Abstract
The evaluation of mechanical properties for the analysis of existing masonry buildings still represents a current problem, especially when retrofit interventions have to be planned on the historical-artistic heritage. Due to the huge cost of laboratory destructive tests, the need of knowledge is today often achieved through the combined use of investigations carried out with different methodologies. The actual tendency should be to increasingly privilege, to the point of almost exclusive use, non-destructive investigations, which can be easily and widely used on buildings, replacing destructive or partially destructive tests. At present, however, relationships among mechanical characteristics (modulus of elasticity, strength) to be reliably estimated using on site measurements are not available. This is especially true for natural stone masonry like those widely diffused in Southern Italy. In this context, an extensive set of experimental data obtained on yellow tuff masonry from the Campania Region is presented and discussed in this paper.
Introduction
The Italian building heritage and particularly that in Southern Italy has been subjected in the last years to an increasing interest in order to assure its conservation. Differently from monumental buildings and contexts, protected by restoration laws and codes, the diffusion and vulnerability of traditional building heritage has been shown during the recent seismic events.
The study of the ancient monuments is one of the higher moments in the learning process of the building’s practice. A critical reading is useful to recognize the modern stratifications of the building’s texture.1,2 In the case of Naples, a city that owns one of the greatest and most valuable Italian historical centers, the effects produced by the earthquake of 1980 and the absence of conservation on large parts of the buildings are nowadays visible. 3 The Neapolitan case is a particular one: diffused knowledge on architectural, urbanistic and environmental characters has been often associated with a low knowledge of materials mechanical properties. 4
In some cases many of these buildings have been subjected to heavy interventions, such as concrete injections or applications of concrete slabs, independently from their historical artistic interest.5,6 Different methods of numerical analysis are now available for their preservation and safety, but the validity of the results, excluding the cases in which sharp constitutive assumptions are considered, 7 strongly depends on the values used for the mechanical properties of the masonry.8,9 The determination of the latter is still a weak point in the entire procedure for assessing the static and seismic safety of buildings, despite the strong development in recent decades of techniques for their on site determination.10 –13 In addition, there is also a partial lack of values in the literature, generally depending on the strong heterogeneity of masonry structures in terms of both materials and construction techniques.
Knowledge of the mechanical characteristics is essential to assess the vulnerability to both ordinary gravitational actions and seismic ones14,15; the characteristics to be considered are not only reduced to the maximum resistance to compression and the corresponding deformation, but extend to the evaluation of the bearing capacity of the masonry in the post-elastic phase, which can be expressed as the deformation corresponding to an aliquot of the maximum capacity. 14 This is especially true in case of structures with peculiar behavior that can be analyzed by means of specific models, such as towers, arches and vaults.7,16,17 Despite their complex realization, on-site tests continue to have a fundamental role in estimating the mechanical properties of masonry,18,19 also in view of the problems that characterize the execution of laboratory tests on samples taken on site or on new samples. 20 Sampling of existing buildings is, in fact, often impractical because of the disturbance caused to the specimens during sampling and transport to the laboratory, and, in most cases, impossible when the building has historical importance that prevents any form of damage, even for diagnostic purposes. The production of samples from scratch in the laboratory is unlikely to reproduce the same characteristics of the historical walls on site. This is particularly relevant for natural stone walls, 21 characterized by stone blocks of multiple materials, by different ways of working the blocks, by mortars with different strengths and often altered by time. 22
Most of the recent regulatory codes that explicitly deal with existing structures,23 –25 rightly make the type of structural analysis and mechanical parameters of calculation depend on the level of knowledge gained for the masonry structures, or on the number and type of experimental research performed. Moreover, the analysis of materials is fundamental to prepare the most suitable intervention techniques. 26
Obviously, the most important parameters are elastic modulus and strength (mainly compression and shear, although many calculation codes make greater use of tensile). As is well known, it is often difficult to estimate the compressive strength of masonry by combining the strength of mortar and bricks. 27
In order to get to know the mechanical characteristics in the best possible way, it is necessary to carry out an in-depth analysis of the different types and, therefore, the material of the masonry textures of the existing heritage of Southern Italy and, in particular, to focus on those of the Neapolitan heritage.
