Construction and Building Materials 156 (2017) 387–401

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Compressive strength of confined column with Fiber Reinforced Mortar

(FRM): New design-oriented-models
Alessio Cascardi ⇑, Fabio Longo, Francesco Micelli, Maria Antonietta Aiello
Dept. of Engineering for Innovation, Univ. of Salento, Lecce, 73100, Italy

h i g h l i g h t s

State of the art on the confinement of masonry and concrete with FRM.

State of the art on analytical prediction of the compressive strength of FRM-column.

Two new DOMs for the FRM-confined column.

Two simplified versions of the proposed DOMs.

a r t i c l e i n f o a b s t r a c t

Article history: As well known the confinement is an effective intervention for increasing the compressive strength of
Received 20 May 2017 an axially loaded column. During the decades, different materials have been adopted in order to pro-
Received in revised form 26 July 2017 vide confining action, such as steel, reinforced concrete and latest high-performance fibers. Fibers
Accepted 1 September 2017
application resulted a powerful solution mainly because of the high ratio between mechanical prop-
Available online 9 September 2017
erties and weight and the absence of electrochemical corrosion. On the other hand, fiber needs to be
impregnated with a matrix in order to be applied around the column and to exert an effective con-
Keywords:
fining pressure. Polymers manifest excellent performance in this role, but unfavorable result if
Design-oriented-model
Confinement
breathability of the structural element is a key-issue as well as reversibility and compatibility of
FRM the new materials with the existing substrates (e.g. in historical masonry construction). In this scenar-
FRCM. TRM io, the interest in non-polymeric matrix has been increased in the last years, as the use of a cement/
Masonry lime based mortar.
Concrete In the present paper, a state-of-art regarding experimental programs on concrete and masonry sam-
Column ples confined by Fiber Reinforced Mortar (FRM) and subjected to pure compressive test is preliminary
reported. Therefore, a multiple linear regression analysis was implemented in order to identify the pos-
sible influence of the interaction between the properties of the matrix and those of the fibers on the
effectiveness of the confining pressure. Two new analytical formulations were assessed and discussed
in the present study (the first related to concrete and the second related to masonry columns), evidenc-
ing the importance of taking into account the characteristics of the confining mortar besides those of
fibers. A deep comparison (experimental vs. predicted) of the proposed formulae with Design-Oriented-
Models (DOM) from the literature was also provided in terms of accuracy, precision and correlation.
Finally, simplified relationships are also furnished and addressed to a possible contribution for design
guidelines on the topic.
Published by Elsevier Ltd.

1. Introduction by the transverse dilatation strain. The goal of the confinement

consists in limiting that cracking phenomenon by providing a
The mechanism of compression failure of axially loaded hydrostatic pressure that may counteract the lateral deformation.
elements is well-know (e.g. in [1]). The failure occurs due to the Modern solutions for column confinement are those related to
propagation of a continuous macroscopic splitting crack caused the use of Fiber Reinforced Polymer (FRP) and Fiber Reinforced Matrix
(FRM) systems; these systems may guarantee an increasing lateral
⇑ Corresponding author. pressure up to the tensile strength of the confining material is
E-mail address: alessio.cascardi@unisalento.it (A. Cascardi).
attained.

http://dx.doi.org/10.1016/j.conbuildmat.2017.09.004
0950-0618/Published by Elsevier Ltd.
388 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401

The mechanical behavior of FRPs and FRMs are significantly dif- In the case of the FRP confinement a post-peak hardening behavior
ferent under the tensile action [2]. For FRP systems the hypothesis is expected up to the FRP rupture. Definitely, the FRP system is
of the perfect bond between fibers and matrix is assumed and they more performable with respect to the FRM one both in terms of
result characterized by a linear elastic constitutive law up to fail- compressive strength and ultimate strain (see Fig. 1). Therefore, a
ure. Conversely, the tensile behavior of FRM is a bit more complex, key-role can be attributed to the adopted matrix and to the bond
having three different phases (see [3]). In the first phase, the inor- level between matrix and fibers: in the case of the FRM, the dam-
ganic matrix bears basically the load until first cracking occurs. In age of the mortar and the possible mortar/fibers debonding
the second phase, due to the cracking evolution, stresses are trans- involves a reduction of the confining pressure as well as of the final
ferred from the matrix to the reinforcing fibers, with some debond- compressive strength; while for the FRP the high performance of
ing at the fiber-matrix interface. In the third and last phase, the the polymeric matrix and the perfect bond between the matrix
composite system behaves almost linear until failure; at this stage, and the fibers allow the composite to remain undamaged up to
fibers mainly bring the load and the crisis occurs by a progressive the failure, while the damage of the column implies a decay of
rupture of the roving fiber filaments and the debonding at the the column stiffness. ‘‘Based on this observation, the analytical
fiber-matrix interface. The strength of the FRM is strongly affected interpretation of the phenomenon should take into account also
by the type of the adopted fibers, the fiber sizing, the mesh layout the mechanical properties of the matrix and, in this sense, the
and the properties of the matrix. effect of the loss of the bond stress transfer to the fiber-grid due
On the other hand, the behavior of a confined column is to the crack opening may be implicitly considered.”
strongly dependent on the tensile performance of FRMs; in fact, Several researches available in the literature confirm that the
tensile stresses are generated within the jacketing due to the lat- FRPs behave elastic until rupture and their confining action can
eral expansion of the column caused by the axial load. Important be easily assumed growing pseudo-linearly with the increase of
differences regarding the performance of FRPs and FRMs confined the longitudinal loading, as reported in [5–7]. On the contrary,
columns are qualitatively illustrated in Fig. 1, where the axial the mechanical behavior of the FRM is affected by the brittle
stress-strain curves referred to unconfined, FRM-confined and behavior of the matrix. In other words, the confining system
FRP-confined columns are reported (see also [4,14]). Three signifi- (FRM) damage as well as the column damage during the loading
cant stress-values are identified and named r1, r2 and r3 in the Fig- phase may involve a decrease of the confining pressure after the
ure. If the axial stress is less than r1, the column is un-damaged first crack appears in the matrix (agreeing with [8]). However, if
because of stress level is less than the unconfined column com- a high performance confining mortar is used and a high bond level
pressive strength (r1). When beside the damage of the column also between fibers and matrix is guaranteed, the FRM tends to behave
damage of the FRM system takes place (stress level equal to r2), as an FRP since the effective stresses-migration to the fiber-grid;
the inorganic matrix cracks, after this stage, the mechanical the constitutive law of such FRM-jacket can be compared with that
response greatly depends on the properties of the matrix and of of an FRP (see [9–11]).
the interface fibers-matrix; at this stress level, the FRP-confined On the basis of the above considerations, new analytical models
column shows still an ascending trend (undamaged FRP system). have been developed in this paper in order to provide design for-
The premature damage of the mortar, when its mechanical proper- mulae for the prediction of the compressive strength of concrete
ties are very poor, produces a softening phase in the FRM-confined or masonry column (circular or square cross-section) confined
curve; which ends with ultimate strain values anyway greater than with FRM systems. A collection of the experimental data is pre-
that of unconfined column. The behavior following the matrix sented and exploited for understanding the influence of matrix
cracking is affected by the bond between fibers and matrix itself; and fibers on the increase of the compressive strength due to FRMs
indeed, if a perfect bond exists the stresses are gradually trans- and to assess a validate the proposed models.
ferred to the fiber mesh while the crack phenomenon evolves up
to the complete damage of the mortar. Otherwise, if the level of
the bond at the interface fibers-matrix is very low the stress may 2. Background
be suddenly transferred to the fibers that may be subjected to
uneven tensile stresses leading to a possible premature failure at Fiber Reinforced Polymer is a composite material made of a poly-
a level of load significantly lower than the mesh tensile strength. mer matrix reinforced with long fibers. The fibers are usually glass,
The stress value equal to r3 corresponds to the ultimate condition carbon, or aramid (see [12]), although other fibers such as basalt,
of the FRP-confined column or rather the FRP failure (brittle rup- steel, flax and hemp have been used and addressed by researchers
ture of the fibers and consequent debonding from the substrate). and practitioners (e.g. in [13–18]). The polymer is usually an epoxy,

Fig. 1. Sketch of the FRP vs. FRM confinement behavior.

