Experimental Study on the Damaged Pillars of the Noto Cathedral

Experimental study on the damaged pillars of the Noto Cathedral

L. ~inda',A. saki', R. De ~enedictis*& S. ~rin~ali~ I DIS - Dept. of Structural Engineering, Politecnico di Milano, Italy 2 Designer

Abstract

The paper describes the on site and laboratory investigation carried out on the remaining pillars of the Noto Cathedral, in order to verify their state of conservation in view of the rebuild of the church, and how the designers had to take the decision of demolishing them.

1 Introduction

In December 1990 an earthquake hit the Eastern part of Sicily damaging old and contemporary buildings in different towns. Noto, known as the "Baroque city" was among them and several of its most beautihl buildings were seriously hurt. Also the Church of St. Nicolo, the Cathedral had damages to the vault, the lateral domes and to the pillars, apparently no more than other buildings. Provisional structures and scaffoldings were set up to sustain the damaged parts waiting for the repair and strengthening intervention. The partial sudden collapse occurred on March 13, 1996 fortunately without any casualty, left the Noto community astonished by the loss of one of its most famous buildings.

The church had been built in different phases suffering several casualties from 1764 over a previous smaller church opened in 1703 to the public and demolished in 1769170 as the new Cathedral was growing. The Cathedral was opened in 1776. In 1780 the dome collapsed and the church was reopened in l8 18. In 1848 the dome collapsed again under an earthquake and then it was rebuilt and the church reopened again in 1862 but the dome was not completely finished until 1872. In 1950 the Cathedral was restored with new renderings and paintings and the timber roof substituted with a concrete structure; the work continued until 1959.

90 Strrcctrrral Srudir~,Rrpam atrd Marnrorancr of Hrriragr Aldzirrcruw V111

The losses caused by the collapse were the following: pillars of 4 the right part of the central nave and one of the 4 pillars sustaining the main

dome and the transept, the complete roof and vault of the Figure 1 : Plan of the remains and of the tested elements central nave. three quarter of the drum and dome with the lantern, the roof and vault of the right part of the transept and part of the small domes of the right nave (Fig. 1). The extensive experimental and numerical investigation carried out after the removal of the ruins by a team of experts together with the designers [l], [2] clearly showed that the collapse started from one of the pillars, due to the damaged situation they were bearing before the earthquake. Taking into account the clear weakness of the collapsed pillars, the designers asked for a further careful investigation on the remaining pillars of the central nave which also where damaged by the collapse. The first idea was to repair and preserve these pillars during the reconstruction of the Cathedral. The paper will describe the on site and laboratory investigation carried out and how the designers had to take the decision of demolishing them. Other cases of similar damages occurred in Sicily will also be described showing the importance of investigation in order to prevent future failures.

2 Investigation on the remaining parts of the collapsed pillars:

Layout of the section and material characteristics

After the study of the collapse mechanism, carried out also on site during the removal of the ruins [3], the attention of the consultants was focused on the careful study of the peculiar features of the collapsed pillars.

The removal by layers of the components of the collapsed pillars allowed to understand the poor technique of construction used for them. Layers of large round river stones with thick mortar joints, where the mortar appeared very weak and dusty, were found in the core of the structure, surrounded by an external leaf made with regular blocks of more compact limestone at the base of the pillars (Fig. 2). Since only the base had remained after the collapse and the symmetric pillars were still covered by plaster, the hypothesis was made at first that this limestone had been used for the external part of the whole pillars. This material, compact but not very strong, came from sedimentary carbonatic depositions which can be found in the area and are still used as quarries for the building industry [4]. Inside the rubble filling also pieces of a material full of voids were found which was called travertine; this material is of the same nature of the limestone, but deposited in the presence of turbulent

waters and it is rich in voids of various shape and dimensions which previously contained vegetarian and organic parts

later on dissolved. The height of the blocks varies from 25 to 30 cm and the thickness, small compared to the pillar dimensions, is

ranging from 25 (stretcher) to 40 cm (header). No really effective connection was realised between the external leaf and the core. The stones of the pillar strips supporting the arches, vault and domes