This paper presents an experimental campaign on tuff masonry, partly performed in laboratory and partly performed in situ. The results of both the test sets are discussed and compared, in order to achieve better knowledge about the behavior of a masonry heritage so largely diffused in Southern Italy.
Masonry chronotypes
The research conducted on Neapolitan yellow tuff masonry has identified three fundamental constructive chronotypes: the first one, named cantieri masonry (realized mainly in the Modern and Contemporary Age), the second one, named filari di bozzette masonry (realized in the XVIII century) and a third one named filari di blocchetti masonry (corresponding to buildings realized in XIX and the first half of the XX century).
During the XVI and the XVII century, the pieces of tuff were supplied on site according to size conditioned by the extraction system. According to the terminology suggested by the royal pragmas, among the pieces of current use, they normally produced the spaccatelle, that is equal to sort of half spaccata (cm 35 × 25 × 13 around), and the so-called pietre rustiche, quadrangular irregular stones. It must be remembered the asche, rather discards of workmanship used for filling the voids and the spaccatoni, long and low thickness elements. 28
In the first decades of XVI century, the masonries manifested discontinuity in comparison to those of the preceding century and the Aragonese walls are one of the greatest examples of civil work of XV century. The building texture is composed by a nucleus with yellow tuff pieces and externally covers by piperno slabs (with a height of 50 cm) that determined the height of the cantieri, including many spaccatoni and blocks placed overhead with connection function. The good quality mortar is rich of aggregates of volcanic origin, fragments of terracotta, piperno and limestone masonry not well cooked.
The analysis of the building textures, reported in Table 1, show the differences between the technique called filari, in comparison to that cantieri previously adopted in Naples.
Masonry chronotypes during XV to XX centuries in the Neapolitan area.
The essay of De Cesare 28 is an useful starting point to reflect on the tuff masonry building tradition, although the fundaments of requirements, criteria and guidelines can be traced back to the treatises of Pollio 29 and Plinii.30,31 The publication of characteristics and texture of tuff masonry will only take place in the nineteenth century contextually with the work of many French treatises, see for example Quatremere de Quincy. 32 Despite the particularity due to the local availability and workmanship, the building specifications in the Neapolitan area in the XIX century remained strongly linked to the 19th-century prescriptions of the rest of Europe, perhaps linked to the political story of the area, subjected to French and Spanish domination.
Typical building mortar was composed of hydrated lime putty, with addition in some cases of arena, a pure fine aggregate extracted in the tuff caves, and pozzolana of the best quality. In the construction phase, the mix design contained 1/3 of slaked limes, 1/3 of arena and 1/3 of pozzolana. The mortar mixture was performed by an expert and special worker in order to get a perfect loose mortar, without the help of the water (contained in the lime putty). 33
The stones were chosen to have homogeneous texture without pumices, lapilli and other inclusions, the soft rock was easily workable with the mason’s working tools.34,35 Quarry excavation techniques have not made much progress in recent times, since we can with sufficient certainty affirm that the block currently on the market are not strongly different from those excavated in the last three centuries, although the development of transportation and handling has incremented the production. Consequentially, the attribution of chronotypes can be made with sufficient reliability in the post-medieval tuff masonry, while the most ancient masonries present a large variability of textures, so that a rough classification can be made distinguishing regular and irregular assemblies.
Materials and methods
Masonry strength estimation can, at present, be carried out by means of various destructive, partially destructive and/or non-destructive investigation methods. Among these, some allow the direct determination of the parameters of interest, while others require the use of analytical expressions or correlations, whose validity has often a limited field of application. Since on-site non destructive tests on masonry present a wide range of uncertainties,36 –38 recently the use of semi destructive tests like the flat jack ones has been widely diffused.
Following the analysis of documents and literature briefly resumed in the previous sections, in this paragraph a detailed description of the laboratory tests performed on materials, tuff and masonry, and masonry elements, columns and squat walls, is reported in this paragraph, together with a large set of results on tuff masonry obtained by in situ flat jack tests.
Laboratory tests on mortar specimens
The mechanical characteristics of mortars are the starting point to understand the behavior of a masonry building, 39 especially in tuff masonry where blocks are made by a soft and low-strength rock. The mortar specimens were prepared using three different mixes based on lime, cement and the traditional Neapolitan pozzolana (Figure 1).