 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401 389

vinyl-ester or polyester thermosetting resin, phenol formaldehyde 15,12 MPa to 67,10 MPa. The matrix was typically a cement-
resins are also in use (see [12]). The concept of load-sharing based mortar, with a thickness tmat between 4 mm and 120 mm
between the matrix and the reinforcing constituent (fibers) is cen- and a compression strength fmat ranging from 2,25 MPa to
tral for understanding the mechanical behavior of the composite. 67 MPa. Different types of fibers were investigated (basalt = B, car-
An external force applied to a composite is partly sustained by bon = C, glass = G, steel = S, PBO = P), with an elastic modulus Ef
the matrix and partly by the fibers (see [19]). The proportion from 52 GPa (basalt) to 270 GPa (carbon) and a thickness tf from
depends on the volume fraction, the shape and the orientation of 0,0455 mm to 0,58 mm. The geometrical percentage of the fiber-
the fibers as well as on the elastic properties of both constituents. reinforcement qf, calculated as 4tf/D, ranges from 0,04% to 2,63%,
The externally bonded FRP reinforcement systems can be dis- highlighting the great variety of fiber configuration explored. Hol-
tinguished in pre-impregnated systems and in-situ systems. The low columns aren’t included in the mentioned database.
first ones are pre-manufactured strips, shells or jackets, which The masonry specimens have a rectangular cross section, with
are applied through adhesives on the substrate. The in-situ sys- dimensions A and B from 102 to 380 mm, and an unconfined
tems, which are widespread in reinforcement techniques, are com- strength fmo from 1,19 MPa to 44,5 MPa. Also in the case of
posed of dry fibers installed through resin that, at the same time, masonry confinement the most utilized matrix was a cement-
constitutes the composite matrix. The mechanical contribution of based mortar, with a thickness tmat from 5 to 15 mm and a com-
resin is totally neglected in this case, but it is fundamental to keep pression strength fmat from 0,55 to 32,6 MPa. Basalt, carbon, glass,
the fibers united and to permit a uniform stress distribution PBO and steel fibers were investigated, with an elastic modulus Ef
between them (affording to [20]). Concerning the in-situ impreg- from 59,76 GPa to 270 GPa and a thickness tf from 0,043 mm to
nated composite, the area of the dry fibers can be used for the com- 1,8 mm. The geometrical percentage of fiber qf varies from 0,05%
putation of the FRP mechanical parameters due to the difficulty in to 2,22%, exhibiting even in this case a great variability of this
measuring the real volume of the resin (according to [12]). parameter. The available data referred to masonry columns appear
The Fiber Reinforced Mortar is a composite material which dif- still limited if compared with data collected for concrete columns.
fers from the FRP because of a non-polymeric matrix is adopted. In Table 1, test data from [14,37–49] were chronologically recorded
The fiber’s role consists mainly in ensuring the post-cracking per- (concrete specimens) and similarly in Table 2, test data from [51–
formance of the FRM system. The bond action between matrix 58,60] (masonry specimens). Databases include both large and
and fibers is crucial to the strength of the composite (see [21]). small scale specimens; an analogous behavior was observed vary-
However, further factors can affect the FRM-capacity; such as ing the size of the specimens, for instance a similar failure mode
matrix composition (cement, lime, gypsum, pozzolana, additives, was recorded (e.g. in [44,47,54]). In fact, a typical crack pattern
short fibers presence, etc.), fiber type (carbon, glass, PBO, steel, within the cross-section firstly occurred, characterized by splitting
basalt, etc.), fiber size, fiber distribution (unidirectional, grid, tex- lines within the columns [12]. Ordinarily, the failure mode is due to
tile etc.) and orientation (0°/90°, 45°/+45°, 45°/+90°/+45°, fiber-matrix separation, which is caused by not sufficient impreg-
etc.). Contrarily to the FRP, the FRM is characterized by a non- nation of the matrix. This mechanism leads to a fiber-matrix
linear constitutive relationship because of the premature cracking debonding at first and a tensile failure of the fiber consequently.
of the matrix. The FRM stress-strain law was evaluated and the Moreover, the jacket failure manifested often a corner effect, when
influence of the adopted matrix was also underlined by experimen- referring to square cross-sections that caused the rupture of the
tal and analytical outcomes (e.g. in [22–33]). confining system and the subsequent FRM-detachment. Further-
The use of FRM for strengthening and retrofitting applications more, a mineral mortar, with the addition of short fibers (for
may have several advantages [34]. Moreover, cement binders are increasing the tensile strength), was adopted as matrix for the
less sensitive than epoxy matrixes to fire hazards, high tempera- FRM-confinement of concrete cylinders in [44]; in this case, a typ-
ture environments and freeze–thaw effects. The new mortar- ical hardening post-peak range (see also Fig. 1) was obtained, sim-
based matrix is able to ‘‘breathe” as much as the existing mortar ilar to that characterizing the FRP-confined columns. behavior.
in the masonry avoiding the moisture formation (see [35]). The In Table 2, a default value of E0 = 1000 fmo has been adopted
compatibility of the mortar with the substrate should be consid- according to Eurocode 6, [62]. Even if this factor (1000) is sug-
ered in terms of mechanical, chemical and aesthetic properties as gested by the Eurocode 6, it can be said, based on abundant exper-
appropriate to the philosophy of the conservation of cultural her- imental evidence (e.g. in [63–66]), that it leads to a significant
itage, according to [36]. overestimation of the masonry’s Young modulus. According to
many experiments, the value of the Young’s modulus of the
2.1. Experimental database masonry may take significant lower values ranging between 100
and 750 times the compressive strength of the masonry itself.
In order to achieve the goal of the present paper, a deep study of The variability of databases’ parameters is drawn in Figs. 2 and
the state-of-art regarding FRM-confinement was carried out. Two 3; in which the investigated quantities (vertical-axis) are reported
databases containing the results of 276 compression test per- per each group of samples of the collected databases; in fact, vari-
formed on FRM-confined specimens and available in the literature ables referred to similar specimens have been averaged and thus
were analyzed, the former referring to 231 concrete columns and 97 and 33 values for concrete and masonry columns, respectively
the latter to 45 masonry columns. The specimens with similar (horizontal-axis) are shown in the mentioned Tables. In particular,
characteristics have been grouped and the related experimental the ratio between the value referred to a group of similar samples
result were averaged; therefore, 97 and 33 groups of specimens and the maximum corresponding value utilized in the whole data-
were considered for the concrete and masonry database, respec- base is reported (color scale) in order to graphically appreciate the
tively. For each specimen, geometrical and mechanical characteris- dispersion ranges of the tested materials and geometries (column
tics referred to the substrate, matrix and fibers were collected. The and reinforcement). As concerns the concrete’s database, the
increased compressive strength of the FRM-confined specimens is parameters with the highest variability are the fiber’s thickness,
reported in Tables 1 and 2. with a Coefficient of Variation (CoV) equal to 1,08; and the matrix’s
The concrete specimens have a circular cross section, with a thickness, with a CoV almost equal to 1,06. On the contrary, low
diameter D ranging from 113 mm to 200 mm, or a quadratic/rect- values of CoV were recorded for the dimensions of the columns
angular cross section, with dimensions A (length) and B (width) and the mechanical properties of the substrate (e.g. the CoV of D
from 100 mm to 400 mm, and an unconfined strength fco from is equal to 0,41 and that of Eco equal to 0,17). The high CoVs are
390 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401

Table 1
FRM-confined concrete specimens.

