You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/282442389

New tools for non-linear analysis of masonry buildings

Conference Paper · August 2010

CITATIONS READS

26 412

3 authors, including:

Fulvio Parisi Elia Acconcia


University of Naples Federico II University of Naples Federico II
161 PUBLICATIONS   2,410 CITATIONS    6 PUBLICATIONS   86 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Special Issue of Engineering Failure Analysis on "Damage and Failure Analysis of Civil Structures and Infrastructures: Physics-based versus Data-driven Methods" View
project

All content following this page was uploaded by Fulvio Parisi on 03 October 2015.

The user has requested enhancement of the downloaded file.


New Tools for Non-Linear Analysis of Masonry Buildings

N. Augenti & F. Parisi


University of Naples Federico II, Italy

E. Acconcia
Structural Engineer, Italy

ABSTRACT:
A new computer program for design and assessment of masonry buildings is presented. Starting from the
structural modelling, both gravity loads and seismic hazard at the construction site are defined. Design spectra
can be derived at different limit states and linear equivalent analysis can be run to assess the seismic
performance in a simplified way.
The program is particularly suitable for non-linear seismic analysis being it based on a spread plasticity macro-
element modelling of masonry walls and response-controlled pushover procedures. Geometrical and mechanical
non-linearity sources are considered through evolutionary strength domains and force-displacement curves.
Performance-based verifications are carried out for both pier panels, namely the vertical macro-elements of load-
bearing walls, and spandrel panels, namely the horizontal macro-elements, according to Eurocode 8 and the
Italian building code. Seismic performance can be controlled at both global and local levels through a number of
parameters and building maps related to each analysis step.

Keywords: Masonry buildings, Macro-elements, Response-controlled pushover analysis, Computer program

1. INTRODUCTION

The development of effective tools for non-linear analysis of masonry structures is presently one of
the most research and professional needs, especially in the field of existing constructions. Even though
refined finite element (FE) programs can be employed to investigate the mechanical behaviour of
masonry structures through different numerical strategies, their use in prediction analyses is still
critically discussed as they require high computational effort and expert engineering judgement in the
interpretation of numerical results. The main open issues in FE techniques are related to: (1)
geometrical modelling in the case of rubble stone, or multi-leaf, masonry; (2) the assumption of
reliable values for mechanical parameters of masonry, to be calibrated on an experimental basis; and
(3) the correlation between the stress/strain fields in the continuum model and the performance limit
states conventionally defined in terms of member drifts in the frame of performance-based earthquake
engineering (PBEE). The identification of the finite elements to be monitored for capturing the
achievement of the limit drifts is often ambiguous and needs for repeated averaging operations. Due to
these motivations, in the last decade a number of simplified methods (Magenes and Della Fontana
1998, Augenti 2000) and computer programs (Galasco et al. 2009) based on macro-element modelling
of the structure have been developed to perform seismic analyses on standard masonry buildings.
Amongst the others, the RAN method by Augenti (2000) includes several geometrical and mechanical
non-linearity sources in the structural model in order to get reliable predictions of the actual behaviour
under earthquake loading. The original version of RAN method has been recently extended to non-
linear incremental static (pushover) analysis of masonry buildings by means of a distributed plasticity
modelling of masonry components and response-controlled algorithms (Augenti and Parisi 2009a).
In this paper, after the last version of RAN method is briefly discussed with special emphasis to non-
linear analysis, a new innovative computer program called RAN code is presented. Finally, non-linear
analyses of a masonry building according to Eurocode 8 (EC8) (CEN 2004) and the Italian building
code (IBC) (IMIT 2008) are discussed.
2. THE RAN METHOD FOR PUSHOVER ANALYSIS OF MASONRY BUILDINGS

The RAN method allows to assess seismic performance of masonry buildings composed by load-
bearing walls with openings. The main features of the RAN method are: (1) the modelling of each
masonry wall as a system of two-dimensional elements called ‘macro-elements’; (2) the mechanical
characterisation of vertical and horizontal macro-elements through limit strength domains and force-
displacement curves which depend on geometrical and mechanical non-linearity phenomena; and
(3) the updating of both strength and deformation characteristics as forces/displacements change
throughout each wall during the structural analysis. Masonry is considered as a homogeneous, no-
tensile-resistant material with limited ductility in uniaxial compression.
In the presence of rigid floor diaphragms and tie-elements (e.g., reinforced concrete ring beams or
steel ties) at each floor level, the whole building is idealised as a ‘box system’ whose global
earthquake resistance is chiefly provided by the walls parallel to the loading direction. In the absence
of tie-elements, each masonry wall can be analysed separately from the others through linear or non-
linear seismic analysis.

