Académique Documents
Professionnel Documents
Culture Documents
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · ce/papers 1 (2017), No. 2 & 3
https://doi.org/10.1002/cepa.395 wileyonlinelibrary.com/journal/cepa 3404
| 3405
Romania is the only country in the Europe that introduced BRBs into the National Anti-Seismic
Design Code, P100-1/2013 [4], since January 1st, 2014. Project based experimental qualification is
needed when using BRBs in real structures. Therefore, the main objective of this research is the pre-
qualification of a set of BRBs, covering a range of capacities suitable for low-rise and mid-rise
buildings. The experimental program and the specimens IDs are summarized in Table 1.
Within this paper, the numerical pre-testing program is presented for the conventional BRBs. The
outcomes of the numerical analyses were used to develop and design the final engineering solutions
that are going to be manufactured and then tested in view of prequalification. Several parameters
were analysed in order to identify the optimal configuration that will lead to the desired behaviour
of the BRB. Parameters as core aspect ratio, gap size, critical elastic axial load of the buckling-
restraining mechanism (Pe), friction coefficient were investigated. The performance of the BRBs
was analysed in terms of compression adjustment factor (β), core axial strain, cumulative inelastic
deformation (CID), and global performance.
2 FEM MODELLING
2.1 Calibration
The conceptual development of the BRB devices had its basis on numerical investigations
performed on finite element models, using Abaqus/CAE 6.14.4. The numerical model adopted was
calibrated against experimental data available within the technical literature [5]. The numerical
simulations performed with the FE model adopted had the purpose of providing information on
specific input parameters (vertical and horizontal core to concrete gap, contact law definition,
friction coefficient) that will be necessary to model and design the experimental specimens. Using
the BRB’s geometry from [5], a similar finite element model was constructed (see Figure 1Figure
1).
All of the components (steel core, infilled concrete, steel casing) were directly modelled using 3D
finite elements. The unbonding layer was modelled by the use of a small gap (through thickness
direction, 1 mm; through width direction 0.2 mm); and a contact law (normal behaviour set as
“hard” contact; tangential behaviour set as penalty/friction = 0.1). Due to the complex nonlinearity
and type of loading, cyclic, the Dynamic Explicit solver was used. To reduce the dynamic effects, a
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)
3406 |
mass scaling factor equal to 0.0001 and a time of 79 s was used, assuring the quasi-static
application of the load.
To predict a proper spatial stress state, all the components were modelled with tridimensional
deformable solid elements. The discretization was performed using C3D8I hexahedral, 8 node
linear variation of displacements along the element’s edges. This type of finite element was used in
order to prevent shear locking and hourglass effects that can appear when using first order finite
elements loaded in bending (C3D8, C3D8R). Partitions were done to allow for a structural
discretization of the parts.
The secondary system (steel casing and infilled concrete) was modelled using elastic material
models. Therefore, the Young’s moduli (E), the Poisson’s coefficients (ν), and the material’s
densities (ρ) were needed. The following values were used to define the concrete (Ec = 21000 MPa,
ν = 0,18, ρ =2,5 E-9 tonne/mm3) and the steel behaviour (Es = 210000 MPa, ν =0,3, ρ = 7,85 E-9
tonne/mm3). The material associated with the dissipative core was chosen with respect to the type of
loading that was applied: monotonic or cyclic. For the case of monotonic loading, isotropic
hardening plastic model was used, based on Hubert von Mises yield criterion (fy = 282 MPa, fu =
510 MPa, fr = 490 MPa). In the case of cyclic loading, combined isotropic/kinematic hardening
plasticity model was used, which consists of two components: a nonlinear kinematic behaviour
(superposition of several backstresses, m = 5: (1) C1 = 25000 MPa, g 1 = 500; (2) C2 = 21000 MPa,
g 2 = 375; (3) C3 = 5950 MPa, g 3 = 120; (4) C4 =935 MPa, g 4 = 25; (5) C5 = 300 MPa, g 5 = 0);
and a nonlinear isotropic behaviour (Qinfinity = 60 MPa and b = 4). Using this approach, a good
prediction can be obtained, including the Baushinger effect.
A general contact, “All with self”, was defined, using a friction coefficient equal to 0,1 to account
for the friction forces that may take place between the exterior surface of the core and the interior
surface of the concrete. A “Tie” constraint was also defined between the exterior surface of the
concrete and the interior surface of the steel casing. Boundary conditions were defined at the end of
the steel core (fixed ends). The load was applied as a displacement controlled, using the “smooth
step” method.
After performing the analyses, a good correlation between the experimental and the numerical
results was observed. The numerical model was able to capture all the main characteristics of the
tested BRB: initial elastic stiffness, yield stress and displacement, strain hardening, the energy
dissipated. The comparison between experimental and numerical results is presented in Figure 2.