Author Label A B D fco Eo fmat Emat tmat Fiber n tf Ef rfu qf fcc

(mm) (mm) (mm) (MPa) (GPa) (MPa) (GPa) (mm) (mm) (GPa) (MPa) (%) (MPa)
Triantafillou et al. A_MI2 – – 150 15,24 19,52* 8,56 14,63 4 C 2 0,047 225 3350 0,25 20,77
(2006) A_MII2 – – 150 15,24 19,52* 30,61 27,66 4 C 2 0,047 225 3350 0,25 23,88
A_MI3 – – 150 15,24 19,52* 8,56 14,63 6 C 3 0,047 225 3350 0,38 26,5
A_MII3 – – 150 15,24 19,52* 30,61 27,66 6 C 3 0,047 225 3350 0,38 27
B_MII2 – – 150 21,81 23,35* 30,61 27,66 4 C 2 0,047 225 3350 0,25 27,36
B_MII3 – – 150 21,81 23,35* 30,61 27,66 6 C 3 0,047 225 3350 0,38 32,44
Ombres (2007) HCR1a-A – – 150 67,1 32,36 2,25 7,5 12 C 1 0,047 240 3404 0,13 70,18***
HCR2a-A – – 150 67,1 32,36 2,25 7,5 18 C 2 0,047 240 3404 0,25 71,77***
CR-1a-B – – 150 27,05 20,54 2,25 7,5 12 C 1 0,047 240 3404 0,13 32,03***
CR-2a-B – – 150 27,05 20,54 2,25 7,5 18 C 2 0,047 240 3404 0,25 34,87***
CR-3a-B – – 150 27,05 20,54 2,25 7,5 24 C 3 0,047 240 3404 0,38 36,79***
CR-2a-C – – 150 19,42 25,18 2,25 7,5 18 C 2 0,047 240 3404 0,25 29,00***
CR-3a-C – – 150 19,42 25,18 2,25 7,5 24 C 3 0,047 240 3404 0,38 31,43***
CR-4a-C – – 150 19,42 25,18 2,25 7,5 30 C 4 0,047 240 3404 0,50 35,09***
R-Q11 150 150 – 17,42 23,79 2,25 7,5 12 C 1 0,047 240 3404 0,09 20,02
R-Q12 150 150 – 17,42 23,79 2,25 7,5 18 C 2 0,047 240 3404 0,18 21,46***
R-Q24 150 150 – 17,42 23,79 2,25 7,5 24 C 3 0,047 240 3404 0,27 26,90***
R-Q27 150 150 – 17,42 23,79 2,25 7,5 30 C 4 0,047 240 3404 0,35 29,73***
R-T1-1-R1 150 300 – 27,05 20,54 2,25 7,5 12 C 1 0,047 240 3404 0,06 28,5
R-T1-1-R3 150 300 – 27,05 20,54 2,25 7,5 18 C 2 0,047 240 3404 0,11 30,53
R-T1-1-R5 150 300 – 27,05 20,54 2,25 7,5 24 C 3 0,047 240 3404 0,17 33,33***
Di Ludovico et al. S2 – – 150 15,52 19,36 30 27,39* 8 B 1 0,046 91 1814 0,12 22,66***
(2008) S6 – – 150 17,83 19,36 30 27,39* 8 B 1 0,046 91 1814 0,12 25,32***
S8 – – 150 17,83 19,36 30 27,39* 12 B 2 0,046 91 1814 0,25 28,35***
Di Ludovico et al. S3 – – 150 15,52 19,70* 30 27,39* 15 B 2 0,050 91 1814 0,27 22,35
(2010) S10 – – 150 17,83 21,11* 30 27,39* 10 B 1 0,050 91 1814 0,13 20,03***
S13 – – 150 17,83 21,11* 30 27,39* 15 B 2 0,050 91 1814 0,27 23
S4 – – 150 15,52 19,70* 30 27,39* 10 B 1 0,050 72 1440 0,13 22,5
S5 – – 150 15,52 19,70* 30 27,39* 15 B 2 0,050 72 1440 0,27 22,81
S14 – – 150 17,83 21,11* 30 27,39* 10 B 1 0,050 72 1440 0,13 25,32***
S16 – – 150 17,83 21,11* 30 27,39* 15 B 2 0,050 72 1440 0,27 28,35***
S6 – – 150 15,52 19,70* 30 27,39* 10 B 1 0,050 72 1440 0,13 19,71
S7 – – 150 15,52 19,70* 30 27,39* 15 B 2 0,050 72 1440 0,27 22,5
S18 – – 150 17,83 21,11* 30 27,39* 10 B 1 0,050 72 1440 0,13 22,91***
S20 – – 150 17,83 21,11* 30 27,39* 15 B 2 0,050 72 1440 0,27 26,745***
Garcìa et al. (2010) M1-1 – – 150 21,8 23,35* 31,5 28,06* 10 B 1 0,0865 52 894 0,23 26,20***
M2-1 – – 150 21,8 23,35* 31,5 28,06* 15 B 2 0,0865 52 894 0,46 27,21***
C1-1 – – 150 21,8 23,35* 22,4 23,66* 10 B 1 0,0865 52 894 0,23 28,67***
C2-1 – – 150 21,8 23,35* 22,4 23,66* 15 B 2 0,0865 52 894 0,46 28,78***
Ortlepp et al. B1 150 150 – 29 26,92* 67 40,93* – G 1 0,07 74 1231 1,32 33,4***
(2011) B6 – – 150 29 26,92* 67 40,93* – G 1 0,07 74 1231 1,87 35
B7 100 100 – 29 26,92* 67 40,93* – G 1 0,07 74 1231 1,98 35
B8 300 300 – 29 26,92* 67 40,93* – G 1 0,07 74 1231 0,66 40
C1 150 150 – 29 26,92* 67 40,93* – C 1 0,07 223 1032 1,32 35***
C6 – – 150 29 26,92* 67 40,93* – C 1 0,07 223 1032 1,87 39
C7 100 100 – 29 26,92* 67 40,93* – C 1 0,07 223 1032 1,98 37
C8 300 300 – 29 26,92* 67 40,93* – C 1 0,07 223 1032 1,21 41
D1 150 150 – 29 26,92* 67 40,93* – C 1 0,18 204 4000 0,34 41***
D6 – – 150 29 26,92* 67 40,93* – C 1 0,18 204 4000 0,48 57
D7 100 100 – 29 26,92* 67 40,93* – C 1 0,18 204 4000 0,51 57
D8 300 300 – 29 26,92* 67 40,93* – C 1 0,18 204 4000 0,17 38
Colojanni et al. CF31M – – 200 17 20,61* 31,17 15 30 C 1 0,047 240 4800 0,09 19,8
(2012) CF32M – 200 200 17 20,61* 31,17 15 60 C 2 0,047 240 4800 0,19 22,5
CF33M – – 200 17 20,61* 31,17 15 90 C 3 0,047 240 4800 0,28 23
S15F4M 200 200 – 17 20,61* 31,17 15 120 C 4 0,047 240 4800 0,27 22,25***
R15F2M 200 400 – 17 20,61* 31,17 15 60 C 2 0,047 240 4800 0,08 20,35***
De Caso et al. BGP-A-1 – – 152 20,4 22,58* 2,49 7,89* 8 G 2 0,25 72 3240 1,32 26,85***
(2012) BGP-H-1 – – 152 20,4 22,58* 31,1 27,88* 8 G 2 0,25 72 3240 1,32 30***
1B-1 – – 152 21,7 23,29* 2,49 7,89* 8 G 1 0,25 72 3240 0,66 26,3***
2B-1 – – 152 21,7 23,29* 2,49 7,89* 12 G 2 0,25 72 3240 1,32 34,73***
4B-1 – – 152 21,7 23,29* 2,49 7,89* 20 G 4 0,25 72 3240 2,63 47,9***
Trapko (2012) 20M1_1 – – 113 23,23 24,1* 32,5 28,50* 10 P 1 0,0455 238 4300 0,16 32,57***
20M2_1 – – 113 23,23 24,1* 32,5 28,50* 15 P 2 0,0455 238 4300 0,32 42,72***
20M3_1 – – 113 23,23 24,1* 32,5 28,50* 20 P 3 0,0455 238 4300 0,48 56,94***
Colojanni et al. CA1-2L – – 154 25 25,00* 16 6 8 P 2 0,0455 270 5800 0,24 31,23***
(2014) CA4-3L – – 154 25 25,00* 16 6 12 P 3 0,0455 270 5800 0,35 36,57***
CB2-2L – – 200 25 25,00* 16 6 8 P 2 0,0455 270 5800 0,18 31,17***
 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401 391

Table 1 (continued)

Author Label A B D fco Eo fmat Emat tmat Fiber n tf Ef rfu qf fcc

(mm) (mm) (mm) (MPa) (GPa) (MPa) (GPa) (mm) (mm) (GPa) (MPa) (%) (MPa)
CB1-3L – – 200 25 25,00* 16 6 12 P 3 0,0455 270 5800 0,27 33,55***
SB2-2L 200 200 – 25 25,00* 16 6 8 P 2 0,0455 270 5800 0,13 28,27***
SB1-3L 200 200 – 25 25,00* 16 6 12 P 3 0,0455 270 5800 0,19 30,17***
Colajanni et al. CF2M-a – – 200 16,66 20,41* 31,17 15 8 C 2 0,047 240 4800 0,19 20,83
(2014) CF3M-a – – 200 16,66 20,41* 31,17 15 12 C 3 0,047 240 4800 0,28 22,67
S15F2M 200 200 – 16,66 20,41* 31,17 15 8 C 2 0,047 240 4800 0,13 19,11***
R15F2M 200 400 – 16,66 20,41* 31,17 15 8 C 2 0,047 240 4800 0,08 18,88***
S15F4M 200 200 – 16,66 20,41* 31,17 15 16 C 4 0,047 240 4800 0,27 22,75v**
R15F4M 200 400 – 16,66 20,41* 31,17 15 16 C 4 0,047 240 4800 0,17 21,02***
Ombres (2014) CRP1-I – – 152 15,4 19,62* 30,4 30,4 6 P 1 0,0455 270 5800 0,12 29,03***
CRP2-I – – 152 15,4 19,62* 30,4 30,4 9 P 2 0,0455 270 5800 0,24 30,94***
CRP3-I – – 152 15,4 19,62* 30,4 30,4 12 P 3 0,0455 270 5800 0,36 35,40***
CRP4-I – – 152 15,4 19,62* 30,4 30,4 15 P 4 0,0455 270 5800 0,48 52,74***
CR-1-C – – 150 29,26 19,62* 30,4 30,4 6 P 1 0,0455 270 5800 0,12 54,9
CR-2-C – – 150 29,26 19,62* 30,4 30,4 9 P 2 0,0455 270 5800 0,24 51,45
CR-3-C – – 150 29,26 19,62* 30,4 30,4 12 P 3 0,0455 270 5800 0,36 55,94
Thermou et al. A3  2h1 – – 150 15,12 19,44* 22,1 8,03 7 C 1 0,58 130 2479 1,55 22,27***
(2015) 1
A3  2m1 – – 150 15,12 19,44* 22,1 8,03 7 C 1 0,12 130 2479 0,32 25,66***
1
A3  2l1 1 – – 150 15,12 19,44* 22,1 8,03 7 C 1 0,06 130 2479 0,16 22,80***
A12  h1 – – 150 15,12 19,44* 22,1 8,03 7 C 1 0,58 130 2014 1,55 25,30***
1
A12  m1 – – 150 15,12 19,44* 22,1 8,03 7 C 1 0,12 130 2014 0,32 26,64***
1
A12  l1 1 – – 150 15,12 19,44* 22,1 8,03 7 C 1 0,06 130 2014 0,16 22,55***
B3  2m2 – – 150 26,2 19,44* 22,1 8,03 7 C 1 0,12 130 2479 0,32 36,07***
1
B3  2l2 1 – – 150 26,2 19,44* 22,1 8,03 7 C 1 0,06 130 2479 0,16 35,93***
B12  m2 – – 150 26,2 19,44* 22,1 8,03 7 C 1 0,12 130 2014 0,32 40,42***
1
B12  l2 1 – – 150 26,2 19,44* 22,1 8,03 7 C 1 0,06 130 2014 0,16 36,21***
*
Hongyu et al. R2 200 200 – 45,5 33,72 52,8 36,33 10 G 1 0,07 180 4100 0,10 55,75
(2016) R3 200 200 – 45,5 33,72 52,8 36,33* 15 G 2 0,07 180 4100 0,20 60
R4 200 200 – 45,5 33,72 53,8 36,67* 20 G 3 0,07 180 4100 0,30 63,75
R5 200 200 – 45,5 33,72 54,8 37,01* 15 G 2 0,07 180 4100 0,20 59,5
R6 200 200 – 45,5 33,72 55,8 37,35* 15 G 2 0,07 180 4100 0,20 59,75

- Data not cited in the reference.

