POLITECNICO DI TORINO

Master’s Degree in Civil Engineering

Master’s Degree Thesis

Cyclic Non-linear Analysis of Complex Reinforced Concrete

Structures with GLRC_HEGIS Model

Supervisors Candidate
Prof. Rosario CERAVOLO
Michele MASSARELLA
Dir. HDR Silvano ERLICHER
Ph.D. Miquel HUGUET AGUILERA

Academic year: 2019/2020

Summary

The following master’s thesis was written during a period of foreign mobility, with a duration of
5 months, which was spent at Egis, a French group leader in the engineering sector. In particular,
the internship was carried out in the Egis Industries (nuclear department of EGIS group), within
the EAS (Egis Advanced Solutions) team. The internship aimed at validating GLRC_HEGIS
nonlinear constitutive model developed by Miquel Huguet Aguilera for the software Code_Aster,
by means of a several simulations aimed at reproducing the behaviour of real case studies.
This constitutive law is able to reproduce the main nonlinear phenomena that characterise the
behaviour of RC (reinforced concrete) elements: concrete damage, concrete cracking, bond-slip
stress and yielding of steel reinforcement bars. The validation of the model has been carried out
through three main analyses: (i) comparison with experimental results of reinforced concrete walls
subjected to horizontal cyclic loads in plane shear, (ii) the comparison with other non-linear global
models available in Code_Aster and (iii) comparison with experimental results of reinforced
concrete elements subjected to pure out-of-plane bending. The first considered experimental test
includes the four walls designed and tested in the framework of the French national research project
CEOS.fr (Behaviour and Assessment of Special Structures. Cracking - Shrinkage). Shear-wall
specimens were designed in order to simulate the behaviour of the nuclear power plant walls
submitted to the seismic action. The four walls are subjected to horizontal cyclic loading in
plane shear (reversing for three walls and non-reversing for one) and differ from each other or
by the strength class of the concrete used, or by the amount of reinforcement or by the type of
load applied. The numerical results obtained were compared with the available experimental
ones, in terms of force-displacement curves and crack opening on the walls surface. Then, the
GLRC_HEGIS global model was compared to other global models for RC plates available in
Code_Aster (GLRC_DM, DHRC and EIB) already implemented in Code_Aster and has been
analysed the case of cyclic pure flexion test. In particular, the behaviour of a reinforced concrete
plate subjected to constant cyclic bending moment was studied, implementing the results already
obtained with two different load cases. At first, the number of cycles was increased from one to
six and subsequently an increasing moment was applied. The comparison with the other models
mainly concerned the energy dissipated during the load cycles and the applied rotation-moment
curves. To extend the results obtained so far, the model was used to simulate an experimental test
of a three-point RC beam subjected to pure bending. The numerical results have been compared
with the experimental ones and important conclusions have been reached. In particular, by
studying both behaviours (membrane and flexural) it is noted that the general behaviour of the
model is governed mainly by the concrete damage. In fact, for all the models the same non-linear
parameters were used except those relating to the damage to the concrete. That means that
concrete damage is the only phenomena that behaves differently in membrane and in bending
solicitations. This aspect makes it possible to standardize the model and further eliminate the
small uncertainty that governs the choice of parameters that must be used.

2
 Alla mia famiglia,
ai miei amici
e a tutte le persone che credono in me.
Acknowledgements

Firstly I would thank Egis Industries and Politecnico di Torino for allowing me to carry out these
months of internship in which the following master thesis has been prepared. In particular a
special thanks to Director Silvano Erlicher and Miquel Huguet-Aguilera, my company supervisors,
who have always encouraged and followed me in my work. I especially thank my academic tutor
Prof. Rosario Ceravolo for supporting me during my thesis period. I thank all the guys who
have always helped me on this path with technical advice and suggestions, in particular a special
thank you to Olivier, Jean, Thibaut, David, Giuseppe and Nicola. Thank you again all the
guys of Egis Industries who welcomed me right away on this adventure and with whom I shared
pleasant moments inside and outside the work environment. An infinite thanks to my family who
has always supported me during these years and to all my friends, constant and indispensable
presence to better face any challenge.

4
Table of Contents

List of Tables 8

List of Figures 9

Acronyms 12

1 INTRODUCTION 13

2 GLRC_HEGIS model 14
2.1 Helmholtz free energy surface density [3] . . . . . . . . . . . . . . . . . . . . . . . 14
2.1.1 Helmholtz free energy surface density as a function of the state variables . 14
2.1.2 Thermodynamic forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2 Local constitutive laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.1 Concrete damage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.2 Cracking of the concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.3 Steel-concrete bond-slip . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.2.4 Yielding of steel reinforcement bars . . . . . . . . . . . . . . . . . . . . . . 23
2.3 Numerical implementation [2] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3.1 DKTG Finite Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3 CEOS.FR 25
3.1 Experimental test description [7, 8, 9, 10] . . . . . . . . . . . . . . . . . . . . . . 25
3.1.1 Experimental test results [11] . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2 Numerical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.1 Non-linear analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3 Comparison between experimental and numerical results . . . . . . . . . . . . . . 32
3.3.1 Crack opening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.2 Envelope curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3.3 Crack pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4 CODE_ASTER TESTS 47
4.1 Cyclic pure flexion test [12] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.1.1 Constant rotation over time . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.1.2 Incremental rotation over time . . . . . . . . . . . . . . . . . . . . . . . . 51
6
5 THREE-POINT BENDING TEST ON RC BEAM 53
5.1 Experimental test description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.1.1 Experimental test results . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.2 Numerical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.2.1 Non-linear analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.3 Comparison between experimental and numerical results . . . . . . . . . . . . . . 57
5.4 Study of mesh sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.5 Comparison with membrane behaviour model . . . . . . . . . . . . . . . . . . . . 61

6 CONCLUSION 66

A 67

B 69

7
List of Tables

3.1 Main characteristics of the specimens. . . . . . . . . . . . . . . . . . . . . . . . . 26

3.2 GLRC_HEGIS parameters for CEOS.fr RC walls. . . . . . . . . . . . . . . . . . 31

4.1 GLRC_HEGIS parameters for CAS_TEST. . . . . . . . . . . . . . . . . . . . . . 48

5.1 Mechanical characteristic of the concrete. . . . . . . . . . . . . . . . . . . . . . . 54

5.2 Mechanical characteristic of the steel. . . . . . . . . . . . . . . . . . . . . . . . . 54
5.3 GLRC_HEGIS parameters for RC beam. . . . . . . . . . . . . . . . . . . . . . . 56
5.4 GLRC_HEGIS parameters for RC beam (third model). . . . . . . . . . . . . . . 61

B.1 Parameters of GLRC_HEGIS model. . . . . . . . . . . . . . . . . . . . . . . . . . 69

B.2 Internal variables of the GLRC_HEGIS model. . . . . . . . . . . . . . . . . . . . 70

8
List of Figures

2.1 Uniaxial concrete response with damage evolution. . . . . . . . . . . . . . . . . . 17

2.2 Displacement discontinuity at cracks. . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3 Second crack in a RC panel: cracked RC panel geometry and relative orientation
between the two cracks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Relation between normal concrete stress at cracks gn and crack opening wn . . . . 20
2.5 Aggregate interlock considered by Walraven. . . . . . . . . . . . . . . . . . . . . . 20
2.6 Experimental cyclic aggregate interlock test vs. analytical results of Contact
Density Model [5]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.7 Gambanova aggregate interlock law [6]. . . . . . . . . . . . . . . . . . . . . . . . 21
2.8 Tangential-to crack concrete stress at cracks gt as a function of the tangential crack
displacement wt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.9 Transformation from crack openings to steel-concrete slip in (a) x direction and
(b) y direction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.10 Typical force-displacement curve for a RC beam with a displacement-controlled test. 23
2.11 Relationship between real and reference DKQG and DKTG elements. . . . . . . 24

3.1 Dimensions of the specimen [7] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3.2 Steel frame of the specimen [7] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3 Section in the middle of steel frame and of the specimen (a) and reinforcement and
part of formwork (b) [7]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.4 Measured load time histories: reversing cycling load (left), non reversing cycling
load (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.5 Arrangement of sensors for walls A, B, C and D respectively. . . . . . . . . . . . 27
3.6 Experimental global force-displacement curves. . . . . . . . . . . . . . . . . . . . 28
3.7 Experimental crack orientation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.8 FE model of the RC walls (a) and 3D visualization (b). . . . . . . . . . . . . . . 29
3.9 Real boundary conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.10 Modelled boundary conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.11 Experimental vs. numerical global force-displacement curves. . . . . . . . . . . . 32
3.12 Experimental vs. numerical crack opening for wall A at sensors D18, D19 and D20. 33
3.13 Experimental vs. numerical crack opening for wall B at sensors D1 to D11. . . . 35
3.14 Experimental vs. numerical crack opening for wall C at sensors D5, D6 and D7. . 36
3.15 Experimental vs. numerical crack opening for wall D at sensors D1 to D8. . . . . 38
3.16 Experimental vs. numerical envelope curves. . . . . . . . . . . . . . . . . . . . . . 39
3.17 Experimental vs. numerical crack pattern at F = ±900 kN. . . . . . . . . . . . . 40
3.18 Experimental vs. numerical crack pattern at F = ±1200 kN. . . . . . . . . . . . 40
3.19 Experimental vs. numerical crack pattern at F = ±1500 kN. . . . . . . . . . . . 41
9
3.20 Experimental vs. numerical crack pattern at F = ±1800 kN. . . . . . . . . . . . 41
3.21 Experimental vs. numerical crack pattern at F = ±2100 kN. . . . . . . . . . . . 42
3.22 Experimental vs. numerical crack pattern at F = ±2400 kN. . . . . . . . . . . . 42
3.23 Experimental vs. numerical crack pattern at F = ±2700 kN. . . . . . . . . . . . 43
3.24 Experimental vs. numerical crack pattern at F = ±3000 kN. . . . . . . . . . . . 43
3.25 Experimental vs. numerical crack pattern at F = ±3300 kN. . . . . . . . . . . . 44
3.26 Experimental vs. numerical crack pattern at F = ±3600 kN. . . . . . . . . . . . 44
3.27 Experimental vs. numerical crack pattern at F = ±3900 kN. . . . . . . . . . . . 45
3.28 Experimental vs. numerical crack pattern at F = ±4200 kN. . . . . . . . . . . . 45
3.29 Numerical values of the residual cracks after F = ±4200 kN. . . . . . . . . . . . . 46

4.1 Geometry of the considered RC plate. . . . . . . . . . . . . . . . . . . . . . . . . 47

4.2 FE model and boundary conditions. . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.3 Loading history with constant rotation. . . . . . . . . . . . . . . . . . . . . . . . 49
4.4 Moment-rotation comparison. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.5 Dissipated energy comparison. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.6 GLRC_HEGIS: Crack opening evolution (a) and dissipated energy components. 50
4.7 Loading history with constant rotation. . . . . . . . . . . . . . . . . . . . . . . . 51
4.8 GLRC_HEGIS: Moment-rotation (a) and dissipated energy (b) comparison. . . . 51
4.9 GLRC_HEGIS: Crack opening evolution (a) and dissipated energy components (b). 52

