Sei sulla pagina 1di 8

Curvature Ductility of Masonry Spandrel Panels

F. Parisi & N. Augenti


University of Naples Federico II, Italy

ABSTRACT:
Non-linear response of masonry walls with openings to in-plane seismic excitations depends considerably on the
capability of spandrel panels, namely the horizontal structural components above the openings, to withstand
large deformations. Whilst bending moment capacity of masonry spandrels has been studied both theoretically
and experimentally, curvature ductility of their end cross-sections has not been yet analysed.
This paper deals with the development of moment-curvature relationships for different constitutive models by
taking into account cracking, strain-softening, and strain ductility of masonry. Resistance and deformation
parameters of interest are also discussed. It is shown that curvature ductility does not depend on the compression
level if the given axial force is not larger than one-half of the maximum compressive resistance. Conversely, it is
significantly affected by strain ductility and is always larger than this material parameter. If strain-softening is
included in the constitutive law, moment capacity degradation increases with the axial force.

Keywords: Masonry, Spandrel panel, Moment-curvature analysis, Curvature ductility

1. INTRODUCTION

Seismic performance of masonry buildings is currently evaluated at both global and local levels by
means of three-dimensional models and sub-systems separately (Tomaževič 2000). The overall
response of the masonry structure under horizontal actions depends on the in-plane behaviour of
individual walls. Conversely, partial simplified models are employed to predict response of masonry
walls to out-of-plane earthquake excitations.
In-plane seismic behaviour of masonry walls with openings is significantly affected by the capability
of spandrels to withstand large deformations (Augenti 2000). In the context of a macro-element
discretization of the bearing walls the spandrel panels, namely the horizontal structural components
above the openings (Fig. 1), drastically influence coupling between piers and, thus, the distribution of
internal forces throughout the structural system.

Pier panel
Macro-element
idealisation Spandrel panel

Joint panel

Figure 1. Identification of the spandrel panel in a perforated masonry wall

When performing non-linear incremental analysis, one must account for the relationship between
moment capacity and the given flexural curvature, which depends on several factors related to, inter
alia, the evolutionary mechanical behaviour of masonry. The latter does not behave linearly even at
small deformations due to local phenomena such as micro-cracking, strength degradation due to strain-
softening, and dilatancy. Therefore, a sound characterization of moment-curvature relationship must
include the constitutive law of masonry and is to be derived by deformation-based analysis. If the
stress-strain relationship is sensitive to strain-softening in the post-peak range, moment capacity
reduces under increasing average curvature due to the contraction of the axial force - bending moment
interaction diagram.
While flexural resistance of unreinforced masonry (URM) spandrels has been recently studied both
experimentally and theoretically (Augenti 2007, Cattari and Lagomarsino 2008, Gattesco et al. 2009),
curvature ductility of their end cross-sections has not been yet investigated. This topic is extremely
important to better estimate seismic performance of masonry buildings at different earthquake
intensity levels, especially in the case of existing constructions where spandrels do not typically have
reinforced concrete beams, or tie-rods, at the top and well-bonded lintels above the openings.
Moment-curvature analysis is a valuable tool to understand how moment capacity of a cross-section
changes as the average sectional curvature increases. To this end, a set of constitutive laws were
processed to define curvature ductility of masonry spandrels through a deformation-based approach.
Both geometrical and mechanical non-linearity sources were included in the computations.

2. MOMENT-CURVATURE ANALYSIS: PROCEDURE AND CONTROL PARAMETERS

A rectangular cross-section with thickness s and gross width H was analysed under combined bending-
compression. Assuming a zero-tensile strength for masonry and planar cross-sections after flexural
deformation, moment-curvature analysis was performed accounting for the variation of the effective
cross-section due to cracking. Since the reduction in the cross-section width changes with the given
axial force, an iterative incremental procedure consisting in the application of a monotonically-
increasing maximum axial strain was employed for deriving moment-curvature relations. Denoting by
h the distance of the neutral axis from the most compressed edge, axial force and moment capacity
resulting from a given linear distribution of axial strains were obtained through translational and
rotational equilibrium equations across the effective width. For both cracked and uncracked cross-
sections, axial force and moment capacity were expressed as functions of the neutral axis-to-
compressed edge distance normalised to the gross width, i.e. h = h / H , and strain ductility demand,
με,d = εmax/εy, corresponding to the given maximum axial strain, εmax.
In order to get generalised equations able to predict moment capacity M at a given axial force N and
sectional curvature, φ , for any rectangular cross-section and compressive strength fhd (defined along
the mortar bed joints), the axial force and moment capacity were normalised as follows:

N M N e M
N= ; = N ⋅e = ⋅ = (2.1)
Nm H Nm H Nm ⋅ H

being Nm = fhd · s · H the maximum resisting axial force and e the axial force eccentricity from the
cross-section centroid. Also, the maximum axial strain was normalised to h , so moment-curvature
analysis allowed to plot M / H versus φ H at different levels of normalised axial force N . In
particular, the following values of given axial force were considered: 0.1, 0.25, 0.5, and 0.75. The
product of the sectional curvature by the cross-section width was evaluated at both yielding and
ultimate limit states and was denoted by φy and φu , respectively. To understand the behaviour of URM
cross-sections, the following control parameters were computed for each compression level:
(1) the ratio αM = Mmax/My between the maximum resisting moment and the yielding moment, which
allowed to quantify the cross-section overstrength after masonry yielding;
(2) the ratio βM = Mmax/Mu between the maximum resisting moment and ultimate moment, which is a
capacity index associated with hardening or softening behaviour in the plastic range;
(3) the ratio γM = Mu/My , which describes the overall post-yielding behaviour of the cross-section;
(4) the average curvature ductility μφ = φu/φy ;
(5) the curvature capacity Cφ = (φu − φy)/φy = μφ − 1; and
(6) the yielding and ultimate curvature ratios:
φy φu
ρφ y ( N ) = ; ρ φu ( N ) = (2.2)
φy ( 0.1) φu ( 0.1)

which describe the change in sectional curvature when the cross-section is subjected to
normalised axial force levels larger than 0.1.

3. DEVELOPMENT OF MOMENT-CURVATURE RELATIONSHIPS

Moment-curvature analysis was carried out by computing h and M / H under a given axial force at
monotonically-increasing sectional curvature steps. The computations were performed by using
closed-form equations derived in the elastic, elastic-plastic, and plastic ranges. To investigate the
influence of the compressive behaviour of masonry, the analyses were conducted separately for the
elastic-perfectly plastic (EPP), EC6 (CEN 2005), Turnšek-Čačovič (1970), and Augenti-Parisi (2010)
constitutive models. In the following, the main analysis results are presented and discussed.

3.1. Case #1: EPP Constitutive Model

For the EPP stress-strain relationship, yielding and ultimate strains were assumed to be 1‰ and 3.5‰,
respectively, resulting in a strain ductility of masonry equal to 3.5. As well-known, the cross-section is
fully resistant in combined flexure-compression at low curvature levels, while it reaches cracking at
large curvature demands. Based on the EPP model, for an axial force increasing in the range [0, Nm/2],
moment capacity increases whereas ultimate curvature reduces. When the given axial force increases
from Nm/2 to Nm, both moment capacity and ultimate curvature reduce. Conversely, cracking curvature
always increases with the given axial force. Figure 2 shows the moment-curvature diagrams plotted
for different values of axial force. One can note that the moment-curvature diagram corresponding at
N/Nm = 0.75 is partially overlapped to that traced for N/Nm = 0.25. Nevertheless, both yielding and
ultimate curvatures are lower for the larger normalised axial force.

0.14 0.05 1
N/Nm = 0.1 -1.11
0.12 φ y = 0.0004 · (N/Nm)
N/Nm = 0.25 0.04 -1 0.8
N/Nm = 0.5 φ u = 0.003 · (N/Nm)
0.1 -1.11
N/Nm = 0.75 ρφy = 0.082 · (N/Nm )
0.03 0.6
M/(Nm · H)

0.08 -1
ρφu = 0.1 · (N/Nm)
φ ·Η

ρφ

0.06 0.02 0.4


0.04
0.01 0.2
0.02

0 0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.2 0.4 0.6 0.8
φ ·H N/Nm

(a) (b)