Mortar cube after the compressive test.
The compositions of the three mortars are reported in Table 2. The second one has been prepared according to the traditional ratios used in Neapolitan area for a commonly used structural mortar, the other two contain a small amount of cement, to avoid a strongly different behavior of masonry with respect to the traditional mortar M2. The water dosage in the three cases was chosen to obtain a workable mortar, to reproduce the yard workmanship. Curing of the specimens was made in tap water. The tests were performed according the indications of RR.DD. 11/06/1939, 40 UNI EN 1015-11, 41 on prisms 40 × 40 × 160 mm3 for flexural tests and on cubes 40 × 40 × 40 mm3 for compressive ones.
Ratios of mortars’ components.
The mechanical characteristics reported in Table 3, where
Mechanical test results on mortars.

Mechanical characteristics of M2 mortar versus curing time. 42
Laboratory tests on tuff specimens
The building heritage of central-southern Italy is mainly characterized by structures in masonry made with tuff blocks extracted from local quarries, often named Neapolitan Yellow Tuff, and reduced to units suitable to be part of different textures as shown in Table 1.
Tuff is a soft rock of volcanic origin (Figure 3), whose physical and mechanical characteristics are in general variable with place and excavation depth. In the Neapolitan area there are several active quarries and the material properties have been analyzed by the authors in a large experimental campaign. The uniaxial tensile strength typically varies between 0.5 and 5.0 MPa (Figure 4), while the compressive elastic modulus ranges from 800 to 3000 MPa; most frequent values (80% of cases) between 1000 and 2000 MPa. The authors have performed direct tensile tests on prismatic specimens, indicated with Pr_i in Figure 4. The dimensions of the samples have been fixed among those indicated by the UNI standard, 46 in relation to the size of the stone grains: dimensions for both direct tensile and compressive tests are 70 × 70 × 140 mm. To enhance the rupture in the middle part of the sample, a slight reduction (1 cm along two opposite faces of the specimen) of the middle cross section has been realized (Figure 3, right). The tensile elastic modulus obtained is slightly higher than 100 MPa (Figure 4).

Direct tensile test on tuff specimens (left) and specimens after the test (right).

Direct tensile strength of tuff specimens.
On the basis of a preliminary survey of existing standards, reference standards for mechanical characterization tests have been identified. The compressive tests have been carried out according to UNI EN 1926. 45 The load device used is a mechanical one, capable of applying compressive forces in displacement control and equipped with a load cell capable of measuring the applied load.
The samples, indicated with Pr_i in Figure 5 (dimensions 70 × 70 × 140 mm) have been subjected to a monotonous load process, with displacement control and load velocity less than or equal to 0.5 MPa/s (Figure 5). Between specimen faces and loading plates no material was placed to avoid end friction. Vertical displacement transducers and horizontal strain gage have placed on the four faces of the specimen, so that both Young modulus (on the right side of Figure 6) and Poisson coefficient (obtainable by data on the left side of Figure 6) have been determined.

Compressive test on tuff specimen.

Diagram σ-ε for tuff specimens.
The mean mechanical properties of tuff blocks are summarized in Table 4, where
Mean mechanical characteristics of tuff elements.
The stress strain diagrams depicted in Figure 6 for only six of the total number of tests performed give an overview of the tuff constitutive behavior.
Laboratory tests on tuff masonry elements
The laboratory tests were carried out on yellow tuff masonry columns with square bases (400 × 400 mm2). The size of the standard blocks (as they were produced in the quarry) was reduced, with a single cut, to 120 × 200 × 400 mm3, to let the building of a square masonry prism, with height/base ratio equal to 3 (Figure 7). To assure interface conditions equal to the real ones, the blocks have been put in place with the cutted side on the external face of the column.

Compressive tests on tuff masonry columns (left) and detail of the acquisition setup (right).
The type of pozzolanic mortar used were M1 and M2 of Table 2. Three columns for every mortar mix, composed by ten layers of blocks, were subjected to uniaxial compressive stress. The applied load was evaluated by means of a load cell (5000 KN), while displacements were detected by a 400 mm inductive displacement transducer on every side of the column. Omega transducers over the horizontal and vertical joints and strain gages on the tuff blocks on two opposite faces of the column were placed.
The results of the compressive tests on the masonry columns are shown in Figure 8. On the right side of the graph the longitudinal strains (obtained by the mean value of the displacement transducers on the four sides of the column) are reported, while the transversal ones (mean values on two opposite sides) are on the left side. The blue lines correspond to the M1 columns, the red ones to the M2 ones: longitudinal (dashed curves) and transversal strain (continuous lines) were evaluated.

