Study on the Stability of the Abbey of San Costanzo Al Monte (Cuneo, Italy)

Structural Studies, Repairs and Maintenance of Heritage Architecture IX 135

Study on the stability of the Abbey of San Costanzo al Monte (Cuneo, Italy)

G. Pistone & A. Violante Dipartimento di Ingegneria Strutturale e Geotecnica, Politecnico di Torino, Turin, Italy

Abstract

In connection with the strengthening and restoration project for the functional rehabilitation of the Abbey of San Costanzo al Monte, near Cuneo, in Piedmont, Italy, the authors studied the static behaviour of the structure in order to evaluate its current state of stability and identify the actions needed to bring back the construction to within acceptable safety margins pursuant to the applicable national regulations. This is a highly valuable structure, unique in this area, whose construction is believed to date back to the eighth-ninth century and to have been commissioned by the Langobard sovereigns of the time. In the course of the early centuries of its existence (from the tenth to the fourteenth century), the abbey suffered from the onslaught of hostile peoples that plundered the area, it experienced brief periods of renascence, and finally relapsed into a state of disrepair, which continued uninterruptedly into the seventeenth and the eighteenth century, until, in 1800, the abbey was deconsecrated and fell into a state of utter degradation. Though it experienced a condition of neglect for such a long period of time, during which no maintenance works were performed, the structure of the abbey still retains all the original masonry parts, albeit many of them are in extremely poor conditions. Large portions of the roof have been missing for at least two centuries, and its absence accelerated the deterioration of the material of the building. The damaging process was also heightened by the geographic position of the abbey, in a mountain area where freezing cold winter weather surely had adverse effects on the materials. Moreover, it should be noted that in the latest seismic charts produced in Italy, the site on which the abbey rises is classified as medium seismic (Alpine zone). In spite of all these adverse circumstances, it is still possible to appreciate the remarkable set-up of the building, which enabled it to withstand a thousand year long process of disruption. Keywords: stability, structure, masonry.

1 Introduction

The conditions of the abbey of S. Costanzo al Monte [1], in the commune di Villar S. Costanzo (CN) at the foot of the Alps, were investigated by means of a number of numerical models designed to assess the stability of the building before undertaking strengthening and restoration works to bring it in line with the margins of safety specified by the applicable standards. In view of the significance of the building, safety conditions must be achieved through sparing and non-obtrusive interventions, preferably reversible, in keeping with present-day restoration criteria. The analysis was developed in stages, and namely: a detailed examination to determine the conditions of the original structure and the subsequent additions to it, in preparation for the static analyses to be performed was followed by the acquisition of data on damages; finally, a numerical model of the structural geometry of the Abbey was produced. The results of these studies are described below, followed by considerations on the stability of the building in its present-day conditions.

2 Overall survey and damage assessment

The church is organised in two levels, save for the part immediately behind the main entrance, which is full height. Both the lower and upper storeys have a nave and two side aisles divided into a front and a rear portion. The lower storey, partly underground on the side adjacent to the mountain slope, has a front portion divided into a central nave culminating in three stone cross vaults and side aisles topped by barrel vaults. In the rear portion, the nave and aisles are all topped by barrel vaults, the ones furthest back covering the apses area. The overall ground plan, as described above, is partitioned into various environments by non load-bearing walls of different sizes. The upper storey, much higher than the crypt level, has a front part topped by a wooden ceiling, consisting of trusses in the nave area; the nave and the side aisles are divided by masonry partitions, opened at floor level on either side by three large, slightly pointed arches resting on short columns with stone capitals. The rear section of the side aisles, whose ends contain the apses, is topped by three cross vaults; the same pattern is repeated in the nave, whose central span is topped by a segmented dome resting on an octagonal drum linked to the impost quadrilateral by pendentives, all such elements being made of facing stones. The flooring is made of stone slabs. (fig. 1) The exterior walls, adjoining the sidewalls of the aisles, are part of adjacent buildings erected at a later stage. In particular, the walls in contact with the abbey wall facing the valley perform a key role, protecting the wall against settling and inflection phenomena. This favourable effect is even more pronounced due to the presence of an effective connection between the masonry walls, as in this particular case. The verification of connecting elements was extended to all the walls of the upper (church) level set at right angles to one another.