Figure 2. (a) Moment-curvature diagrams and (b) variation of sectional curvatures and curvature ratios with
normalised axial force for EPP constitutive model

Moment capacity, sectional curvatures and the related control parameters defined in Sect. 2 are listed
in Tables 3.1 and 3.2. They were computed at the actual yielding, peak, and ultimate limit states of the
cross-section without performing bilinear or trilinear approximation of the moment-curvature
diagrams. It is underlined that the EPP constitutive model produces equal values of maximum and
ultimate resisting moment of the cross-section (that is, αM equal to γM and βM equal to unity).
Hardening in moment-curvature response increases with the axial force, as shown by the values of αM
and γM . This outcome is obviously consistent with the limit strength domains at the elastic limit state
(ELS) and ultimate limit state (ULS), where the difference between Mu and My increases in the range
[0, Nm/2] and then reduces at larger axial forces. Regression analyses on sectional curvatures and
curvature ratios at both yielding and ultimate states allowed to get simple equations of the form:

φ = a⋅(N ) ; ρ φ= c ⋅ ( N )
b d
(2.3)

where the values of a, b, c, and d are shown in Figure 2(b). In particular, for the EPP model, d was
found to fall in the interval [−1.11, −1]. The rather high strain ductility assumed for masonry resulted
in curvature ductility equal to 6 for N/Nm falling in the range [0, 0.5] (Tab. 3.2). At larger axial forces,
curvature ductility reached 8 since the cross-section was fully resistant at ELS and cracked at ULS.
Sectional curvature rapidly decreases as the normalised axial force increases. Indeed, for N/Nm = 0.25,
it is 40% of that associated with N/Nm = 0.1 reaching only 20% for N/Nm = 0.5.

Table 3.1. Moment capacity and related control parameters for EPP constitutive model
N My H M max H Mu H αM βM γM
0.10 0.043 0.045 0.045 1.04 1 1.04
0.25 0.083 0.093 0.093 1.12 1 1.12
0.50 0.083 0.124 0.124 1.49 1 1.49
0.75 0.042 0.091 0.091 2.19 1 2.19

Table 3.2. Sectional curvatures and related control parameters for EPP constitutive model
N φy H [%] φu H [%] μφ Cφ ρφy N ( ) ρφu N ( )
0.10 0.50 3.00 6.00 5.00 1.00 1.00
0.25 0.20 1.20 6.00 5.00 0.40 0.40
0.50 0.10 0.60 6.00 5.00 0.20 0.20
0.75 0.05 0.40 8.00 7.00 0.10 0.13

3.2. Case #2: EC6 Constitutive Model

The stress-strain relationship provided by EC6 is non-linear elastic - perfectly plastic. Assuming axial
strains at yielding and ultimate state equal to 2‰ and 3.5‰, respectively, strain ductility was 1.75.
Figures 3(a) and 3(b) show the moment-curvature diagrams and the variation of curvatures and
curvature ratios with the normalised axial force.

0.14 0.05 1
N/Nm = 0.1 φy = 0.0013 · (N/Nm) -1.01
0.12 N/Nm = 0.25 0.04 φu = 0.0028 · (N/Nm)-1 0.8
N/Nm = 0.5
0.1 -1
N/Nm = 0.75 ρφy = 0.1 · (N/Nm)
0.03 0.6
M/(Nm · H)

-1.01
0.08 ρφu = 0.098 · (N/Nm)
φ·H

ρφ

0.06
0.02 0.4
0.04
0.01 0.2
0.02

0 0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.2 0.4 0.6 0.8
φ ·H N/Nm

(a) (b)

Figure 3. (a) Moment-curvature diagrams and (b) variation of sectional curvatures and curvature ratios with
normalised axial force for EC6 constitutive model
The moment-curvature response is rather linear up to ULS for N/Nm = 0.75. The variation of sectional
curvatures with the axial force does not differ from that observed for the EPP model. The parameters
in Eqn. (2.3) were also derived via regression analysis. Yielding curvature was found to be 2.7 and 3.4
times that computed for the EPP model for N/Nm equal to 0.1−0.5 and 0.75, respectively. Conversely,
ultimate curvature was about 0.95 times that corresponding to the EPP constitutive equation.
The presence of a perfectly plastic branch produces the same maximum and ultimate moments and,
hence, βM equal to unity and αM equal to γM (Tab. 3.3). The latter ratios were lower than those
computed for the EPP constitutive model showing a lower moment capacity increase in the post-
yielding range. This was because the assumption of the EC6 constitutive equation causes a yielding
bending moment larger than for the EPP model. Their difference increases from 2 to 31 per cent as the
normalised axial force raises in the interval [0, Nm/2].