5.1 Test layout. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

5.2 Beam’s section (a) and sensors arrangement (b). . . . . . . . . . . . . . . . . . . 54
5.3 Measured load time history. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.4 Global experimental force-displacement curves at half (a), one quarter and three-
fourth of the length (b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.5 Experimental steel deformation given by sensors J1-J6. . . . . . . . . . . . . . . . 55
5.6 FE model of the RC beam. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.7 Global experimental vs. numerical force-displacement curves at half (a), one
quarter and three-fourth of the length (b). . . . . . . . . . . . . . . . . . . . . . . 57
5.8 Experimental vs. numerical steel deformation given by sensors J1-J6. . . . . . . . 57
5.9 Trend of the main internal variables and dissipated energy components. . . . . . 58
5.10 Second FE model of the beam. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.11 Global experimental vs. numerical force-displacement curves at half (a), one
quarter and three-fourth of the length (b) for the second model. . . . . . . . . . . 59
5.12 Experimental vs. numerical steel deformation given by sensors J1-J6 for the second
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.13 Trend of the main internal variables (a) and dissipated energy components (b) for
the second model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.14 Third FE model of the RC beam. . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.15 Global experimental vs. numerical force-displacement curves at half (a), one
quarter and three-fourth of the length (b) for the third model. . . . . . . . . . . 62
5.16 Experimental vs. numerical steel deformation given by sensors J1-J6 for the third
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.17 Trend of the main internal variables (a) and dissipated energy components (b). . 63
5.18 Experimental vs. numerical crack pattern at F = 90 kN (Cycle 5). . . . . . . . . 64
5.19 Experimental vs. numerical crack pattern at F = 150 kN (Cycle 6). . . . . . . . 64
5.20 Experimental vs. numerical crack pattern at F = 185 kN (Cycle 7). . . . . . . . 64
5.21 Numerical values of the crack openings at the end of test. . . . . . . . . . . . . . 65
10
Acronyms

DHRC
Dissipative Homogenised Reinforced Concrete
DKTG
Discrete Kirchhoff Triangle Generalized

EIB
Endommagement Isotrope Beton

FE
Finite Elements

GLRC
Global Reinforced Concrete Models

RC
Reinforced Concrete

12
Chapter 1

INTRODUCTION

Egis is a major international group in the construction engineering and mobility services sectors
whose unique global service range encompasses infrastructure consulting, engineering and op-
eration. Through its capacity for innovation, it responds to the climate emergency and to the
greatest challenges of our time by offering solutions and acknowledged know-how in the areas of
transportation and mobility, sustainable city construction, buildings, water, the environment and
energy.
The need for innovation in non-linear calculations and the need to have models capable of
describing their behaviour when submitted to seismic action, led Egis to develop a non-linear
global model capable of simulating the global behaviour of reinforced concrete shell elements
under cyclic load. During my internship I had the opportunity to learn and use this model and
thus to expand the company’s knowledge about it. In the first period of my internship I analysed
and subsequently improved the results obtained by one of my predecessors on the study of a
reinforced concrete plate subjected to horizontal cyclic loads within the French CEOS.fr program
in order to simulate the behaviour of the nuclear power plant walls submitted to the seismic
action. Subsequently, the study was extended to analyse the behaviour of the model in the case
of pure bending: at first the model was compared with other non-linear global model available in
Code_Aster through simple tests made available directly by the code developer, extending the
results already obtained to longer and increaser loading cycles. Afterwards, the comparison was
made with a three-point experimental bending test emphasizing the behaviour of the model in
the case of pure bending and analysing how each individual non-linear parameter affects it.
The finite element code used was developed by EDF and is mainly used for the analysis of
structures and thermomechanics for studies and research. Code_Aster [1] is an open source free
software and therefore it allows to add / develop new functions and for this reason it was possible
to implement the GLRC_HEGIS model in it. It is also developed in the Python environment
and this allows to operate with greater ease especially in the post processing phase.

13
Chapter 2

GLRC_HEGIS model

GLRC_HEGIS is a global constitutive model developed by Miquel HUGUET-AGUILERA [2] in

order to describe the response of reinforcement concrete buildings submitted to seismic actions.
To limit the computational time it’s necessary to use plate and shell element to model concrete
and reinforcement steel bars (global modelling). In order to use the aforementioned model, it is
necessary to define a series of linear and non-linear parameters capable of taking into account all
the phenomena considered in the model.
In particular, this model currently takes into consideration four non-linear phenomena:

• Concrete damage
• Cracking of the concrete
• Steel-concrete bond slip
• Yielding of steel reinforcement bars

The global behaviour of the model is described by 18 internal variables that become 47 if
we consider the parametric variables (top and bottom layer), the variables related to energy
dissipation and those of post-treatment. The all intern variables are listed in Appendix B.
The constitutive model is formulated in the framework of Thermodynamics of Irreversible
Processes. At the beginning it has been obtained using an analytical homogenization of a RC
panel submitted to in-plane forces considering only one crack direction; afterwards, the model
has been extended to take into account a second crack direction and finally out-of-plane bending
moments (Figure 2.3).
The Helmholtz free energy surface density and the phenomena taken into consideration by
the model will be described below.

2.1 Helmholtz free energy surface density [3]

2.1.1 Helmholtz free energy surface density as a function of the state
variables
The Helmholtz free energy surface density defines the elastic part of the material behaviour and
depends on the state variables of the model:
• Observable variables, which define the linear elastic behaviour:
14
 GLRC_HEGIS model

– The generalized membrane strain Ô, composed of three components Ôxx , Ôyy and Ôxy
– The generalized curvature κ, composed of three components κxx , κyy and κxy
• Internal variables, which describe the non-linear behaviour of the model:
– The crack displacement w = (wn , wt ) in the normal (crack opening) wn and in the
tangential wt directions with respect to the crack:
∗ For the first family of cracks of the top layer: wt1
∗ For the second family of cracks of the top layer: wt2
∗ For the first family of cracks of the bottom layer: wb1
∗ For the second family of cracks of the bottom layer: wb2
– The steel concrete inelastic slip v p = (vxp , vyp ) in the x vxp and y vyp reinforcement
directions:
∗ For the top layer: v pt
∗ For the bottom layer: v pb
– The steel plastic strain εps = (εps
x , εy ) in the x εx and y εy reinforcement directions:
ps ps ps

∗ For the top layer: εpst

∗ For the bottom layer: εpsb
– Concrete damage variable d:
∗ For the top layer: dt
∗ For the bottom layer: db
Thus a total of 18 internal scalar variables, together with the observable strain variables,
define the Helmholtz free energy surface density, which can be expressed as the sum of three
different terms:

ψ 0 = ψ lin (Ô, κ; wt1 , wt1 , wt2 , wb1 , wb2 , εpst , εpsb , dt , db ) + ψ nl (wt1 , wt2 , wb1 , wb2 , v pt , v pb , εpst , εpsb , dt , db )
(2.1)
with ψ lin depending on the observable variables:

4 3 4 ØA Ø 3 4 3 βm 4 3 4 3 βm 4 B
1 Ô
3 4 3 mm
A Amf Ô Ô Bγ Ô Cγ
ψ lin
= : : + − : β
· wγ − : ·ε psβ
2 κ Amf Af f κ γ
κ Bβf
γ κ Cβf
γ
β
(2.2)
and ψ nl,t and ψ nl,b determining the coupling between the internal variables:

Ø1 1 1 Ø
ψ nl,β = wβγ ·Dβγ ·wβγ + v pβ ·Eβ ·v pβ + ·εpsβ ·Fβ ·εpsβ − wβγ ·Gβγ ·v pβ +wβ1 ·Hβ ·wβ2 (2.3)
γ
2 2 2 γ

where γ = 1,2 denotes the first and second family of cracks for each layer. It is remarked that:
• The fourth order tensor Amm , Amf and Af f depend on dt and db
• The three order tensors Bβm
γ and Bβf
γ depend on d
β

• The two order tensors Dβγ , Eβ , Gβγ and Hβ depend on dβ

15
 GLRC_HEGIS model

and all of them, together with the three order tensors Cβm
γ and Cβf
γ and the two order tensors
F are determined uniquely from:
β

• Geometrical properties of the section;

• Material elastic properties;

• Crack pattern geometry;

2.1.2 Thermodynamic forces

The constitutive law of the model is the relationship between the generalized efforts:

• in-plane stress resultant N = (Nxx , Nyy , Nxy )

• out-of-plane generalized bending moment M = (Mxx , Myy , Mxy )

and the observable variables: the generalized membrane strain Ô and the generalized curvature
κ. By definition, this relationship is obtained by derivation of the Helmholtz free energy surface
density:

∂ψ 0
3 4
N= =A mm
:Ô+A mf
:κ− β χ Bχ · w χ + C
q q βm β βm psβ
·ε
∂Ô
(2.4)
∂ψ 0
3 4
M= =A mf
:Ô+A :κ− β
ff
χ Bχ · w χ + C
q q βf β βf psβ
·ε
∂κ

2.2 Local constitutive laws

The evolution of the 18 scalar internal variables presented in the previous section is governed
by 18 scalar threshold equations depending on the thermodynamic forces of the model. These
threshold functions are the description of each of the four non-linear physical phenomena retained
for the description of the RC plates behaviour: concrete cracking, concrete damage, steel yielding
and the relative steel-concrete slip.

2.2.1 Concrete damage

Concrete damage is considered as isotropic and it is introduced as in internal scalar positive
variable, d which can evolve only at high stress states associated to compression. Concrete
Young’s modulus Ec is reduced through a decreasing convex function ζ(d):

Ec (d) = Ec ζ(d) (2.5)

Concrete is modelled using an isotropic elastic damage constitutive law characterised by

a stiffness tensor Cc (d), defined by the Young’s modulus Ec , the Poisson’s ratio νc and the
decreasing convex function ζ(d). The relationship between concrete membrane stress σ c and the
plane components of the strain tensor Ôc is given by:

σ c = Cc (d) : Ôc (2.6)
16
 GLRC_HEGIS model

which can be also expressed as:

1 νc 0
 c     
σxx Ôcxx
 c  Ec ζ(d)  νc 1 0  . Ôyy 
σyy  = (2.7)
  c 
1 − νc2  
c
σxy 0 0 1−νc
2 Ôcxy

The chosen convex decreasing function is defined as:

1 + γd d
ζ(d) = (2.8)
1+d
And was inspired by the constitutive models GLRC_DM et DHRC. One of the properties of
this function is that it tend towards infinity to γc when d approaches +∞ :

lim ζ(d) = γc (2.9)

d→+∞

The damage function used ζ(d) assures a damage evolution phase with a constant tangent
slope γd Ec and a bilinear behaviour of the concrete with:

c
ε Ec
 for σ c ≤ σd
σ =
c 3
σd
4 (2.10)
σd + εc −
 γd Ec for σ c > σd
Ec

Figure 2.1: Uniaxial concrete response with damage evolution.