The data acquisition stage was concluded by determining the deviations from the vertical, especially in the piers and the main walls of the church. As can be clearly seen, even from a simple visual examination, the north and south longitudinal walls are slanted outwards. Internally, the loss of verticality is even more evident in the piers supporting the heavy outer structure of the dome; their tops, in fact, are appreciably displaced towards the exterior, transversely to the nave, as borne out by the opening of the arches linking the columns.

Figure 1: Plan and sections of the abbey.

A decisive aspect in the definition of a model that will faithful reproduce the behaviour of a building is the acquisition of up-to-date information on cracking conditions: this provides the basis to account for the displacements that have already occurred in a building, assess their severity and possibly interpret their evolution. The main types of damage observed were as follows: at crypt level, the vaults in the proximity of the apse area, especially the side vaults, appeared extensively damaged, with many cracks, including sizeable ones, cutting across them and mostly running parallel to the main axis of the church. Presumably, they were caused by the thrust of the vaulted elements, not sufficiently absorbed by the supporting parts. At the higher level, the vaults display lesions that propagate into the side and partition walls; these can be ascribed not solely to the thrust of the vaults, but also to seismic events and other causes, such as the settling of the foundations

WIT Transactions on The Built Environment, Vol 83, © 2005 WIT Press www.witpress.com, ISSN 1743-3509 (on-line) 138 Structural Studies, Repairs and Maintenance of Heritage Architecture IX brought about by landslides that struck the mountain. The vaults also display forms of deterioration due to the infiltration of rainwater, originated by the bad conditions of the roofing. As for the exterior of the building, both sidewalls, north and south, display some evident cracks running horizontally atop the side openings and the pilasters, in the lower part of the outer structure of the dome. Such lesions are not visible from inside the church and are surely (fig 2).

Figure 2: Cracks due to the compression and bending forces.

Finally, the facade, around the main entrance, is subject to an outward tilting phenomenon as borne out by several cracks, running along nearly vertical paths, both in the facade and in the upper portion of the partitions separating the nave and the aisles (fig 3).

Figure 3: Cracks on facade and nearly vertical paths

The material used for the construction of the abbey consists of different types of masonry. Relatively compact parts, made up of regular, carved stone blocks, bound together with lime mortar, are observed in the sections under the heaviest loads and, in particular, in the apse vaults (upper part), the dome and the vault reinforcing arches. A less valuable material was used in the remaining zones, mostly consisting of roughly squared stones or pebbles bonded with lime mortar; many voids can be observed in the masonry structure in such zones. As mentioned above, in many areas the material is badly deteriorated, both on account of its long life and on account of the state of degradation that has characterised the building for centuries.

3 Numerical models

The numerical model was defined with the aim of evaluating the current conditions of the abbey and its behaviour both under its dead weight and under external actions in keeping with the applicable Italian regulations. The special features and the complexity of the masonry structure required the definition of a numerical model reflecting the spatial behaviour of the structure. The structure of the building was reproduced by means of solid elements whereby all masonry parts were represented initially as having the same mechanical properties; the model was loaded first with the dead weight of the structure, this being the most adverse action in a building with a bearing structure made of masonry. The mechanical properties assigned to the masonry have been derived from earlier experience as well as from the literature on masonry structures of similar type, from the same period and the same geographical area. The values used were: Elastic module E = 1500 MPa Tangential elastic modulus G = 652 MPa Poisson’s coefficient 1/m = 0.15 Specific weight γ = 23 KN/m3 The entire model is made up of 33,000 elements and 62,000 nodes generating ca 180,000 equations (computation code COSMOS/M). The analysis of the monumental complex was developed starting from a situation that reflected the contingent situation: then, based on this initial setup, several hypotheses regarding other possible situations were developed. With reference to the results obtained from the individual models we can make the following considerations. a. The first model produced reproduces the entire complex plus the walls of the adjoining constructions. From the data obtained, it can be seen that the action of dead weight gives rise to stress gradients steadily increasing towards the bottom; from the sectional view we find that compressive stresses are highest in the piers of the nave, both at the upper and at the lower level. The values observed, ranging from 0.5 to 0.6 MPa, with peaks of 0.9 MPa at the foundations level, can be rated as still compatible with the actual situation of the masonry in serviceability conditions, for which we