Table 3.3. Moment capacity and related control parameters for EC6 constitutive model
N My H M max H Mu H αM βM γM
0.10 0.044 0.045 0.045 1.01 1 1.01
0.25 0.090 0.093 0.093 1.03 1 1.03
0.50 0.109 0.121 0.121 1.11 1 1.11
0.75 0.062 0.086 0.086 1.38 1 1.38

Table 3.4. Sectional curvatures and related control parameters for EC6 constitutive model
N φy H [%] φu H [%] μφ Cφ ρφy N ( ) ρφu N ( )
0.10 1.33 2.83 2.12 1.12 1.00 1.00
0.25 0.53 1.13 2.12 1.12 0.40 0.40
0.50 0.27 0.57 2.12 1.12 0.20 0.20
0.75 0.17 0.38 2.18 1.18 0.13 0.13

3.3. Case #3: Turnšek-Čačovič Constitutive Model

The constitutive equation proposed by Turnšek and Čačovič includes strain-softening of masonry, so it
allows to get a better simulation of monotonic compressive behaviour also in the inelastic range. Since
yielding and ultimate strains were assumed to be 2‰ and 3.5‰, respectively, strain ductility was 1.75.
Figures 4(a) and 4(b) show the moment-curvature diagrams and the variation of curvatures and
curvature rations with the normalised axial force.

0.14 0.05 1
N/Nm = 0.1 φy = 0.0014 · (N/Nm) -1
0.12
N/Nm = 0.25 0.04 φu = 0.0028 · (N/Nm)-1 0.8
0.1 N/Nm = 0.5 -1
ρφu = ρφy = 0.1 · (N/Nm)
N/Nm = 0.75
0.03 0.6
M/(N m · H)

0.08
φ·H

ρφ

0.06 0.02 0.4


0.04
0.01 0.2
0.02

0 0 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.2 0.4 0.6 0.8
φ ·H N/Nm

(a) (b)

Figure 4. (a) Moment-curvature diagrams and (b) variation of sectional curvatures and curvature ratios with
normalised axial force for Turnšek-Čačovič constitutive model
Moment capacity increases with sectional curvature up to ULS for N/Nm = 0.1 while strength
degradation is significant when the cross-section is subjected to larger axial forces. Again, the
parameters in Eqn. (2.3) were estimated through regression analysis.
The main response parameters of the masonry cross-section are listed in Tables 3.5 and 3.6. In this
case, αM is slightly larger than γM , because moment capacity increases as yielding develop throughout
the cross-section (hardening ratio reaches 1.07 at N = Nm/2). On the other hand, γM is almost equal to
βM as strength degradation is nearly identical to hardening. Sectional curvature at ELS is larger than
those corresponding to the constitutive models considered in the previous sections. Ultimate curvature
reduces as the given axial force increases and is equal to that evaluated for the EC6 model.
Nevertheless, curvature ductility was found to be lower for the Turnšek-Čačovič model (Tab. 3.6).

Table 3.5. Moment capacity and related control parameters for Turnšek-Čačovič constitutive model
N My H M max H Mu H αM βM γM
0.10 0.044 0.045 0.045 1.01 1.00 1.00
0.25 0.090 0.092 0.091 1.02 1.01 1.01
0.50 0.111 0.118 0.115 1.07 1.03 1.03
0.75 0.063 0.079 0.070 1.24 1.12 1.11

Table 3.6. Sectional curvatures and related control parameters for Turnšek-Čačovič constitutive model
N φy H [%] φu H [%] μφ Cφ ρφy N ρφu N ( ) ( )
0.10 1.42 2.84 1.99 0.99 1.00 1.00
0.25 0.57 1.13 1.99 0.99 0.40 0.40
0.50 0.28 0.57 1.99 0.99 0.20 0.20
0.75 0.19 0.38 2.01 1.01 0.13 0.13

