A., Pampanin, S. and Calvi, G. M. (2005). "Concept and development of hybrid
solutions for seismic resistant bridge systems." Journal of Earthquake Engineering 9(6): 899-
921
Palermo, A., Pampanin, S. and Marriott, D. (2007). "Design, modeling, and experimental
response of seismic resistant bridge piers with posttensioned dissipating connections." Journal
of Structural Engineering 133(11): 1648-1661
Pampanin, S. (2003). "Alternative design philosophies and seismic response of precast concrete
buildings." Structural Concrete 4(4): 203-211
Pang, J. B. K., Eberhard, M. O. and Stanton, J. F. (2010). "Large-bar connection for precast
bridge bents in seismic regions." Journal of Bridge Engineering 15(3): 231-239
Pantelides, C. P. and Gergely, J. (2008). "Seismic Retrofit of Reinforced Concrete Beam-

Column T-Joints in Bridge Piers with FRP Composite Jackets". SP-258: Seismic Strengthening
of Concrete Buildings
Park, Y.-J. and Ang, A. H. S. (1985). "Mechanistic Seismic Damage Model for Reinforced
Concrete." Journal of Structural Engineering 111(4): 722-739.
Parra-Montesinos, G. J., Wight, J. K. and Setkit, M. (2010). "Earthquake-Resistant Coupling

Beams Without Diagonal Reinforcement." Concrete International 32(12): 36-40
Paulay, T. (1971). "Coupling beams of reinforced concrete shear walls." Journal of Structural
Engineering 97(3): 843-862
Paulay, T. and Binney, J. R. (1974). "Diagonally reinforced coupling beams of shear walls".
Shear in Reinforced Concrete. Detroit, Michigan, American Concrete Institute: 579-598
Pegon, P. "Présentation des joints dilatants." from http://www-

cast3m.cea.fr/html/Pegon/elemjoint.pdf
PCI Bridge Design Manual (1997). "Fabrication and Construction"
Priestley, M., Calvi, G. and Kowalsky, M. (2007). "Displacement-Based Seismic Design of

Structures", IUSS Press.
Priestley, M. J. N. (1991). "Overview of PRESSS research program." PCI Journal 36(4): 50-57.
R.10
References
Proença, J., Romba, J., Barros Viegas, J. and Vieira, A. (2002). "Experimental Development
Stages of an Innovative Earthquake-Resistant Precast Frame System". 12nd European
Conference on Earthquake Engineering (12ECEE), London, UK
R
Richard, B. and Ragueneau, F. (2013). "Continuum damage mechanics based model for quasi
brittle materials subjected to cyclic loadings: Formulation, numerical implementation and
applications." Engineering Fracture Mechanics 98(0): 383-406
Richard, B., Ragueneau, F., Cremona, C. and Adelaide, L. (2010). "Isotropic continuum
damage mechanics for concrete under cyclic loading: Stiffness recovery, inelastic strains and
frictional sliding." Engineering Fracture Mechanics 77(8): 1203-1223
Rodrigues, H., Arêde, A., Varum, H. and Costa, A. (2013). "Experimental evaluation of
rectangular reinforced concrete column behaviour under biaxial cyclic loading." Earthquake
Engineering & Structural Dynamics 42(2): 239-259
Romão, X. (2012). "Deterministic and Probabilistic Methods for Structural Seismic Safety
Assessment". Porto, Faculdade de Engenharia da Universidade do Porto
S
Santos, R. (2012). "Avaliação da aderência aço-betão em elementos estruturais de B.A.".
Departamento de Engenharia Civil. Aveiro, Tese de Mestrado, Universidade de Aveiro.
Schlaich, M. (2012). "Die neuen Brücken der deutschen Bahn (english translation) ". 22
Dresdener Brückenbausymposium. Dresden, Germany
Schokker, A. J., West, J. S., Breen, J. E. and Kreger, M. E. (1999). "Interim Conclusions,
Recommendations and Design Guidelines for Durability of Post-Tensioned Bridge
Substructures".Center for Transportation Research.The University of Texas at Austin
Snyman, M. F., Bird, W. W. and Martin, J. B. (1991). "A Simple Formulation of a Dilatant
Joint Element Governed by Coulomb Friction." Engineering Computations 8(3): 215-229
Sobrino, J. A. and Murcia, J. (2007). "Structural Analisys of High Speed Rail Bridge
Substructures. Application to Three Spanish Case Studies". Track-Bridge Interaction on High-
Speed Railways, Porto, Portugal
Stanton, J., Eberhard, M. and Steuck, K. (2006). "Rapid construction details for bridges in
seismic zones". 22th US-Japan Bridge Engineering Workshop
R.11
References
T
Tamai, S. (2014). "Landscape Bridges - A Technique for a stable roadbed".
Kunskapsseminarium, Stockholm, Sweden
Tassios, T. P. (1979). "Properties of bond between concrete and steel under load cycles
idealizing seismic actions." Bulletin d’information du CEB,(131): 65-122
Tassios, T. P., Moretti, M. and Bezas, A. (1996). "On the behavior and ductility of reinforced
concrete coupling beams of shear walls." ACI Structural Journal 93(6): 711-720
Taylor, D. P. "Fluid Lock-Up Devices - A Robust Means to Control Multiple Mass Structural
Systems Subjected to Seismic or Wind Inputs". Inc. Taylor Devices. North Tonawanda, NY,
Taylor Devices, Inc.
Tegos, I. A. and Penelis, G. G. (1988). "Seismic Resistance of Short Columns and Coupling
Beams Reinforced with Inclined Bars." ACI Structural Journal 85(1): 82-88
Tran, V. (2012). "Drilled Shaft Socket Connections for Precast Columns in Seismic Regions",
University of Washington
U
UIC (2010). "High speed rail - Fast track to sustainable mobility"
V
Vamvatsikos, D. and Cornell, C. A. (2002). "Incremental dynamic analysis." Earthquake
Engineering & Structural Dynamics(31): 491-514
VanGeem, M. (2006). "Achieving Sustainability with Precast Concrete." PCI Journal 51(1): 42-
61
Vavel, A. (2004). "Une innovation pour les ouvrages ferroviaires". Construction