can estimate a failure stress not lower than ca 2 MPa in the gross section (face walls + filling with irregular material). It should also be noted that, in a real structure, the stress peaks observed in the model may be attenuated by the plastic properties of the material. A greater cause for concern is the tensile stress level (0.3 MPa) observed externally, in the pilasters next to the openings in the dome and in the north and south walls (sides facing the mountain and the valley). Such stress values are not compatible with the masonry. It should be underscored that this observation finds confirmation in the real structure, where it is evidenced by clearly discernible cracks cutting across the wall blocks (see § Overall survey and damage assessment). The model also revealed another phenomenon, which is typical of this type of construction, consisting of a marked bulging/outward displacement of the main facade. This phenomenon is due to the thrust generated by the double series of longitudinal inner arches, originally loaded with the enormous weight of the outer dome structure, and, along their development, also by the dead weight of the building: whereas at the far end this thrust is effectively countered by the presence of the massive and rigid apse, on the facade side there is really nothing that can effectively oppose it. This interpretation is corroborated by earlier studies on similar buildings described in the literature. The assumption of this phenomenon being due to the tremendous weight of the outer structure of the dome and its drum is further confirmed by the deformation diagram of the longitudinal section, where all the piers, that begin at the dome and develop along the nave, are slanted in the same direction as the facade (fig. 4).

Figure 4: The model shows the displacement of the facade and tensile stress in the pilasters.

b. The second item checked was the contribution to stability afforded by the adjoining construction; the checking process was performed by simply removing from the previous model (a) the elements reproducing the walls that did not belong to the church.

From a comparison between the results obtained from model (a) and those obtained from model (b), we find that the contribution of these elements is not decisive, and yet it is valuable. Such conclusion can be accounted for by the fact that the adjoining constructions are markedly lower that the church walls they are in contact with, leaving the latter free to bend outwards over a sizeable portion of their surfaces. c. Another aspect worthy of note is the influence of the fact that the abbey was built on a slope. Accordingly, the results were compared with those obtained for a building erected on level ground and it was ascertained that the slant of the land did not have substantial effects on the stability of the church. After the first cycle of analysis, designed to identify and assess the overall situation, a new cycle was undertaken to explore some of the critical aspects that had been revealed by the initial model (a). In particular, the analysis focused on the behaviour of the outer structure of the dome and the impact of this element on the entire abbey complex. d. The new set of analyses was conducted by means of model (b), which is on the conservative side compared to model (a). In particular, some elements adopted in model (b), referred to as contact elements, have the peculiarity of transferring only the compressive forces (not the tensile forces); such elements were placed at the points where the model revealed localised tensile stresses at the connection between the cross vaulted system of the side aisles (at church level) and the north and south walls. Such contact elements were added solely to the upper part of the arch-vault system, without undermining the effectiveness of the support (fig.5).

Figure 5: Contact elements placed where the model revealed localised tensile stresses at the connection between the cross vaulted and the wall.

The aim was to eliminate the effects of the bending force that is generated in the numerical model by the restraint between the zone above the arch and the vertical wall, which do not exist in the real structure, in view of the inability of the masonry to resist tensile forces; this bending

force leads to an increase in the tensile stresses in the pilasters already described when discussing model (a). This change entailed a 20% reduction in the tensile stress in the outer wall, now down to ca 0.18 MPa. The non-negligible proportion of residual tensile stress, as confirmed by the survey, is actually due to the (non counterbalanced) thrust of the lateral vaulted system, which, in its turn, receives the thrust arising from the weight of the dome and its drum. e. Having ascertained that the tensile stresses in the pilasters of the dome walls are mostly due to the thrust of the vaulted system, it was deemed necessary to determine in which proportions this should be ascribed to the action of the arches underlying the dome as opposed to the entire system of vaults surrounding the dome. To this end, 4 non-extensible braces were inserted at the imposts of the dome arches. No appreciable reductions in the tensile stresses in the outer pilasters were observed, yet, this attempt showed that the concentrated load in the central zone gave rise to load-effects having repercussions all the way to the outer enclosure of the building. This observation should be kept in mind, especially in the case of seismic load-effects, directly proportional to the masses involved; in such cases, the weak effect observed in the purely elastic field can be amplified to an extent that is far from proportional and, broadly speaking, can be multiplied according to ratio between the mass of the dome and the mass of the side vaults (approx. 1 : 5 - 6)