3.4. Case #4: Augenti-Parisi Constitutive Model

Recently, the authors carried out a number of compression tests on stonemasonry prisms aimed at
deriving two different constitutive models for the principal material directions of masonry with regular
pattern, namely those parallel and orthogonal to the mortar bed joints (Augenti and Parisi 2010). Since
the spandrels in laterally-loaded masonry walls are subjected to compression forces mainly along the
mortar bed joints, the constitutive model considered in this study was that derived for this material
direction. In such a case, strain ductility of masonry was equal to 6. Figure 5(a) shows the moment-
curvature diagrams derived for this constitutive model.

0.14 0.1 1.0


N/Nm = 0.1 φy = 0.0013 · (N/Nm)-1
0.12
N/Nm = 0.25 0.08 φu = 0.0083 · (N/Nm)-1 0.8
0.1 N/Nm = 0.5 -1
N/Nm = 0.75 ρφu = ρχy = 0.1 · (N/Nm)
M/(Nm · H)

0.08 0.06 0.6


φ·H

ρφ

0.06
0.04 0.4
0.04
0.02 0.2
0.02

0 0 0.0
0 0.02 0.04 0.06 0.08 0.1 0 0.2 0.4 0.6 0.8
φ ·H N/Nm

(a) (b)

Figure 5. (a) Moment-curvature diagrams and (b) variation of sectional curvatures and curvature ratios with
normalised axial force for Augenti-Parisi constitutive model
Yielding and maximum bending moments are not considerably different from those evaluated for the
other constitutive models (Tab. 3.7). Strain-softening significantly affects the moment-curvature
response especially for an axial force larger than Nm/2. Curvature ductility was found to be 6.53 for
any compression force level (Tab. 3.8). Whilst yielding curvature is lower than those computed for the
EC6 and Turnšek-Čačovič constitutive models, ultimate curvature is about 2.9 and 3.4 times higher for
N/Nm equal to 0.1−0.5 and 0.75, respectively. Regression analysis was carried out to estimate the
coefficients in Eqn. (2.3). The law of variation of both yielding and ultimate curvatures is not different
from the one derived for the Turnšek-Čačovič constitutive model (Fig. 5(b)).

Table 3.7. Moment capacity and related control parameters for Augenti-Parisi constitutive model
N My H M max H Mu H αM βM γM
0.10 0.044 0.045 0.042 1.01 1.06 0.95
0.25 0.089 0.092 0.076 1.03 1.21 0.86
0.50 0.107 0.119 0.056 1.11 2.13 0.52
0.75 0.060 0.080 0 1.23 ∞ 0

Table 3.8. Sectional curvatures and related control parameters for Augenti-Parisi constitutive model
N φy H [%] φu H [%] μφ Cφ ρφy N ρφu N ( ) ( )
0.10 1.27 8.31 6.53 5.53 1.00 1.00
0.25 0.51 3.33 6.53 5.53 0.40 0.40
0.50 0.25 1.66 6.53 5.53 0.20 0.20
0.75 0.20 1.29 6.53 5.53 0.15 0.15

4. CLOSING REMARKS

Moment-curvature analysis allowed to characterise the non-linear response of URM cross-sections at


the interface between spandrel panels and the adjoining piers. In Figures 6(a) and 6(b), yielding and
ultimate curvatures are plotted against the normalised axial force for the constitutive models
considered in this study. Such curvatures clearly depend on the values assumed for yielding and
ultimate axial strains of masonry. The assumption of the Augent-Parisi model produces large
curvatures at ULS, but moment capacity vanishes at a normalised axial force equal to 0.64 due to
strain-softening. Nevertheless, in laterally-loaded masonry walls with openings, spandrel panels are
typically subjected to low axial forces and to large bending moments and shear forces.