Results of the compressive tests on the tuff columns.
The laboratory performed in the second case concerned compression tests carried out on the reproduction of three different types of masonry, largely diffused in the Neapolitan area, considered related to different historical periods. The tests were carried out on squat tuff masonry panels, which roughly reproduced the typical Neapolitan tuff masonry (Figure 9). 46

Results of the compressive tests on the squat walls. 47
The yellow tuff blocks come, as in the case of the tuff blocks of the columns, from a quarry in Chiaiano in one of the cases (filari walls) and from a quarry in the center of the city in the other two (cantieri and bozzette walls). In this case, each of the masonry type was associated with a different pozzolanic mortar, whose compositions and use in the tests is shown in Table 2. 46 For the compressive tests an oleodynamic testing setup able to perform controlled displacement tests up to 3000 KN was used. On the two opposite faces of the tested walls three inductive displacement transducers (420 mm) were placed. The obtained stress strain curves make reference to the mean value of the six transducers. 48
The highest values, in this case, are those related to cantieri masonry and are probably due to the intrinsic resistance of the mortar and tuff and their consequent homogeneous of behavior as a single composite material. In the case of bozzette masonry, the lower compressive strength is due to the mortar.
In the case of blocchetti a filari masonry, however, the lower compressive strength is due to the tuff elements.
In conclusion, for the panels the results of the experimental tests showed good mechanical behavior in compression and an appreciable ductile behavior of the tuff masonry in the post-peak range, due to the end confinement of the panel,49 –52 without strong differences in the elastic behavior of the three different types of panel. The behavior is similar to that observed by Sandoli et al. 53 on scaled squat columns. On the contrary, the slender specimens corresponding to the columns tested by the authors have shown a more fragile behavior, with limited deformation in the post-elastic range. In general, the different geometric configuration of the specimen has a strong influence on the behavior of masonry, and the inverse increase of Young modulus with respect to strength versus specimen shape has been observed in case of fiber reinforced concrete, see for example Serafini et al. 54
In situ tests on tuff masonry
The flat jacks technique allow to evaluate both the on-site tension (single jack test) and the first cracking and “breaking” tension, as well as the elasticity modules (double jack test). The most widespread methodology is the double flat jacks,55 –58 thanks to the availability of elastic parameters. Although in some cases good correlations have been found, 59 the investigation should make it possible to directly determine the strength and deformability of the masonry portion under test, but in reality, due to the constraint conditions (localization of the specimen, confinement conditions, distance of openings) and the masonry texture, it often leads to values that are not fully representative of the actual strength of the masonry.
The performed tests described in this section refer to a very large sample of data relating to the masonry of the Campania region. 60 The data concern different types of regular and irregular (mainly regular) tuff masonry and represent a sufficiently large and statistically significant sample, such as to be considered representative of the types of masonry present in the Campania territory, belonging to different historical periods and made with different construction techniques. Since only in a limited number of cases the difference among masonry chronotypes has been possible, two types of masonry, regular and irregular, have been distinguished. In Figure 10 some of the obtained σ-ε diagrams are reported. The diagrams relative to the three chronotypes described in Section “Masonry chronotypes” are compared to the diagrams obtained for regular and irregular masonries, for which a chronotype has not been detected. It must be noted that in general the masonries present in the existing buildings, especially the oldest ones, have been subjected during the centuries to interventions, partial rebuilding and substitutions, not always datable with absolute certainty. As a consequence, the original masonry texture is in some cases not recognizable. In general, in the flat jack tests, while the maximum stress presents a wide range of values, the Young modulus variation is more limited. Comparing in fact the diagrams and the values of compressive strength (2.85 N/mm2 cantieri, 1.71 N/mm2 irregular blocks), it is immediately clear how this type of test, in particular, involves low reliability for the maximum stress detection (Figure 11).

Bozzette (left), Filari (center) and Cantieri (right) specimens after the compressive tests. 48

Example of stress strain diagrams for Flat-Jack tests.
Specifically, the validity of this test is strongly influenced by the type of masonry on which it is performed.61,62 If in the case of regular block masonry, the position of the flat jacks, residing on the mortar’s layers, allows crack forming at higher intervals, in the case of irregular masonry, the position of the flat jacks will certainly reside on the stone blocks and not on the mortar’s layers leading to fractures that will almost always occur at lower intervals.
The discrepancy is more evident when reference to the entire test set is made. In Figures 12 and 13 a picture of the distribution of elastic modulus and compressive stress for a set of 300 flat jack tests on tuff masonry in real existing buildings in the Campania region is given.

