On the geology and the geotechnical properties of pyroclastic flow deposits of the Colli Albani
- First Online:
- Cite this article as:
- Cecconi, M., Scarapazzi, M. & Viggiani, G.M.B. Bull Eng Geol Environ (2010) 69: 185. doi:10.1007/s10064-009-0250-x
- 231 Views
This paper presents the results of an experimental investigation into the engineering geology properties of three pyroclastic deposits from the Colli Albani volcanic complex, typical of the subsoil of Rome (Italy). In their natural state, these materials are coarse-grained weak rocks, generally unsaturated in situ. The mechanical properties of the material are related to the geological origin of the deposits, their formation environment and mechanisms. These are also revealed by peculiar micro-structural features. A technical sheet for the description and classification of these deposits is proposed in the paper. The experimental investigation consisted mainly of identification and classification tests, one-dimensional compression and direct shear tests on saturated and dry samples. Due to their nature, non standard techniques had to be developed for sample preparation and testing. The main findings on compressibility, shear strength and creep properties of the materials are presented in the paper. Special attention is given to the definition of the failure envelope at relatively low confining stress and to the assessment of the influence of saturation on shear strength parameters.
KeywordsPyroclastic depositsTechnical classificationPhysical propertiesCompressibilityShear strength
L’article présente les résultats de travaux expérimentaux relatifs aux propriétés géotechniques de trois dépôts pyroclastiques issus de la formation volcanique de Colli Albani, typique du sous-sol de Rome (Italie). A l’état naturel, ces matériaux sont des roches tendres à grain grossier, généralement non saturées in situ. Les propriétés mécaniques de ces matériaux doivent être mises en rapport avec l’origine géologique des dépôts, les contextes environnementaux et les processus de formation, ces derniers étant révélés par des caractéristiques micro-structurales particulières. Une fiche technique est proposée dans l’article pour la description et la classification de ce type de dépôt. Les travaux expérimentaux ont consisté principalement en des tests d’identification et de classification, des essais de compression simple et des essais de cisaillement direct sur des échantillons saturés ou secs. Du fait de la nature de ces matériaux, des techniques particulières ont dû être mises en œuvre pour la préparation des échantillons et les essais. Les principaux résultats relatifs à la compressibilité, la résistance au cisaillement et les caractéristiques de fluage de ces matériaux sont présentés dans l’article. Une attention particulière est donnée à la définition de la courbe de rupture sous relativement faible confinement et à l’évaluation de l’influence de la saturation sur les paramètres de résistance au cisaillement.
Mots clésDépôts pyroclastiquesClassification techniquePropriétés physiquesCompressibilitéRésistance au cisaillement
Pyroclastic deposits cover large parts of Central and Southern Italy, totalling some 8–9,000 km2. Volcanoclastic deposits of southern Italy—both airfall and flow deposits—have been extensively studied in the literature (e.g.: Croce et al. 1961; Pellegrino 1967; Evangelista and Aversa 1994; Del Prete et al. 1998; Esposito and Guadagno 1998). In central Italy, the main volcanic complexes in the area of Rome are the Monti Sabatini, 30 km to the north-west, and the Colli Albani, about 25 km to the south-east of the city. The rocks from these volcanic complexes are very significant in the development of the Rome area, creating fertile soils but also high, steep slopes which are easily quarried for building materials. Pozzolanas have been continuously exploited since Roman and Etruscan times to produce hydraulic mortars and cement; they are still often used as sub-grade for roads and tennis courts.
This study of the geotechnical properties of the “soft” pyroclastic rocks (or pozzolanas) beneath Rome was initiated some ten years ago, driven by the need to analyse the stability of sub-vertical cuts and the underground cavities frequently encountered in many parts of the city. Mining activities over the centuries have resulted in a complex network of underground cavities and in some cases sudden collapses have occurred with the formation of sinkholes at the surface (Bernabini et al. 1966; Lembo Fazio and Ribacchi 1990), but numerous underground cavities have not yet been surveyed or mapped. Open pozzolana quarries are also common in the south east of Rome, with sub-vertical excavation faces which can be as high as 20–25 m.
Cecconi (1999), Cecconi and Viggiani (2000, 2001), and Cecconi et al. (2003) describe the main outcomes of a detailed study of the mechanical properties of the Pozzolana Nera, belonging to the III cycle of the Tuscolano-Artemisia eruptive phase of the Colli Albani (De Rita 1988). Their work was based on saturated samples, but as these materials are generally above the water table and only partly saturated in situ (Sr = 0.4/0.5), further research and experimental work on the hydraulic and mechanical properties of the partially saturated material was carried out for both basic research and engineering applications. Some of this work was presented by Cattoni (2003) and Cattoni et al. (2007).