17
 GLRC_HEGIS model

2.2.2 Cracking of the concrete

For the definition of cracks in RC panels, the crack opening wn and the tangential crack
displacement wt are necessary in order to define the complete concrete displacement fields at
crack. As you can see in the Figure 2.2 the crack displacement vector is defined as w = (wn , wt ).
In a RC plate submitted to any loading originating tension stress in at least the first principal
direction, cracking onset occurs when the stress reaches the concrete tensile strength:

Fcr (σ c ) = σ1c − fct ≤ 0 (2.11)

Figure 2.2: Displacement discontinuity at cracks.

The crack pattern evolves with the increasing load until reaching a stabilized crack pattern.
If the applied load is cyclic two different families of cracks can be distinguished, characterized by
the average crack orientations θr1 and θr2 and spacings sr1 and sr2 , see Figure 2.3. In order to
limit the model to a maximum of two crack families, it is assumed that their orientations are
separated at least by 60◦ .

|θr1 − θr2 | > π/3 (2.12)

When a crack family appears following Equation (2.2.2), the crack orientation is fixed as the
orthogonal direction to the first principal stress direction. Moreover, the average crack spacing is
calculated from the crack orientation with the expression that was first given by Vecchio and
Collins (1986):
4−1
sin |θr | cos θr
3
sr = + (2.13)
srx sry
where srx and sry are the theoretical average crack spacings in the equivalent RC tie-beams
in the x and y directions.
18
 GLRC_HEGIS model

Figure 2.3: Second crack in a RC panel: cracked RC panel geometry and relative orientation
between the two cracks.

A crack displacement w originates a number of phenomena which create concrete stresses in

the crack surface. The most analysed are:

• Bridgind stress: normal stress with respect to the crack depending on the normal crack
opening wn , which governs the post-peak concrete tensile behaviour
• Aggregate interlock: tangential stresses caused by the resistance that concrete cracks show
when a tangential wt slip is applied. Crack surface is rough because of concrete aggregates
which are not cracked since their tensile strength is higher than fct .
• Dilatancy stress: concrete normal stress at crack due to the tangential displacement wt .
• Dowel action: resistance of steel bars to deform transversally to their axis, and therefore
opposing to the crack tangential displacement.

The first two will be analysed in the following paragraphs.

Normal crack opening: bridging stress

Bridging stress allow to describe the relation between the normal concrete stress at cracks gn and
the crack opening wn with the following remarks

• The envelope curve Gn (wn ) is equal to the fracture energy dissipated:

Ú +∞
Gn (wn )dwn = Gf (2.14)
0

• Unloading is done elastically (constant crack opening) until compressive stresses are reached.
Under compressive stress, crack recloses until a reclosing value which is a fraction of the
maximum historical crack opening.
• Reloading is first done elastically (constant crack opening) until tensile stresses are reached;
at this moment, crack reopens.

19
 GLRC_HEGIS model

In this model, it is considered that:

• The unloading slope corresponding to the reclosing crack phase for compressive stresses is
αu Ec , where αu it refers to crack unload slope ratio.

• The crack reclosing value is αr wnmax , where wnmax is the historical maximum crack opening
and αr is the crack reclosing ratio.

• When reloading, the envelope curve is attained at the point corresponding to the unloading
onset.

In this model a linear bridging stress law Gn (wn ) is retained:

2Gf

f2
fct − ct wn for 0 ≤ wn ≤


Gn (wn ) = 2Gf fct (2.15)
2Gf
0 for ≤ wn


fct
That explained above is shown in the Figure 2.4

Figure 2.4: Relation between normal concrete stress at cracks gn and crack opening wn .

Tangential crack displacement: aggregate interlock

There are several studies in the literature about aggregate interlock;
Walraven [4] propose an analytical study of the aggregate interlock effect in plain concrete
considering spherical aggregates that crash with the remaining concrete in the opposite crack
surface, Figure 2.5. The analysis considers the frictional forces between aggregate particles and
matrix when sliding occurs, and the deformation of the hardened cement matrix.

Figure 2.5: Aggregate interlock considered by Walraven.

20
 GLRC_HEGIS model

The Contact density Model of Li et al. [5] (Figure 2.6) adopts the following expression for the
aggregate interlock stress τagg :

(wt /wn )2
τagg = τLIM sign(wt ) (2.16)
1 + (wt /wn )2
where τLIM is the maximum aggregate interlock stress estimated as:

τLIM = 3.83fc1/3 (2.17)

Figure 2.6: Experimental cyclic aggregate interlock test vs. analytical results of Contact Density
Model [5].

Similar to the previous one is the model proposed by Gambanova [6] but the peaks of
both aggregate interlock and dilatancy stresses are not reached asymptotically but at a limited
tangential displacement wt , Figure 2.7

Figure 2.7: Gambanova aggregate interlock law [6].

The definition of the aggregate interlock law in the constitutive model is based on the following
observation concerning the aggregate interlock phenomenon just explicated:

• For tangential displacement values in RC structures wt ≤ 0.5mm the loading stress-

tangential displacement curve shown in Figure 2.6 and Figure 2.7 is roughly linear.
21
 GLRC_HEGIS model

• The experimental unloading of Figure 2.6 is done at roughly constant tangential displacement
wt until 0 tangential stress is reached.

• The cyclic response of Figure 2.6 is symmetric in both senses of loading.

Starting from these definitions the obtained tangential stress-tangential displacement wt curve
is shown in Figure 2.8

Figure 2.8: Tangential-to crack concrete stress at cracks gt as a function of the tangential crack
displacement wt .

where T1 accounts for the aggregate interlock stiffness and T0 for the initial shear resistance.

2.2.3 Steel-concrete bond-slip

Bond stresses τ = (τx , τy ) are transmitted from the x and y reinforcement bars to concrete and
they occur when there is a relative slip s = (sx , sy ).It is assumed, the average tension stiffening
bond stress τ 0 is proportional to the visible and measurable quantity slip at cracks v = (vx , vy )
by means of the global bond stiffness diagonal tensor K τ :
Ø
τ 0 = Kτ · v = τ
Kαα vα eα (2.18)
α

The steel-concrete slip at cracks is geometrically related to the crack displacement w =

(wx , wy ):

vx = |wn sin θr − wt cos θr |vy = |wn cos θr + wt sin θr | (2.19)

Therefore, the slip v is linearly related to the crack displacement w by means of the geometrical
matrix M vw (θr ):

v = M vw (θr ) · w (2.20)
where the following definition for the matrix M vw (θr ), graphically shown in Figure 2.9, is
retained:

− cos θr sign(θr )
3 4
|sin θr |
M vw
(θr ) = (2.21)
cos θr sin θr

22
 GLRC_HEGIS model

Figure 2.9: Transformation from crack openings to steel-concrete slip in (a) x direction and (b)
y direction.

Moreover, in the formulation of the global constitutive model the bond-slip stress appears
under the form of an average tension stiffening effect (Figure 2.10). These phenomenon is
considered in the law through a "tension stiffening" coefficient kt .
For further details on the values of these latter coefficients see Appendix A.

Figure 2.10: Typical force-displacement curve for a RC beam with a displacement-controlled

test.

2.2.4 Yielding of steel reinforcement bars

The steel reinforcement bars are supposed to be a one-dimensional medium and to carry only
longitudinal forces. Their behaviour is modelled using an elastic - plastic constitutive law:

σ sα = Es (εsα
αα − εα )eα ⊗ eα
ps
(2.22)
where:
• ⊗: Tensor dyadic product
• σ sα :Membrane stress tensor
• Es : Steel Young’s modulus
• εsα
αα , εα : Total and plastic longitudinal strain for steel bars in α = x, y direction.
pα

In this model the yielding of steel reinforcement bars is assumed to be located and concentrated
at the crack crossing and the variable steel plastic strain εps = (εps x , εy ) is introduced, which
ps

regroups the plastic strains at cracks of x and y steel bars.

23
 GLRC_HEGIS model

2.3 Numerical implementation [2]

The numerical implementation of the model has been done in Code_Aster FE software, developed
by EDF using the Discrete Kirchhoff Triangle Generalized (DKTG) finite element. The numerical
integration of the constitutive model in each Gauss point of the DKTG finite element it was made
through the multicriteria return mapping algorithm which will not be elaborated on in this thesis.

2.3.1 DKTG Finite Element

The Discrete Kirchhoff Triangle Generalized (DKTG) finite elements are isoparametric flat shell
elements that do not take into account the curvature of the structure. They are adapted to
thin slender structures, where the transverse distortion energy can be considered as negligible
with respect to the other deformation energies. An unique integration point is considered in the
plate thickness, so they are adapted to constitutive models formulated in generalised stresses and
strains, adapted to global modelling approaches for RC plates. The three nodes of the DKTG
elements, or the four of the DKQG quadrangle variant element of DKTG, have 5 degrees of
freedom: the translations and two in-plane rotations. They are isoparametric elements, and the
equivalence between the user-defined axes x, y and the reduced axes η, µ , is shown in Figure
2.11.

Figure 2.11: Relationship between real and reference DKQG and DKTG elements.

24
Chapter 3

CEOS.FR

The considered experimental test includes the four walls designed and tested in the framework of
the French national research project CEOS.fr (Behaviour and Assessment of Special Structures.
Cracking - Shrinkage).

3.1 Experimental test description [7, 8, 9, 10]

Shear-wall specimens were designed and tested in order to analyse shear cracking in thick RC
walls under cyclic loading (earthquake). The specimens measure 4.20m, x 1.05m, x 0.15m and
they are equivalent, with a geometrical scale of 1/3, to the real measures of the wall used in
nuclear facilities. Two vertical rebars of 25mm and 32mm have been added to avoid the cracking
due to bending. Furthermore the height/length ratio of 1/4 ensure a low slenderness of the wall.
The redistribution of the shear stresses in the wall is guaranteed by two horizontal thick concrete
beams with a high reinforcement ratio connected to its upper and bottom edges. The main
characteristics of the specimens are set out in Table 3.1

Figure 3.1: Dimensions of the specimen [7]

The specimen is horizontally and vertically connected at the frame only at the edges of the
bottom beam. The vertical support being secured by post-stressed bars, in order to avoid uplift
for the highest values of the loading assumed for the test. The steel frame, the detail of the
section and the reinforcement position are set out in Figure 3.2 and Figure 3.3
The load is applied by two hydraulic jacks placed 100mm above the top of the wall by means
300kN force increments. The applied loading history was cycling and reversing for wall A, B
and D while for wall C the loading history was non-reversing. In the first case three cycles were
applied at each force amplitude how can it be seen in Figure 3.4.
25
 CEOS.FR

Wall 1 Wall 2 Wall 3 Wall 4

Nature of test Variation of concrete Reference Variation of loading Variation of reinforcement
Concrete class C25 C40 C40 C40
Reinforcement Φ10@100 Φ10@100 Φ10@100 Φ8@80
Steel ratio / direction 1.05% 1.05% 1.05% 0.84%
Cyclic load history Reversing Reversing Non reversing Reversing

Table 3.1: Main characteristics of the specimens.

Figure 3.2: Steel frame of the specimen [7]

(a)

(b)

Figure 3.3: Section in the middle of steel frame and of the specimen (a) and reinforcement and
part of formwork (b) [7].