Moderne/Annuel Ouvrages d'art: 33-36
Vila Pouca, N. (2001). "Simulação Numérica da Resposta Sísmica de Elementos Laminares em

Betão Armado". Porto, Tese de Doutoramento, Faculdade de Engenharia da Universidade do
Porto
Vision2000 (1995). "Performance-based seismic engineering".SEAOC. Sacramento, CA
R.12
References
W
Wang, J., Ou, Y., Chang, K. and Lee, G. C. (2008). "Large-scale seismic tests of tall concrete
bridge columns with precast segmental construction." Earthquake Engineering and Structural
Dynamics 37(12): 1449–1465
Y
Yee, A. A. (1991). "Design considerations for precast prestressed concrete building structures
in seismic areas." PCI Journal 36(3): 40-55
Yee, A. A. P. (2001). "Social and environmental benefits of precast concrete technology." PCI
Journal 46(3): 14-19
Yee, A. A. P. and Eng, H. D. (2001). "Structural and economic benefits of precast/prestressed

concrete construction." PCI Journal 46(4): 34-43
Z
Zacher, M. and Baeßler, M. (2005). "Dynamic behavior of ballast on railway bridges". Advance
Course Dynamics of high speed railway bridges, Porto
Zareian, F. and Medina, R. A. (2010). "A practical method for proper modeling of structural
damping in inelastic plane structural systems." Computers & Structures 88(1–2): 45-53
WEB LINKS
1. http://www.railtemperature.com/ (Last visit on Jun 2015)
2. http://en.structurae.de/ (Last visit on Jun 2015)
3. http://www.lusas.com/ (Last visit on Jun 2015)
4. http://www.rff.fr/ (Last visit on Jun 2015)
5. http://www.nishimatsu.co.jp/ (Last visit on Jun 2015)
6. http://cms.asce.org/ (Last visit on Jun 2015)
7. http://www.ioa.fr/ (Last visit on Jun 2015)
8. http://horsost.blogs.upv.es/ (Last visit on Jun 2015)
9. http://www.vde8.de/ (Last visit on Jul 2015)
10. http://www.tucrail.be/ (Last visit on Jul 2015)
11. https://www.flickr.com (Last visit on Jul 2015)
12. http://www.pedelta.es/ (Last visit on Jul 2015)
13. http://www.lgvrhinrhone.com (Last visit on Sep 2015)
14. http://www.peikko.ca/ (Last visit on Mar 2016)
15. http://www.pfeifer.de/ (Last visit on Mar 2016)
16. https://www.smcon.co.jp/en/ (Last visit on Mar 2016)
17. http://www.uic.org/ (Last visit on Mar 2016)
R.13
References
18. http://www-cast3m.cea.fr/ (Last visit on Jan 2017)

19. http://www.csiamerica.com/products/sap2000 (Last visit on Jan 2017)
20. http://www.liebherr.com/ (Last visit on Feb 2017)
21. http://www.mageba.ch/ (Last visit on Feb 2017)
22. www.sika.pt (Last visit on Feb 2017)
R.14
ANNEXES
Annex A – Modeling Parameters
A
MODELING PARAMETERS
A.1. MOMENT - CURVATURE ANALYSES (CHAPTER 4 AND CHAPTER 5)
Table A.1 – Concrete modeling parameters for Moment – Curvature analyses