The present paper focuses on the results of a recent comparative geological and geotechnical investigation of two additional pyroclastic flow deposits of the Colli Albani, locally known as Pozzolanelle and Conglomerato Giallo, belonging respectively to the IV and II cycles of the Tuscolano-Artemisia eruptive phase of the Colli Albani (De Rita et al. 1988).
As pointed out by Karner et al. (2001), because of the many uses of this material over the long history of the area, a confusing nomenclature for the rock units of the Colli Albani has built up over the years. The volcanic rocks from the Colli Albani have often been identified on the basis of the geographic location of outcrops and quarries or their colour; more rarely on the basis of their physical and mechanical properties, in part because of the difficulty of intact sampling of these materials for laboratory testing (Cecconi 1998). Often these units have been mis-identified because of similar physical characteristics and ambiguous terminology. From a strictly geological point of view, these units can be distinguished unequivocally only through geochemical or radio-isotopic analysis, as done, for example, by Karner et al. (2001). The technical description sheet proposed here is simply an aid to a standard description of pyroclastic soils and to facilitate a better qualitative and semi-quantitative assessment of some of the main properties of the material.
Measurement of the physical properties and mechanical characteristics of the two materials under investigation was intentionally carried out using common laboratory techniques, including conventional one-dimensional and uniaxial compression tests, indirect tension tests and direct shear tests. Due to the pyroclastic nature of the materials, special non-standard procedures were developed for the preparation of the natural samples, including rotary coring of frozen samples. Some of the experimental results obtained on reconstituted and natural samples of Pozzolanelle and Conglomerato Giallo are compared with those obtained in a much wider experimental programme on the Pozzolana Nera.
However, in this study, the aim was not to give average values of the mechanical properties but rather to identify some of the problems that may be encountered in the laboratory characterisation of these materials and to guide the choice of suitable testing programmes. In an attempt to relate the microscopic features and the macroscopic response of the two materials, optical and electron scanning microscopy was also undertaken.
The Colli Albani volcanic complex is in the southernmost part of Lazio (Central Italy). The most important structure is the central caldera (Giordano et al. 2006) which was formed about 600,000 years ago (De Rita et al. 1995). Different periods of activity have been recognised, which resulted in the deposition of 300 km3 of materials, mainly pyroclastic, which outcrop extensively to the south of the river Aniene and east of the river Tiber. The presence of magmatic and lahar deposits indicates that the volcano is still active although quiescent, with fumarole activity (Funiciello et al. 2003). The last explosive eruption was at the time of Tullio Ostilio (673 BC–641 BC).
Despite several stages of volcanic activity, the different explosive phases are homogeneous from the geochemical point of view (Fornaseri et al. 1963; Ferrara et al. 1985; Turi et al. 1991; Serri et al. 1991) and are effectively uniform in terms of petrochemical composition (Trigila et al. 1995).
At the macro-scale, Pozzolanelle is massive and chaotic. Diano (2005) defined four different facies of the eruption unit, the fourth of which, forming the top unit of the Tuscolano-Artemisio lithosome, is mainly dark grey to dark red/brown, ashy, and reaches a maximum thickness of about 4 m. The composition is tephritic-phonolitic (Freda et al. 1997; Gaeta et al. 2006).
The Pozzolanelle is uncemented and although the lower portion can occasionally be lithified, generally it presents transitional features with the Tufo Lionato, also belonging to the Villa Senni eruption unit. This material is lithoidal with an extensively zeolitised matrix rich of chabasite and phillipsite. Underlying the Tufo Lionato is the deposit Pozzolana Nera which consists of a thin level of fallout materials (scoriaceous lapilli with lithic fragments and crystals of leucite and pyroxene) at the bottom of an ashy level about 0.3 m thick. In its central portion, the deposit is massive and chaotic, mainly formed by ignimbrites. Its maximum thickness is approximately 20 m. The matrix consists mostly of dark grey to black fine ash shards, and crystals such as leucite, clinopyroxene and biotite (Giordano et al. 2006).
The Conglomerato Giallo is a succession of pyroclastic deposits with particle sizes ranging from ash to lapilli. It consists primarily of rounded altered yellow and red scorias with secondary volcanic lithics and clinopyroxene, biotite and altered leucite crystals (mm to cm in size). This deposit varies in grading and forms in somewhat chaotic banks or lenses which vary in thickness from 2–3 m to a maximum of about 20 m in the area of the Basilica of S. Paolo fuori le Mura. Both the geometry and the dominance of the juvenile/scoriaceous component suggest a syn-eruptive origin, probably linked to remoulding of the scoria cones formed along peri-calderic fractures following the eruption of the Pozzolane Rosse. The most likely depositional mechanism is debris flow or hyperconcentrated flow inside a wide depression at the top of the Pozzolane Rosse in the north western sector of the Colli Albani volcanic complex.