Figure 2: A collapsed pillar have no connection either to the internal masonry or to the other parts of the external leaf (Figs. 3 and 4). The inner part of the pillars represents

the 55% of the entire section, while in the pillars sustaining the dome it is the 58%. This part is a rubble masonry made with irregular stones but it could be seen from

the ruins that approximately up to the half of the total height was made with large round river pebbles. The courses of these Figure 3: Horizontal pillar section stones are rather irregular without any transversal connection or small stones to fill the -- voids and with thick mortar joints. Nevertheless every two courses of the external leaf (about 50 cm) a course made with small stones and mortar was inserted in order to obtain a certain horizontality (Figs. 2 and 4). Scaffolding holes were left everywhere, some crossing the whole section. The mortar appeared to be very weak made with lime and a high fraction of very small calcareous 1 aggregates. Also the bond between the mortar and the 'r stones was very weak; in fact it was possible to remove stones and pebbles from the interior of the i pillars without any difficulty and with the stones being completely clean. This poor technique of construction and the use of the weak limestone (actually called "Noto stone") typical in the Noto region, was probably the cause of the damages to the pillars of the Cathedral, even if a clear crack pattern was reported to have appeared only after the 1990 earthquake. The walls were built -L.- similarly; nevertheless, the internal part was made Figure 4:Reconstruction of a pillar section. with smaller sharp stones alternated with a slightly

92 Strrcctrrral Srudir~,Rrpam atrd Marnrorancr of Hrriragr Aldzirrcruw V111

stronger mortar, in some way a better masonry. Some stones were sampled from pillars and walls and mortar samples were

taken from horizontal, vertical joints and from the interior of the masonry (Fig. 2). The samples were sent to the DIS Laboratory in Milan and tested in order to

find the material characteristics [l], [5]. The investigation (Fig.-. 5) has shown Figure 5: Detail of the foundation that the foundations df pillars and walls were sufficiently well constructed; rubble walls but with enough load carrying capacity for the weight of the above structures. The soil was a sort of natural compact silt and clay thick layer from where also the aggregates of the mortars were taken. Up to this point of the investigation even if the weakness of the material used seemed to be the cause of the high damage suffered from the earthquake, but it was not still clear why the pillars reached the collapse.

3 Survey of the remaining pillars

The left pillars, still covered with a thick plaster, seemed to have suffered minor damages; only small and diffused vertical Figure 6:Large crack in a pillar cracks were present on the plaster. and example of a crack filled with mortar. Nevertheless the doubt that the damage could be deeper inside and perhaps even present before the 1990 earthquake, suggested to cany out on these pillars to a more accurate survey. As the plaster made at the end of the works done in the fifties was partially removed, a series of vertical large cracks were found, some of which filled with the gypsum mortar used for the plaster (Fig. 6). This finding gave the authors the first evidence that the damage was already present in the fifties. The pre-existing crack pattern was clearly a damage from compressive stresses, a long range damage dating probably even long time before the fifties. The lesson after the collapse of the Civic Tower in Pavia and the subsequent research taught the authors that the damage would probably have progressed even without the earthquake, which only accelerated the collapse. After the recognition of the damages, the removal of the plaster from all the pillars was planned in order to survey the crack pattern. Fig. 7 shows a reconstruction of the crack pattern of the pillar before removing the plaster from all .the pillars. As it possible to observe, the cracks are diffused and interesting the whole prospect, with a concentration in the corners.

4 Laboratory testing

On the materials sampled on site physical, chemical, petrographic- mineralogical and mechanical tests were carried out in Milan at the DIS

Laboratory. The aim was to characterise the materials of a typical (CC') transversal section (Fig. 8) of the Cathedral [5].

The chemical and mineralogical analyses were canied out following a procedure set up in [6] on the mortars sampled from all the pillars and walls at different height. The mortars contain a

high percentage of CaC03 showing that they are based on hydrated lime with a slightly high content of soluble silica but Figure 7:Prospect of pillar PIE and with fine aggregate size distribution. survey of the crack pattern.

Table 1 gives the results on the mortar of pillar PIA together with the crack repair composition. Fig. 9 gives an example of grain size distribution of aggregates. The soil was also examined and it appears of being composed by more than the 87% of

Table 1. Chemical analysis and bulk density of pillar P1 A mortars. PILLAR PlLLAR

PIA PI A Figure 8: Transversal Section CC' 13CNMP'3 Filling of cracks calcium carbonate, by 8% of different

silicates and for the remaining 5% by alcali, allumine, iron, gypsum, etc. The grain size distribution of the soil shows that it is composed for the 8% by clay, the 72% by silt and the 20% by sand, a

very fine material. K20 0.98 1.06 Some compressive tests were carried so3 0.64 39.02 out on cylindrical samples of the two Loss on ienition 4 l 1 1 lA 97 stones, limestone and travertine; their

CO? 40.50 1 1 .92 texture is shown in Fig. 10. The tests Ins. Res. 4.20 1.73 Soluble Silica. 0.66 0.30 performed on the calcarenite show that Cl 0.025 0.028 its strength when saturated at constant Bulk density l ,3 13 kg/m3 1,489 kg/m3 mass (1 1.56 ~/mm')drops dramatically