To measure the relative displacement between top and bottom massif of specimens, sensors
type LVDT were inserted, fixed to a rigid bar hinged on the lower beam and free at the top.
Details and arrangement of the sensors are shown in Figure 3.5
26
 CEOS.FR

5000
Force Force
4000
4000
2000
3000
Force [kN]

Force [kN]
0
2000
−2000
1000
−4000
0
0 25 50 75 100 125 150 175 0 5 10 15 20 25 30
Time [instant] Time [instant]

Figure 3.4: Measured load time histories: reversing cycling load (left), non reversing cycling
load (right).

Figure 3.5: Arrangement of sensors for walls A, B, C and D respectively.

3.1.1 Experimental test results [11]

In the next session the main results obtained from the experimental tests will be shown. The
global force-displacement curves are defined by the horizontal applied force vs. horizontal relative
(between the top and bottom beams) displacement the wall refer to rigid sensors with bar. Figure
3.6 shows the experimental global force displacement curves for the walls. Each cycle causes skew
cracking depending on the direction of the applied force; when the force direction is reversed, this
set of cracks closes and another family of cracks opens symmetrically with respect to the vertical
direction. Crack pattern of walls correspond to a load equal to ±4200 kN are shown in Figure
3.7. The average inclination angle and spacing between cracks are not significantly affected by
the direction of thrust.

27
 CEOS.FR

4 4
A B
2 2
Applied Force (MN)

Applied Force (MN)

−4 −2 0 2 4 −4 −2 0 2 4
0 0

−2 −2

−4 −4
Experimental Experimental
Displacement (mm) Displacement (mm)

4 4
C D
2 2
Applied Force (MN)

−4 −2 0 2 4 Applied Force (MN) −4 −2 0 2 4

0 0

−2 −2

−4 −4
Experimental Experimental
Displacement (mm) Displacement (mm)

Figure 3.6: Experimental global force-displacement curves.

(b) Wall B
(a) Wall A

(c) Wall C (d) Wall D

Figure 3.7: Experimental crack orientation.

28
 CEOS.FR

3.2 Numerical Model

The modelling and the mesh are the same for all walls. The model was already made and was
only slightly modified by adding the sensor bar for measuring displacements. The wall and the
horizontal thick concrete beams are modelled by plate DKTG plate finite elements in Code_Aster
FE software while the steel frame is modelled by beam elements. The average pitch of the mesh is
about 25 cm. For the calculation of displacements, the rigid bar that supports the LVDT sensors
shown in the Figure 3.5 was also modelled (Figure 3.8 (b)).

(a) (b)

Figure 3.8: FE model of the RC walls (a) and 3D visualization (b).

The test bench is made integral with the reinforced concrete specimen only in the edge of the
lower beam. The vertical support being secured by two Dywidag post-stressed bars, in order to
avoid uplift for the highest values of the loading assumed for the test. A load of 500 kN is applied
on each bar. To ensure instead the horizontal congruence of the displacements there are two
metal profiles HEB 700 type. The details of the one just described are visible in the Figure 3.9.

Figure 3.9: Real boundary conditions.

29
 CEOS.FR

In the numerical model the Dywidag post-stressed bars are modelled by equivalents loads
distributed over three nodes (Figure 3.10 (a)). As can be seen in Figure 3.10 (b) the horizontal
bond wall-frame is ensured by the rigid discrete elements DIS_T finite elements in Code_Aster
FE software

(a) Post-stress equivalent loads (b) Model of the bond wall-frame

Figure 3.10: Modelled boundary conditions.

3.2.1 Non-linear analysis

Having a low damage the two horizontal thick concrete beams are modelled with an elastic while
the CEOS wall has a non-linear behaviour given by the GLRC_HEGIS law. The parameters
of GLRC_HEGIS law used are the same in the four walls except for those dependent on the
characteristics of concrete or on the amount of the steel reinforcement. Taking into account the
problem of the concrete size effect, the experimental values of fct are reduced by 2/3. The values
of the the fracture energy derive from the experimental tests and they are reduced with the square
of the fct reduction, in order to preserve the post-peak slope of bridging stress curve; Concrete
damage is supposed to appear in compression at σd = fc /4 and the asymptotic damage slope is
set to γd = 0.3, so (A.3) gives a damage threshold of k0 . The theoretical average crack spacings
are calculated with (A.2) which differ due to the different cover in each direction. Between cracks,
a relatively low tension stiffening effect is supposed to be developed in reason of the cyclic loading:
kt = 0.2. The bond-slip stiffness is estimated to Kl = 1011 Pa/m. In reason of the cyclic loading
and the evolution at the same time of normal wn and tangential wt crack displacements, the
values of the cyclic bridging stress parameters αr = 0.05, αu = 0.05 are retained. Not being a
relevant phenomenon and to avoid convergence problems, the tangential crack opening was not
considered and therefore the parameter T0 is set very large (T0 = 100 MPa), while the parameter
T1 is estimated to 10 GPa/m. Table 3.2 summarizes all the parameters used.

30
 CEOS.FR

Parameters Unit Wall A Wall B Wall C Wall D

ρc [Kg/m ] 3
2167 2259 2259 2259
Ec [GP a] 19.9 27.4 27.4 27.4
νc [−] 0.183 0.13 0.13 0.13
Es [GP a] 200 200 200 200
Asx [cm /m]
2
7.854 7.854 7.854 6.283
Asy [cm /m]
2
7.854 7.854 7.854 6.283
z [mm] 55 55 55 55
fsy [M P a] 555 555 555 555
fct [M P a] 1.47 2.37 2.37 2.37
Gf [J/m2 ] 71 80 80 80
αd [−] 0.05 0.05 0.05 0.05
αr [−] 0.05 0.05 0.05 0.05
kt [−] 0.2 0.2 0.2 0.2
T0 [M P a] 100 100 100 100
T1 [GP a/m] 10.0 10.0 10.0 10.0
k0 [P a] 687 1277 1277 1277
Kl [P a/m] 10 11
10 11
10 11
1011
γd [−] 0.3 0.3 0.3 0.3
Φx [mm] 10 10 10 8
Φy [mm] 10 10 10 8
srx [mm] 124 124 124 124
sry [mm] 138 138 138 138

Table 3.2: GLRC_HEGIS parameters for CEOS.fr RC walls.

31
 CEOS.FR

3.3 Comparison between experimental and numerical re-

sults
The experimental loading history of Figure 3.5 is applied on the FE model of the RC walls.
Figure 3.11 shows the global force-displacement curves comparison between experimental and
the numerical results respectively for the wall A, B, C and D. As can be seen in the figure, both
for all the walls, the numerical curve represents the experimental results very well especially in
terms of initial stiffness and envelope. For the wall B and D there is a slight underestimation of
the residual displacements but overall the law GLRC_HEGIS provides results very close to the
experimental ones both in the case of reversible and non-reversible cyclic loading.

4 4
A B
2 2
Applied Force (MN)

Applied Force (MN)

−4 −2 0 2 4 −4 −2 0 2 4
0 0

−2 −2

−4 Experimental −4 Experimental
Numerical Numerical
Displacement (mm) Displacement (mm)

4 4
C D
2 2
Applied Force (MN)

Applied Force (MN)

−4 −2 0 2 4 −4 −2 0 2 4
0 0

−2 −2

−4 Experimental −4 Experimental
Numerical Numerical
Displacement (mm) Displacement (mm)

Figure 3.11: Experimental vs. numerical global force-displacement curves.

3.3.1 Crack opening

Figure 3.12, 3.13, 3.14, 3.15 shows the comparison between the experimental and numerical values
of crack opening. The sensors arrangement and the comparison between the experimental and
the numerical results are shown in the following images.

32
 CEOS.FR

(a) Sensors arrangement of wall A [7]

3 Experimental D18 3 Experimental D19

Numerical D18 Numerical D19
2 2

1 1
Applied Force (MN)

Applied Force (MN)

−150 −100 −50 0 50 100 150 200 −150 −100 −50 0 50 100 150
0 0

1 1

2 2

3 3
Crack opening (μm) Crack opening (μm)

Experimental D20 3
Numerical D20
2

1
Applied Force (MN)

−100 −50 0 50 100 150

0

−1

−2

−3
Crack opening (μm)

Figure 3.12: Experimental vs. numerical crack opening for wall A at sensors D18, D19 and
D20.

33
 CEOS.FR

(a) Sensors arrangement of wall B [8]

Experimental D1 Experimental D2
Numerical D1 4 Numerical D2 4

2 2
Applied Force (MN)

Applied Force (MN)

−200 −100 0 100 200 300 −200 −100 0 100 200 300 400
0 0

−2 −2

−4 −4

Crack opening (μm) Crack opening (μm)

Experimental D3 Experimental D4
Numerical D3 4 4 Numerical D4

2 2
Applied Force (MN)

Applied Force (MN)

−200 −100 0 100 200 300 400 −200 −100 0 100 200 300 400 500
0 0

−2 2

−4 4

Crack opening (μm) Crack opening (μm)

Experimental D5
4 N merical D5 4

2 2
Applied Force (MN)

Applied Force (MN)

−100 0 100 200 300 400 500 600 700 −200 −100 0 100 200 300 400
0 0

−2 2

−4 Experimental D6 4
Numerical D6
Crack opening (μm) Crack opening (μm)

34
 CEOS.FR

Experimental D7 Experimental D8
4 Numerical D7 4 Numerical D8

2 2
Applied Force (MN)

Applied Force (MN)

−200 −100 0 100 200 300 400 −200 −100 0 100 200 300
0 0

2 2

4 4

Crack opening (μm) Crack opening (μm)

E perimental D9 Experimental D10

4 Numerical D9 4 Numerical D10

2 2
Applied Force (MN)

Applied Force (MN)

−200 0 200 400 600 800 1000 −100 0 100 200 300 400
0 0

−2 −2

−4 −4

Crack opening (μm) Crack opening (μm)

Experimental D11
4 Numerical D11

2
Applied Force (MN)

−200 −100 0 100 200 300 400

0

−2

−4

Crack opening (μm)

Figure 3.13: Experimental vs. numerical crack opening for wall B at sensors D1 to D11.

35
 CEOS.FR

(a) Sensors arrangement of wall C [9]

5 5
Experimental D5 Experimental D6
Numerical D5 Numerical D6
4 4
Applied Force (MN)

Applied Force (MN)

3 3

2 2

1 1

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250
Crack opening (μm) Crack opening (μm)

5
Experimental D7
Numerical D7
4
Applied Force (MN)

0
0 50 100 150 200 250 300
Crack opening (μm)

Figure 3.14: Experimental vs. numerical crack opening for wall C at sensors D5, D6 and D7.

36
 CEOS.FR

(a) Sensors arrangement of wall D [10]

Experimental D1 Experimental D2
Numerical D1 4 4 Numerical D2

2 2
Applied Force (MN)

Applied Force (MN)

−200 −100 0 100 200 −100 −50 0 50 100 150 200 250
0 0

−2 −2

−4 −4

Crack opening (μm) Crack opening (μm)

Experimental D3 Experimental D4
4 N merical D3 Numerical D4 4

2 2
Applied Force (MN)

Applied Force (MN)

−100 0 100 200 300 400 500 600 700 −400 −300 −200 −100 0 100 200 300 400
0 0

−2 −2

−4 −4

Crack opening (μm) Crack opening (μm)

Experimental D5 Experimental D6
4 Numerical D5 4 Numerical D6

2 2
Applied Force (MN)

Applied Force (MN)

−100 0 100 200 300 400 500 0 200 400 600 800
0 0

2 −2

4 −4

Crack opening (μm) Crack opening (μm)

37
 CEOS.FR

E perimental D7 Experimental D8
Numerical D7 4 4
Numerical D8

2 2
Applied Force (MN)

Applied Force (MN)

−300 −200 −100 0 100 200 300 400 500 −100 0 100 200 300 400
0 0

−2 2

−4 4

Crack opening (μm) Crack opening (μm)

Figure 3.15: Experimental vs. numerical crack opening for wall D at sensors D1 to D8.