2.1. Macro-Element Modelling

The macro-element discretisation of each masonry wall with openings allows to simulate the main
characteristics of the in-plane seismic response. By extending the contour lines of the openings, one
can identify the following types of structural components (Fig. 1): ‘pier panels’, which provide load-
carrying capacity to both gravity and horizontal loads; ‘spandrel panels’, which distribute gravity loads
and provide coupling between adjoining piers under horizontal actions; and ‘joint panels’, which link
pier and spandrel panels together. A group of spandrel and joint panels at the same level is referred to
as ‘spandrel’, while a set of pier and joint panels along the same vertical masonry strip is called ‘pier’.

Spandrel panel

Pier panel

Joint panel

Pier

Spandrel

Opening

Figure 1. Identification of macro-elements within a regular masonry wall with openings

Since joint panels are assumed to be rigid and infinitely resistant in the seismic analysis, just strength
and displacement capacity of pier and spandrel panels must be characterised. To this end, different
strategies are employed to perform linear and non-linear analyses. It is also underlined that the macro-
element idealisation of the masonry wall enables to define different mechanical properties for piers
and spandrels.

2.2. In-Plane Behaviour of Pier Panels

The large amount of experimental data on the in-plane resistance of pier panels under different
boundary conditions has allowed to derive several theoretical formulations (Magenes and Calvi 1997).
In-plane shear resistance against diagonal cracking and bed-joint sliding mechanisms is estimated via
limit strength domains derived from local plasticity failure criteria (Turnšek and Čačovič 1970).
Augenti and Parisi (2009c) proposed a spread plasticity macro-element whose flexural strength is
defined through evolutionary shear force - axial force domains including the constitutive model
adopted for masonry. Force-displacement curves were also developed through a deformation-based
approach by limiting the ultimate axial strain at the end sections (Augenti and Parisi 2009b).
Figure 2 shows a typical force-displacement curve for an unreinforced masonry (URM) pier panel
failing in flexure, while Figures 3(a), 3(b) and 3(c) show the geometry evolution of the resisting
domain inside the panel under increasing deformations.
Given an axial force, as the maximum axial strain at the end sections is monotonically increased, also
the relative lateral displacement increases due to shearing and flexural deformations, as well as
rocking. Accordingly, the lateral resistance of the panel changes resulting in the following branches of
the force-displacement curve:
- The first branch is associated with an uncracked panel in elastic conditions (Note that only
material non-linearity at small deformations affects the flexural strength.). The effective resistant
region, denoted by Ω, has the gross geometry of the macro-element (Fig. 3(a)).
- The second branch describes the evolution of the lateral resistance in the presence of mechanical
non-linearity in the elastic range and geometrical non-linearity due to cracking. At this stage, the
effective part of the panel can be considered as the union of the sub-regions Ω1 and Ω2 which are
related to the uncracked and cracked parts, respectively (Fig. 3(b)).
- The third and the fourth branches are related to a cracked and yielded panel (Fig. 3(c)). When the
maximum axial strain at the end section reaches yielding strain of masonry, a plastic deformation
localization occurs within a further region, denoted by Ω3 , resulting in a hardened behaviour of
the panel. At larger displacements, strain-softening of masonry significantly affects the in-plane
lateral response and the hardening due to damage localization is progressively ‘covered’ resulting
in a flexural strength degradation. The panel is assumed to fail when the ultimate axial strain is
attained (Note that masonry softening induces a contraction of the flexural strength domain.).

V
Vmax ULS N
(s)

Vu ELS 3 4 (s) M
(s)

Ve V
2
Vl (i)
(i) V
1 M (i)
N
O δl δe δ (Vmax) δu δ

Figure 2. Typical force-displacement curve for an URM pier failing in flexure

z z z
εz,max εz,min εz,max εz,max
Ω3
Ω2

Ω2
Ω H/2
ze
Ω1 zn
Ω1 zn

G x G x G x
B B B
(a) (b) (c)

Figure 3. Evolution of the macro-element with the deformation state: (a) uncracked panel in the elastic range;
(b) cracked panel in the elastic range; and (c) cracked panel in the plastic range
Lateral resistance and displacement capacity are thus derived by integrating axial strains within the
panel and summing up the contribution of shearing deformations and rocking response.
If shear strength associated with diagonal cracking or bed-joint sliding is reached at a given shear
force level, a plastic plateau with limited ductility is added to the force-displacement curve.