Chapter 5
Beton_uni Chapter 4
SP_PC02A SP_PC02B SP_PC02C SP_M02C SP_F01 SP_F02 SP_F03
YOUN (x109) 33.00 36.00 37.00 36.00 41.00 40.00 45.00 37.00
NU 0.20 0.20 0.20 0.20 0.20 0.20 0.20 0.20
STFC (x106) 38.00 37.00 45.00 49.00 51.00 47.00 57.00 47.00
EZER 0.0022 0.0022 0.0022 0.0022 0.0022 0.0022 0.0022 0.0022
STFT (x106) 2.90 3.00 3.40 3.10 3.00 2.80 3.50 3.70
ALF1 0 0 0 0 0 0 0 0
OME1 0 0 0 0 0 0 0 0
ZETA 200.00 200.00 200.00 200.00 200.00 200.00 200.00 200.00
ST85 0 0 0 0 0 0 0 0
TRAF 20.00 20.00 20.00 20.00 20.00 20.00 20.00 20.00
FACL 0.75 0.75 0.75 0.75 0.75 0.75 0.75 0.75
FAMX 20.00 20.00 20.00 20.00 20.00 20.00 20.00 20.00
STPT 0 0 0 0 0 0 0 0
RHO 2400 2400 2400 2400 2400 2400 2400 2400
FAM1 2.00 2.00 2.00 2.00 2.00 2.00 2.00 2.00
FAM2 20.00 20.00 20.00 20.00 20.00 20.00 20.00 20.00
A.1
Table A.2 – Steel modeling parameters for Moment – Curvature analyses

Chapter 5
Acier_uni Chapter 4
SP_PC02A SP_PC02B SP_PC02C SP_M02C SP_F01 SP_F02 SP_F03
YOUN (x109) 200.00 200.00 200.00 200.00 200.00 200.00 200.00 200.00
NU 0.30 0.30 0.30 0.30 0.30 0.30 0.30 0.30
STSY (x106) 500.00 504.00 504.00 504.00 504.00 504.00 504.00 504.00
EPSU 0.2500 0.2500 0.2500 0.2500 0.2500 0.2500 0.2500 0.2500
STSU (x106) 580.00 590.00 590.00 590.00 590.00 590.00 590.00 590.00
ESPH 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001
RHO 2400 7850 7850 7850 7850 7850 7850 7850
FALD 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00
A6FA 620 620 620 620 620 620 620 620
CFAC 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50
AFAC 0.006 0.006 0.006 0.006 0.006 0.006 0.006 0.006
ROFA 20.00 20.00 20.00 20.00 20.00 20.00 20.00 20.00
BFAC (x10-3) 6.15 6.15 6.15 6.15 6.15 6.15 6.15 6.15
A1FA 18.50 18.50 18.50 18.50 18.50 18.50 18.50 18.50
A2FA 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15
A.2
A.2. REFINED 2D FEM ANALYSES (CHAPTER 6)
Table A.3 – Concrete Modeling Parameters for 2D FEM analyses – Regular Concrete
Chapter 6
Damage_tc
SP_M02 SP_M03 SP_M04 SP_PC02A SP_PC02B SP_F01
YOUN (x109) 36.00 36.00 40.00 36.00 37.00 40.00
NU 0.20 0.20 0.20 0.20 0.20 0.20
RHO 2400.00 2400.00 2400.00 2400.00 2400.00 2400.00
GVAL 250.00 290.00 240.00 200.00 220.00 180.00
FTUL (x106) 3.80 4.40 3.60 3.00 3.40 2.80
REDC 0 0 0 0 0 0
FC01 (x106) -20.50 -17.00 -21.50 -17.00 -20.00 -23.50
RT45 1.20 1.20 1.20 1.20 1.20 1.20
FCU1 (x106) -51.00 -43.00 -54.00 -42.50 -51.50 -58.50
STRU (x10-3) -15.00 -15.00 -15.00 -15.00 -15.00 -15.00
EXTP (x10-3) -5.00 -5.00 -5.00 -5.00 -5.00 -5.00
STRP (x106) -43.00 -38.00 -45.00 -38.00 -46.00 -52.00
EXT1 (x10-3) -2.30 -2.30 -2.30 -2.20 -2.30 -2.40
STR1 (x106) -48.00 -40.00 -50.00 -40.00 -49.00 -55.00
EXT2 (x10-3) -4.50 -4.50 -4.80 -4.50 -4.50 -4.80
STR2 (x106) -48.00 -40.00 -50.00 -40.00 -49.00 -55.00
NCRI 2 2 2 2 2 2
A.3
Table A.4 – Concrete Modeling Parameters for 2D FEM analyses – High Ductility Concrete
Chapter 6
Damage_tc
SP_M02 SP_M03 SP_M04 SP_PC02A SP_PC02B SP_F01
YOUN (x109) 36.00 36.00 40.00 36.00 37.00 40.00
NU 0.20 0.20 0.20 0.20 0.20 0.20
RHO 2400.00 2400.00 2400.00 2400.00 2400.00 2400.00
GVAL 250.00 290.00 240.00 200.00 220.00 180.00
FTUL (x106) 3.80 4.40 3.60 3.00 3.40 2.80
REDC 0 0 0 0 0 0
FC01 (x106) -20.50 -17.00 -21.50 -17.00 -20.00 -23.50
RT45 1.20 1.20 1.20 1.20 1.20 1.20
FCU1 (x106) -51.00 -43.00 -54.00 -42.50 -51.50 -58.50
STRU (x10-3) -25.00 -25.00 -25.00 -25.00 -25.00 -25.00
EXTP (x10-3) -8.00 -8.00 -8.00 -8.00 -8.00 -8.00
STRP (x106) -43.00 -38.00 -45.00 -38.00 -46.00 -52.00
EXT1 (x10-3) -2.30 -2.30 -2.30 -2.20 -2.40 -2.40
STR1 (x106) -45.00 -37.00 -46.50 -37.00 -45.50 -52.00
EXT2 (x10-3) -7.30 -7.30 -7.30 -7.20 -7.30 -7.30
STR2 (x106) -45.00 -37.00 -46.50 -37.00 -45.50 -52.00
NCRI 2 2 2 2 2 2
A.4
Table A.5 – Steel Modeling Parameters for 2D FEM analyses

