You are on page 1of 14

Available online at www.sciencedirect.

com

Advances in Engineering Software 39 (2008) 301–314
www.elsevier.com/locate/advengsoft

Numerical characterization of the structural behaviour of the Basilica
of Pilar in Zaragoza (Spain). Part 1: Global and local models
a,* a,1 b,2
L.E. Romera , S. Hernández , J.M. Reinosa
a
Structural Mechanics Group, School of Civil Engineering, University of Coruña, Campus de Elviña, 15071 La Coruña, Spain
b
Department of Mechanical Engineering, Industrial Engineering, University of Coruña, C/Mendizábal, 15403 Ferrol, Spain

Available online 30 April 2007

Abstract

The Basilica of Pilar, located in the city of Zaragoza, is one of the most famous Spanish temples. Several domes of this church contain
frescoes painted by Francisco de Goya, which have suffered damages in the past decades due to various pathologies. The frescoes of the
Regina Martirum dome are being restored at the present time. In this paper, we present a set of structural models of the entire temple and
local models of the Regina Martirum dome by discussing and comparing the obtained results. The main objective of the research was to
identify the current structural state of the church, its safety level and the relationship between the structural behaviour and the damages
observed. The behaviour of the main material used in the construction, brick masonry, was simulated first with linear elastic isotropic
models and later with several non-linear isotropic models including brittle behaviour in tension and plastic behaviour in compression. In
addition, we consider the historical construction and the reinforcement works added to the structure in its history, with the purpose of
testing the finite element models by reproducing the pathologies described from the beginning of the current Baroque construction. 
2007 Elsevier Ltd. All rights reserved.

Keywords: Historical buildings; Masonry; Structural models; Visualization; Finite element analysis; Non-linear material

1. The building and its history in XVII century when the Baroque church project was set
up by the Spanish architect Ventura Rodriguez. In addition
The Basilica of Pilar is one of the most important wor- a reorientation to the main axis of the temple took place
ship places in Spain. The actual temple is a large complex and part of the new construction was erected on terrain
brick masonry construction with rectangular plan of about taken to the adjacent river. Fig. 1a presents a sketch
100 m long and 70 m wide. Their construction process was describing the position of each building along the history
difficult and took long time [1]. In the current site, a small and some plans; Fig. 2 shows several views of the actual
Visighotic chapel was first built more than a thousand Basilica.
years ago; then later on, it became a Romanesque church The temple is composed of three longitudinal naves
with cloister that suffered after the Gothic enlargement. (Fig. 1b). The central one contains the main dome of circu-
Nevertheless, all the aforementioned constructions were lar design and double shell, and two more domes of ellipti-
small in size compared with the huge expansion taken place cal shape (Fig. 1d). The two lateral naves have eight more
domes placed at a second level, the Regina Martirum dome
belongs to this second level, located next to the central
*
Corresponding author. Tel.: +34 981167000x1404; fax: +34 dome (Fig. 1c). Surrounding the principal naves there is a
981167100. series of chapels and rooms with up to 11 additional domes
E-mail addresses: lromera@udc.es (L.E. Romera), jreinosa@udc.es at the third lower level. The domes at the third level are
(J.M. Reinosa).
URL: http://caminos.udc.es/grupos/mmcte (L.E. Romera).
located between the 16 external buttresses and the four
1
Tel.: +34 981167000x1404; fax: +34 981167100. towers, of more than 90 m height, that mark the corners
2
Tel.: +34 981167000x3209; fax: +34 981337410. of the building and form the external perimeter. The

0965-9978/$ - see front matter  2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.advengsoft.2007.01.009
302 L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314

Fig. 1. Plans: (a) evolution of the worship site; (b) inferior top view; (c) front section BB for the Regina Martirum dome; (d) front section CC for the
elliptical dome.