2.3. In-Plane Behaviour of Spandrel Panels

The in-plane lateral response of masonry walls with openings is drastically affected by coupling
provided by spandrels. Although few experimental data are currently available on these structural
elements, a theoretical model has been proposed by Augenti (2007) to predict flexural strength without
imposing any boundary condition. The failure domain employed in the RAN method is three-
dimensional and was derived by simple equilibrium equations. Figure 4 shows the dimensionless
interaction domain of a spandrel panel having dimensions 90×140×60 cm3. Both axial and shear forces
(denoted by N and V, respectively) are normalised to the maximum axial force Nm, while the load
eccentricity e is normalized to the height H of the panel.

Figure 4. Typical three-dimensional failure domain of a spandrel panel failing in flexure

2.4. Response-Controlled Pushover Analysis

Incremental non-linear static analysis is carried out by monotonically increasing the horizontal
absolute displacement at a control point of the building. The procedure formulated by Augenti and
Parisi (2009a) includes both inherent (structural) and accidental eccentricities of the seismic actions to
estimate torsionally-induced displacements. At each displacement increment at the control point, the
deformed shape associated with a pre-defined lateral force pattern (e.g., modal or uniform) is
determined through the Pegasus method. The magnitude (not the shape) of the horizontal forces is
changed by a load multiplier derived through a total updating scheme in a way to avoid any relation
between equilibrated solutions corresponding to subsequent displacement steps.
Spread plasticity modelling of macro-elements, as well as the formulations employed to evaluate their
strength and deformation capacity, are perfectly consistent with the response-controlled pushover
analysis. In fact, even if the computational strategy adopted in the RAN method is presently force-
based and non-adaptive, it allows to control either the global response of the masonry structure, or the
local response of the macro-elements, as damage occurs and develops within them. Furthermore, no
pre-defined, multi-linear force-displacement curves are used but they are conditioned upon the overall
response of the structure by means of equilibrium and kinematic compatibility equations.
The global collapse of the building is assumed to occur when the lateral resistance of a storey level
vanishes.
3. THE RAN CODE

In order to carry out non-linear seismic analysis on masonry buildings modelled with spread plasticity
macro-elements, the authors developed a new computer program called RAN code by implementing
not only the last version of the RAN method, but also the simplified seismic hazard assessment
procedure given in IBC (IMIT 2008) and the derivation of design acceleration spectra.
Seismic hazard at a site is estimated in terms of peak ground acceleration at the bedrock, ag, and the
following spectral shape parameters: maximum amplification factor of the horizontal acceleration
response spectrum, Fo; and upper limit of the period of the constant spectral acceleration branch on
type A ground, TC*. These hazard parameters are associated with a given return period of the design
earthquake which, in turns, is derived from a (temporal) reference period VR and the probability of
exceeding the ag value within VR, denoted by PVR. Whilst VR is defined as the nominal life of the
structure (VN) times the importance coefficient for the construction (CU), PVR changes with the limit
state of interest in accordance with the general PBEE methodology. To account for amplification
effects due to local soil conditions, ag is multiplied by a soil factor defined as the product of a
stratigraphic amplification factor, SS, by a topographic amplification factor, ST. Peak ground
acceleration at a site is then given by S times ag.
Based on the seismic source zone model ZS9 by Meletti et al. (2008) (Fig. 5(a)), the hazard parameters
were computed by the Italian National Institute of Geophysics and Volcanology at each node of a
regular grid having 5 km spacing and covering the whole Italian territory with 10751 nodes. If the site
is not a node of the reference grid, the nodes of the elementary mesh containing the site must be
identified. The hazard parameters are then computed as weighted average of the ones at the reference
nodes by the distances of the nodes from the site. RAN code enables the analyst to estimate the seismic
hazard parameters at any site and for each limit state of interest, once geographical coordinates are
entered (or the site name is selected) along with VN and CU. Figure 5(b) shows the program window
related to the code-based automated hazard assessment, as well as the derivation of design acceleration
spectra at the limit states of interest for seismic safety verifications. If another national or international
seismic code like as EC8 is chosen, the user can set the hazard parameters by hand.