Following the results just shown, the subsequent considerations can be made:

• For all sensors there is an imprecision due to the "zero point", value at which the sensors
have been installed (just after opening the first crack), which is estimated between 50 and
100 µm.
• In the reversing cyclic load tests, the negative crack opening measured by the sensors
represents the compression deformation of the concrete, however the numerical and the
experimental values are not comparable.
• As can be seen in Figure 3.11 and in Figure 3.13 some sensors detect two or more cracks
and then the two families of curves are not similar and comparable: this is detectable for
the sensors D5, D6 and D7 related to wall B and sensors D3, D6 and D7 related to wall D.

38
 CEOS.FR

3.3.2 Envelope curves

Figure 3.16 shows the experimental and the numerical envelope curves of displacements for all
walls. Both experimental and numerical envelope underline that the wall A presents a less stiffness
than other walls but it is more ductile in fact it has a greater displacement capacity. This is due
to a lower class of concrete. The behaviour of the other walls is quite close as they differ only
for the amount of the steel reinforcement. They have less capacity in terms of displacement but
reach a greater level of effort.

2
Applied Force (MN)

−6 −4 −2 0 2 4 6
0

−2

Numerical results
Wall A
Wall B
−4 Wall C
Wall D
Displacement (mm)

Figure 3.16: Experimental vs. numerical envelope curves.

3.3.3 Crack pattern

Figures from 3.17 to 3.28 show the crack pattern related to wall B for twelve steps of loading. As
can be seen, the advancement of the numerical crack opening is very close to the experimental
one with a slightly overestimation of the crack spacing. In particular the crack spacing displayed
directly in the figures is the srx,y parameter of the GLRC_HEGIS model calculated through
(A.2). To note the presence of the second family of crack due to the inversion of loading. For a
better interpretation of the results, the scale relative to the crack opening has not been shown.
For an evaluation of the residual crack opening after F = ±4200 kN refer to the Figure 3.29.

39
 CEOS.FR

Figure 3.17: Experimental vs. numerical crack pattern at F = ±900 kN.

Figure 3.18: Experimental vs. numerical crack pattern at F = ±1200 kN.

40
 CEOS.FR

Figure 3.19: Experimental vs. numerical crack pattern at F = ±1500 kN.

Figure 3.20: Experimental vs. numerical crack pattern at F = ±1800 kN.

41
 CEOS.FR

Figure 3.21: Experimental vs. numerical crack pattern at F = ±2100 kN.

Figure 3.22: Experimental vs. numerical crack pattern at F = ±2400 kN.

42
 CEOS.FR

Figure 3.23: Experimental vs. numerical crack pattern at F = ±2700 kN.

Figure 3.24: Experimental vs. numerical crack pattern at F = ±3000 kN.

43
 CEOS.FR

Figure 3.25: Experimental vs. numerical crack pattern at F = ±3300 kN.

Figure 3.26: Experimental vs. numerical crack pattern at F = ±3600 kN.

44
 CEOS.FR

Figure 3.27: Experimental vs. numerical crack pattern at F = ±3900 kN.

Figure 3.28: Experimental vs. numerical crack pattern at F = ±4200 kN.

45
 CEOS.FR

Figure 3.29: Numerical values of the residual cracks after F = ±4200 kN.

On the basis of the results just obtained and shown, it is possible to make the following
considerations:

• For the first load cycles the crack pattern reflects the experimental one very well.
• After the fifth loading step the numerical results go towards a homogenization, in fact from
the seventh step it is complicated to extrapolate significant differences in the results.
• In general it is possible to see a slight underestimation of the crack spacing in fact the
distance between two experimental cracks is on average about 15/20 cm while in the
numerical model a crack spacing value of about 12 cm has been used. The size of the mesh
contributes significantly to this underestimation.
• At the penultimate loading step (F = ±4200 kN) the crack pattern is fully developed and
almost homogeneous on the whole wall (Figure 3.29).

46
Chapter 4

CODE_ASTER TESTS

In this chapter the GLRC_HEGIS model is compared to other global models for RC plates
GLRC_DM, DHRC and EIB already implemented in Code_Aster. In particular will be analysed
the case of cyclic pure flexion test.

4.1 Cyclic pure flexion test [12]

The global performance is analysed in pure flexion test deepening the results already obtained
by Miquel HUGUET-AGUILERA [2] in fact, the same model has been analysed considering at
first several load cycles and then an increase of the same over time applied to the relatively high
reinforced RC plate of Figure 4.1.

Figure 4.1: Geometry of the considered RC plate.

The dimension of plate are 1m x 1m and h = 0.1m. The concrete is characterised by a Young’s
modulus of Ec = 32.3Gpa, a Poisson’s ratio of νc = 0.2, a tensile strength of fct = 3.4M pa and a
post-cracking slope equal to 30% of the initial elastic modulus. The plate is reinforced by rebars
of 10mm spaced by 10cm in the top and bottom layers at both x and y direction with a cover of
10mm. The steel of reinforcement is characterized by a Young’s modulus of Es = 200Gpa and a
yielding stress of fsy = 570M pa. The complete set of GLRC_HEGIS parameters is summarised
in Table 4.1.
Figure 4.2 shows the FE model of the considered RC plate and the applied boundary conditions.

47
 CODE_ASTER TESTS

Parameters Unit Value Parameters Unit Value

ρc [Kg/m3 ] 2500 αd [−] 0.05
Ec [GP a] 32.3 αr [−] 0.05
νc [−] 0.2 kt [−] 0.33
Es [GP a] 200 T0 , T1 [M P a], [GP a/m] -
Asx [cm2 /m] 7.85 k0 [P a] 4816
Asy [cm /m]
2
7.85 Kl [P a/m] 1011
zinf [mm] 35 γd [−] 0.3
zsup [mm] 35 Φx [mm] 10
fsy [M P a] 570 Φy [mm] 10
fct [M P a] 3.4 srx [mm] 160
Gf [J/m2 ] 143 sry [mm] 160

Table 4.1: GLRC_HEGIS parameters for CAS_TEST.

Figure 4.2: FE model and boundary conditions.

48
 CODE_ASTER TESTS

4.1.1 Constant rotation over time

The time loading history of Figure 4.3 is applied on the FE model. The imposed constant rotation
creates a homogeneous stress state with a cyclical flexion of the lower and upper fibres.

0.8
0.6

Applied rotation R0 x 10−3 (-)

0.4
0.2
0.0
−0.2
−0.4
−0.6
−0.8
0 5 10 15 20 25
Pseudo-Time (s)

Figure 4.3: Loading history with constant rotation.

By comparing the results of Figure 4.4, the following considerations can be made:

• The elastic stiffness is different because of different model’s assumptions.

• Unlike the other laws, GLRC_HEGIS has a bilinear behaviour in the unloading phase. this
is due to taking into account in the model the crack re-closure phenomenon.
• The hysteresis surface described by GLRC_HEGIS it’s larger than that one described by
the other global models.

GLRC_HEGIS 15
DHRC
GLRC_DM
EIB 10
Bending moment MYY (kNm)

5
−0.006 −0.004 −0.002 0.000 0.002 0.004 0.006
0

−5

−10

−15
Rotation RY (-)

Figure 4.4: Moment-rotation comparison.

As noted previously, taking into account the crack reclosing phenomenon ensures that the
energy dissipated by the GLRC_HEGIS model is much larger and above all increasing than that
dissipated by the other two global models. In particular for the DHRC and GLRC_DM models
the dissipated energy remains constant after the opening of the first and second crack families
49
 CODE_ASTER TESTS

(Figure 4.5). Moreover, as can be seen in the Figure 4.6 (b) in this test, energy is only dissipated
by crack opening-reclosing mechanism by GLRC_HEGIS model.

GLRC_HEGIS
100 DHRC
GLRC_DM

80

Dissipated energy (J/m^2)

60

40

20

0
0 5 10 15 20 25
Pseuso-Time (s)

Figure 4.5: Dissipated energy comparison.

The crack opening evolution obtained with GLRC_HEGIS in the top and bottom layers is
shown in Figure 4.6 (a). Because of the applied rotation history, it is the bottom layer which
cracks firstly, reaching a maximum crack opening of approximately 16µm. When the applied
rotation changes its sign, this crack recloses until its residual crack opening while the top layer
cracks. Due to the symmetry of the applied load and of the RC plate, the obtained maximum
opening of 17µm in this layer is similar to the previous one, although it is slightly higher because
of the previous degradation of the plate. The results obtained in terms of crack opening are
exactly identical to those obtained for a single load cycle [2].

bottom layer Crack opening-reclosing

20.0 top layer 100 Total
17.5
80
Dissipated energy (J/m^2)

15.0
Crack opening (μm)

12.5
60
10.0

7.5 40

5.0
20
2.5

0.0 0
0 5 10 15 20 25 0 5 10 15 20 25
Pseudo-Time (s) Pseudo-Time (s)

(a) (b)

Figure 4.6: GLRC_HEGIS: Crack opening evolution (a) and dissipated energy components.

50
 CODE_ASTER TESTS

4.1.2 Incremental rotation over time

In this time loading history (Figure 4.7) the applied rotation is increasing over time. The flexion
of the fibres occurs in the same way as the previous load with an increase of the phenomena over
time.

1.00
0.75
Applied rotation R0 x 10−3 (-)
0.50
0.25
0.00
−0.25
−0.50
−0.75
−1.00
0 5 10 15 20 25
Pseudo-Time (s)

Figure 4.7: Loading history with constant rotation.

As can be seen in the Figure 4.8 the hysteresis surface in this case grows over time due the
increase of the crack opening and the residual crack and the same considerations as above can be
made. For the same reasons also the energy dissipated for the DHRC and GLRC_DM models
grows over time.

GLRC_HEGIS 20 160 GLRC_HEGIS

DHRC DHRC
GLRC_DM 15 140 GLRC_DM
EIB
Bending m ment MYY (kNm)

10 120
Dissipated energy (J/m^2)

5 100
−0.0075 −0.0050 −0.0025 0.0000 0.0025 0.0050 0.0075
0 80

−5 60

−10 40

−15 20
−20 0
Rotation RY (-) 0 5 10 15 20 25
Pseuso-Time (s)

(a) (b)

Figure 4.8: GLRC_HEGIS: Moment-rotation (a) and dissipated energy (b) comparison.

51
 CODE_ASTER TESTS

bottom layer 160 Crack opening-reclosing

40
top layer Total
140
35
120

Dissipated energy (J/m^2)

30
Crack opening (μm)

25 100

20 80

15 60

10 40

5 20

0 0
0 5 10 15 20 25 0 5 10 15 20 25
Pseudo-Time (s) Pseudo-Time (s)

(a) (b)

Figure 4.9: GLRC_HEGIS: Crack opening evolution (a) and dissipated energy components (b).