Acier_uni Chapter 6 - All specimens
Acier_ancrage Column reinforcement Stirrups Beam Reinforcement
9
YOUN (x10 ) 190.00 190.00 180.00
NU 0.30 0.30 0.30
STSY (x106) 504.00 504.00 601.00
EPSU 0.2500 0.2500 0.2500
STSU (x106) 590.00 590.00 702.00
ESPH 0.001 0.001 0.003
RHO 7850 7850 7850
FALD 4.00 4.00 4.00
A6FA 620 620 620
CFAC 0.50 0.50 0.50
AFAC 0.006 0.006 0.006
ROFA 20.00 20.00 20.00
BFAC (x10-3) 6.15 6.15 6.15
A1FA 18.50 18.50 18.50
A2FA 0.15 0.15 0.15
Table A.6 – Bond - Slip Modeling Parameters for 2D FEM analyses

Chapter 6
Acier_ancrage
SP_M02 SP_M03 SP_M04 SP_PC02A SP_PC02B
s1 (mm) 0.79 0.86 0.79 0.90 0.82
s2 (mm) 2.37 2.57 2.37 2.70 2.45
s3 (mm) 7.59 8.21 7.59 8.64 7.84
τpeak (MPa) 16.77 16.01 17.32 15.21 16.77
τres (MPa) 6.71 6.40 6.93 6.08 6.71
Table A.7 – Joint Modeling Parameters for 2D FEM analyses
Joint_Dilatant SP_PC02A SP_PC02B
Kn (MPa/mm) 65.30 47.60
Ks (MPa/mm) 18.00 19.10
A.5
A.3. GLOBAL 2D BRIDGE IDA ANALYSES (CHAPTER 7)
Table A.8 – Modified Ibarra-Medina-Krawinkler modeling parameters
Backbone Hysteretic behavior and degradation
Parameters Monolithic Precast Parameters Monolithic Precast
K0 124487653 123677379 FprPos 0.35 0.25
as_Plus 0.002 0.007 FprNeg 0.35 0.25
as_Neg 0.002 0.007 A_pinch 0.00 0.20
My_Plus 66567 42000 Lamda_S 0.32 1.00
My_Neg -66567 -42000 Lamda_C 0.32 1.00
theta_p_Plus 0.015 0.012 Lamda_A 0.32 1.00
theta_p_Neg 0.015 0.012 Lamda_K 0.32 1.00
theta_pc_Plus 0.050 0.600 c_S 1.00 1.00
theta_pc_Neg 0.050 0.600 c_C 1.00 1.00
Res_Pos 0.200 0.200 c_A 1.00 1.00
Res_Neg 0.200 0.200 c_K 1.00 1.00
theta_u_Plus 0.180 0.900 D_Plus 1.00 1.00
theta_u_Neg 0.180 0.900 D_Neg 1.00 1.00
A.6
Annex B – Mesh Density
B
MESH DENSITY
Following the modeling strategy presented in Chapter 6, focused on using 2D finite

elements under a plane stress assumption for simulation of the double column pier tests, a mesh
efficiency study was developed, aiming to define an adequate mesh density for balanced result
quality and calculation times. The adopted methodology firstly considered characterization of
the main 2D pier dimensions using around 10 eight-node rectangular finite elements, which
would lead to targeted mesh densities of 0.035m and 0.055m, regarding beam height and
column width, respectively. Those values served as guidance for the establishment of a wider
range of mesh density values, enabling the definition of the following four different models,
wherein constant mesh density is considered:
▪ Model A: Maximum element size of the mesh of 1.00cm;