Figure 5. (a) Seismic source zone model ZS9 by Meletti et al. (2008) for the Italian territory and (b) code-based
estimation of seismic hazard parameters for seismic analysis into RAN code

Once the seismic hazard parameters have been evaluated or defined, the structural model must be
completely built up by entering the values of the main mechanical properties of the masonry types and
tensile resistant elements within spandrels. A number of planar and three-dimensional views allow to
control all dimensions entered by the user for each masonry wall.
Structural analysis under gravity loads and safety verifications against out-of-plane failure modes
under local seismic actions can be carried out. In this analysis stage, one can evaluate demand-to-
capacity ratios related to ultimate limit state in the absence of earthquake loading according to IBC
and Eurocode 6 (CEN 2005). This computation is particularly important to detect the in-plan and
vertical distribution of axial forces within masonry piers and, thus, to investigate their influence on
potential failure modes under seismic actions.
Linear seismic analysis can be then performed and further safety verifications can be carried out under
the horizontal forces derived from design acceleration spectra, including torsional response of the
building due to both structural and accidental eccentricities at each floor level.
Non-linear seismic analysis is run by limiting yielding and ultimate axial strains of masonry in the
uniaxial constitutive model proposed by Turnšek and Čačovič (1970) and implemented into RAN code
for step-by-step double integration of strains within each macro-element as the analysis goes on. The
main analysis output is a set of pushover curves associated with two lateral force distributions: the
modal pattern and the uniform pattern. Each curve describes the overall inelastic behaviour of the
structure and is scaled through the first mode participation factor and approximated as linear elastic -
perfectly plastic model according to the N2 method (Fajfar 2000) to compute the lateral stiffness,
lateral resistance, and displacement capacity of the equivalent single-degree-of-freedom system.
Finally, seismic performance at each limit state of interest is assessed by comparing the displacement
capacity of the building to the displacement demand corresponding to the expected earthquake.

4. NON-LINEAR SEISMIC ANALYSES OF A MASONRY BUILDING

Pushover analyses of a three-storey masonry school building located in Gesualdo (Avellino, Italy), a
town placed near the Irpinia fault system, are discussed in this section.
Since school buildings are subjected to high concentration of people, the importance factor was set to
1.5 (i.e., utilization class III according to IBC). With reference to life safety (LS) limit state, the return
period of the design earthquake was estimated in 712 years so the following hazard parameters were
considered in the analysis: ag = 0.316 g; Fo = 2.285; and TC* = 0.393 s. Site amplification effects were
take into account by the stratigraphic amplification factor SS = 1.11 (type B ground) and the
topographic amplification factor ST = 1. Therefore, the characteristic periods of the design acceleration
spectrum at LS limit state were: TB = 0.174 s; TC = 0.521 s; and TD = 2.874 s.
Figure 6(a) shows the typical plan of the building which is characterised by a central open space and
eccentric location of the staircases. The three-dimensional structural model is showed in Figure 6(b).
33.00

37.00
(a) (b)

Figure 6. (a) Typical plan of the building and (b) three-dimensional structural model into RAN code

To preliminarily analyse the non-linear seismic response of the building, no accidental eccentricity
was applied to the centres of mass along the height. Given that a non-uniform distribution of gravity
loads was detected due to the variation of live loads in the plan, twelve different angles of incidence
(denoted by α) were considered for the horizontal actions in order to investigate the variation of both
seismic demand and capacity parameters with the earthquake loading direction.
Figures 7(a) and 7(b) show pushover curves corresponding to several angles of incidence for modal
and uniform load patterns, respectively. It is underlined that the initial elastic response predictions are
the same resulting in similar displacement demands. Conversely, the maximum and ultimate base
shears of the structure are very different as the failure modes of the macro-elements drastically
changed from an angle of incidence to another.

x103 x103
40 40
α=0 α=0
α=30° α=30°
α=60° α=60°
30 30
α=90° α=90°
α=120° α=120°
Vb [kN]

Vb [kN]
α=150° α=150°
20 α=180°
20 α=180°
α=210° α=210°
α=240° α=240°
10 α=270° 10 α=270°
α=300° α=300°
α=330° α=330°
0 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
d [mm] d [mm]
(a) (b)

Figure 7. Pushover curves corresponding to several angles of incidence for (a) modal load pattern and
(b) uniform load pattern

