You are on page 1of 5

Non-linear homogenized and heterogeneous

Fe models for FRCM reinforced masonry


walls out-of-plane loaded
Cite as: AIP Conference Proceedings 2116, 420009 (2019); https://doi.org/10.1063/1.5114436
Published Online: 24 July 2019

Elisa Bertolesi, Gabriele Milani, and Bahman Ghiassi

AIP Conference Proceedings 2116, 420009 (2019); https://doi.org/10.1063/1.5114436 2116, 420009

© 2019 Author(s).
Non-Linear Homogenized and Heterogeneous FE Models for
FRCM Reinforced Masonry Walls Out-Of-Plane Loaded
Elisa Bertolesi1, Gabriele Milani2,a, Bahman Ghiassi3

ICITECH, Universitat Politècnica de Valencia, Camino de Vera s/n 46022, Valencia, Spain
1

Department of Architecture, Built environment and Construction engineering (ABC), Politecnico di Milano,
2

Piazza Leonardo da Vinci 32, 20133 Milano, Italy


3Faculty of Engineering, University of Nottingham, Coates Building, University Park, Nottingham NG7 2RD, UK
a
gabriele.milani@polimi.it

Abstract. Two distinct non-linear FE modeling techniques are compared to have an insight into the efficacy of FRCM
reinforcement for masonry subjected to out-of-plane loads. In particular, both a micro-modeling technique and a
homogenization approach are compared. The first approach is a tridimensional heterogeneous procedure where
constituent materials (bricks, joints, reinforcing mortar and reinforcing grid) are modeled separately. The second
technique is a consolidated two-step homogenization where the meso-scale homogenization problem is solved
discretizing the elementary cell with few elastic constant stress triangles (bricks) and non-linear interfaces (joints). The
non-linear structural analyses are performed replacing masonry at the macro-scale with an assemblage of rigid elements
interconnected by non-linear homogenized springs (HRBSM modelling). Both models are directly implemented in the
commercial software Abaqus. Advantages and limitations of the two approaches are discussed in detail –especially as far
as the rather different computational effort required by the two strategies is concerned- with reference to their ability in
reproducing global force-displacement curves and crack patterns of some reinforced wallettes in simple bending.
Keywords: Homogenization approach; Heterogeneous approach; FE numerical model; Masonry strengthening; FRCM
retrofitting; masonry in bending.

INTRODUCTION
With the aim of both preserving and upgrading in seismic area masonry, in the last decades various studies have
been conducted to analyze the effects of different kinds of reinforcing techniques when applied to traditional
supports [1]-[3]. In this framework, several reinforcing materials have been proposed to retrofit masonries, the most
successful being FRP (Fiber Reinforced Polymer) and FRCM (Fiber Reinforced Cementitious Matrix) composite
materials. Nowadays FRP is not anymore considered suitable for the reinforcement of historical constructions, due
to durability, permeability and excessive stiffness issues. This is the reason why now an innovative cement-based
strengthening material, known in the technical literature as FRCM, is spreading [1]. FRCM relies into a cementitious
plaster reinforced with a fiber grid. The presence of continuous long fibers embedded into a cementitious matrix
ensures the improvement of the global mechanical behavior of weak masonries in addition to the preservation of
traditional supports. Full reversibility is another advantage when compared with FRPs. Unfortunately, the use of a
cement based matrix, which contains aggregates, prevents the deep impregnation of the fibers thus resulting in a
different failure mechanism with respect to FRP composites. From a numerical point of view, such peculiar behavior
can be considered adopting sophisticated modeling strategies that may be computationally too demanding when
large scale structures have to be analyzed. Here, two numerical strategies with increasing level of accuracy –namely
a heterogeneous 3D and a homogenization 2D approach- are adopted and compared in the prediction of the behavior
of full scale walls reinforced with FRCM and subjected to vertical bending and experimentally tested by Nanni and
co-workers in [4][5]. Advantages and limitations of the two approaches are discussed in detail –especially as far as
the rather different computational effort required by the two strategies is concerned- with reference to their ability in
reproducing (i) global moment-displacement curves, (ii) failure mechanisms and (ii) crack patterns found at
collapse.