▪ Model B: Maximum element size of the mesh of 2.50cm;
▪ Model C: Maximum element size of the mesh of 5.00cm;
▪ Model D: Maximum element size of the mesh of 10.00cm;
In addition, two plane stress analyses were performed with every mesh, regarding
evaluation under both monotonic and cyclic loading conditions, using the same constitutive
models that were presented for the main applications in Chapter 6. Considering that high finite
element strains are expected for reasonable lateral displacements (due to, for example, low
concrete tensile capacity), the loading increments and element size should ideally be calibrated
in order to guarantee that the non-linear response calculated at the gauss point level is described
by a sufficient number of values. Furthermore, accurate numerical simulations should also take
element size into account in the definition of the constitutive properties, in order to reduce mesh
dependency. Correspondingly, fine and coarse meshes should be analyzed under different
loading conditions, in order to guarantee adequate result quality and reliability. However, if that
B.1
were the case in this study, then direct comparison would not be possible, since every model
would require a different loading history and constitutive parameters. Within this context, it was
decided that accounting for constant loading step and modeling parameters was the best
approach, which can lead to varying degrees of convergence difficulty on different models.
Nonetheless, it also enables comparison in a straightforward manner, which was deemed
suitable for this task.
With that in mind, the current study focused on the following premises:
▪ Base FEM meshes representing the same specimen (SP_M02);

▪ The same loading histories were adopted for the four models, using a displacement
increment of 0.2mm;
▪ A maximum displacement limit of 50 mm was considered for both the monotonic
and cyclic analyses (roughly corresponding to a 2.00% drift level) in order to
prevent the occurrence of large non-linear incursion and near-collapse effects,
aiming to avoid the resulting convergence problems;
The previously described strategy guarantees that all the models were subjected to the same
exact number of loading steps, which enables direct comparison of calculation times. The
respective results are summarized in Table B.1.
Table B.1 – Mesh refinement calculation times
Parameters Monotonic Cyclic

(250 loading increments) (2272 loading increments)
Model 1 2 3 4 1 2 3 4
Mesh density
1.0 2.5 5.0 10.0 1.0 2.5 5.0 10.0
(cm)
Mesh size
31620 4958 1264 368 31620 4958 1264 368
(Number of 2D elements)
Total calculation time

N/A 96 21 20 N/A 443 94 64
(min)
Avg. time per element
N/A 1.16 1.00 3.26 N/A 5.36 4.46 10.43
(sec)
Avg. time per increment
N/A 23.04 5.04 4.80 N/A 11.70 2.48 1.69
(sec)
The first thing that comes to mind by analyzing Table B.1 is that no results were available
for Model A. That is because its model size was too large due to the fine mesh density, leading
to the inability to complete the analyses under the wall clock time of 20 hours, as set on the
computer cluster. In addition, average calculation time per element and average calculation time
per loading increment can serve as an efficiency measurement of the analyses. In that regard,
considering that all the analyses used the same loading histories, it is not surprising that Model
B.2
D exhibited the smallest average time per loading increment (4.80 and 1.69 seconds for
monotonic and cyclic analyses, respectively), as it was also the smallest model. Despite that, the
relative increase in calculation time regarding Model C is less than 50% for cyclic analyses, and
only 5% for monotonic analyses, which is important to note since Model C is around 250%
larger than Model D. The same efficiency cannot be observed with Model B, however, as the
decrease in the mesh density parameter leads to an exponentially larger increase in mesh size,
which slows down the overall analysis time by a considerable amount (380% and 600% for
monotonic and cyclic loading conditions, respectively), when compared to Model D times. On
another note, Model D actually shows the lowest values regarding calculation time per element,
with an average value that is between two to three times larger than the equivalent value in
Models B and C. That means that although the model is considerably small (since it is built
using the larger 0.10m 2D elements), its mesh is not refined enough to capture the local
deformations of this structure, requiring more iterations per load increment to reach a balanced
state than the more refined alternatives. As it stands, Model C displayed the fastest calculation
times per element, followed closely by Model B.
It is also interesting to note that the average calculation time parameters indicate higher
efficiency for cyclic analyses, which is curious since they typically involve more complex
behavior, leading to more lengthy iterations for each loading increment. In reality, that
observation could be related to the target displacement limit set for the analyses, which
corresponds to a demand level equivalent to a loading stage between yielding and peak
response. In that case, it also means that elastic loading and unloading constitute a considerable
part of the overall analyses, where the computation of FEM relevant quantities is easier.
Figure B-1 illustrates the Force – Displacement plots obtained from the monotonic and
cyclic results of the analyses. As expected, numerical results between the three models are
generally equivalent until the onset of cracking and, particularly, yielding. When that occurs, the
finer mesh models (B and C) exhibit a slight force reduction when in comparison the coarse
mesh model (D). Nonetheless, the overall differences between all the models are small, and
even smaller between Models B and C.
B.3
250 250
Model B Model B
Model C 200 Model C
Model D Model D
200 150
100
Force (kN)
Force (kN)
150 50
0
100 -50
-100
50 -150
Cracking Yielding -200
Cracking Yielding
0 -250
0 10 20 30 40 50 60 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60
Displacement (mm) Displacement (mm)
a) Monotonic loading b) Cyclic loading

Figure B-1 – Force-Displacement relations for the mesh refinement analyses
With that in mind, Mesh C was considered to enable the best compromise between result
quality and computational effort and was, therefore, adopted as the basis for the analyses whose
results are presented in Chapter 6.
B.4
Annex C – IDA Output Data
C
INCREMENTAL DYNAMIC
ANALYSES
C.1. GROUND MOTIONS