Distribution of elastic modulus.

Distribution of compressive strength.
The data sample is sufficiently huge to be significant from a statistical point of view, as it consists of a large number of data relating to walls of the same type, mostly built with the same construction technique, although in different eras.
Reference to the results obtained in the experimental tests described in the previous sections have been reported on the same graphs. It can be easily seen that the tests on the squat tuff masonry panels give values of elastic moduli, although slightly lower than the most frequent value, somewhat in the range defined by the on site tests. On the contrary, the value obtained from the columns is outside of the interval defined by the flat jack tests. This can be explained considering the geometry of the samples in the two cases: the masonry panels are significantly squat and confined, while the columns are very slender specimen and in the central part the confinement due to the end friction can be considered totally absent.
Reverse behavior can be observed in the case of maximum compressive strength distribution. The columns present very low values and the squat tuff masonry panels higher ones, largely outside of the interval defined by the flat jack tests, in this case also due to the geometry and end condition of the specimen. The different geometry of the specimens, together with the different confinement conditions, play a key role in the performance, as it has been elsewhere underlined for different no-tension materials.54,63 –66
Figure 14 presents the variation of Young modulus versus compressive strength in the mechanical tests examined. Smaller dots represent the double flat jack set of tests, red ones for regular masonry and green ones for irregular one, large blue dots are for the laboratory tests on masonry walls, blue triangle for the tuff columns test results. Due to the limited number of flat jack tests on irregular masonry, a discussion cannot be made, except the fact that these results represent less than 1% of the total number of the on site data. The historical studies reported in Section “Masonry chronotypes” confirm that all the masonry typologies built in the last centuries regard regular masonries, so that a distinction among them is of historical interest only, since the mechanical characteristics can be considered belonging to the “regular” macrocategory. The experimental data of the columns fall outside the range defined by the regular masonry, because, although the arrangement of the blocks can be considered regular, the geometrical slender configuration of the tested column does not give to the structural element sufficient strength and ductility. This is confirmed by the laboratory tests performed by Aiello et al. 66 on unreinforced squat columns, whose test results are reported in Figure 14 with purple diamonds. In this last case the high observed strength are mainly due to the high strength of the tuff blocks.

Young modulus versus compressive strength.
The experimental data are compared with two correlation formulas. As it can be seen, the relation provided by the Eurocode 6, included in the Italian codes actually in force, 24 overestimates the Young modulus, while the relation of the FEMA document 67 seem more consistent with the experimental data.
It should be underlined that codes reference values for mechanical properties of tuff masonry, while can be independent on the masonry texture (regular/irregular), cannot be independent on the type of structural element taken into considerations: isolated columns, cloisters and in general slender walls present limited ductility range, differently from masonry portions placed in large building walls, where all the surrounding elements are a kinematic constraint. 68
Final remarks
The results of an experimental campaign on yellow tuff masonry from the Campania Region, performed both in laboratory and both on site, has been presented in this paper. Laboratory tests have been carried out on building materials, tuff blocks and mortar, and structural elements, walls and columns, while on site tests regard flat jack tests on existing tuff masonry buildings. Despite the limited number of laboratory tests on structural elements, several consideration about their reliability and comparability of the results with those obtained by a large set of in situ tests can be made.
The main observation regards the geometry of the tested specimens: slenderness, confinement and in general end conditions strongly influence the performance of the masonry. Non slender and squat masonry walls present lower elastic modulus, larger ductile post-peak range and higher compressive strength with respect to slender masonry specimens, independently on the masonry texture and organization of the blocks.
If a comparison is made with the in situ tests, squat tuff masonry panels present values of elastic moduli corresponding, although slightly lower than the central value, to the most frequent values. On the contrary, the value obtained from the columns is outside the range of data defined by the flat jack tests. Reverse behavior can be observed in the case of maximum compressive strength distribution. The columns present very low values and the squat tuff masonry panels higher ones, largely outside of the results interval defined by the flat jack tests, where the surrounding wall makes a sort of confinement on the tested element.
Footnotes
Acknowledgements
The scientific support of Professor Giorgio Frunzio to the tests on materials and columns and that of Eng. Andrea Basile, Director of Tecnolab s.r.l. (Naples), in performing the flat jack tests is gratefully acknowledged.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work has been financially supported by University of Campania “Luigi Vanvitelli” and University of Sannio at Benevento.