94 Strrcctrrral Srudir~,Rrpam atrd Marnrorancr of Hrriragr Aldzirrcruw V111

on respect to the strength measured when dried at constant mass (17.98 ~lrnm~).The compressive strength of the

travertine is very low and can vary from 4 to 6 or more NI&. In order to know the response of the two stones to the elastic

waves, the ultrasonic velocity was determined by transmission -6.0 -4.0 -2.0 0.0 2.0 4.0 6.0 strain [@mm] on stone blocks. The two Figure l l:Double flat jack carried out on the materials show very different behaviour. In fact, in the case of external and internal part of PIA the limestone (calcarenite) the values are almost constant, between 2912 dsand 3157 dswith an average of 3068 ds. The values of the travertine are more scattered, with a measured velocity between 1325 mls and 3548 ds, and an average of 1823 ds. The scattering of the data is due to the presence of large voids, randomly distributed in the material, and confirms the results of the mechanical tests. Injectability tests proposed in [7] were carried out in laboratory on materials sampled from the internal part of the pillars and walls, and from the collapsed pillars of the Cathedral [4]. Grout 100 injection was controlled directly on 90 site as well [g]. 80 Finally it was decided that :: injection cannot help improving the g ,, behaviour of the damaged pillars. 40 30 5 NDT Evaluation 20 CNMABSlA 10 0 5.1 Flat-jack tests 0.01 0.10 I .OO 1000 10000 (mm) A single flat-jack test was carried Figure 9: Grain size distribution out on the &lar PIE in order to know the state of stress in it simply due to the dead load of the pillar itself and a value of 0.85 ~lmm' was found at a height of 3.00m.

Taking into account the missed weight of arches, vaults and dome in the collapse, it is easy to make the hypothesis that the pillars must have been under a non negligible state of stress. Double flat-jack tests were also Figure 10: Calcarenite (Stone of Noto) and carried out on pillars PIE and PIA travertine.

and on the external walls of the Cathedral. A double flat jack test was also carried out in the inner part of pillar PIA in order to check the behaviour of its weakest part. Figure l I shows the difference between the external leaf (CNJID) and the core (CNJ2D) which had a much higher deformability and lower strength.

5.2 Application of sonic pulse velocity test to pillars

As a confirmation of the state of damage, and also a calibration of the procedure, sonic pulse velocity tests were carried out on the remains of the collapsed pillars, as well. It is well known that ultrasonic frequencies can not be used on rubble walls due to the high attenuation caused by joints, voids and hornogeneities.

Nevertheless, being travertine and calcarenite so different the ultrasonic velocities measured in laboratory on single blocks, were very useful.

Fig. 12 localises as an example the test position on the pillar PIE and in the correspondent collapsed pillar called PE. Measurements were taken at different Figure 12: Geometry of the pillars PI E and PE and heights. It was localisation of the sonic tests. impossible to position equal levels for all the pillars due to the presence of safety scaffolding. Nevertheless it was clear that the material of the external blocks was changing from the base (calcarenite) to the top of the pillars (travertine). The measurements were also carried out on some parts of the external walls as a comparison. Fig. 13 shows the average values found for the left pillars and for the external pillar and wall called MlB. Low velocity values were systematically recorded in all the tested pillars of the Cathedral from about 1.00-1.50 m on, that is above the limestone base. The values reported on Fig. 14 are average values over the measurements carried out in the two orthogonal directions. The pillar PlB shows the lowest values of the sonic velocities recorded at each level compared to the other P l i pillars. The pillar state is in fact characterised by a very serious damage, as described by the crack pattern of Fig. 7.

6 Design decisions

The accurate and detailed survey carried out by a multidisciplinary team was very helpful for the designers who had to take many difficult decisions. The crack pattern survey revealed large vertical crack already present and filled with

96 Strrcctrrral Srudic~,Repam atrd Marnrorancc of Heritage Aldzirccruw V111

gypsum mortars in the sixties when the timber roof of the Cathedral was

500 replaced by a concrete roof. These

damages indicate, together with the laboratory results that the material 6... 300 c3 used for the construction was very Q c weak and damaged by long term effects; the collapse perhaps could go have taken place in a longer time without the earthquake. Fig. 15 shows as a confirmation 0 400 800 1200 1600 the state of damage of one of the sm~cveloclty [rnbec] pillars as observed after the complete Figure 13: Vertical distribution of the removal of the plaster. sonic velocity measured on The left hand pillars could not be preserved due to the high state of the pillars and the walls. , damage caused by the weak technique of construction and the weak materials; therefore it was decide to demolished them and rebuilt them, together with the collapsed ones using

better materials and technique of construction, i.e. hydraulic mortars obtained with hydrated lime and pozzolana, calcarenite stones avoiding

travertine and good connections between the external leaf of the pillars Fig yre 14: Sonic velocity distribution in and the core. the P l A section at 90 cm. The soil and foundation seem to be acceptable everywhere; the only exception can be made fox the foundation of the central nave pillars, which can eventually follow a new conception.