central dome was erected by the middle of XIX century, • An extensive injection of cement grouting in the soil.
and the first two towers of Pilar square façade were final- • Foundation improvement with reinforced concrete
ized at the end of the XIX century while the towers of Ebro beams linking steel lining at pillars foundation.
river façade were erected in 1940. • Temporary support and steel reinforcement of central
arches and pillars.
2. Structural pathologies and past repairing works • External reinforcement of several tambours by rein-
forced concrete.
The first references to verify problems about fissures in • Installation of inclined supports in tambours under the
arches, tambours and vaults, especially in elliptical domes, central and the elliptical domes, connecting them with
appeared in 1796, and also in 1804 some reports come into external buttresses.
view concerning structural problems in several arches and
vaults. In 1907 after the completion of the southern towers A few years ago other structural pathologies, in a smaller
symptoms of general ruin emerged, and then in 1927 cracks scale of previously mentioned, were observed. They were
of significant size were visible in arches, pillars were out of significant because they could affect frescoes of one of the
their vertical orientation and important relative displace- most brilliant Spanish painters, Francisco de Goya. These
ments were observed in the Regina Martirum dome. In fact pathologies are:
a discontinuity of about 4 cm, located in the connexion
between the dome and its tambour, and throughout a third • Presence of humidity, small fissures and loss of material
of a meridian could be observed in the Regina Martirum in frescoes of Regina Martirum dome (Fig. 5) and other
dome at the present time (Fig. 5b). Therefore, an ambitious domes.
programme of repairing works was approved under the • Cracks in several arches near the Santiago apostle
direction of the architect Teodoro Rı́os in 1927. The reha- chapel.
bilitation task continued until the beginning of the Spanish • Deterioration of masonry towers, especially in the tow-
civil war started in 1936 and after the war, it was resumed ers of square facade.
again to be completed in 1940 (Fig. 3). The most important
features of the restoration (Fig. 4) can be summarized as To determine if the damage of frescoes was due to struc-
follows. tural pathologies, and if this was the case, to outline the
L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314 303

digital model of the complete church was generated by
using information from old blueprints and in situ measure-
ments and observations [4]. This model was used later as
geometrical bases for the development of the structural
models. The development of this geometrical model
(Fig. 7) allowed us to understand better the architectural
composition as well as the construction connectivity
between the components of the temple. The accomplish-
ment of the model was based on the idea that geometry
was as exact as possible, avoiding any simplification that
could alter the structural behaviour and only eliminating
the non-structural elements.
From the geometrical model, the visualization model
(Fig. 8) and several structural models were developed
using finite elements techniques, in the first place by means
of the linear theory and later considering the material non-
linearities. Due to the importance of repairing works exe-
cuted between 1927 and 1940, which avoided the collapse
of the temple, the first structural model was created to
reflect the state of the construction previous to the afore-
mentioned reforms. The comparison of the results
obtained in this model with the available historical descrip-
tions of the pathologies that forced the repairing works,
has allowed us to verify the hypotheses of the structure
and the loads considered in the model. After verifying
the 1927 model, the necessary modifications were made
in a new model corresponding to the present structural
state, to simulate the repairing works and new structures
such as the river towers, which was again recalculated with
the same loads.

4. Structural models

To develop the structural models the codes Cosmos/m
v.2.8 [5,6] and Msc.Marc2003 [7] have been used, combin-
ing several element types depending on the architectural
parts of the construction. Hexahedral and tetrahedron ele-
ments were used for modelling arches, pillars, tambours,
buttresses and soil foundation; whereas beam, bar and
quadrilateral or triangular shell elements were selected
for meshing domes, floors, walls and reinforcements struc-
tures of steel and concrete. The only temple parts not
included, due to their non-structural function, are the cover
system and the chapiters of domes and towers.
For each model, the first version of linear solids and
shell elements with 4 or 3 nodes, and the second version
Fig. 2. The Basilica of Pilar in Zaragoza: (a) external view; (b) interior of quadratic solids and shells with 6–9 nodes were ana-
view of central dome; (c) longitudinal nave. lyzed. The latter mesh with quadratic elements is composed
of 82.000 elements and 140.000 nodes.
necessary actions to repair them, an extensive numerical In the transition areas, in order to reduce the number of
study was carried out [2,3]. elements, an incompatible mesh was used between tetrahe-
dron and hexahedral or between hexahedral and shell ele-
3. Methodology: cad and visualization ments in the central dome (Fig. 9). In order to assure the
displacement compatibility in the non coincident nodes,
The approach used in this study is described in the flow- routines that impose kinematics restriction equations of
chart of Fig. 6. Due to geometrical complexity and the sig- the incompatible nodes have been developed. Considering
nificant structural dimensions, first of all, a detailed 3D the case represented in Fig. 9c, the restriction prevails by
304 L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314

Fig. 3. Outside views in 1939.