International Conference of Numerical Analysis and Applied Mathematics (ICNAAM 2018)


AIP Conf. Proc. 2116, 420009-1–420009-4; https://doi.org/10.1063/1.5114436
Published by AIP Publishing. 978-0-7354-1854-7/$30.00

420009-1
EXPERIMENTAL DATA USED AS BENCHMARK
Unreinforced and FRCM reinforced full scale masonry wall were tested [4][5] in simple vertical bending by Nanni
and coworkers, with the aim of assessing the efficiency of FRCM composites for the reinforcement of masonry
structures out of plane loaded. The full experimental campaign relies into a total of twelve full scale panels, one half
reinforced using a cement-based strengthening technique, the rest tested unreinforced. Two different types of blocks
were used to build the panels, namely concrete blocks and clay bricks, both joined by a 10 mm thick M type mortar.
Panels dimensions are the following: 1422x1220x92 mm3 and 1422x1220x102 mm3 for concrete and clay units,
respectively. FRCM strengthening was installed on the tensile surface through a bi-directional balanced carbon fiber
grid embedded into a 10 mm thick cementitious mortar. The experimental set-up was designed to apply a horizontal
uniformly distributed pressure by means of air bags interposed between samples and a reaction wall. Details on the
experimental set-up are available in [2][3], where the reader is referred to for further details. Maximum horizontal
displacements were registered at the center of the walls. The results, in terms of bending moment versus mid-height
deflection are depicted in FIGURE 1 for unreinforced and FRCM strengthened clay panels. As clearly visible, the
experimental response of the as built walls is characterized by an initial elastic first phase followed by a brittle
collapse. As expected, all the as built panels exhibited a collapse mechanism characterized by the formation of a
well-defined horizontal crack at the mid-height bed joint. In order to faithfully reproduce the experimental setup, for
the unreinforced walls the external pressure was increase up to failure in a single cycle of loading, whereas for the
strengthened walls a series of loading and unloading cycles were applied. Whilst the unreinforced panels failed for
brittle cracking of the middle horizontal joint, failure of the reinforced specimens occurred with the formation of
several horizontal cracks near the mid-height section that developed in the reinforcing cementitious layer. The
opening of the cracks was accompanied by fabric slippage until the collapse of the walls, which obviously was much
more ductile.

(a) (b)
FIGURE 1. Experimental results: (a) unreinforced clay brick panels and (b) FRCM strengthened walls.

NUMERICAL MODEL AND RESULTS


The first approach considered is a detailed 3D heterogeneous discretization where mortar joints and bricks are
modeled separately by means of eight-noded FEs.
In order to drastically decrease the number of FEs needed in the analyses, only one fourth of the original geometry
of the panels (see FIGURE 2) was discretized, resulting into a mesh constituted by 48466 and 34586 FEs,
respectively for clay and concrete walls. The final meshes were conceived in such a way to allow the presence of at
least two FEs along the thickness of both head and bed joints. As far as the FRCM reinforcement is concerned, truss
elements were used for the fiber net, whereas cementitious mortar was again modelled with eight-noded brick
elements disposed into two rows. Globally, the performance of the model was evaluated comparing the results
obtained with the experimental ones on moment-deflection curves experimentally provided, where monitored
displacements where those registered out-of-plane in the mid-height section.

420009-2
The second numerical approach considered to fit masonry behavior in simple vertical bending is the simplified two-
step homogenization procedure formulated in [6]-[8]. In the first step, the arrangement of bricks and mortar joints is
analyzed at the meso-scale. The unit cell is discretized with 24 elastic constant stress triangles (bricks), whereas
mortar joints are reduced to inelastic zero-thickness interfaces. Xu-Needleman exponential holonomic constitutive
relationships as well as piece-wise linear laws between micro-stresses (normal σ and tangential τ) and total jumps of
displacements are adopted. Where possible, numerical parameters adopted fit experimental data, in particular as far
as initial stiffness, peak tensile (fnI) and tangential strength (ftI).

Equivalent truss
(a) (b) (c)
106 Experimental test #1
8 Experimental test #2
Bending moment M [Nmm]

Experimental test #3
Detailed hetergeneous model
6 Homogenized approach

0
0 5 10 15
Deflection [mm]
(d) (e)
FIGURE 2. (a), (b) and (c) heterogeneous approach. (a) FE 3D heterogeneous discretization used to model 1/4 of the
unreinforced concrete block specimens; (b) deformed shape with indication of cracks (tensile damage maps) obtained at failure
for the clay brick panel (magnified 400 times); (c) detail of deformed shape and indication of cracks (on both mortar joints and
FRCM) for the reinforced case. (d) homogenized HRBSM approach, unreinforced case, deformed shape at collapse with
indication of interfaces with cracks. (e) comparison between experimental and numerical bending-moment deflection curves.