By comparing the results obtained for the two load cases the following considerations can be
made:

• The hysteresis surface increases over time for all models due to the damage caused by the
increasing applied rotation.
• The energy dissipated by GLRC_HEGIS at the end of load is about 60% larger with the
increasing rotation.
• For the increasing load the crack opening is not constant; at the end of load the increase of
the crack opening is more than double compared to the previous case; the residual crack
also grows over time.
• For both load cases energy is only dissipated by crack opening-reclosing mechanism by
GLRC_HEGIS model.

52
Chapter 5

THREE-POINT BENDING
TEST ON RC BEAM

In this chapter will be analysed the behaviour of the GLRC_HEGIS model in the case of out-of-
plane pure bending. In particular the numerical model will be compared with the experimental
results of a three-point bending test concerning a RC beam.

5.1 Experimental test description

The benchmark [13] consist of a three-point bending test on a RC beam under cyclic loading in
order to analyse the global behaviour until the rupture. The beam has a rectangular section of
0.20m x 0.50m and a length of 5.40m. The distance between the supports is 5.00m while the load
is applied by an hydraulic jack in the middle of the beam. The main reinforcement consists of
two lower bars HA32 and two upper bars HA8. The shear reinforcement consists of stirrups HA8
spaced by 100mm. Three LVDT sensors are placed one quarter, half and three quarters of the
length respectively and they measure global displacements of the beam. Two further comparators
are placed at the level of the supports to monitor their movements. The following figures show
the test and the beam layout (Figure 5.1 and Figure 5.2).

Figure 5.1: Test layout.

53
 THREE-POINT BENDING TEST ON RC BEAM

(b)
(a)

Figure 5.2: Beam’s section (a) and sensors arrangement (b).

The principal mechanical characteristics of the concrete and the steel reinforcement are
summarized in the following tables:

Material Class E[Gpa] ρ [Kg/m3 ] fc [M pa] fct [M pa]

Concrete C35 37.2 2350 36.08 3.68

Table 5.1: Mechanical characteristic of the concrete.

Material Type E[Gpa] ρ [Kg/m3 ] fy [M pa]

Steel 500 HA 197.2 7800 466

Table 5.2: Mechanical characteristic of the steel.

The loading is carried out by a hydraulic unit feeding a jack arranged on a cross. The jack
used is 1000 kN capacity (+/- 125 mm stroke for an output signal of +/- 10V). Figure 3.17 shows
the applied load history of the static cycling load.

300

250
Applied force [kN]

200

150

100

50

0
0 50 100 150 200 250
Pseudo-time [s]

Figure 5.3: Measured load time history.

54
 THREE-POINT BENDING TEST ON RC BEAM

5.1.1 Experimental test results

The sensors placed at one quarter, half and three quarter of the total length measure the
displacements for each of the eight load cycles. After the rupture the deformation control
allows the trace the curves also for the post-break stretch. The deformations of the lower
reinforcement bars were also measured using sensors J1-J6. The global experimental results of
the force-displacement curves and the steel deformation are shown below.

300 300

250 250
Applied Force (KN)

Applied Force (KN)

200 200

150 150

100 100

50 50
Experimental_50 Experimental_25,75
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Displacement (mm) Displacement (mm)

(a) (b)

Figure 5.4: Global experimental force-displacement curves at half (a), one quarter and three-
fourth of the length (b).

300 300 300

250 250 250

Applied Force (KN)

Applied Force (KN)

Applied Force (KN)

200 200 200

150 150 150

100 100 100

50 50 50
Experimental J1 Experimental J2 Experimental J3
0 0 0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 0.0016 0.0000 0.0005 0.0010 0.0015 0.0020 0.0025 0.0030
Steel deformation (-) Steel deformation (-) Steel deformation (-)

300 300 300

250 250 250

Applied Force (KN)

Applied Force (KN)

Applied Force (KN)

200 200 200

150 150 150

100 100 100

50 50 50
Experimental J4 Experimental J5 Experimental J6
0 0 0
0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 0.0016
Steel deformation (-) Steel deformation (-) Steel deformation (-)

Figure 5.5: Experimental steel deformation given by sensors J1-J6.

55
 THREE-POINT BENDING TEST ON RC BEAM

5.2 Numerical model

The beam is modelled by plate DKTG plate finite elements in Code_Aster FE software. The
mesh is composed of 2D quadrangular elements. The boundary conditions simulate the real
supports of the beam and a linear distribution of the applied concentrated force was used. The
mesh, horizontal with elements equidistant from each other, was used to simulate the case of pure
bending. The FE model used for the beam is shown in Figure 5.6.

Figure 5.6: FE model of the RC beam.

5.2.1 Non-linear analysis

The beams is modelled with a non-linear behaviour given by GLRC_HEGIS. Taking into account
the problem of the concrete size effect, the experimental value of fct is reduced by 2/3. The value
of the the fracture energy is calculated with (A.1) and it is reduced with the square of the fct
reduction, in order to preserve the theoretical post-peak slope of bridging stress curve. Concrete
damage is supposed to appear in compression at σd = fc /6 and the asymptotic damage slope is
set to γd = 0.2, so (A.3) gives a damage threshold of k0 = 389 Pa. The theoretical average crack
spacings are calculated with (A.2), obtaining srx,sup = 186 mm, srx,inf = 89 mm and sry = 138
mm. Furthermore, for srx,inf it was used a value equal to 100 mm obtained from the available
images of the crack state; Between cracks, a relatively low tension stiffening effect is supposed to
be developed in reason of the cyclic loading: kt = 0.2. The bond-slip stiffness is estimated to
Kl = 1011 Pa/m. In reason of the cyclic loading and the evolution at the same time of normal
wn and tangential wt crack displacements, the values of the cyclic bridging stress parameters
αr = 0.05, αu = 0.05 are retained. Aggregate interlock parameters are estimated to T0 = 0.01
MPa and T1 = 10 GPa/m. Table 5.3 summarizes all the parameters used.

Parameters Unit Value Parameters Unit Value

ρc [Kg/m3 ] 2350 αr [−] 0.05
Ec [GP a] 37.2 kt [−] 0.2
νc [−] 0.2 T0 [M P a] 0.01
Es [GP a] 195 T1 [GP a/m] 10
Asx,sup [cm2 /m] 5.03 k0 [P a] 389
Asx,inf [cm2 /m] 80.40 Kl [P a/m] 1011
Asy [cm /m]
2
5.03 γd [−] 0.2
zinf [mm] 206 Φx , sup [mm] 8
zsup [mm] 218 Φx , inf [mm] 32
fsy [M P a] 466 Φy [mm] 8
fct [M P a] 2.45 srx,sup [mm] 186
Gf [J/m2 ] 62 srx,inf [mm] 100
αd [−] 0.05 sry [mm] 138

Table 5.3: GLRC_HEGIS parameters for RC beam.

56
 THREE-POINT BENDING TEST ON RC BEAM

5.3 Comparison between experimental and numerical re-

sults
The Figures 5.7 and 5.8 show the comparison between the experimental and numerical results
obtained with the GLRC_HEGIS model. For the force-displacement curves the loading and
unloading phases coincide with a slight underestimation of the residual displacements and a
overestimation of the maximal displacement at the peak. It should be emphasized that for the
global force-displacement curves the analyses only concerned the loading and unloading phases
and were not studied in depth to study post-peak behaviour.

300 300

250 250
Applied Force (KN)

Applied Force (KN)

200 200

150 150

100 100

50 50
Experimental_50 Experimental_25,75
GLRC_HEGIS GLRC_HEGIS
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Displacement (mm) Displacement (mm)

(a) (b)

Figure 5.7: Global experimental vs. numerical force-displacement curves at half (a), one quarter
and three-fourth of the length (b).

300 300 300

250 250 250

Applied Force (KN)

Applied Force (KN)

Applied Force (KN)

200 200 200

150 150 150

100 100 100

50 50 50
GLRC_HEGIS GLRC_HEGIS GLRC_HEGIS
Experimental J1 Experimental J2 Experimental J3
0 0 0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 0.0016 0.0000 0.0005 0.0010 0.0015 0.0020 0.0025 0.0030
Steel deformation (-) Steel deformation (-) Steel deformation (-)

300 300 300

250 250 250

Applied Force (KN)

Applied Force (KN)

Applied Force (KN)

200 200 200

150 150 150

100 100 100

50 50 50
GLRC_HEGIS GLRC_HEGIS GLRC_HEGIS
Experimental J4 Experimental J5 Experimental J6
0 0 0
0.000 0.002 0.004 0.006 0.008 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.000000.000250.000500.000750.001000.001250.001500.001750.00200
Steel deformation (-) Steel deformation (-) Steel deformation (-)

Figure 5.8: Experimental vs. numerical steel deformation given by sensors J1-J6.

57
 THREE-POINT BENDING TEST ON RC BEAM

As regards the deformation of the steel for sensors J2 and J3, the numerical results lead to
an underestimation of the final deformation and to a slightly lower initial drop. The opposite
consideration can be done for the J6 sensor. For the sensors J4 and J5 we note that there is an
overestimation of the residual plastic steel deformation but with a good approximation in the
first phase. The proximity of the J1 sensor to the supports means that the numerical model does
not lead to results comparable with the experimental one.
Figure 5.9 (a) shows the trend over time of the main internal variables of the GLRC_HEGIS
model. We can note that the Damage of upper layer is larger than that one of lower layer
because it’s subject to compression and at all times the difference between crack opening and the
steel-concrete sliding represent the steel yielding.

Damage variable and Pla tic teel deformation x 10−3 (-)

Crack opening 14 Num Steel-concrete slip
Bond slip concrete-steel 1750 Num Crack opening-reclosing
400 Damage lower layer 12 Num Concrete damage
Damage upper layer Num Plastic steel deformation
Crack opening and bond lip (μm)

1500
Plastic steel deformation Experimental
10 Num Total

Dissipated energy (J)

300 1250
8
1000
200 6
750
4 500
100
2 250

0 0 0
0 50 100 150 200 3 4 5 6 7
Time ( ) Cycle

(a) (b)

Figure 5.9: Trend of the main internal variables and dissipated energy components.

By analysing instead the components of the dissipated energy (Figure 5.9 (b)) it is noted that
in the first elastic phase the energy is dissipated mainly by the concrete damage phenomenon;
after the plasticization of the steel, the dissipation of energy is instead governed by the plastic
steel deformation. The dissipated energy calculated directly as the internal variable of the model
was compared with the experimental one, calculated from the force-displacement curve. The
comparison was made for each load cycle up to the penultimate one, since the last cycle only
consists of a load phase, it is not possible to calculate the experimental dissipated energy. The
figure shows how the numerically calculated dissipated energy is slightly less than the experimental
one. This can also be seen through the comparison of the force-displacement curves. The area
subtended by the numerical curve is smaller than that subtended by the experimental curve.