Fig. 4. Repairing works: (a) foundation reinforcements; (b) steel bonds of tambours; (c) discharge arches.

imposing the following equations on the displace- ple was generated (Fig. 11) with all repairing works
ments of node s1 of the hexahedron and node t1 of the included. The results obtained by the actual state model,
tetrahedron: with soil properties enhanced due to injections of cement
X
nes X
net grouting and reinforced concrete beams linking steel cais-
ut 1 ¼ N hi ðxt1 Þus and us1 ¼ N ti ðxs1 Þut ð1Þ sons at pillars foundations, were almost identical to those
i¼1 i¼1 obtained with fixed basement. Thereby, the soil was elimi-
where N hi ðxt1 Þ and N ti ðxs1 Þ are the interpolation functions nated from the current state model.
of the hexahedron and the tetrahedron particularized to Local models, with shell elements of variable thickness
the coordinates of nodes t1 and s1, respectively. or solid elements, of Regina Martirum dome are presented
In the unions between C1 type elements (shells and in Fig. 12. The dome has a diameter of 10.6 m and an angle
beams) and C0 elements (solid), all degrees of freedom have between the tambour and the lantern of 66.6, its thickness
been considered linked, by means of new attach coupling varies from 0.8 m in the lower part to 0.25 m in the upper
equations between the displacements of one line defined part; the lantern thickness is about 0.7 m. To take into
by two nodes in solids elements, and the nodal displace- account the interaction with the rest of the temple, the final
ments and rotations of shell and beam elements situated displacements obtained in the dome base, or in the tam-
in this line [8,9]. bour, from the global models have been imposed as kine-
The first global model intended to represent the behav- matics constraints to the dome base in the local model,
iour of the temple before the repairing works started in along with the gravitational loads and the external weights
1927 is presented in Fig. 10 with the unfinished two towers that act on the dome. This approach is exact for linear
of the riverside. Due to the poor quality of the alluvial sed- analysis, although it introduces some error level in non-lin-
imentary soil next to the Ebro river, the surrounding ter- ear analysis. In any case, the results are better than those
rain was included in the model by means of a compatible with fixed base dome and we consider that it allows us to
mesh of prismatic solid elements. After verifying the 1927 obtain one of the first non-linear approaches of its behav-
model, a new one representing the current state of the tem- iour and the collapse load.
L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314 305

Fig. 5. Frescoes of Regina Martirum dome: (a) general view with presence of humidity; (b) old crack and loss of frescoes; (c) detail of a previous
restoration work; (d) fissures detail.

Fig. 6. Flowchart of methodology.

Fig. 7. Geometrical model.

4.1. Considered loads • Dead loads due to lanterns and spires in the upper part
of domes and towers, introduced as nodal force loads.
The loads were assumed to be static and composed of: • Uniform thermal load with DT = ±15 C in the whole
temple.
• Self weight of the structure. • Thermal gradient in the local models of Regina Martirum
• Dead loads due to the cover systems of the whole temple dome produced by a temperature difference of ±15 C
and domes, introduced as pressure loads. between the inner and the outer surface of the dome.
306 L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314

Fig. 8. Visualization model.

Fig. 9. Details of transitions: (a) between tetrahedrons and hexahedrons; (b) between shell and solids; (c) example of application of kinematics coupling.

Fig. 10. The 1927 structural model: (a) 3D view; (b) section of the elliptical dome; (c) top view of terrain; (d) top view of the entire model.
L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314 307

Fig. 11. Actual state model: (a) 3D view; (b) longitudinal section; (c) reinforcements of tambours; (d) steel reinforcements of central arches and pillars.

Fig. 12. Local models of Regina Martirum dome: (a) shell model with a part of the tambour; (b) solid model.