The homogenized stress-strain curves for the in-plane case are obtained solving a particularly simple
homogenization problem formulated on the unit cell, as specified comprehensively in [6]-[8]. Out-of-plane behavior
here used is obtained assuming a Kirchhoff-Love plate hypothesis and hence by a simple on-thickness integration.
During the second step, performed at a structural level. The non-linear structural analyses are performed replacing
masonry at the macro-scale with an assemblage of rigid elements interconnected by non-linear homogenized
springs, i.e. using a so-called HRBSM modelling approach (the reader is referred to [6]-[8] for further details).
Thanks to the impressive reduction of variables needed in the homogenization approach, the entire panels can be
easily modelled, differently from the heterogeneous approach, where only ¼ of the mesh is considered to speed up
computations, see FIGURE 2. The resulting discretization is composed by 504 rigid quadrilateral elements

420009-3
interconnected by homogenized springs exhibiting inelastic behavior with softening for both the reinforced and
unreinforced case. HRBSM discretization is directly imported into the commercial software Abaqus [9], where all
the proposed simulations have been performed using for interfaces a concrete damage plasticity model (CDP) to
describe both tensile and compressive damaging processes, assuming dilatation angle ψ=10°, flow potential
eccentricity e=0.1 and Kc=0.667.
A satisfactory agreement is found comparing the numerical predictions and the experimental results obtained, using
both the numerical strategies. In particular, it is worth mentioning how the homogenization model is able to properly
capture the elastic branches as well as the maximum bending moments in both the reinforced and unreinforced case,
see FIGURE 1(a) and FIGURE 2(e). In general, very good results are obtained on the global bending-moment
deflection curves obtained also with the sophisticated tri-dimensional heterogeneous approach. In this latter case, the
response of the models resulted slightly less stiff with respect to the experimental envelope. Crack patterns obtained
with both approaches are again in excellent agreement with experimental evidences. As expected, in the
unreinforced case failure was a consequence of a crack spreading on the central mortar joint, whereas when dealing
with the FRCM reinforced case, damage spreads laterally starting from the mid-height section, contributing in an
increase of both the load carrying capacity and the ductility of the reinforced system. Failure occurs for progressive
slippage phenomena occurring at the interface between fibers and cementitious matrix. No tensile rupture of the
equivalent trusses was experienced in all the simulations performed. Globally, both the numerical strategies are able
to describe the behavior of the masonry panels analyzed, even if with different levels of detail.
CONCLUSIONS
The paper presented a series of numerical results obtained using two different strategies for FRCM reinforced
masonry panels in simple vertical bending. The numerical predictions were benchmarked with reference to some
existing experimental results obtained at the University of Miami on unreinforced and FRCM reinforced masonry
panels in vertical bending and subjected to uniform distributed load incremented up to collapse. The first approach
considered was a 3D heterogeneous procedure where constituent materials (bricks, joints, reinforcing mortar and
reinforcing grid) were modeled separately. The second technique was a consolidated two-step homogenization
where the meso-scale homogenization problem was solved discretizing the elementary cell with few elastic constant
stress triangles (bricks) and non-linear interfaces (joints). The non-linear structural analyses were performed with a
so-called HRBSM model, i.e. replacing masonry at the macro-scale with an assemblage of rigid elements
interconnected by non-linear homogenized springs. Both models shown a promising agreement with experimental
evidences, both in reproducing global force-displacement curves and crack patterns observed experimentally.
REFERENCES
1. T.C. Triantafillou, A new generation of composite materials as alternative to fiber reinforced polymers for
strengthening and seismic retrofitting of structures. Proc. Composite materials. A vision for the future, L.
Nicolais, M. Meo & E. Milella eds., Springer-Verlag London Limited, 2011.
2. F.G. Carozzi, C. Poggi, E. Bertolesi, G. Milani, Ancient masonry arches and vaults strengthened with TRM,
SRG and FRP composites: Experimental evaluation. Composite Structures, 187, 466-480, 2018.
3. E. Bertolesi, G Milani, F.G Carozzi, C. Poggi, Ancient masonry arches and vaults strengthened with TRM,
SRG and FRP composites: Numerical analyses. Composite Structures, 187, 385–402, 2018.
4. S. Babaeidarabad, F. De Caso, A. Nanni, Out-of-plane behavior of URM walls strengthened with fabric-
reinforced cementitious matrix composite. J. Compos. Constr., 18(4), 04013057, 2014.
5. S. Babaeidarabad, G. Loreto, D. Arboleda, A. Nanni, FRCM-strengthened CMU masonry walls subjected
to out-of-plane load. The Masonry Society Journal, 32(1), 2014.
6. E. Bertolesi, G. Milani, P.B. Lourenço, Implementation and validation of a total displacement non-linear
homogenization approach for in-plane loaded masonry. Comput. Struct., 176, 13-33, 2016.
7. G. Milani, E. Bertolesi. Quasi-analytical homogenization approach for the non-linear analysis of in-plane
loaded masonry panels. Construction and Building Materials, 146(15), 723–743, 2017.
8. E. Bertolesi, G. Milani, C. Poggi, Simple holonomic homogenization model for the non-linear static
analysis of in-plane loaded masonry walls strengthened with FRCM composite. Compos. Struct., 158, 291-
307, 2016.
9. ABAQUSTM. Finite Element Analysis, v6.6. Theory Manual. SIMULIA, Inc.: Maastricht, 2006.

420009-4

You might also like