*
Data not cited in the reference and determined as according to [61].
**
Data not cited in the reference and determined as according to [62].
***
Average data of two or more identical specimens.

Table 2
FRM-confined masonry specimens.

Author Label A B D’ fmo Emo fmat Emat tmat Fiber n tf Ef rfu qf fcc
(mm) (mm) (mm) (MPa) (GPa) (MPa) (GPa) (mm) (mm) (GPa) (MPa) (%) (MPa)
Di Ludovico et al. GRM-1 220 220 311 3,67 3,67** 25 8 8 G 1 0,043 72 1440 0,06 6,00***
(2008)
Yilmaz et al. S-2B-L(1) 360 360 509 1,2 0,14 – 0,37 15 B 2 0,047 220 1814 0,07 1,85
(2013) S-2B-L(2) 360 360 509 1,45 0,2 – 0,37 15 B 2 0,047 220 1814 0,07 1,70
Ombres (2014) C-1-e0 250 250 354 6,97 6,97** 32,6 15 6 C 1 0,0455 240 4800 0,05 14,64
Carloni et al. W_C-1 200 90 219 44,5 6,64 28,4 26,65* 8 P 1 0,092 270 5800 0,17 52,60
(2015) SQ_C-1 102 100 143 28,7 5,52 28,4 26,65* 8 P 1 0,092 270 5800 0,26 36,10
RECT_1_C-1 102 152 183 25,4 3,26 28,4 26,65* 8 P 1 0,092 270 5800 0,20 38,40
RECT_2_C-1 102 204 228 10,8 3,09 28,4 26,65* 8 P 1 0,092 270 5800 0,16 36,50
Incerti et al. CP_RC170_01 380 380 537 12,05 9,57 6,52 12,77* 10 C 2 0,047 220 2363 0,07 13,09***
(2015) CP_RV320_01 380 380 537 12,05 9,57 6,52 12,77* 10 G 2 0,06 60 991 0,09 12,75***
CP_GV160_01 380 380 537 12,05 9,57 6,52 12,77* 10 C 2 0,088 200 2800 0,13 12,83***
Theofanis (2015) C1_1_R20 240 240 339 1,19 1,19** 5,27 11,48* 5 C 1 0,047 220 4800 0,06 2,74***
C2_1_R10 240 240 339 1,19 1,19** 5,27 11,48* 10 C 2 0,047 220 4800 0,11 4,60
C3_1_R10 240 240 339 1,19 1,19** 5,27 11,48* 15 C 3 0,047 220 4800 0,17 6,03
Fossetti M. et al. M1_BF_1 230 230 325 4,31 4,31** 0,55 3,71** 10 B 1 0,05 89 3000 0,06 7,80
(2016) M1_SW_1 230 230 325 4,31 4,31** 0,55 3,71** 10 S 1 1,8 190 2845 2,22 5,60
M3_BF_1 230 230 325 9,14 9,14** 4,54 10,65* 10 B 1 0,05 89 3000 0,06 10,88
M3_SW_1 230 230 325 9,14 9,14** 4,54 10,65* 10 S 1 1,8 190 2845 2,22 9,32

(continued on next page)

392 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401

Table 2 (continued)

Author Label A B D’ fmo Emo fmat Emat tmat Fiber n tf Ef rfu qf fcc
(mm) (mm) (mm) (MPa) (GPa) (MPa) (GPa) (mm) (mm) (GPa) (MPa) (%) (MPa)
Mezrea et al. S-2B-L(1) 360 360 509 1,42 0,18 1,4 5,92** 10 B 2 0,046 91 1814 0,07 1,78***
(2016) S-2B-F(1) 360 360 509 1,5 0,23 9,83 15,68* 10 B 2 0,046 91 1814 0,07 1,68***
R-2B-L(1) 360 630 726 1,37 0,16 1,4 5,92** 10 B 2 0,046 91 1814 0,05 1,43***
R-2B-F(1) 360 630 726 1,43 0,18 9,83 15,68* 10 B 2 0,046 91 1814 0,05 1,46***
Witzany et al. CB_CFRCM_W 290 290 410 3,45 3,45** 15 C 1 0,047 240 4800 0,05 8,13***
(2016)

- Data not cited in the reference.

*
Data not cited in the reference and determined as according to [61].
**
Data not cited in the reference and determined as according to [62].
***
Average data of two or more identical specimens.

Fig. 2. Variables distribution of the FRM-confined concrete database.

Fig. 3. Variables distribution of the FRM-confined masonry database.

mostly related to the wide difference between the minimum and In the masonry’s database, the fiber’s thickness CoV was even
maximum value of some parameters, while the color map in higher, with a value of 2,09. Moreover, a higher variety of the geo-
Fig. 2 clearly illustrates that most of the concrete specimens were metrical and mechanical characteristics of the substrate can be
confined with quite the same thickness of the fiber grid and the observed by database’s records, since the CoV of D0 was 0,44 and
matrix; on the other hand, different values of the Ec0 and Ef have that of fmo was 1,03. The reason can be attributed to the different
been deeply experienced. Moreover, the color legend points out size and different types of the examined masonry samples (see
that a large range of Ef has been used in FRM-confinement of con- [51], where yellow Neapolitan Tuff bricks were analyzed, and
crete specimens especially if compared with the limited variability [52], where historical clay bricks were used in specimen’s con-
of Emat. struction). Similarly, to the concrete database, the Ef value resulted
 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401 393

respect to the main diagonal. Poor correlation is indicated by light

red color (e.g. in D vs. E0) while a significant correlation is evi-
denced by light green (e.g. in fco and E0). The matrix is forced to
be quadratic because it presents the same number of variables in
rows and columns. In Figs. 4 and 5, the CMPs evidence the impor-
tance of the parameters related to the matrix, such as fmat and Emat
(R2 = 0,65 and R2 = 0,53 for concrete and masonry column respec-
tively), which are correlated to the compressive strength of the
FRM-confined column; in addition the influence of the fiber Young
modulus, Ef, and of the mechanical properties of materials utilized
for realizing the column, E0 and fco, is confirmed. On the basis of
this preliminary analysis it is evident as the strength prediction
of FRM confined columns should take into account all these vari-
ables as also already mentioned in Section 2.
Fig. 4. Correlation Matrix Plot referred to the variables of the FRM-confined
concrete.
2.2. Design-oriented-models (DOM)

A review of the existing analytical relationships proposed to

evaluate the compression strength of the FRM-confined columns
is presented in this section and summarized in Table 3. The first
DOM about FRM-confined concrete is dated 2006 [37] and cali-
brated on the basis of a performed experimental campaign. The
authors executed axial test on textile-reinforced-mortars confining
concrete columns aiming to increase the axial capacity of concrete,
and compared obtained results with those obtained by the FRP-
confinement. Both cylinder and prismatic samples were tested
and in all cases the effectiveness of the strengthening technique
has been confirmed (increase of loading capacity and ductility). A
confinement model was also developed by extending FRP-
confinement typical models, while calibrating effectiveness factor
depending on the innovative jacketing system. The same approach
Fig. 5. Correlation Matrix Plot referred to the variables of the FRM-confined
was adopted by Ortlepp et al. late in 2011 (see [41]), varying the
masonry. cross-section geometry; obtained results evidenced the influence
of matrices used. In 2012, De Caso et al. studied the effect of differ-
ent types of fiber aiming to investigate also the fire-resistance (see
to be more investigated with respect to the Emat value (see Fig. 3) [43]). In 2013, the first international Code regarding Fabric-
because different types of fiber were often applied with the same Reinforced Cementitious Matrix was published by ACI [67], provided
matrix. analytical relationships for predicting the confinement effect. Test
The Correlation Matrix Plot (CMP) of the considered variables results from Bournas et al. 2007 [8], Abegaz et al. 2012 [68], De
(see Table 1 and Table 2) related to the compressive strength of Caso et al. 2012 [43], etc. are referred in that guideline furnishing
the confined column is reported in Fig. 4 and in Fig. 5 for the con- a linear correlation between compressive strength of the FRM-
crete and masonry column, respectively. A color map makes the reinforced concrete and the ultimate confining pressure (for both
understanding of the relation between the variables easier: the circular and rectangular cross-section). In [47] an experimental
range spaces between dark green (=total correlation = identity) program on cement based high strength composite material for
and dark red (=null correlation). The distribution ranges of the vari- confinement of concrete columns is reported; the proposed tests
ables analyzed are shown on the main diagonal (according to Fig. 2 allowed to assess a new Design-Oriented Model, that was refined
and Fig. 3), while correlations are symmetrically reported with in 2016 by the same authors [50].