B
3628
3850
A B A
2
Core Core
45
2
Concrete Concrete
193,7
CHS
99
Unbonding Unbonding
CHS
2
20 2 180,5 6,6
180,5 6,6 193,7
The pre-testing numerical program was perform by using some of the information obtained from
the calibration tests (friction coefficient 0,1; gap 1 mm; FE type C3D8I; contact laws; material
definition approaches). Different types of solvers were preliminarily used in order to establish the
proper one to be used for the whole pre-testing program. Static General, Dynamic Implicit (quasi-
static), and Dynamic Explicit were initially used. Due to the complex nonlinearity (material,
geometry (imperfections), contact laws) the first two solvers took a large amount of computational
time. Therefore, the Dynamic Explicit solver was used, with similar results as the implicit solvers.
To validate the results, the output energies were compared assuring that a quasi-static analysis is
being performed: artificial strain energy and kinetic energy were kept to a minimum when
comparing to internal energy and external work (less than 1%). It must be mentioned that mass
scaling was used, with a target increment of 5E-5.
The FE model used in the experimental program consists of three parts (steel core, concrete, CHS
tube with two end-caps), gaps and interactions between parts. The core was modelled using C3D8I
finite element type, with a size of 10 mm for the plastic zone and 20 mm for the elastic zone.
Several C3D6 elements were used to facilitate the mesh propagation for the transition zones. The
concrete part was modelled using C3D8R elements due to its less significant contribution to the
global behaviour of the BRB. The tube, including a cap at each end, was modelled using shell
elements, S4R, with five integration points per thickness. The thickness of the CHS varied
depending on the specimen. The assembly, the independent discretized constitutive parts and the
gaps are presented in Figure 4. Gaps are placed near the transition zones to allow for free movement
of the core in compression.
When modelling BRBs, interactions are very important. Therefore, a general contact was defined,
and the contact domain consisted of two selected surface pairs that were assigned different contact
properties as follows: the core – to – concrete interaction was defined as having the tangential
behaviour defined as “Penalty” with a friction coefficient of 0.1 and the normal behaviour set to
“Hard” contact; the concrete – to- steel casing interaction had the same properties except the
friction coefficient which was set to 0.4 in to account for the partial composite behaviour.
A coupling constraint was defined at each end of the core, by connecting a reference point to a
surface using the “Continuum distributing” coupling type, allowing for free transversal deformation
of the selected surface.
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)
3408 |
= + + ;
The BRB was considered pinned at both ends, while the load was applied in displacement control.
In order to include imperfections into the numerical model, a previous buckling analysis, where the
two contact pairs were redefined as “Tie” constraints, was performed. The first buckling mode was
scaled to the length of the BRM divided by 1000, LBRM/1000 = 3,7 mm, and then used as
“*Imperfection” for the main model.
For the concrete part, just the elastic behaviour definition was assigned, using the following
parameters: Ec = 29108 MPa, ν = 0.2; ρ = 2.5E-9 tonnes/mm3. Plastic behaviour was also used
within preliminary analyses, but no difference in results was noticed, just an increased in
computational time (and convergence problems), therefore just the elastic definition was used.
The steel casing and the core were modelled using the combined isotropic/kinematic hardening
plasticity model. The model consists of a nonlinear kinematic and a nonlinear (exponential law) or
multilinear (tabular data) isotropic hardening component. A very good prediction can be obtained
with this model for both monotonic and cyclic analyses. Due to the fact that in [5] the material used
to manufacture the steel core was not a European mild carbon steel, as is the one used within this
research program, another reference was used for calibration of the material’s parameters [6]. After
performing the calibration (see Figure 5), the input parameters to be used within this numerical
program, for both kinematic and isotropic components, were determined by scaling up the
calibrated parameters obtained from the experimental tests performed by [6].
a) b)
Figure 5. Calibration of material model for monotonic and cyclic
The calibration procedure described in [7] was used. The input parameters for kinematic hardening
were determined using the “Stabilized” data type option. Yield stress – plastic strain data pairs were
introduced, and five backstresses were requested for better prediction. Small adjustments were
necessary for finer prediction of kinematic behaviour. In the following analyses, the above
calibrated backstresses were used: (1) C1 = 18518 MPa, g 1 = 954; (2) C2 = 13855 MPa, g 2 = 184;
(3) C3 = 966,2 MPa, g 3 = 15; (4) C4 =480 MPa, g 4 = 4; (5) C5 = 200 MPa, g 5 = 0,1). For the
isotropic behaviour, a multilinear isotropic softening (up to the end of plateau), followed by
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)
| 3409
isotropic hardening was used. The use of softening was necessary in order to obtain the plateau. The
isotropic behaviour input curve is presented in Figure 6.