The weight of non structural elements on the upper part eses of the soil mechanical properties distribution were
of domes is important due to the use of metals like cop- used to take into account the possible different behaviour
per or lead in the spires. In the case of Regina Martirum of the original sedimentary soil and the soil taken from
dome, the total vertical force on the lantern is about the Ebro river, with more reduced mechanical properties
630 kN, and 1.5 kPa for the pressure due to the cover (soil 2 in Table 1).
system. The principal masonry pillars, with cross-sectional
dimensions of about 6 m · 6 m, have been considered mas-
4.2. Material models sive in the models, however they were analyzed with
reduced elasticity modulus in several cases due to possible
In the linear models, the brick masonry [10,11], the soil, existence of a inner filled nucleus surrounded by masonry.
the steel and concrete reinforcements have been considered After carrying out linear analysis, several non-linear
as linear and isotropic materials with their properties materials were considered for the brick masonry (Fig. 13)
shown in Table 1. In the 1927 model, two different hypoth- and the soil of 1927 model. In all models, brick masonry
308 L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314

Table 1 5. Numerical results
Linear material properties
E (kPa) m a (m/Cm) c (kN s2/m3) The results of the 1927 model are presented first, fol-
Masonry 4.5 · 106 0.1 1 · 105 1.8 lowed by the results of the actual state model and the local
Soil 1 2.5 · 106 0.3 1 · 105 – Regina Martirum dome.
Soil 2 1.0 · 105 0.3 1 · 105 –
Pillars 2.25 · 106 0.1 1 · 105 1.8
Concrete 3.0 · 107 0.2 1 · 105 2.4 5.1. The 1927 model
Steel 2.1 · 108 0.3 1.2 · 105 7.85
Considering linear material, a uniform soil of type 1 and
only self weight and dead loads, maximum vertical dis-
was simulated as an isotropic material, with brittle behav- placements reaches 2.41 · 102 m in towers and central
iour in tension and linear in compression (model 1 in dome, principal stresses field reaches maximum values of
Fig. 13). Moreover, the possibility of plastic behaviour in 719 kPa for tensile stresses, and 3300 kPa for compressive
compression (model 2 in Fig. 12) was considered using an stresses (Fig. 14). Both values surpass the allowable range
elastoplastic Mohr–Coulomb linear model [12]. For the for masonry.
soil, an elastic perfectly plastic Drucker–Prager model Table 2 shows values of maximum principal tensile and
was used. compressive stresses in each one of the following structural
Material model 1 was programmed inside the Cosmos/ elements: pillars and arches, towers, buttress, tambours,
m code, and it was applied to the global temple model only vaults, domes and the Regina Martirum dome. If an elas-
for solid elements using a smear-crack approximation [13] toplastic material soil model is used, with usual parameters,
because of the difficulties of implementing this type of plastic areas appear under several pillars, increasing move-
models in shell typologies. In each non-linear iteration ments and stresses.
and for every Gauss point of integration, the possibility The main cause of these significant values is the exces-
of cracking perpendicular to the principal tension stress sive size and weight from the central dome and the square
was checked out. Once the crack is open, its direction is towers. The construction of the central dome imposes lon-
fixed, in another words, the possibility of rotation of the gitudinal and transversal rotations (global axis X and Y in
fissures once opened at increasing load levels is not Fig. 14b) to the tambours of second level domes and
considered. increases the stresses in arches, and similarly the interaction
Material model 2 was applied only to the local analysis between the square towers and the rest of the Basilica tend
of Regina Martirum dome modelled with solids elements to produce horizontal movements in towers, a loss of ver-
using the non-linear finite element analysis code ticality, and high tensile stresses in their connection with
Msc.Marc2003. In order to apply this material model, it the temple. Slender elements, like vaults, connected with
required us to work with only solid elements in the global pillars and arches of central dome suffer very high tensile
model, which was not a possible solution considering all and compressive stresses.
the existing domes and vaults. For this reason, a local solid Fig. 15 shows membrane and top face principal tensile
model with solid elements of Regina Martirum dome was stresses for Regina Martirum dome, and the maximum
developed. principal tensile stresses developed in tambours. The tem-
For the material parameters used in both models, a typ- ple stresses distribution agrees with the general description
ical value range of similar historical brick masonry struc- of structural damages at that time, with their maximum
tures obtained from bibliography [14,15] was considered. values in arches and vaults surrounding the central dome,
A parameter variation study of results concerning the value and with important tensile stresses in the base of several
range was performed as well. A gypsum mortar with very domes tambours, especially in the elliptic ones. Fig. 16
low resistance under tension is commonly used with the shows the cracking results obtained from non-linear mate-
masonry. The used values are indicated in the results. rial analysis using material model 1, assuming a limit ten-

Fig. 13. Non-linear material models.
L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314 309

Fig. 14. The 1927 model with linear materials: (a) vertical displacements (m); (b) resultant displacements (m) on deformed shape (scale 1000); (c) principal
tensile stresses (kPa); (d) principal compressive stresses (kPa).