7 The dismantling of the

remaining pillars

The substitution of the left pillars takes place in alternate order demolishing one pillar at a time and reconstructing it. Before this operation, the vaults of the left nave are supported by a stiff steel structure (Fig. 16). Figure 15: State of damage observed after

The dismantling of every single the removal of the plaster

pillar is carried out In successive steps, demohshing stone by stone every single course.

The use of the local travertine is confirmed, as well as the serious state of damaged and the lack

of connection between Figure 16: Steel structure supporting the vaults. the stone leaf and the core. The stones showed passing through cracks or deep cracks; when lifted by the workers, the blocks often broke, revealing the internal large voids.

Furthermore the fractures in the external surface could be observed also in the internal rubble, even if less readable for the high inhomogeneity of the masonry. They in fact can follow the boundary between the mortar and the but also go through every single stone (Fig. 17). Figs. 18 a,b,c show as an example of the sequence of demolition of one course of pillar PIB. It can be seen that the vertical and horizontal joints are so week that the dismantling can be carried out by hands. Furthermore the inside of the stones which are practically all broken or fissured is full of voids and very Figure 17: of fractures weak. This explains why the pillar did not bear the state of stress for n long time. Creep phenomena have certainly developed during thc lif'c of the pillars. lowering their strength.

8 Conclusions

After the lesson learned Figure 18: Example of a sequence of demolition of from the collapse of the one course of pillar P I B. Noto Cathedral, the experience showed that other buildings could have been under the same damage. In fact in Noto itself the Church of the SS. Crocefisso was found under a very high state of damage having all the pillars with a dangerous vertical crack pattern. After that the church of SS. Annunziata in Ispica (Eastern Sicily) was

98 Strrcctrrral Srudir~,Rrpam atrd Marnrorancr of Hrriragr Aldzirrcruw V111

found in similar conditions, the pillars of the Syracuse Cathedral also show similar damages, as St. Nicolb L'Arena in Catania. When observing similar situations a first crack pattern survey can be useful to interpret the damage cause; then a monitoring of the structure must be carried out if the situation is not evolving very quickly. Also NDT as sonic or radar tests and tomography can also help in finding the state of damage of the structure. Laboratory tests are also needed to characterjse the materials, masonry and masonry components.

Acknowledgement

Authors wish to thank for their contribution M. Antico, M. Cucchi, M. Iscandri C. Tedeschi and C. Tiraboschi.

References

[l] Binda, L., Baronio, G., Gavarini, C., De Benedictis, R. & Tringali, S., Investigation on materials and structures for the reconstruction of the partially collapsed Cathedral of Noto (Sicily), Proc. STREMAH 99, Dresden,

Germany, pp. 323-332, 1999. [2] Tringali, S., De Benedictis, R., La Rosa, R., Russo, C., Bramante, A., Gavarini, C., Valente, G., Ceradini, V., Tocci, C., Tobriner, S., Maugeri, M., Binda, L. & Baronio G., The reconstruction of the Cathedral of Noto, Proc. lnt. Synzp. on Earthquake Resistant Engineering Structures (ERES II),

Catania, pp. 499-5 10, 1999. [3]De Benedictis, R., Tringali, S., Gavarini, C., Binda, L. & Baronio G., Methodology applied to the removal of the ruins and to the survey of the remains after the collapse of the Noto Cathedral in Sicily, Proc. STREMAH 99, Dresden, Germany, pp. 529-538, 1999.

[4] Binda, L., Baronio, G., Tiraboschi, C. & Tedeschi, C., Experimental Research for the Choice of Adequate Materials for the Reconstruction of the Cathedral of Noto, Construction Building Materials, to appear. [5] Baronio, G., Binda, L., Tedeschi, C. & Tiraboschi C., Characterization of the

Materials Used in the Construction of the Noto Cathedral, Construction Building Materials, to appear. [6] Baronio, G. & Binda, L., Experimental approach to a procedure for the investigation of historic mortars, Proc. 9Ih Int. BricWBlock Masonry Con$, Berlin, pp. 1397-1405, 1991.

[7] Binda, L., Modena, C. & Baronio, G., Strengthening of masonries by injection technique, Proc. 6'h NaMC, Vol. I, Philadelphia, pp. 1-14, 1993. [g] Binda, L., Saisi, A. & Tiraboschi, C., Application of sonic tests to the diagnosis of damage and repaired structures, NDT&E Int. Journal, 34(2), pp.

123-138,2001.