Abaqus input: true σ − ε
600
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)
3410 |
Uniaxial Behavior: F - D
1500
Yield Point, 6.47, 878.00
1000
500
Force [kN]
0
Legend:
-500 Yield Point
T
-1000 C
Cyclic
-1500
-200 -150 -100 -50 0 50 100 150 200 250 300 350
Displacement [mm]
450 CR73-g2-f.1
0
-450
-900
-1350
-1800
-150 -100 -50 0 50 100 150
Displacement [mm]
450
0
-450
-900
-1350
-1800
-150 -100 -50 0 50 100 150
Displacement [mm]
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)
3412 |
450
0
-450
-900
-1350
-1800
-150 -100 -50 0 50 100 150
Displacement [mm]
450 CR73-g12-f.15
0 CR73-g12-f.2
-450
-900
-1350
-1800
-150 -100 -50 0 50 100 150
Displacement [mm]
450 LP.eps.3.0%
0 LP.eps.3.5%
-450
-900
-1350
-1800
-150 -100 -50 0 50 100 150
Displacement [mm]
Figure 12. Cyclic behaviour of CR73-g12-f.1 at different levels of axial core strain
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)
| 3413
3 CONCLUSIONS
Finite element based numerical analyses were performed on two “conventional” BRBs with
rectangular and square cores. Calibration of numerical model based on experimental tests was
performed in order to calibrate representative input parameters that define the BRB’s behaviour. A
pre-testing numerical program was performed, in order to determine the optimal conceptual design.
From the parametric study, it can be concluded that optimal results can be obtained: by using a
nonuniform gap layout (1-2-2-1 mm) around the rectangular steel core (critical); by using a BRM
with a lateral stiffness greater than 1.5 (non critical); by using a compact square core shape (non
critical); by assuring a friction coefficient less than 0.15 (critical). The experimental loading
protocol should be constructed using an axial core strain demand not greater than 2,75% as criterion
for determining the deformation in BRB corresponding to the design story drift (∆bm).
ACKNOWLEDGMENTS
The research leading to these results has received founding from the MEN-UEFISCDI grant
Partnerships in priority areas PN II, contract no. 99/2004 IMSER: Implementation into Romanian
seismic resistant design practice of buckling restrained braces.” This support is greatly
acknowledged.
REFERENCES
[1] Xie, Q., “State of the art of buckling-restrained braces in Asia”, Journal of Constructional Steel
Research 61, pp. 727-748, 2005.
[2] CEN - European Committee for Standardization. "Eurocode 8: Design of structures for earthquake
resistance. Part 1: General rules, seismic actions and rules for buildings”, Eurocode 8-1/2008, 2008
[3] CEN - European Committee for Standardization. "Anti-seismic devices”, EN 15129:2010, 2010
[4] P100-1/2013, „National Seismic Design Code – Part I – Design Provisions for Buildings”, 2013
[5] Dunai, L., Zsarnóczay, A., Kaltenbach, L., Kálló, M., Kachichian, M., Halász, A., "Type testing of
Buckling Restrained Braces according to EN 15129 – EWC800 – Final report", 2011
[6] Kaufmann E.J., Metrovich B., Pense A.W., “Characterization of Cyclc Inelastic strain behavior on
properties of A572 Gr. 50 and A913 Gr. 50 Rolled Sections”. ATLSS Report No. 01-13, American
Institute of Steel Construction, December, 2001
[7] ABAQUS “ABAQUS Documentation”, Dassault Systèmes, Providence, RI, USA, 2014
[8] Watanabe, A., Hitomi, Y., Saeki, E., Wada, A., Fujimoto, M., “Properties of brace encased in buckling
restraining concrete and steel tube”, Proc. 9th World Conf. Earthquake Engineering, vol. 4, pp. 719-
724, 1988
[9] Black, C., Makris, N., Aiken, I., "Component Testing, Stability Analysis and Characterization of
Buckling-Restrained Unbonded BracesTM". PEER Report 2002/08. Pacific Earthquake Engineering
Research Center, College of Engineering, University of California, Berkeley, 2002
[10] Iwata, M., Murai, M., “Buckling-restrained brace using steel mortar planks; performance evaluation as
a hysteretic damper”, Earthquake Engineering and Structural Dynamics, 35, pp. 1807-1826, 2006
[11] ANSI/AISC 341-10, “Seismic Provisions for Structural Steel Buildings”, American Institute of Steel
Construction, Inc. Chicago, Illinois, USA, 2010
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin ∙ CE/papers (2017)