Table 2 to be a possible origin, with approximately 1/6 of the base
1927 model: maximum principal tensile (P1) and compressive stresses (P3) of the temple on a soil of poor quality. This distribution is
in several parts of the model
agreed with the terrain taken from the adjacent river
P1 (kPa) P3 (kPa) described in the historical introduction. In these conditions
Pillars and arches 719 3300 a distribution of tensile stresses with maximum values up to
Towers 177 1261 1300 kPa were obtained in the dome, coincident with the
Buttresses 583 784
Tambours 493 1599
direction and the position where real fissure is located.
Uniform thermal loads were applied imposing a con-
M B M B stant temperature of ±15 C in all nodes of the temple,
Vaults 653 1265 1277 1033 the same temperature in the soil surface and a linear vari-
Domes 301 280 712 694 ation to 0 C (temperature for zero stress) in the lower soil.
Regina Martirum dome 169 265 513 635
Their effect is less important than the gravitational ones,
In shell elements membrane (M) and bending (B) values are indicated fundamentally maximum tensile and compressive stresses
separately.
appear in connections between towers and the basilica,
and in some perimeter walls, which are coincident with fis-
sile cracking stress rcr = 150 kPa, a residual stress of 10% sures that are opened and closed at the present time based
of the previous value rres = 0.1rcr, an unloading module on the temperature variations. Stresses in domes due to
of elasticity of 10% of the linear value Es = 0.1E, and a uniform thermal effects are low, for the Regina Martirum
retention shear factor of 0.01 for open cracks. It is dome the maximum values are 41 kPa/53 kPa for princi-
observed that the cracks begins in arches and later progress pal membrane tensile and compressive stresses respectively,
fundamentally along the tambours of the elliptic domes. and 68 kPa/28 kPa for principal bending.
The fissure of the Regina Martirum dome, mentioned in
the introduction, which was later repaired, does not appear 5.2. Actual state model
clearly in the previous results. After working with several
hypotheses of differentials movements, a distribution of For the actual state model with linear analysis, the
properties of the terrain represented in Fig. 17 is considered sequential application of the repairing works allowed us
310 L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314

Fig. 15. Stress results (kPa): (a) principal tensile stresses in tambours; (b) Regina Martirum dome: principal membrane tensile stresses; (c) Regina
Martirum dome: principal tensile stress in top face.

Fig. 16. Percentage of integration cracked in solid elements: (a) 3D view; (b) top view.

to check their relative effect, and we found it especially sig- for tensile stresses in arches keystone, and to 1125 kPa
nificant in the case of the foundation works and steel lining for compressive stresses in pillars basement. The maximum
of pillars and arches. Considering the global model with value of the local tensile stress appears in the tower connec-
linear material properties and all the rehabilitation works tions with the main temple with a value of 467 kPa.
applied, maximum vertical displacements decreases to Table 3 shows values of maximum principal tensile and 
0.76 · 102 m in towers and central dome, and stresses compressive stress in each one of the structural elements
field in masonry decreases to maximum values of 173 kPa and its variation with respect to the results of the 1927
L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314 311

Fig. 17. Model with two different types of soils: (a) variation of soil properties; (b) maximum principal tensile stresses in Regina Martirum dome (kPa).

Table 3
Actual state model: maximum principal tensile (P1) and compressive stresses (P3) in several parts of the model
P1 (kPa) P3 (kPa)
Pillars and arches 173 (76%) 1125 (66%)
Towers 83 (53%) 888 (30%)
Buttresses 52 (91%) 723 (8%)
Tambours 176 (64%) 622 (64%)
M B M B
Vaults 184 (72%) 365 (71%) 581 (55%) 591 (43%)
Domes 147 (51%) 246 (12%) 540 (24%) 608 (12%)
Regina Martirum dome 89 (47%) 246 (7%) 423 (18%) 580 (9%)
In shell elements membrane (M) and bending (B) values are indicated separately. The reduction percentage with respect to the values of the 1927 model is
indicated between parentheses.