Magnitude (Mw)
SelEQ scaling
Duration (s)
Sa1 (g) - A
Sa1 (g) - D
Sa1 (g) - B
Sa1 (g) - C
parameter
PHA (g)
Year
Name
1-ImperialValley-06.dat 1979 6.5 0.29 38.96 0.60 0.65 0.67 0.76 0.28
2-ChalfantValley-02.dat 1986 6.2 0.28 21.98 3.40 0.42 0.49 0.39 0.24
3-TaiwanSMART1(40).dat 1986 6.3 0.34 29.00 1.84 0.76 0.68 0.64 0.41
4-LomaPrieta.dat 1989 6.9 0.22 39.64 0.79 0.76 0.59 0.64 0.35
5-Chi-ChiTaiwan.dat 1999 7.6 0.23 90.00 3.61 0.56 0.62 0.69 0.26
6-Northridge-01.dat 1999 7.6 0.19 40.00 0.33 0.55 0.49 0.37 0.40
7-TaiwanSMART1(40).dat 1986 6.3 0.29 29.13 1.13 0.53 0.35 0.38 0.37
8-Chi-ChiTaiwan-05.dat 1999 7.6 0.26 86.00 2.90 0.45 0.49 0.61 0.29
11-SuperstitionHills-02.dat 1987 6.5 0.39 22.12 2.50 0.49 0.61 0.75 0.37
15-LomaPrieta.dat 1989 6.9 0.34 39.95 3.77 0.57 0.79 0.69 0.45
16-BigBear-01.dat 1992 6.5 0.24 60.01 2.70 0.45 0.40 0.34 0.31
17-Northridge-01.dat 1999 7.6 0.23 40.00 2.69 0.56 0.46 0.56 0.27
18-Northridge-01.dat 1994 6.7 0.25 40.00 2.86 0.59 0.76 0.79 0.35
20-VictoriaMexico.dat 1980 6.3 0.24 15.58 3.44 0.54 0.40 0.60 0.27
A - Macheda Viaduct
B - Palheta Viaduct
C - Viaduct over Degebe River
D - Viaduct over Ribeira da Laje
C.1
0.40 0.40
0.30 0.30
0.20 0.20
0.10 0.10
PGA (g)
PGA (g)
0.00 0.00
-0.10 0 10 20 30 40 -0.10 0 5 10 15 20 25
-0.20 -0.20
-0.30 -0.30
-0.40 -0.40
Duration (s) Duration (s)
1-ImperialValley-06 2-ChalfantValley-02
0.40 0.40
0.30 0.30
0.20 0.20
0.10 0.10
PGA (g)
PGA (g)
0.00 0.00
-0.10 0 5 10 15 20 25 30 -0.10 0 10 20 30 40
-0.20 -0.20
-0.30 -0.30
-0.40 -0.40
3-TaiwanSMART1(40) 4-LomaPrieta
0.40 0.40
0.30 0.30
0.20 0.20
0.10 0.10
PGA (g)
PGA (g)
0.00 0.00
-0.10 0 20 40 60 80 -0.10 0 10 20 30 40
-0.20 -0.20
-0.30 -0.30
-0.40 -0.40
5-Chi-ChiTaiwan 6-Northridge-01
0.40 0.40
0.30 0.30
0.20 0.20
0.10 0.10
PGA (g)
PGA (g)
0.00 0.00
-0.10 0 5 10 15 20 25 30 35 -0.10 0 20 40 60 80
-0.20 -0.20
-0.30 -0.30
-0.40 -0.40
7-TaiwanSMART1(40) 8-Chi-ChiTaiwan-05
0.40 0.40
0.30 0.30
0.20 0.20
0.10 0.10
PGA (g)
PGA (g)
0.00 0.00
-0.10 0 10 20 30 40 -0.10 0 10 20 30 40 50 60
-0.20 -0.20
-0.30 -0.30
-0.40 -0.40
9-ImperialValley-06 10-Chi-ChiTaiwan
C.2
0.40 0.40
0.30 0.30
0.20 0.20
0.10 0.10
PGA (g)
PGA (g)
0.00 0.00
-0.10 0 5 10 15 20 25 -0.10 0 20 40 60 80
-0.20 -0.20
-0.30 -0.30
-0.40 -0.40
11-SuperstitionHills-02 12-Chi-ChiTaiwan-04
0.40 0.40
0.30 0.30
0.20 0.20
0.10 0.10
PGA (g)
PGA (g)
0.00 0.00
-0.10 0 20 40 60 80 -0.10 0 20 40 60 80 100
-0.20 -0.20
-0.30 -0.30
-0.40 -0.40
13-Chi-ChiTaiwan.dat 14-Chi-ChiTaiwan-06
0.40 0.40
0.30 0.30
0.20 0.20
0.10 0.10
PGA (g)
PGA (g)
0.00 0.00
-0.10 0 10 20 30 40 -0.10 0 10 20 30 40 50 60
-0.20 -0.20
-0.30 -0.30
-0.40 -0.40
15-LomaPrieta 16-BigBear-01
0.40 0.40
0.30 0.30
0.20 0.20
0.10 0.10
PGA (g)
PGA (g)
0.00 0.00
-0.10 0 10 20 30 40
-0.10 0 10 20 30 40
-0.20 -0.20
-0.30 -0.30
-0.40 -0.40
17-Northridge-01 18-Northridge-01
0.40 0.40
0.30 0.30
0.20 0.20
0.10 0.10
PGA (g)
PGA (g)
0.00 0.00
-0.10 0 20 40 60 80 100 -0.10 0 5 10 15 20
-0.20 -0.20
-0.30 -0.30
-0.40 -0.40
19-ImperialValley-06 20-VictoriaMexico
C.3
C.2. DAMAGE MEASURE 1: STRUCTURAL PERFORMANCE
0.050 0.050
16th fractile 16th fractile
EDP - Peak Hinge Rotation (rad)
0.045 50th fractile