Table 3
Predictive models.

DOM for FRM-confined concrete columns

Author (year) Compression strength Confining pressure Reference
Triantafillou et al. (2006) 1;27 ke ntf Ef ef [37]
f cc
¼1þ 1; 9ðff lu Þ f lu ¼ D0
f co co

Ortlepp et al. (2011) 2 3 ke ntf Ef ef [41]

f cc
¼ 1 þ 0; 27ðff lu Þ þ 5; 55ðff lu Þ  3; 51ðff lu Þ f lu ¼ D0
f co co co co

De Caso et al. (2012) 0;775 2nt f Ef ef [43]

f cc
¼ 1 þ 2; 87ðff lu Þ f lu ¼ D0
f co co

ACI (2013) f cc 2nt f Ef ef [67]

f co
¼ 1 þ 3; 1ðff lu Þ f lu ¼ D0
co

Ombres (2014) f cc
¼1þ 5; 286ðff lu Þ f lu ¼ ke qf Ef ef [47]
f co co

Ombres and Mazzuca (2016) f cc

¼ 1 þ 0; 913ðf Þf 0;5 lu f lu ¼ 12 ke qf Ef ef [50]
f co co

DOM for FRM-confined masonry columns

Author (year) Compression strength Confining pressure Reference
Mazrea et al. (2016) f m ¼ f mo þ kflu f lu ¼ 12 ke qf Ef ef [58]
With:
gm
k ¼ 1000
394 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401

A DOM for the prediction of the compressive strength of
b is the general calibrated dimensionless coefficient.
masonry column confined by FRM system was assessed recently
ei is the i-value of the scatter between the predicted (yi) and
by Mezrea et al. [58]. The proposal was based on the FRP- experimental (Yi) value.
confinement model provided the CNR-DT200 R1 [12] for masonry
column, which includes a specific property of the masonry, namely Eq. (4) includes the square power of the variables and the pro-
the specific weight gm. duct between two variables to take into account a possible interac-
tion effect (see [69]). By considering a model with two explanatory
3. Proposed models variables (x1 and x2), the multiple linear regression produces a
regression surface E(yi) that can be represented in a 3D space as
The compressive strength of the confined column is assumed schematically illustrated in Fig. 6. The linear regression is com-
equal to the compressive strength of the unconfined one plus a puted in both x1 and x2 directions, thus a series of b coefficients
contribute provided by the jacket, depending on the confining is obtained, furnishing the shape of the E(yi) surface. Subsequently,
pressure modified by an effectiveness coefficient, k (see Eqs. (1)– the punctual scatter ei (=Yi-yi) is evaluated in order to check if the
(3)) according to the approaches available in the literature [59]. resulting surface plot is consistent with the experimental data (the
0;5
less is the scatter the more representative is the correlation
f cc ¼ f co þ kf lu ð1Þ surface).
where: The assumptions of the analysis were:

2nt f Ef ef 1. Data are normally distributed.

f lu ¼ for circular cross-section ð2Þ
D 2. A linear relationship is assumed between the dependent vari-
able and the independent variables.
2nt f Ef ef
f lu ¼ for square=rectangular cross-section ð3Þ 3. The residuals are homoscedastic (= all random variables have
D0 finite variance).
The proposed models aim to define a function able to predict
the actual value of k factor by considering not only parameters On the basis of the analysis reported in Section 1.2 (see CMPs),
related to the proprieties of the fibers but also those characterizing the following variables were utilized (see Eqs. (5) and (6)) for
the inorganic matrix. Among the possible regression functions that determining the target value of k (see Eq. (1)); therefore, in the
may describe the relationship between the dependent variable and analyzed case ki=yi and the corresponding regression surface is
the explanatory variables, a linear function has been chosen; in called k. The correlation between the parameter related to the fiber
particular, a Multiple Linear Regression (MLR). The polynomial (=x1) and the effectiveness of the FRM-confined column is well-
regression model is a particular type of multiple linear regression known; moreover, the correlation with respect to the variable
model and it is reported in Eq. (4). regarding the matrix (=x2) of the FRM has been proved in Figs. 4
and 5.
Eðyi Þ ¼ b0 þ b1 xi1 þ b2 xi2 þ b3 x2i1 þ b4 xi1 xi2 þ b5 x2i2 þ ei ð4Þ
Ef
where: x1 ¼ q f ð5Þ
E0

Y i is the i-value experimentally obtained.
f mat

yi is the i-point of the E-surface corresponding to the theoretical x2 ¼ qmat ð6Þ
f0
dependent variable.

xi is the i-value of one of the independent variable (at least two
independent variables are needed in MLR). 3.1. DOM for concrete column

The MLR was first used for the concrete column database (see
Table 1) and the obtained regression surface is shown in Fig. 7

Fig. 6. Scheme of the multiple linear regression surface E(yi). Fig. 7. 3D surface plot of k against x1 and x2 for the concrete database.
 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401 395

Fig. 8. 3D contour plot of k against x1 and x2 for the concrete database.

Fig. 9. 3D surface plot of k against x1 and x2 for the masonry database.

while an up-view is also drown in Fig. 8. It is evident as the FRM-
confining action is affected not only by the proprieties of the fibers
(see x1), but also by the kind of matrix (see x2). In fact, the effective-
ness of the jacketing action (= k = color legend) has the peak value
in the corner corresponding to a combination of the mechanical
proprieties of the fibers (elastic modulus) and the matrix (com-
pressive strength). Indeed, by maintaining the geometrical per-
centage of reinforcement constant (qf and qmat), when using very
performable fibers (in terms of hoop strength) but a matrix of poor
quality (in terms of compressive strength) the effectiveness of the
confinement dramatically drops because of the premature cracking
of the mortar (see green zone in Figs. 7 and 8). On the contrary,
when a high-performance matrix is adopted (in terms of compres-
sive strength) in combination with a low Young’s modulus of the
fibers, a significant effectiveness is reached confirming the impor-
tance of the matrix’s role on the confining action (see Fig. 8). On the
other hand, if the same type of fiber and matrix is considered, but
the amount of material increases with respect to dimensions of the
cross-section, a reasonable increase of effectiveness is observed
related to both the thickness of fibers and that of the mortar
included in the parameter qf and qmat , respectively. Definitely, Fig. 10. 3D contour plot of k against x1 and x2 for the masonry database.

the parameter used in the MLR-analysis, namely x1 and x2, seems

having a significant influence on the effectiveness of the FRM- consequence, the influence of specimens with prismatic cross-
confinement; in fact, for x1 ranging between 0,08 and 0,18 and x2 section (especially for the rectangular ones) is less important with
ranging between 1,4 and 3,5 the k parameter assumes values from respect to the cylindrical specimens in order to reach the
11 to 16. accurate DOM. Further experimentations regarding the prismatic
The multiple linear regression furnishes the relationship cross-section of concrete columns are needed in order to confirm
between x1 and x2 variables and the k-target by fitting a theoretical or not the proposed model.
linear equation on the basis of available data. It was used for eval-
uating both the trend and predicting the target value. In fact, the
3.2. DOM for masonry column
MLR architecture allowed to evaluate the analytical formulation
expressed in Eq. (7), which represents the 3D regression surface,
Likewise, the MLR was implemented for the masonry samples
herein named k.
(see Table 2). The new regression surface was generated as
reported in Figs. 9 and 10. Two regions of maximum effectiveness
k ¼ 4; 05  3; 07x1 þ 2; 62x2  254ðx1 Þ2 þ 48; 72x1 x2  1; 47ðx2 Þ2 (with peak value of 7.8) can be identified (red zone), both corre-
sponding to a combination of the performance offered by fibers
ð7Þ
and matrix (tensile modulus of the fiber less than 100 times the
Negative and, therefore, not admissible results (= confinement elastic modulus of the masonry with a confining mortar’s compres-
that promotes the lateral expansion of the column instead of lim- sion strength about 3–5 times higher than the compressive
iting it) of the k-coefficient are obtained for x1 less than 0,005 and strength of the masonry). On the other hand, two regions of low
x2 less than 0,2 that basically correspond to a negligible FRM- confinement effectiveness may be found when poor or very high
retrofitting and thus to an almost absent confining action. In compressive strength of the matrix is adopted (mainly referred
Table 1, the collected experimental data mainly refers to to the elastic modulus of the fibers from 300 to 600 times higher
cylindrical and square cross-sections specimens and just in a few than the elastic modulus of the masonry and a compression
cases to prismatic cross-section (rectangular and square). As a strength of the matrix comparable to that of the masonry or at
396 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401