0.02 0.1
EPP model EPP model
EC6 model 0.08 EC6 model
0.015
Turnsek-Cacovic model Turnsek-Cacovic model
Augenti-Parisi model 0.06 Augenti-Parisi model
φy · H

φu · H

0.01
0.04

0.005
0.02

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
N/Nm N/Nm
(a) (b)

Figure 6. (a) Yielding curvature and (b) ultimate curvature versus normalised axial force for different
constitutive models
Therefore, high capability of withstanding large curvature demands is fully ensured to spandrels
resulting in large energy dissipation capacity. The large difference in ultimate curvature predictions
between the Augenti-Parisi constitutive model and the others is highlighted in Figure 7(a), where
ultimate moment capacity is the same and ultimate curvature values are identified with dashed lines.
Ultimate curvature computed for the former model is at least 2.7 times those corresponding to the
other stress-strain relationships considered in this study.
Figure 7(b) shows the variation of curvature ductility with strain ductility of masonry. It is worth
noting that curvature ductility was found between 1.14 and 1.25 times strain ductility when the latter
was set to 1.75 (EC6 and Turnšek-Čačovič constitutive models). This ratio increased to 1.71 for N/Nm
falling in the range [0.1, 0.5] and to 2.28 for N/Nm equal to 0.75 when strain ductility was assumed to
be 3.5 (EPP model). Finally, curvature ductility was just 1.09 times strain ductility for the constitutive
equation by Augenti and Parisi, because strain ductility of masonry was even equal to 6. In general,
the increase in ductility from material to cross-section level was constant with the normalised axial
force and curvature ductility was not lower than 2. If strain-softening is included in the constitutive
law, strength degradation increases with the axial force and cannot be neglected at normalised axial
forces larger than 0.1. Conversely, moment-curvature diagram is affected by sectional hardening even
if a perfectly-plastic behaviour is assumed for masonry.
Since curvature ductility of spandrels is rarely considered in simplified methods of seismic analysis,
its inclusion may lead to a significant overstrength in URM buildings and hence to force-reduction (or
behaviour) factors larger than those currently used within equivalent linear seismic analysis.

EPP model EC6 model N/Nm = 0.1 N/Nm = 0.25


Turnsek-Cacovic model Augenti-Parisi model N/Nm = 0.5 N/Nm = 0.75
0.06 10
9
8
0.04 7
M/(Nm · H)

6
μφ

5
0.02 4
3
2
0 1
0 0.02 0.04 0.06 0.08 0.1 1 2 3 4 5 6 7
φ ·H με
(a) (b)

Figure 7. (a) Comparison between moment-curvature diagrams for different constitutive models and
(b) variation of curvature ductility with strain ductility for different normalised axial forces

REFERENCES

Augenti, N. (2000). Il calcolo sismico degli edifici in muratura, UTET, Turin, Italy (in Italian).
Augenti, N. (2007). Resistenza delle “fasce di piano” di edifici in muratura sollecitati da azioni sismiche. Twelfth
Italian National Conference on Earthquake Engineering. Paper No. 7 (in Italian).
Augenti, N. and Parisi, F. (2010). Constitutive models for tuff masonry under uniaxial compression. Journal of
Materials in Civil Engineering (in press).
Cattari, S. and Lagomarsino, S. (2008). A strength criterion for the flexural behaviour of spandrels in un-
reinforced masonry walls. Fourteenth World Conference on Earthquake Engineering. Paper No. 05-04-0041.
CEN (2005). Eurocode 6: Design of masonry structures. Part 1-1: General rules for reinforced and
unreinforced masonry structures. ENV 1996-1-1, Comité Européen de Normalisation, Brussels, Belgium.
Gattesco, N., Macorini L., Clemente, I. and Noè, S. (2009). Shear resistance of spandrels in ancient brick-
masonry buildings. Thirteenth Italian National Conference on Earthquake Engineering. Paper No. S4-18.
Tomaževič, M. (2000). Earthquake-resistant design of masonry buildings, in Series on Innovation in Structures
and Construction - vol. 1, A.S. Elnashai and P.J. Dowling, eds., Imperial College Press, London, U.K.
Turnšek, V. and Čačovič, F. (1970). Some experimental results on the strength of brick masonry walls. Second
International Brick & Block Masonry Conference. 149-156.

Potrebbero piacerti anche