58
 THREE-POINT BENDING TEST ON RC BEAM

5.4 Study of mesh sensitivity

To study the sensitivity of the mesh used, a second model was created consisting of a single
element in the direction transverse to the axis. The results, listed below, are perfectly identical
to those obtained previously with the first model.

Figure 5.10: Second FE model of the beam.

300 300

250 250
Applied Force (KN)

Applied Force (KN)

200 200

150 150

100 100

50 50
Experimental_50 Experimental_25,75
GLRC_HEGIS GLRC_HEGIS
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Displacement (mm) Displacement (mm)

(a) (b)

Figure 5.11: Global experimental vs. numerical force-displacement curves at half (a), one
quarter and three-fourth of the length (b) for the second model.

The two models lead to exactly identical results because the parameters used are essentially
the same. In fact, the parameters that govern the model do not change according to the number
of transversal elements (one or two). For simplicity and computational times (slightly less) it
is better to use the simpler model. Being used, the latter, only to study the sensitivity of the
mesh, in this thesis and in the next sections, reference will always be made to the first model
(two transversal elements).

59
 THREE-POINT BENDING TEST ON RC BEAM

300 300 300

250 250 250

Applied Force (KN)

Applied Force (KN)

Applied Force (KN)

200 200 200

150 150 150

100 100 100

50 50 50
GLRC_HEGIS GLRC_HEGIS GLRC_HEGIS
Experimental J1 Experimental J2 Experimental J3
0 0 0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 0.0016 0.0000 0.0005 0.0010 0.0015 0.0020 0.0025 0.0030
Steel deformation (-) Steel deformation (-) Steel deformation (-)

300 300 300

250 250 250

Applied Force (KN)

Applied Force (KN)

Applied Force (KN)

200 200 200

150 150 150

100 100 100

50 50 50
GLRC_HEGIS GLRC_HEGIS GLRC_HEGIS
Experimental J4 Experimental J5 Experimental J6
0 0 0
0.000 0.002 0.004 0.006 0.008 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.000000.000250.000500.000750.001000.001250.001500.001750.00200
Steel deformation (-) Steel deformation (-) Steel deformation (-)

Figure 5.12: Experimental vs. numerical steel deformation given by sensors J1-J6 for the second
model.
Damage variable and Pla tic teel deformation x 10−3 (-)

Crack opening 14 Num Steel-concrete slip

Bond slip concrete-steel 1750 Num Crack opening-reclosing
400 Damage lower layer 12 Num Concrete damage
Damage upper layer Num Plastic steel deformation
Crack opening and bond lip (μm)

1500
Plastic steel deformation Experimental
10 Num Total
Dissipated energy (J)

300 1250
8
1000
200 6
750
4 500
100
2 250

0 0 0
0 50 100 150 200 3 4 5 6 7
Time ( ) Cycle

(a) (b)

Figure 5.13: Trend of the main internal variables (a) and dissipated energy components (b) for
the second model.

60
 THREE-POINT BENDING TEST ON RC BEAM

5.5 Comparison with membrane behaviour model

A third model has been studied with a different mesh characterized by a membrane behaviour.
To better characterize the non-linear parameters of the material, three layers were identified; the
upper and the lower having a width equal to twice the distance between the steel bar and the
external edge of the concrete. The remainder defines the central layer. To avoid convergence
problems the boundary conditions are slightly different from the previous model, in fact the nodes
are fixed and the force is applied on the axis of the shell element. All the parameters used for
each layer are summarized in the Table 5.4. To underline that all non-linear parameters are equal
except for k0 (for the calculation a σd equal to double that of the first model was used) and γd
(0.2 for the first model and 0.3 for this one). The third FE model of the beam is shown in Figure
5.14

Parameters Unit T. layer C. layer B. layer

ρc [Kg/m ] 3
2350 2350 2350
Ec [GP a] 37.2 37.2 37.2
νc [−] 0.2 0.2 0.2
Es [GP a] 195 195 195
Asx [cm2 /m] 7.85 0 91.4
Asy [cm /m]
2
5.03 5.03 5.03
zinf [mm] 62 62 56
zsup [mm] 62 62 56
fsy [M P a] 466 466 466
fct [M P a] 2.45 2.45 2.45
Gf [J/m ] 2
62 62 62
αd [−] 0.05 0.05 0.05
αr [−] 0.05 0.05 0.05
kt [−] 0.2 0.2 0.2
T0 [M P a] 100 100 100
T1 [GP a/m] 10 10 10
k0 [P a] 2050 2050 2050
Kl [P a/m] 1011 1011 1011
γd [−] 0.2 0.2 0.2
Φx [mm] 8 0 32
Φy [mm] 8 8 8
srx [mm] 121 100 100
sry [mm] 120 120 120

Table 5.4: GLRC_HEGIS parameters for RC beam (third model).

61
 THREE-POINT BENDING TEST ON RC BEAM

Figure 5.14: Third FE model of the RC beam.

Comparing the results of the two models (first and third) it can be seen that also for the third
model (Figure 5.15) there is an underestimation of the residual displacement (the same as the
first model). For the curve relative to the central point of the beam, unlike the first model, there
is a slightly underestimation of the displacement relative to the peak of the seventh cycle but
at the end of the test the numerical displacement and the experimental ones coincide perfectly
(Figure 5.15 (a)). For the curve relative to a quarter and three quarters of the length of the
beam, there is a slight underestimation of the displacements and a slope slightly greater than the
experimental one (Figure 5.15 (b)).

300 300

250 250
Applied Force (KN)

Applied Force (KN)

200 200

150 150

100 100

50 50
Experimental_50 Experimental_25,75
GLRC_HEGIS GLRC_HEGIS
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Displacement (mm) Displacement (mm)

(a) (b)

Figure 5.15: Global experimental vs. numerical force-displacement curves at half (a), one
quarter and three-fourth of the length (b) for the third model.

Figure 5.16 shows the comparison between experimental and numerical results in terms of
steel deformation relative to the third analysed model. The results obtained are in line with
those previously seen relating to the first model. The comparison for sensors J2, J3, J4, J5 and
J6 is very similar for both models. For the sensors J4 and J5 is also present in this case an
overestimation of the residual plastic deformation almost equal to that estimated in the first
model. Having modelled the boundary conditions as fixed nodes the results obtained for the
sensor J1 are worse than those previously seen. To underline that the sensors J4 and J5 show the
plasticization of the steel due to the high deformations reached in the middle of the beam.

62
 THREE-POINT BENDING TEST ON RC BEAM

300 300 300

250 250 250

Applied Force (KN)

Applied Force (KN)

Applied Force (KN)

200 200 200

150 150 150

100 100 100

50 50 50
GLRC_HEGIS GLRC_HEGIS GLRC_HEGIS
Experimental J1 Experimental J2 Experimental J3
0 0 0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 0.0016 0.0000 0.0005 0.0010 0.0015 0.0020 0.0025 0.0030
Steel deformation (-) Steel deformation (-) Steel deformation (-)

300 300 300

250 250 250

Applied Force (KN)

Applied Force (KN)

Applied Force (KN)

200 200 200

150 150 150

100 100 100

50 50 50
GLRC_HEGIS GLRC_HEGIS GLRC_HEGIS
Experimental J4 Experimental J5 Experimental J6
0 0 0
0.000 0.002 0.004 0.006 0.008 0.000 0.002 0.004 0.006 0.008 0.010 0.0000 0.0005 0.0010 0.0015 0.0020
Steel deformation (-) Steel deformation (-) Steel deformation (-)

Figure 5.16: Experimental vs. numerical steel deformation given by sensors J1-J6 for the third
model.

Analysing the internal variables and the dissipation of energy in the two different models
(Figure 5.9 and Figure 5.17), it is noted that the general trend over time is very similar but with
slightly higher value of the crack opening in the membrane behaviour model (vertical mesh). As
regards the components of energy dissipated, the same considerations made previously are valid.
It should be noted that unlike the first model in the figure, the internal variable relating to the
damage of the lower layer is not present since, being calculated on the outer edge of the mesh,
the compression damage of the concrete is trivially zero. Unlike the first model, there are also
differences in the components of the dissipated energy. In this case the numerically estimated
dissipated energy is slightly lower than that estimated by the first model because the component
of plastic steel deformation is smaller. Also in this case all this can be seen in the comparison of
the two force-displacement curves. For the third model the area subtracted from the numerical
curve is smaller and in fact the energy dissipated is lower.
Damage variable and Pla tic teel deformation x 10−3 (-)

1000 Crack opening Num Steel-concrete slip

Bond slip concrete-steel 10 1750 Num Crack opening-reclosing
Damage upper layer Num Concrete damage
Plastic steel deformation Num Plastic steel deformation
Crack opening and bond lip (μm)

800 1500
Experimental
8 Num Total
Dissipated energy (J)

1250
600
6 1000

400 750
4
500
200 2
250

0 0 0
0 50 100 150 200 3 4 5 6 7
Time ( ) Cycle

(a) (b)

Figure 5.17: Trend of the main internal variables (a) and dissipated energy components (b).

63
 THREE-POINT BENDING TEST ON RC BEAM

Figures 5.18 5.19 5.20 show the advancement of the crack pattern over loading. The comparison
is made between the numerical and the experimental result at 90, 150 and 185 kN, corresponding
to the fifth, sixth and seventh load cycles. We note how the numerical results are very close to
the experimental ones. For each image, the first figure represents the first model, the second
the third and the third the pattern relating to the experimental results. Figure 5.21 shows the
numerical values of the crack opening at the end of the test for the first and third model. It
should be emphasized that the cracks on the ends of the beam (relative to third model) are due
to the application of the boundary conditions in the numerical model.

Figure 5.18: Experimental vs. numerical crack pattern at F = 90 kN (Cycle 5).

Figure 5.19: Experimental vs. numerical crack pattern at F = 150 kN (Cycle 6).

Figure 5.20: Experimental vs. numerical crack pattern at F = 185 kN (Cycle 7).

64
 THREE-POINT BENDING TEST ON RC BEAM

Figure 5.21: Numerical values of the crack openings at the end of test.

Based on the results obtained for the two different models it is possible to draw the following
observations:

• The first model better simulates the boundary conditions and the real behaviour of the
beam.
• On the other hand for the pure bending behaviour the calibration of the parameters of the
GLRC_HEGIS model turns out more complicated.
• The results obtained by the different models are quite similar both in terms of force-
displacement curves, both in terms of steel deformation and in terms of dissipated energy.
• The third model allows to obtain many more graphic results also thanks to the functions
implemented in it.

65
Chapter 6

CONCLUSION

This stage allowed me to know and work on the non-linear behaviour of reinforced concrete
through the GLRC_HEGIS model, developed at Egis Industries. My work has allowed to improve
numerical results already obtained (Chapter 3) and to gain valuable information on the behaviour
of the model in the case of pure bending (Chapters 4 and 5). On the basis of the results obtained
and described in the previous chapters, it is possible to make the following conclusions:

• Comparing the model with the other global models already implemented in Code_Aster
(GLRC_DM and DHRC) it is noted how the taking into account of multiple phenomena
not considered by the other global models leads to more complete and better results.
• On a global scale the model describes very well the behaviour of reinforced concrete shell
elements and the latter is mainly governed by damage through the model parameter k0 .
At the local scale the lack of some phenomena, not yet considered in the model, leads to
results not too far from experimental ones.
• In the case of pure bending, I could see that for some parameters the range of use changes
slightly. In particular in my case, the need to have more damage led me to use a value of
σd = fc /6. Future studies for the case of pure bending out of plan could give confirmation
and improve the calibration of the parameters.
• For all the models the same non-linear parameters were used except those relating to the
damage to the concrete (k0 and γd ). This aspect makes it possible to standardize the model
and further eliminate the small uncertainty that governs the choice of parameters that must
be used.
• Multiple margins for improvement are possible for the model described. The addition of the
third family of cracks is being developed by Olivier Lherminier and the case of bending out
of plane still requires multiple studies. With the further addition of the concrete permanent
strain in compression, more refined and better results will be achieved.