Fig. 11. Predicted vs. Experimental peak strength comparison (concrete database).

least 6 times higher). In fact, in both cases the compressive crush of The main difference between Eq. (7) and (8) is related to the pres-
the column occurs before the tensile strength of the fibers is ence of a further peak of effectiveness in the case of masonry col-
attained; once again, the results obtained evidence the influence umns. The FRM-confining action has been demonstrated effective
of the matrix’s strength on the confining action. The saddle func- also in cases for which high tensile strength fibers with respect
tion is characterized by a mid-line of effectiveness corresponding the low compressive strength of the unconfined masonry have
to a compressive strength of the matrix about four times that of been adopted. In these cases, the column cracked very easy and
the masonry (see Fig. 10). The FRM-confinement effectiveness, fast, resulting in sudden large lateral deformations directly con-
denoted by k, assumes peak values over 9 in the area limited by verted in hoop stresses within the fiber.
x1 ranging between 1 and 1,4 and, at the same time, x2 from 0,4
to 1,2. k ¼ 1; 26  12; 38x1 þ 10; 20x2 þ 11; 35ðx1 Þ2 þ 2; 82x1 x2  7; 50ðx2 Þ2
The resulting design-oriented-model (DOM) is expressed by Eq.
(8). The expression is simpler than the Eq. (7) because b3 and b4
ð8Þ
resulted almost equal to zero and, thus neglected, even if the com- Similarly, for the case of the DOM for concrete column, there
bined effect of the matrix and fibers properties is still confirmed. are inconsistent zones in which the effectiveness of the
 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401 397

Fig. 12. Predicted vs. Experimental peak strength comparison (masonry database).

Fig. 13. DOMs (concrete column) comparison in terms of precision, accuracy and correlation.

FRM-confinement tends to zero or negative outcome. The explana- ously cited, as [37,41,43,67,47,50] for FRM-confined concrete and
tion is attributable to the lack of date referring to masonry columns [58] for FRM-confined masonry. Scatter between theoretical and
confined with high-compressive-strength-matrix and low-tensile- experimental results are reported in Figs. 11 and 12. Each box is
strength-fiber. Further experimental investigations are suggested square imposed in order to make the linear correlation under-
for more accurate interpretation of the phenomenon; in this study, standing easier (dot on the diagonal means perfect prediction).
a lower bound value for x1 and an upper bound for x2 are fixed The confining fiber is also distinguishable due to the different
equal to 0,01 and 1,9 respectively. In addition, the different shape color. Concerning concrete (Fig. 11), all the models give satisfactory
of the regression surface related to the masonry columns with results evidencing a higher error only when referring to PBORM
respect to the concrete ones can be due to the square cross- (PBO Reinforced Mortar)-confinement. Anyway, the best correla-
section of the all masonry specimens (see Table 2). In fact, it is tion is provided by the model reported in [50] and by the proposed
well-known that the compressive strength of confined column is f
model. Similar considerations can be done with respect to f cc;exp the
cc;theo
sturdily affected by the failure modes, which are related to the
masonry FRM-confinement. The model [58] and the proposed DOM
shape of the cross-section (see [12]) and strongly influences the
are quite comparable in term of linear correlation and the PBO
effectiveness of the confinement itself.
effect is still the main challenge.
A more detailed analysis of the models has been made and
4. Models comparison results are shown in the Figs. 13and 14. The gray histograms rep-
resent the Average (Av), the Median (Me) and the Mode (Mo) of
The effectiveness of the two proposed DOM was tested also in the statistical frequency of the ratio. The closest are these three
comparison with results obtained by using available models previ- indexes, the more precise is the prediction. Moreover, when Av,
398 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401

Me and Mo values are closer to 1, it means the accuracy of the out- The confining pressure flu was evaluated by relationships
comes and consequently a lower value of the Mean Average Error defined in each design-oriented-model (see [37,41,43,67,47,
(MAE). The MAE was computed as expressed in Eq. (9). 50,58]). Trivially, the contemporary best performance for precision,
accuracy and correlation identify the analytical model that, more
1X n
MAE ¼ jf  f cc;exp j ð9Þ often, furnishes the best prediction of the experimental value of
n i¼1 cc;theo the compressive strength of the confined column with FRM sys-
tem. The Figs. 13 and 14 both evidence that the two proposed
Also, the linear correlation is evaluated by means of the index R2
DOM guarantee the best precision, low scatter (MAE = 0,13 for con-
calculated as reported in Eq. (10).
crete and MAE = 0,45 for masonry) and the highest correlation
Pn
ðf cc;theo  Av Þ
2
(R2 = 0,85 for concrete and R2 = 0,82 for masonry). The values
R2 ¼ Pi¼1 ð10Þ related to the Av, Me and Mo, the MAE and the R2 for the other
 Av Þ
n 2
i¼1 ðf cc;exp
considered models are provided in Figs. 13 and 14. The best com-
bination of contemporary proximity to the unit value of Av, Me and
Mo; the lowest value of the MAE and the highest R2 is attributable
to the proposed model.

5. Simplified relationships

A simplified form of the proposed models is provided in this

section easier to use for practical applications and to be eventually
considered for design codes or guidelines. By confirming the cru-
cial role of the matrix for the FRM-confinement, the effectiveness
of the lateral pressure has been calibrated depending on x2 (see
Section 2 – Proposed model). The importance of the fiber’s influ-
ence (=x1) is anyway considered in the fl formulation. The simpli-
fied relationships are expressed by Eqs. (11)–(14) and (15)–(18)
for the case of concrete and masonry column respectively. These
formulations are based on those provided for FRP-confinement
according to [11]. In particular, the effectiveness factor (k) was cal-
ibrated referring to experimental data and aiming to take into
account the effect of the mortar properties. A linear regression
analysis has been performed between f cc
f c0
and ffc0l ratios. The powers

Table 4
Statistic data of the performance of the simplified models.

Concrete columns Masonry columns

Average 0,99 1,19
Median 0,91 1,02
Mode 0,82 0,95
Fig. 14. DOMs (masonry column) comparison in terms of precision, accuracy and r 0,38 0,47
correlation.

Fig. 15. Normal distribution frequency of the ratio experimental / theoretical value of the compressive strength of the FRM-confined column: (a) concrete, (b) masonry.
 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401 399

Fig. 16. Proposed DOMs vs. Simplified models: (a) concrete; (b) masonry.

of the fl
index have been imposed equal to 0,66 and 0,50 for con- accurate with respect to the DOM but, the most part of the pre-
f c0
crete and masonry column respectively, according to [12]. dictions is on the safe side; while, the two proposals for the
A calibration aimed to minimize the scatters between the theo- FRM-confinement of masonry columns appear to have compara-
retical predictions and the experimental results led to satisfactory ble performances.
accuracy of the proposed relationships as shown in Fig. 15. The
normal Gaussian distribution of the experimental / theoretical 6. Conclusions
ratio indicates that the trend of the proposed simplified models
is close to the unit value, which means that the prediction The Fiber Reinforced Mortar (FRM) has been experienced as a
approaches the actual data. In Table 4 the statistical parameters successful confining technique for both masonry and concrete col-
of the / ratio between the experimental compressive strength umn in the last years. This new generation of composite appears
and the theoretical one are reported for both models. It may be advantageous in applications where some drawbacks arising with
appreciated that values of mean, median and mode are very close, the use of traditional Fiber Reinforced Polymer (FRP) should be over-
this is a significant result in terms of accuracy of the predictions came, such as the removability of the composite system and the
provided by the models. breathability of the existing substrate (especially in the historic
 23 masonry field). Thus, the need of providing analytical formula for
f cc f the design of FRM confined columns in term of compressive
¼1þk l ð11Þ
f c0 f c0 strength.
Relationships utilized for FRP confined columns doesn’t appear
ntf Ef f adequate to predict the compression strength when the confine-
fl ¼ ð12Þ
D ment is obtained by FRM systems. In fact, when using FRPs the
confining action strongly depends on the quality of the fibers,
f c;mat while the mechanical parameters of the polymeric matrix are not
k ¼ 4qmat ð13Þ
f c0 influential. On the contrary, FRM-system proved to be affected by
the characteristics of the inorganic matrix, as the cracking evolu-
4t mat tion within the matrix may significantly reduce the stress migra-
qmat ¼ ð14Þ
D tion from the column to the fiber jacket (i.e. fibers start to slide
" with respect the matrix) with consequent reduction of the confin-
 12 #
f ing pressure. Accordingly, the calibration of models concerning the
f cc ¼ f c0 1þk l ð15Þ use of FRM systems should take into account the influence of the
f c0
matrix.
Two analytical models (DOM) have been assessed and dis-
ntf Ef f
fl ¼ ð16Þ cussed in the paper, referred to concrete and masonry confined
D
columns. A multiple linear regression approach has been imple-
mented in order to consider the combined effects of the fibers
f c;mat
k ¼ 6qmat ð17Þ and the matrix. Two regression surfaces have been defined on
f c0
the basis of experimental databases collected from literature.
The proposed models have been compared with consistent
4t mat
qmat ¼ ð18Þ existing formulae in order to estimate the predictive perfor-
D mance; obtained results showed the high capability to predict
A final comparison has been performed between the proposed the experimental outcome in terms of precision, accuracy and
models (Section 3) and the respective simplified forms. The scat- linear correlation, confirming as the properties of the matrix
ter plot is illustrated in Fig. 16 for both the concrete and play a key-role in the FRM-confinement of concrete and
masonry columns. The concrete simplified models result less masonry columns. Finally, simplified relationships have been
400 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401

also assessed of more immediate use in practical applications; Acknowledgment

the accuracy of predictions resulted satisfactory and in any case
conservative. The authors are grateful to the ReLUIS (Rete dei Laboratori
Universitari di Ingegneria Sismica) Project 2017, PR 5 – ‘‘Materiali
innovativi per applicazioni su costruzioni esistenti” for the financial
List of symbols
support.