Table 4
tional loads or thermal gradient are shown in Table 4.
Stress results in Regina Martirum dome with fixed base: maximum
principal tensile (P1) and compressive stresses (P3), in membrane (M) and Comparing the results for gravitational loads with those
bending (B) obtained for the dome in 1927 model and actual model,
P1 (kPa) P2 (kPa) we can check the effect of global structural displacements
in dome stresses. As it is logical stresses grow when the base
M B M B
movements are increased, and reach its minimum value
Gravitational loads 50 240 395 488 when the base is fixed. In Fig. 18 we can see the different
Thermal gradient (+15) 43 2 99 412
vertical displacements and tangential rotation imposed in
the dome in each analysis. With respect to the model with
model. Reduction in stress is general, and especially impor- the fixed base, maximum membrane tensile stresses are two
tant in pillars, arches, tambours, and vaults, with reduc- times bigger in the actual state model and four times in
tions of the maximum values superior to 50% checking 1927 model; whereas decrements in maximum bending
out the efficiency of restoration works. and maximum compressive stresses, located in the upper
In the Regina Martirum dome, and similarly in the rest part of the dome near the connection with the lantern,
of domes, maximum principal membrane tensile stresses are smaller because the origin of these stresses are mostly
decrease about 50%, whereas maximum principal bending the vertical gravitational loads of the dome, which are iden-
stresses and membrane compressive stresses decreases tical in the three cases.
about 20%. In the case of thermal gradient produced by a tempera-
ture difference of ±15 C between the inner and the outer
surface of the dome, results obtained in the local models
5.3. Local models shows that tensile and compressive membrane stresses are
much lower than those obtained for gravitational loads,
For shell dome models with one part of the tambour but bending compressive stresses have similar maximum
(Fig. 12a), the results considering a fixed base and gravita- values located in the central part of the dome.
312 L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314

Fig. 18. Movements in the base of Regina Martirum dome: (a) origin of angles; (b) comparison of vertical displacements Uz (m) and tangential rotations
hT (rad).

Fig. 19. Local model of Regina Martirum dome: (a) equivalent cracking deformation after collapse; (b) variation of maximum principal tensile stress in
the cracked area.

Considering non-linear material model 2, a local analy- Es = E/8, crushing strain eap = 0.004, retention shear factor
sis of Regina Martirum dome modelled with solids ele- for open crack G = 0.01, cohesive parameter c = 200 kPa,
ments was carried out, with the displacements obtained and internal friction angle / = 40. The collapse takes
at the dome base from the global linear material model place at a value of maximum tensile stress rcr = 70 kPa,
imposed as constraints to the dome base in the non-linear lower than the common value of 100 kPa, but considered
local model. sufficiently closed.
To study the collapse limit behaviour of the dome, Fig. 19 shows the equivalent deformations of cracking
instead of increasing the external loads above their actual and the variation of the principal tensile stress in the node
value, a collapse mechanism was found by varying the where the cracking begins, displaying the stress discharge.
material non-linear parameters with current gravitational The loads are linearly applied in 200 steps, and the collapse
loads. Considering the cracking stress as variable and tak- of the dome takes place in step 197. Fig. 20 shows the
ing reasonable common values for the rest of the parame- equivalent elemental plastic strains, obtained in the step
ters: elastic modulus E = 4.5e6 kPa, poisson coefficient before the collapse. The cracking process begins in step
m = 0.1, specific weight c = 1.8 T/m3, discharge modulus 100, and later the plastification process in step 140, concen-
L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314 313

Fig. 20. Local model of Regina Martirum dome: (a) vectorial value of the main cracking deformation; (b) equivalent plastic deformation after collapse.