0.045 50th fractile
0.040 0.040
0.035 0.035
0.030 0.030
0.025 0.025
0.020 0.020
0.015 0.015
0.010 0.010
0.005 0.005
0.000 0.000
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Intensity Measure, Sa1 (g) Intensity Measure, Sa1 (g)

0.050 0.050

0.045 0.045
0.040 0.040
0.035 0.035
0.030 0.030
0.025 0.025
0.020 0.020
0.015 0.015
0.010 0.010
0.005 50th fractile 0.005 50th fractile
0.000 0.000
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Figure C.1 – Hinge Rotation IDA curves – Monolithic Model
C.4
0.050 0.050

EDP - Peak Hinge Rotation (rad) 84th fractile 84th fractile
0.040 0.040
0.035 0.035
0.030 0.030
0.025 0.025
0.020 0.020
0.015 0.015
0.010 0.010
0.005 0.005
0.000 0.000
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Intensity Measure, Sa1 (g) Intensity Measure, Sa 1 (g)

0.050 0.050
0.045 0.045

0.040 0.040
0.035 0.035
0.030 0.030
0.025 0.025
0.020 0.020
0.015 0.015
0.010 0.010
0.000 0.000
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Figure C.2 – Hinge Rotation IDA curves – Precast Model
C.5
C.3. DAMAGE MEASURE 2: LATERAL DEFLECTION
0.008 0.008
EDP - Peak Angular Rotation (rad)

0.006 0.006
0.005 0.005
0.004 0.004
0.003 0.003
0.002 0.002
0.001 0.001
0.000 0.000
0.0 1.0 2.0 3.0 4.0 5.0 6.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0

0.008 0.008

0.007 0.007
0.006 0.006
0.005 0.005
0.004 0.004
0.003 0.003
0.002 0.002
0.000 0.000
0.0 1.0 2.0 3.0 4.0 5.0 6.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0

Figure C.3 – Angular Rotation IDA curves – Monolithic Model
C.6
0.008 0.008

0.006 0.006
0.005 0.005
0.004 0.004
0.003 0.003
0.002 0.002
0.001 0.001
0.000 0.000
0.0 1.0 2.0 3.0 4.0 5.0 6.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0

0.008 0.008

0.007 0.007
0.006 0.006
0.005 0.005
0.004 0.004
0.003 0.003
0.002 0.002
0.000 0.000
0.0 1.0 2.0 3.0 4.0 5.0 6.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0
Intensity Measure, Sa1 (g) Intensity Measure, Sa 1 (g)

Figure C.4 – Angular Rotation IDA curves – Precast Model
C.7
C.4. DAMAGE MEASURE 3: DERAILMENT CONDITIONS
14.000 14.000
13.000 16th fractile 16th fractile
13.000
10.000 10.000
EDP - SI (m)
EDP - SI (m)
9.000 9.000
8.000 8.000
7.000 7.000
6.000 6.000
5.000 5.000
4.000 4.000
3.000 3.000
2.000 2.000
1.000 1.000
0.000 0.000
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

14.000 14.000
13.000 16th fractile 13.000
12.000 50th fractile 12.000
11.000 84th fractile 11.000
10.000 10.000
EDP - SI (m)
9.000 9.000
EDP - SI (m)
8.000 8.000
7.000 7.000
6.000 6.000
5.000 5.000
4.000 4.000
3.000 3.000 16th fractile
0.000 0.000
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

Figure C.5 – Spectral Intensity IDA curves – Monolithic Model
C.8
14.000 14.000
10.000 10.000
EDP - SI (m)
EDP - SI (m)
9.000 9.000
8.000 8.000
7.000 7.000
6.000 6.000
5.000 5.000
4.000 4.000
3.000 3.000
2.000 2.000
1.000 1.000
0.000 0.000
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