The following symbols are adopted in the present paper:

Conflict of interest

A is the length of the quadratic/rectangular cross-section of the
column. The authors declare that they have no conflict of interest.
f

Av is the average of the statistical frequency of the ratio or f cc;exp
cc;theo
f m;exp
or . References
f m;theo

B is the width of the quadratic/rectangular cross-section of the [1] Z. Bažant, Y. Xiang, Size Effect in Compression Fracture: Splitting Crack Band
column. Propagation, J. Eng. Mech. (1997) 162–172. 10.1061/(ASCE)0733-9399(1997)

D is the diameter of the circular cross-section of the column. 123:2(162).
[2] C. Carloni, T. D’Antino, L.H. Sneed, C. Pellegrino, Role of the matrix layers in the

D0 is the diagonal of the square/rectangular cross-section of the stress-transfer mechanism of FRCM composites bonded to a concrete
column. substrate, J. Eng. Mech. 141 (6) (2014) 04014165.

E is the regression surface. [3] B. Mobasher, A. Peled, J. Pahilajani, Distributed cracking and stiffness
degradation in fabric-cement composites, Mater. Struct. 39 (3) (2006) 317–

Ef is the longitudinal elastic modulus of the FRM-system’s fiber.
331.

Emat is the elastic modulus of the FRM-system’s matrix. [4] A. Sadrmomtazi, M. Khabaznia, B. Tahmouresi, effect of organic and inorganic

Eo is the elastic modulus of the generic unconfined column. matrix on the behavior of FRP-wrapped concrete cylinders, J. Rehabil. Civil Eng.
4–2 (2016) 52–66.

fcc is the compressive strength of the confined concrete column.
[5] M. Spoelstra, G. Monti, FRP-Confined Concrete Model, J. Compos. Constr.

fcc,exp is the experimental value of the compressive strength of (1999) 143–150. 10.1061/(ASCE)1090-0268(1999)3:3(143).
the confined concrete column. [6] A.Z. Fam, Concrete-filled fibre-reinforced polymer tubes for axial and flexural

fcc,theo is the predicted value of the compressive strength of the structural members. A dissertation submitted to the faculty of graduate
studies in partial fulfilment of the requirements for the Degree of Doctor of
confined concrete column. Philosophy. National Library of Canada, 2000 [0-612-53056-6].

fco is the compressive strength of the unconfined concrete [7] A. Cascardi, F. Micelli, M.A. Aiello, Unified model for hollow columns externally
column. confined by FRP, Eng. Struct. 111 (2016) (2016) 119–130, http://dx.doi.org/
10.1016/j.engstruct.2015.12.032.

flu is the confining pressure provided by the jacketing. [8] D.A. Bournas, P.V. Lontou, C.G. Papanicolaou, T.C. Triantafillou, Textile-

fm is the compressive strength of the confined masonry column. reinforced mortar versus fiber-reinforced polymer confinement in reinforced

fm,exp is the experimental value of the compressive strength of concrete columns, ACI Struct. J. 104 (6) (2007).
[9] M. Valdés, G. Concu, B. De Nicolo, FRP strengthening of masonry columns:
the confined masonry column. experimental tests and theoretical analysis, Key Eng. Mater. 624 (2014).

fm,theo is the predicted value of the compressive strength of the [10] E. Tore, M Comert, C. Demir, A. Ilki, M. Marasli, Seismic retrofit of columns
confined masonry column. using basalt mesh reinforced sprayed GRC jacket, in: Proceedings of the 17th
International Congress of the GRCA was held on the 19th–21st April 2015 at

fmat is the compressive strength of the FRM-system’s matrix.
the Radison Blu, in Dubai, 2015.

fmo is the compressive strength of the unconfined masonry [11] A.O. Ates, E. Tore, S. Khoshkholghi, A. Ilki, Sprayed textile reinforced GFRC for
column. retrofitting of sub- standard non-circular concrete columns, in: Proceedings of
the 16th World Conference on Earthquake Engineering, 16WCEE 2017

fo is the compressive strength of the generic unconfined column.
Santiago Chile, January, 2017.

qmat is the geometrical percentage of the applied matrix in the [12] Italian National Research Council, Guide for the design and construction of
FRM-system. externally bonded FRP systems for strengthening existing structures, CNR-

qf is the geometrical percentage of the applied fiber in the FRM- DT200 R1 (Italian Guideline), Technical Document No. 200/2013, Rome, 2013.
[13] P. Casadei, A. Nanni, T. Alkhrdaji, Steel-reinforced polymer: an innovative and
system. promising material for strengthening infrastructures, Concr. Eng. Int. 9 (1)

gm is the specific weight of the masonry. (2005) 54–56.
[14] M. Di Ludovico, A. Prota, G. Manfredi, Structural upgrade using basalt fibers for

k is the dimensionless jacketing action effectiveness parameter.
concrete confinement, J. Compos. Constr. 14 (5) (2010) 541–552.

MAE is the mean average error of the statistical frequency of the [15] A. Borri, G. Castori, M. Corradi, Masonry columns confined by steel fiber
f cc;exp f m;exp composite wraps, Materials 4 (1) (2011) 311–326.
ratio f cc;theo
or f m;theo
.
[16] M.A. Aiello, F. Micelli, R. Angiuli, P. Corvaglia, Masonry circular columns
f confined with glass and basalt fibers, in: Proceedings of the 6th International

Me is the median of the statistical frequency of the ratio f cc;exp or
cc;theo Conference on FRP Composites in Civil Engineering-CICE, 2012.
f m;exp [17] L. Yan, N. Chouw, Behavior of flax FRP tube confined coir fibre reinforced
f m;theo
.
concrete column: bond effect, in: 4th Asia-Pacific Conference on FRP in
f cc;exp Structure Swinburne University of Technology, Melbourne, Australia 11–13

Mo is the mode of the statistical frequency of the ratio f cc;theo
or
December, 2013.
f m;exp [18] A. Sangthongtong, Q. Hussain, A. Pimanmas, Compressive behavior of concrete
f m;theo
.
confined by hemp fiber composite jackets, Mater. Sci. Forum 860 (2016) 148–
2

R is the index of the linear correlation between the experimen- 151.
tal and theoretical data. [19] R. Codispoti, D.V. Oliveira, R. Fangueiro, P.B. Lourenço, R.S. Olivito,
Experimental behavior of natural fiber-based composites used for

tf is the thickness of the FRM-system’s fiber. strengthening masonry structures, in: Conference Papers in Science, vol.

tmat is the thickness of the FRM-system’s matrix. 2013, Hindawi Publishing Corporation, 2013, September.

x is the independent variable in the multiple linear regression. [20] F.I. du Béton, Design and use of externally bonded fiber reinforced polymer
reinforcement (FRP EBR) for reinforced concrete structures, FIB Bulletin 14:

Y is the dependent variable in the multiple linear regression. Externally bonded FRP reinforcement for RC structures, 2001.

b is the calibrate coefficient in the multiple linear regression. [21] E. Erdogmus, Use of fiber-reinforced cements in masonry construction and

e is the error computed as the difference between Y and its pre- structural rehabilitation, Fibers 3 (1) (2015) 41–63.
[22] S.P. Shah, C. Ouyang, Mechanical behavior of fiber-reinforced cement-based
dicted value. composites, J. Am. Ceram. Soc. 74 (11) (1991) 2727–2953.

ef is the ultimate tensile strength of the fiber. [23] B. Mobasher, J. Pahilajani, A. Peled, Analytical simulation of tensile response of

rfu is the ultimate tensile strength of the FRM-system’s fiber. fabric reinforced cement based composites, Cement Concr. Compos. 28 (2006)
77–89.
 A. Cascardi et al. / Construction and Building Materials 156 (2017) 387–401 401