Figure 6: Columns with the elastic modulus were increased. f. One further aspect that might yield misleading indications in the interpretation of the behaviour of the building, in a purely elastic type model such as the one employed, is the dissimilar elastic shortening taking place in the dome columns as opposed to those of the nave, due to the enormous difference in the loads affecting such parts: these differential displacements give rise to horizontal thrusts on the potential thrust-carrying elements. This aspect has been analysed in an earlier study [2].

Accordingly, in order to determine the thrust apart from the elastic effects brought about by the shortening of the columns, the mechanical properties of the four central columns supporting the dome were enhanced so as to make such elements stiffer and hence reduce elastic shortening: in particular, the elastic modulus was increased by 103 (fig. 6). The analysis confirmed that tensile stresses in the zone in tension have been slightly reduced (0.08 MPa), but are still present, which is only logical, as otherwise this would contradict the real situation. Based on this result, it can be concluded that tensile stresses are due almost exclusively to the non-counterbalanced thrust of the vaulted system of the roof, especially in the dome zone.

4 Final considerations on the stability of the structure

The analyses performed revealed a number of significant aspects affecting the stability of the church. First of all, the abbey suffers from a generalised state of disrepair, as attested by the numerous lesions observed between the vertical elements, and in the vaulted systems at both levels. The models revealed extensive tensile stress zones at the systems of arches present in the building and at the openings in the vertical walls. The most important aspect determined from the analyses is the presence of high stress values in the pilasters around the outer structure of the dome. As discussed above, this can be ascribed directly to the vaults of the side aisles and, indirectly, to the considerable impact of the huge body of the outer dome structure on the entire building. This structure is supported by a system of arches and transfers its dead weight to the adjoining elements via a series of thrusts. This mechanism propagates in all possible directions and stops at elements having sufficient stiffness to absorb the phenomenon. The preferential directions of propagation coincide with the main axes of the church. All the models produced show that such thrusts are effectively carried by the apse zone, which, on account of its shape and thickness, is very rigid. The outer sidewalls and the facade behave differently, as they do not possess the degree of stiffness required to carry the forces they are subjected to. The structural inadequacy of such parts causes a fragmentation of the masonry section in the sidewalls and a detachment for the rest of the construction at various points of the facade. In conclusion, it should be underscored that the damages observed in the building can be ascribed for the most part to the conditions of disrepair in which the building was left for decades, and that they can be remedied through extraordinary maintenance works and none too onerous strengthening interventions. On the other hand, the key factor affecting the behaviour of the building is the presence of the huge suspended mass of the outer dome structure having an adverse impact on the surrounding structures, and especially on the vaulted systems, which, in their turn, have negative effects, not having been

WIT Transactions on The Built Environment, Vol 83, © 2005 WIT Press www.witpress.com, ISSN 1743-3509 (on-line) 144 Structural Studies, Repairs and Maintenance of Heritage Architecture IX equipped from the start with appropriate keystones or sufficient buttresses. The problems arising from all this make it hard to define strengthening works that will not alter the original conception of the building and its formal qualities. In other words, the fundamental problem the designers are confronted with – since it is impossible to integrate structural elements, such as the metal keys and masonry buttresses, that were deficient from the start – is to devise an effective system to carry the thrusts which will not alter the aesthetics of the building and at the same time will ensure the required degree of safety.

References

[1] Rovera Giovanni, L’abbazia benedettina di Villar S. Costanzo (712 – 1803) nella storia e nell’arte, Bertello, Borgo S. Dalmazzo Cuneo, 1981. [2] Pistone Giuseppe, Computer modelling of the behaviour of the San Giovanni church in Farigliano, Proceedings of the Tenth International Brick/block Masonry Conference, Alberta, Canada, 5-7 July, 1994.