trating itself in the upper zone of application of the loads on their deterioration degree and their spacing distribu-
coming from spire. For all cases, we consider a cracking tion, in order to determine the security level of the struc-
stress of 75 kPa. ture as well as assuring our assumptions. Also it was
proposed that it is necessary to install some instruments
6. Conclusions to measure movements and fissure openings periodi-
cally, especially in the Regina Martirum dome. Further-
• The previous realization of a detailed geometrical model more, it would be advisable to install instruments, to
allowed us the generation of a complete FEM model determine the thermal loads and their variation.
with parametric meshing. • With respect to numerical simulations in the future
• The study has permitted us to reproduce the historical development, the following improvements are proposed:
pathologies of the basilica numerically, and to check – Consideration of orthotropic behaviour of the
out the efficiency of restoration works done in the past masonry in the linear and non-linear models.
by the architect Teodoro Rı́os between year 1927 and – Use of a reduced rigidity matrix in the contour of the
year 1940. Regina Martirum dome base which replaces the dis-
• The use of linear models as a previous step to possible placements imposed in the non-linear analysis of the
analyses considering more realistic non-linear models, dome.
allows us to comprehend the real structural behaviour, – Application of a non-linear material model with dis-
with less cost and complexity as compared to those of tributed cracking in tension and plasticity to the com-
the non-linear cases. plete model of the temple.
• Numerical models of the present state of the temple
showed that the current structural situation of Regina Acknowledgements
Martirum dome is satisfactory and there is no risk of
pathologies considering that the resistance to tension Special thanks to A. Sánchez, I. Valcarce, J. Cascales
of the brick masonry is superior to 100 kPa. In the rest and P. Loscos, former research assistants of the University
of the temple the structural situation is similar, except in of Coruña, for their contribution to this study.
the southern towers, where the existing compressive
stresses due to its high elevation and the environment References
have deteriorated the brick masonry. They should be
[1] Rı́os T, y Rı́os ST. El Pilar de Zaragoza. CAI ed. 1983. p. 185–212 [in
under observation to control their behaviour. Spanish].
• The obtained results are based on simplified material [2] Romera LE, Hernández S, Reinosa JM. A comprehensive structural
models with theoretical material parameter values study of the Bası́lica of Pilar in Zaragoza (Spain). In: Brebbia CA,
obtained from the extensive existing bibliography. Torpiano A, editors. Structural studies, repairs and maintenance of
Because of the type of construction, its great dimen- heritage architecture IX; 2005. p. 103–13.
[3] Romera LE, Hernández S, Reinosa JM. Análisis del comportamiento
sions, and the expanded temporary extension, it would estructural de la bası́lica del Pilar de Zaragoza. Revista Int de
be useful a future campaign of tests to determine accu- Métodos Numéricos para Cálculo y Diseño en Ingenierı́a
rately the mechanical properties of the masonry based 2005;21(3):139–49 [in Spanish].
314 L.E. Romera et al. / Advances in Engineering Software 39 (2008) 301–314

[4] Hernández S, Romera LE. Computer modelling of the Bası́lica of Oñate E, Lourenço PB, editors. Structural analysis of historical
Pilar in Zaragoza (Spain). In: Brebbia CA, editor. Structural studies, constructions II. Possibilities of numerical and experimental tech-
repairs and maintenance of heritage architecture VIII; 2003. p. 597– niques. CIMNE; 1998. p. 57–92.
606. [11] Molins C. Characterization of the mechanical behaviour of masonry.
[5] Cosmos/m v.2.8. Finite element analysis system: I) user guide, IV) In: Roca P, González JL, Marı́ AR, Oñate E, editors. Structural
advanced modules. Structural Research & Analysis Corporation, CA; analysis of historical constructions. Possibilities of numerical and
2003. experimental techniques. CIMNE; 1997. p. 86–122.
[6] Romera LE, Hernández S, Mosquera A. Análisis estático y dinámico [12] Fanning PJ, Boothby TE. Three-dimensional modelling and full-scale
de estructuras con el programa Cosmos/m v.2.8. La Coruña: Tórculo testing of stone arch bridges. Comput Struct 2001;79:2645–62.
Ediciones; 2003 [in Spanish]. [13] Crisfield MA. Non-linear finite element analysis of solids and
[7] MSC. Marc 2003. Vol. A – Theory and user information, Santa Ana: structures. Advanced topics, vol. II. John Wiley & Sons; 1997.
MSC Software Corporation; 2003. [14] O’Gara M. Modelling and behaviour of masonry walls in fire. In: Bull
[8] Bathe KJ. Finite element procedures. New Jersey: Prentice-Hall, JW, editor. Computational modelling of masonry, brickwork
Inc.; 1996. and blockwork structures. Saxe-Coburg Publications; 2001. p. 143–
[9] Cook RD, Malkus DS, Plesha ME. Concepts and applications of 176.
finite element analysis. John Wiley & Sons; 1989. [15] Griffith MC, Lam NTK, Wilson JL, Doherty K. Experimental
[10] Lourenço PB. Experimental and numerical issues in the modelling of investigation of unreinforced brick masonry walls in flexure. J Struct
the mechanical behaviour of masonry. In: Roca P, González JL, Engrg 2004;130(3):423–32.