[24] R. Contamine, A. Si Larbi, P. Hamelin, Contribution to direct tensile testing of [47] L. Ombres, Concrete confinement with a cement-based high strength
textile reinforced concrete (TRC) composites, Mater. Sci. Eng., A 528 (2011) composite material, Compos. Struct. 109 (2014) 294–304.
8589–8598. [48] Georgia E. Thermou, Konstantinos Katakalos, George Manos, Concrete
[25] A. D’Ambrisi, L. Feo, F. Focacci, Bond-slip relations for PBO-FRCM materials confinement with steel-reinforced grout jackets, Mater. Struct. 48 (5) (2015)
externally bonded to concrete, Compos. B Eng. 43 (2012) 2938–2949. 1355–1376.
[26] I. Colombo, M. Colombo, A. Magri, G. Zani, M. Di Prisco, Tensile behavior of [49] Yin Hongyu et al., Experimental investigation of concrete confinement with
textile: influence of multilayer reinforcement, in: High Performance Fiber textile reinforced concrete, Appl. Mech. Mater. (2015).
Reinforced Cement Composites, Springer, Netherlands, 2012, pp. 463–470. [50] L. Ombres, S. Mazzuca, Confined concrete elements with cement-based
[27] T. D’Antino, C. Carloni, L.H. Sneed, C. Pellegrino, Matrix-fiber bond behaviour in composites: confinement effectiveness and prediction models, J. Compos.
PBO FRCM composites: a fracture mechanics approach, Eng. Fract. Mech. 117 Constr. (2016) 04016103, http://dx.doi.org/10.1061/(ASCE)CC.1943-
(2014) 94–111. 5614.0000755.
[28] L.H. Sneed, T. D’Antino, C. Carloni, Investigation of bond behavior of [51] Marco Di Ludovico et al., Experimental behavior of masonry columns confined
polyparaphenylene benzobisoxazole fiber-reinforced cementitious matrix- using advanced materials, in: The 14th World Conference on Earthquake
concrete interface, ACI Mater. J. 111 (2014) 1–12. Engineering, 2008.
[29] S. De Santis, G. de Felice, Tensile behaviour of mortar-based composites [52] I. Yilmaz et al., External confinement of brick masonry columns with open—
for externally bonded reinforcement systems, Compos. B Eng. 68 (2015) grid basalt reinforced mortar, in: Proceedings of the Fourth Asia-Pacific
401–413. Conference on FRP in Structures (APFIS 2013), Melbourne, Australia. 2013.
[30] F.G. Carozzi, C. Poggi, Mechanical properties and debonding strength of fabric [53] Luciano Ombres, Confinement effectiveness in eccentrically loaded masonry
reinforced cementitious matrix (FRCM) systems for masonry strengthening, columns strengthened by fiber reinforced cementitious matrix (FRCM) jackets,
Compos. B Eng. 70 (2015) 215–230. Key Eng. Mater. 624 (2015). Trans Tech Publications.
[31] I.G. Colombo, M. Colombo, M. di Prisco, Tensile behavior of textile reinforced [54] Christian Carloni et al., Confinement of masonry columns with PBO FRCM
concrete subjected to freezing–thawing cycles in un-cracked and cracked composites, Key Eng. Mater. 624 (2014).
regimes, Cem. Concr. Res. 73 (2015) 169–183. [55] A. Incerti et al., Uniaxial compressive tests on masonry columns confined by
[32] P. Bernardi, D. Ferretti, F. Leurini, E. Michelini, A non-linear constitutive FRP and FRCM, in: Proc. 12th International Symposium on Fiber Reinforced
relation for the analysis of FRCM elements, Procedia Struct. Integrity 2 (2016) Polymers for Reinforced Concrete Structures (FRPRPCS-12) & The 5th Asia-
2674–2681. Pacific Conference on Fiber Reinforced Polymers in Structures (APFIS-2015),
[33] U. Häussler-Combe, J. Hartig, Bond and failure mechanisms of textile Nanjing, China.
reinforced concrete (TRC) under uniaxial tensile loading, Cement Concr. [56] Krevaikas D. Theofanis, Textile reinforced mortar system as a means for
Compos. 29 (4) (2007) 279–289. confinement of masonry structures, in: Proc. In: the 12th International
[34] O. Awani, T. El-Maaddawy, N. Ismail, Fabric-reinforced cementitious matrix: a Symposium on Fiber Reinforced Polymers for Reinforced Concrete Structures
promising strengthening technique for concrete structures, Constr. Build. (FRPRCS-12) & the 5th Asia-Pacific Conference on Fiber Reinforced Polymers in
Mater. 132 (2017) 94–111. Structures (APFIS-2015) Joint Conference, 14–16 December 2015, Nanjing,
[35] P.F.G. Banfill, A.M. Forster, A relationship between hydraulicity and China.
permeability of hydraulic lime, in: International RILEM Workshop on [57] Marinella Fossetti, Giovanni Minafò, Strengthening of masonry columns with
Historic Mortars: Characteristics and Tests; RILEM Publications SARL, BFRCM or with steel wires: an experimental study, Fibers 4 (2) (2016) 15.
Bagneux, France, 2000, pp. 173–183. [58] P. Mezrea, I. Yilmaz, M. Ispir, E. Binbir, I. Bal, A. Ilki, External jacketing of
[36] ISCARSAH – International Scientific Committee for Analysis and Restoration of unreinforced historical masonry piers with open-grid basalt-reinforced
Structures of Architectural Heritage – Recommendations for the analysis, mortar, J. Compos. Constr. (2016) 04016110, http://dx.doi.org/10.1061/
conservation and structural restoration of architectural heritage, Barcelona, 5 (ASCE)CC.1943-5614.0000770.
June 2005, pp. 39. [59] S. Guler, A. Ashour, Review of current design guidelines for circular FRP-
[37] T.C. Triantafillou, C.G. Papanicolau, P. Zissimopoulos, T. Laourdekis, Concrete wrapped plain concrete cylinders, J. Compos. Constr. 20 (2) (2015) 04015057.
confinement with textile-reinforced mortar jackets, ACI Struct. J. 103 (1) [60] Jiří Witzany, Radek Zigler, Stress state analysis and failure mechanisms of
(2006) 28–37. masonry columns reinforced with FRP under concentric compressive load,
[38] L. Ombres, Confinement effectiveness in concrete strengthened with fiber Polymers 8 (5) (2016) 176.
reinforced cement based composite jackets, in: Proc., FRPRCS-8, Patras, Greece, [61] A.Z. Fam, S.H. Rizkalla, Confinement model for axially loaded concrete
2007. confined by circular fiber-reinforced polymer tubes, ACI Struct. J. 98 (4)
[39] M. Di Ludovico, A. Prota, G. Manfredi, Concrete confinement with BRM (2001) 451–461.
systems: experimental investigation, in: Proceedings of the 4th International [62] Eurocode 6: design of masonry structures – Part 1–1: general rules for
Conference on FRP Composites in Civil Engineering–CICE, 2008. reinforced and unreinforced masonry structures, The European Union Per
[40] David García et al., Confinement of medium strength concrete cylinders with Regulation 305/2011.
basalt textile reinforced mortar, ICPIC 2010–13th International Congress on [63] G. Marcari, G. Fabbrocino, P.B. Lourenço, Mechanical properties of tuff and
Polymers in Concrete, 2010. calcarenite stone masonry panels under compression, in: 8th International
[41] R. Ortlepp, A. Lorenz, M. Curbach, Geometry effect onto the load bearing Masonry Conference, 2010, pp. 1083–1092.
capacity of column heads strengthened with TRC, in: Proc., Fib Symp.: [64] Sayari Arash, Mechanical properties of masonry samples for theoretical
Concrete Engineering for Excellence and Efficiency, Prague, Czech Republic, modeling, in: Proc. of the 15th International Brick and Block Masonry
2011. Conference, Florianópolis, Brazil, 2012.
[42] Piero Colajanni et al., Cyclic axial testing of columns confined with fiber [65] A. Costigan, S. Pavía, O. Kinnane, An experimental evaluation of prediction
reinforced cementitiuos matrix, in: Proceedings of the 6th International models for the mechanical behavior of unreinforced, lime-mortar masonry
Conference on FRP Composites in Civil Engineering (CICE’12), 2012. under compression, J. Building Eng. 4 (2015) 283–294.
[43] F. De Caso y Basalo, F. Matta, A. Nanni, Fiber-reinforced cement-based [66] M. Ridwan, I. Yoshitake, A.Y. Nassif, Proposal of design formulae for equivalent
composite system for concrete confinement, Constr. Build. Mater. 32 (2012) elasticity of masonry structures made with bricks of low modulus, Adv. Civil
55–65. Eng. (2017).
[44] Tomasz Trapko, Fibre reinforced cementitious matrix confined concrete [67] ACI (American Concrete Institute), Guide to design and construction of
elements, Mater. Des. 44 (2013) 382–391. externally bonded fabric-reinforced cementitious matrix systems for repair
[45] Piero Colajanni, Marinella Fossetti, Giuseppe Macaluso, Effects of confinement and strengthening concrete and masonry structures, ACI 549.4 R-13, ACI
level, cross-section shape and corner radius on the cyclic behavior of CFRCM Committee 549, Farmington, MI, 2013.
confined concrete columns, Constr. Build. Mater. 55 (2014) 379–389. [68] Adane Z. Abegaz, Advanced fiber reinforced composites as confining systems
[46] Piero Colajanni et al., Concrete columns confined with fibre reinforced for RC columns, Open Access Dissertations, Paper 1122, 2013.
cementitious mortars: experimentation and modelling, Constr. Building [69] S. Borra, A. Di Ciaccio, Statistica, metodologie per le scienze economiche e
Mater. 52 (2014) 375–384. sociali, McGraw Hill, 2004 (in Italian).