66
Appendix A

This appendix is devoted to the estimation of the parameters which, even if their physical
signification is clear, classical tests do not give enough information to determine them. Some
indications about the determination of these parameters are given below:
• In absence of experimental tests values, the fracture energy Gf can be estimated by the
expression given by MC10:
Gf = 73fc0.18 (A.1)
where concrete compressive strength fc is expressed in MPa and fracture energy Gf in
J/m2
• Parameters αu and αr , which define concrete bridging stress cyclic behaviour, may depend
on concrete characteristics, especially on the aggregate size and form. If no tests are carried
out to identify them, it is recommended to use an unloading slope ratio αu ∈ [0.01,0.2] and
crack reclosing ratio αr ∈ [0.01,0.1]. It is remarked that when the applied loading creates
significant tangential crack displacements wt , the crack reclosing ratio should be relatively
high, in order to reproduce the dilatancy effect, which is not directly considered in the
model formulation.
• Aggregate interlock parameters T0 and T1 may depend on the aggregate characteristics
and the type and magnitude of the loading. Recommend values are T0 ∈ [0.01,0.1] MPa
and T1 ∈ [1,20] GPa/m, where the lower bounds of the ranges are related to expected high
values of crack opening wn (or low values of normal concrete stress at cracks) and to cyclic
loadings.
• The theoretical average crack spacing of the equivalent tie beams in the x and y directions
can be estimated with civil engineering codes formulas for maximum crack spacing, after
the transformation to average spacing values by dividing them by 1.7 (for the case of MC10
and EC2). However, it is recommended to use the following optimized formula for average
crack spacing:

srα = 1.37cα + 0.116Φα /ρsα (A.2)

where cα is the concrete cover of steel reinforcement bars in the α direction. Finally, the
experimental values for the the average crack spacing should be used when available.
• The energy release rate threshold k0 is calculated from the concrete stress at the demage
onset σd with:

(1 − γd )σd2
k0 = (A.3)
2Ec
67
 where it is recommended to use γd ∈ [0.2,0.3] for SLS and σd ∈ [fc /4, fc /2].
• The local bond-slip tangent stiffness can be estimated to Kl ∈ [1010 ,1011 ] Pa/m, depending
on the characteristic values of the obtained steel-concrete slip since it should correspond to
the secant stiffness of realistic bond-slip laws as the given by MC10.
• The retained tension stiffening coefficient may vary between kt ∈ [0.1,0.6]:
– kt = 0.6 when calculating crack openings with the same assumptions as MC10 and
EC2.
– kt ≈ 1/3 when the computation is done for representing the monotonic mechanical
behaviour.
– kt ∈ [0.1,0.2] when time dependent effects in concrete are important, or in cyclic
loadings implying bond degradation.

68
Appendix B

DEFI_MATER Values Explication Material

ELAS_HEGIS =
RHO = 2500 Density RC
_F(

ALPHA = RC
AMOR_ALPHA = RC

AMOR_BETA = RC

AMOR_HYST = RC

E_C = Concrete Young’s modulus Concrete

NU_C = Concrete Poisson coeff. Concrete

E_S = Steel Young’s modulus Steel

A_S (A_SX_SUP, A_SY_SUP,

= Steel areas Shell
A_SX_INF, A_SY_INF)

ZINF, ZSUP = Position of steel bars Shell

GLRC_HEGIS =
FSY = Yielding stress of steel Steel
_F(

FCT = Concrete tensile strength Concrete

GF = Cracking energy Concrete

0.01 - 0.2 Slope of discharge of concrete traction

ALPHD = Concrete
(0.05 default) (fraction of E_C)

0.01 - 0.1 Crack reclosure (fraction of

ALPHR = RC
(0.05 default) historical maximum opening)

SR (SRX_SUP, SRY_SUP,
= Theoretical crack spacing RC
SRX_INF, SRY_INF)

K0 = Threshold of energy dissipation Concrete

Asymptotic damage
GAMMD = 0.2 - 0.3 Concrete
(fraction of E_C)

(0.01 - 0.1)
T0 = Aggregate interlock threshold Concrete
MPa

(1 - 20)
T1 = Aggregate interlock slope Concrete
GPa/m

0.1 - 0.6
KT = Coeff. tension stiffening RC
(0.33 default)

KL = 1010 - 1011 Sliding stiffness steel-concrete RC

PHI (PHI_XS, PHI_YS, PHI_XI, PHI_YI) = 1010 - 1011 Diameters of the bars Shell

Table B.1: Parameters of GLRC_HEGIS model.

69
Internal variable Meaning Unit
V1 Normal crack opening, grid 1, top layer m
V2 Tangential crack opening, grid 1, top layer m
V3 Normal crack opening, grid 2, top layer m
V4 Tangential crack opening, grid 2, top layer m
V5 Steel-concrete slip, X direction, top layer m
V6 Steel-concrete slip, Y direction, top layer m
V7 Plastic steel deformation, X direction, top layer -
V8 Plastic steel deformation, Y direction, top layer -
V9 Damage variable, top layer -
V10 Crack spacing, grid 1, top layer m
V11 Crack orientation, grid 1, top layer rad
V12 Maximum historical normal crack opening, grid 1, top layer m
V13 Maximum historical tangential crack opening, grid 1, top layer m
V14 Crack spacing, grid 2, top layer m
V15 Crack orientation, grid 2, top layer m
V16 Maximum historical normal crack opening, grid 2, top layer m
V17 Maximum historical tangential crack opening, grid 2, top layer m
V18 Normal crack opening, grid 1, lower layer m
V19 Tangential crack opening, grid 1, lower layer m
V20 Normal crack opening, grid 2, lower layer m
V21 Tangential crack opening, grid 2, lower layer m
V22 Steel-concrete slip, X direction, lower layer m
V23 Steel-concrete slip, Y direction, lower layer m
V24 Plastic steel deformation, X direction, lower layer -
V25 Plastic steel deformation, Y direction, lower layer -
V26 Damage variable, lower layer -
V27 Crack spacing, grid 1, lower layer m
V28 Crack orientation, grid 1, lower layer rad
V29 Maximum historical normal crack opening, grid 1, lower layer m
V30 Maximum historical tangential crack opening, grid 1, lower layer m
V31 Crack spacing, grid 2, lower layer m
V32 Crack orientation, grid 2, lower layer m
V33 Maximum historical normal crack opening, grid 2, lower layer m
V34 Maximum historical tangential crack opening, grid 2, lower layer m
V35 Energy (per unit area) dissipated by normal crack opening-closing, grid 1 + 2, sup + inf layer J/m2
V36 Energy (per unit area) dissipated by tangential crack opening-closing, grid 1 + 2, sup + inf layer J/m2
V37 Energy (per unit area) dissipated by the inelastic steel-concrete slip, X + Y directions, sup + inf layer J/m2
V38 Energy (per unit area) dissipated by the plastic steel deformation, X + Y directions, sup + inf layer J/m2
V39 Energy (per unit area) dissipated by the compression concrete damage, sup + inf layer J/m2
V40 Energy (per unit area) dissipated total (= V35 + V36 + V37 + V38 + V39) J/m2
V41 Decrease (both per one) of the membrane stiffness of the plate in RC -
V42 Decrease (both per one) of the bending stiffness of the plate in RC -
V43 Stress of steel, X direction, top layer Pa
V44 Stress of steel, Y direction, top layer Pa
V45 Stress of steel, X direction, lower layer Pa
V46 Stress of steel, Y direction, lower layer Pa
V47 Maximum stress in steels Pa

Table B.2: Internal variables of the GLRC_HEGIS model.

70
Bibliography

[1] EDF. Open source on www.code-aster.org, Structures and Thermomechanics Analysis for
Studies and Research, online documentation. https://www.code-aster.org/V2/doc/v14/
fr/index.php?man=commande. 1989–2019 (cit. on p. 13).
[2] M Huguet. «Homogenised stress resultant constitutive model for cracking in RC plates
under seismic loadings». PhD thesis. Ecole Centrale de Nantes, France: Egis industries,
Institut de Recherche en Génie Civil et Mécanique, 2016 (cit. on pp. 14, 24, 47, 50).
[3] M Huguet, M Bourahla, S Erlicher, and P Kotronis. «GLRC_HEGIS global constitutive
model for RC walls and slabs for seismic nonlinear structural analyses». In: Proceedings
of the 16th European Conference on Earthquake Engineering. Thessaloniki, Greece, 2018,
pp. 1–12 (cit. on p. 14).
[4] J Walraven. «Fundamental Analysis of Aggregate interlock». In: Journal of the Structural
Division 107(11) (1981), pp. 2245–2270 (cit. on p. 20).
[5] B Li, K Maekawa, and H Okamura. «Contact density model for stress transfer across cracks
in concrete». In: Journal of the Faculty of Engineering, The University of Tokyo 40(1)
(1989), pp. 9–52 (cit. on p. 21).
[6] P G Gambarova. «Sulla trasmissione del taglio in elementi bidimensionali piani di c.a.
fessurati». In: Proceedings of Giornate AICAP, Bari, Italy, 1983, pp. 141–156 (cit. on p. 21).
[7] P Rivillon and A Gabs. «Rapport d’essais 09 26023877 Voile A». In: CSTB (2011) (cit. on
pp. 25, 26, 33).
[8] P Rivillon and A Gabs. «Rapport d’essais 09 26023877 Voile B». In: CSTB (2011) (cit. on
pp. 25, 34).
[9] P Rivillon and A Gabs. «Rapport d’essais 09 26023877 Voile C». In: CSTB (2011) (cit. on
pp. 25, 36).
[10] P Rivillon and A Gabs. «Rapport d’essais 09 26023877 Voile D». In: CSTB (2011) (cit. on
pp. 25, 37).
[11] P Bisch, S Erlicher, M Huguet, and G Ruocci. «Experimental and theoretical results on
cracking of concrete walls submitted to cyclic shear forces». In: Proceedings of the 16th
European Conference on Earthquake Engineering. Thessaloniki, Greece, 2018, pp. 1–12
(cit. on p. 27).
[12] EDF. SSNS106 – Dégradation d’une plaque en béton armé sous sollicitations variées avec
les lois globales GLRC_DM et DHRC. https://www.code-aster.org/V2/doc/v14/fr/
man_v/v6/v6.05.106.pdf. 2019 (cit. on p. 47).
[13] EDF/ECN. Poutre_ECN. https://cheops.necs.fr/fydex/show/Poutre_ECN/. 2002
(cit. on p. 53).

71