14.000 14.000
13.000 16th fractile 13.000
12.000 50th fractile 12.000
11.000 84th fractile 11.000
10.000 10.000
EDP - SI (m)
EDP - SI (m)
9.000 9.000
8.000 8.000
7.000 7.000
6.000 6.000
5.000 5.000
4.000 4.000
3.000 16th fractile
3.000
0.000 0.000
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

Figure C.6 – Spectral Intensity IDA curves – Precast Model
C.9

Seismic Behavior of Precast Piers On High Speed Railway Bridges

Uploaded by

Copyright:

Available Formats

Seismic Behavior of Precast Piers On High Speed Railway Bridges

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Seismic Behavior of Precast Piers On High Speed Railway Bridges

Uploaded by

Copyright:

Available Formats

D 2017

SEISMIC BEHAVIOR OF PRECAST PIERS ON

ANDRÉ MANUEL BESSA MONTEIRO

Supervisor: António José Coelho Dias Arêde

This work was also supported by: Project POCI-01-0145-FEDER-007457 – CONSTRUCT -

To Professor António Arêde, for accepting me as a member of the LESE research

Thank you all!

Ainda assim, sendo já vasta a experiência de aplicação de sistemas prefabricados em

A informação experimental é posteriormente utilizada para calibrar ferramentas refinadas

1. INTRODUCTION .................................................................................................. 1.1

1.1. BACKGROUND OVERVIEW ....................................................................................... 1.1

1.2. MOTIVATION AND OBJECTIVES................................................................................. 1.3

1.3. THESIS OUTLINE..................................................................................................... 1.4

2. PIERS FOR HSRL BRIDGES AND VIADUCTS ................................................... 2.1

2.1. INTRODUCTION ....................................................................................................... 2.1

2.2. COMMON RAILWAY BRIDGE LAYOUTS ...................................................................... 2.2

2.3. RAILWAY BRIDGE SEISMIC DESIGN .......................................................................... 2.5

2.3.1. General remarks ............................................................................................ 2.5

2.3.2. Railway Performance Requirements ............................................................ 2.10

2.3.3. Seismic Load ............................................................................................... 2.13

2.4. PIER TYPES ON RAILWAY BRIDGES ........................................................................ 2.15

2.4.1. Overview ..................................................................................................... 2.15

2.4.2. Single Columns ........................................................................................... 2.17

2.4.3. Wall-Piers .................................................................................................... 2.22

2.4.4. Multiple Column Piers .................................................................................. 2.25

2.5. FINAL REMARKS ................................................................................................... 2.31

3. PRECAST TECHNOLOGY FOR BRIDGE PIERS ............................................... 3.1

3.1. ADVANTAGES AND DISADVANTAGES OF PRECAST ..................................................... 3.1

3.2. PRECAST CONNECTIONS......................................................................................... 3.4

3.2.1. Foundation-to-element Connections .............................................................. 3.5

3.2.2. Element-to-element Connections ................................................................. 3.11

3.3. BRIDGE PIER PRECAST SYSTEMS AND DESIGN ....................................................... 3.13

3.3.1. Layouts for non-seismic zones .................................................................... 3.14

3.3.2. Layouts for seismic zones............................................................................ 3.19

3.4. FINAL REMARKS ................................................................................................... 3.27

4.2. PIER SEISMIC DESIGN............................................................................................. 4.5

4.2.1. General Considerations and Design Methodology ......................................... 4.5

4.2.3. Design strategy............................................................................................ 4.14

4.2.4. Case Study Viaducts ................................................................................... 4.18

4.2.5. Reinforcement and Detailing Layouts .......................................................... 4.24

4.2.6. Precast Conception ..................................................................................... 4.30

4.3. TEST SPECIMENS ................................................................................................. 4.43

4.3.1. Phase 1 – Monolithic Specimens ................................................................. 4.45

4.3.2. Phase 2 – Precast Specimens ..................................................................... 4.52

4.3.3. Phase 3 – Foundation-to-element connections ............................................ 4.58

4.4. FINAL REMARKS ................................................................................................... 4.62

5. EXPERIMENTAL CAMPAIGN.............................................................................. 5.1

5.1. TEST SETUP........................................................................................................... 5.1

5.1.1. Phase 1 and Phase 2 – Bent Pier Tests ........................................................ 5.1

5.1.2. Phase 3 - Single Column Tests ................................................................... 5.16

5.2. EXPERIMENTAL RESULTS – BENT PIER TESTS ........................................................ 5.18

5.2.1. Phase 1 Observations ................................................................................. 5.18

5.2.2. Phase 2 Observations ................................................................................. 5.35

5.2.3. Results Discussion ...................................................................................... 5.51

5.3. EXPERIMENTAL RESULTS – SINGLE COLUMN TESTS ............................................... 5.67

5.3.1. Phase 3 Observations ................................................................................. 5.67

5.3.2. Result Discussion ........................................................................................ 5.73

5.4. FINAL REMARKS ................................................................................................... 5.80

6. NUMERICAL SIMULATIONS ............................................................................... 6.1