A polar representation of capacity and demand parameters is particularly helpful to investigate the
influence of the angle of incidence on them. Figure 8(a) shows the variation of displacement capacity
and displacement demand with angle of incidence for a modal load pattern. It is worth noting that
displacement demand does not significantly change with the loading direction consistently with
pushover curves in Figure 7(a). On the other hand, displacement capacity is larger for α falling in the
range [0, 90°] and equal to 330°. The inelastic response of the building is thus asymmetric with respect
to the x-axis of the building plan and rather symmetric with respect to the y-axis. These results are
consistent with the fact that the building plan is symmetric with respect to the y-axis. Demand-to-
capacity ratios in terms of displacements are larger than unity for α falling in the range ]210°, 330°[.
The allowable PGA was evaluated either at the maximum base shear (Vb,max), or at the displacement
demand (ddem). Figures 8(b) and 8(c) allows to compare the PGA demand (PGAdem) to the allowable
PGA (PGAalw) at Vb,max and ddem, respectively, for different angles of incidence. In both cases, the
building does not satisfy the code safety level resulting in a seismic safety index, ISS = PGAalw/PGAdem,
lower than unity.
Figures 9(a) and 9(c) demonstrate that seismic performance of the building is poorer for uniform load
pattern, while Figure 9(b) does not considerably differ from Figure 8(b).

(a) (b) (c)

Figure 8. Variation of capacity and demand parameters with angle of incidence (in degrees) at LS limit state for
modal load pattern: (a) displacements; (b) PGAs at Vb,max; and (c) PGAs at ddem
(a) (b) (c)

Figure 9. Variation of capacity and demand parameters with angle of incidence (in degrees) at LS limit state for
uniform load pattern: (a) displacements; (b) PGAs at Vb,max; and (c) PGAs at ddem

5. CONCLUSIONS

RAN code allows to carry out both linear and non-linear seismic analyses on masonry buildings by
means of macro-element modelling of load-bearing walls with openings. Spread plasticity modelling
of macro-elements makes possible the simulation of the non-linear response by means of evolutionary
strength domains and force-displacement curves including geometrical and mechanical non-linearity
sources. Results from response-controlled pushover analysis are processed to carry out performance-
based verifications at both global and local levels via several parameters and indices.

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. (2009a). Non-linear static analysis of masonry structures. Thirteenth Italian National
Conference on Earthquake Engineering. Paper No. S4-01.
Augenti, N. and Parisi, F. (2009b). Force-displacement curves in displacement control. Thirteenth Italian
National Conference on Earthquake Engineering. Paper No. S4-02.
Augenti, N. and Parisi, F. (2009c). Influence of the constitutive law on the flexural strength of masonry panels.
Thirteenth Italian National Conference on Earthquake Engineering. Paper No. S4-03.
CEN (2004). Eurocode 8: Design of structures for earthquake resistance. Part 1 - General rules, seismic actions
and rules for buildings, ENV 1998-1, Comité Européen de Normalisation, Brussels, Belgium.
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.
Fajfar, P. (2000). A nonlinear analysis method for performance-based seismic design. Earthquake Spectra, 16:3,
573-591.
Galasco, A., Lagomarsino, S. and Penna, A. (2009). TREMURI Program: Seismic Analysis of 3D Masonry
Buildings, University of Genoa, Italy.
IMIT (2008). D.M. 14.01.2008: Norme Tecniche per le Costruzioni. Italian Ministry of Infrastructures and
Transportation, Rome, Italy (in Italian).
Magenes, G. and Calvi, G.M. (1997). In-plane seismic response of brick masonry walls. Earthquake Engineering
and Structural Dynamics. 26, 1091-1112.
Magenes, G. and Della Fontana, A. (1998). Simplified Non-linear Seismic Analysis of Masonry Buildings.
Proceedings of the British Masonry Society. 8, 190-195.
Meletti, C., Galadini, F., Valensise, G., Stucchi, M., Basili, R., Barba, S., Vannucci, G. and Boschi, E. (2008). A
seismic source zone model for the seismic hazard assessment of the Italian territory. Tectonophysics. 450:1-
4, 85-108.
Turnšek, V. and Čačovič, F. (1970). Some experimental results on the strength of brick masonry walls. Second
International Brick & Block Masonry Conference. 149-156.

View publication stats

You might also like