Technical classification of pyroclastic soils
The geological complexity of pyroclastic rocks/soils makes it difficult to classify these deposits using conventional systems. The heterogeneity of the soil mass depends on several factors, such as viscosity, temperature, chemical interaction, eruption energy, distance from the magma chamber, etc. These factors may differ within the same depositional flow and they generally vary due to the interaction between the pyroclastic flow, the pre-existent morphology and the ground water. Further complexities may be induced by post-depositional processes such as welding, cementation, fissuring and chemical weathering.
For these materials, therefore, the geological and geotechnical classification systems commonly used for the majority of natural soils—probably less problematic than volcanic soils—may be inadequate or just incomplete. For this reason an original descriptive model for volcanic soils has been developed, applying some of the specific terminology of volcanology (Scandone and Giacomelli 1998).
Items 1–4 in Fig. 3 are self-explanatory and provide the basic information needed for future reference.
Lithoid or semi-lithoid: the pyroclasts (the clasts composing the pyroclastic rock) are strongly welded together and the deposit behaves as a soft rock. This type of deposit is often represented by zeolitized or variously cemented tuffs.
Welded: the pyroclasts are held together even without capillary tension. This soil often has an intermediate behaviour between welded granular soils and soft rocks (pyroclastic flow deposits usually belong to this class).
Granular: the pyroclasts are not linked one to each other even if apparent cohesion may be present. These deposits behave as loose granular soils. Pyroclastic fall deposits may sometimes belong to this class.
Altered: directly deriving from the former classes, it is deeply modified by an alteration process which has partly destroyed the original structure. These materials usually have an intermediate behaviour between fine grained soils and granular soils.
Deeply altered: the product of an alteration process which has completely destroyed the original structure of the volcanic rock. These materials behave as fine grained soils.
Item 6—colour: The colour may give qualitative information on the mineralogical composition of the single pyroclast or of the whole deposit. Colour is defined both for the matrix and the clasts. Tone indicates if the colour is dark or light and gives qualitative and indirect information on the saturation degree, e.g. black to dark green is typical of material containing iron and magnesium or the presence of organic components; red indicates oxidized iron and green reduced iron. It is recommended that the colour is determined on both dry and saturated material to obtain the relation between chromatic variation and saturation degree; natural saturation conditions may also be recorded. If considered useful, the colour of the individual pyroclasts may be indicated in the Notes, i.e.: colour of scorias, colour of crystals etc. The chromatic aspect of the soil can be referred to the Munsell Colour Chart code (Hue, Value and Chroma).
juvenile (produced directly from cooling magma during transport prior to primary deposition). The juvenile component can also be divided into scoria, pumices and crystals;
secondary or cognate (formed during earlier volcanic activity, but which have been detached and ejected with other pyroclastic debris during a later eruption);
accidental (generated by disruption as a direct result of volcanic action, but not formed by previous activity of the volcano, generally belonging to the sedimentary basement: xenolites, mantle fragments, sedimentary).
The mineralogical nature of the components may be described in the Notes on the basis of macroscopical characteristics but further analysis (thin section, diffractometry, microprobe) may be required.
granular sustained: closed texture with interlocked clasts;
intermediate: either frequent contacts without locking or occasional contact points between clasts;
matrix sustained: coarse fraction dispersed in a prevalent matrix.
isotropic: elongated or flat clasts do not have a particular direction but are randomly oriented and dispersed;
anisotropic: elongate or flat clasts have the same orientation, which is parallel to the layer. In this case the declination should be recorded;
imbricated: elongated or flat clasts have the same orientation which is different from the declination of the layer; again the declinations of the clasts should be determined;
Item 11—Grading: determines the percentage of clasts of various sizes: blocks or bombs (d > 64 mm), lapillus (2 < d < 64 mm), coarse ash (0.063 < d < 2 mm), medium to fine ash (d < 0.063 mm).
Item 14—porosity: the porosity of the material can be defined qualitatively as low, medium or high and quantitatively by the calculation of voids ratio (which should be recorded in the Notes). However, the voids ratio calculation does not take the clasts’ internal voids into account. In the case of welded deposits, but also in general, this aspect cannot easily be defined in the preliminary phases.
Item 15—bonding: may be either true (cannot be degraded by water saturation) or apparent (deriving from the suction in partly saturated soils). Small interparticle forces of an electrostatic nature may exist in the clays deriving from the alteration of pyroclastic rocks. True welding is related to syn- and post-depositional processes (cooling, chemical cementation by mineral, zeolites etc) and can be high, medium or low. Apparent bonding can be quantitatively defined as the difference between the cohesion of the soil at the natural water content and that of the same soil in saturated conditions. Apparent and true bonding may coexist.
The technical classification method described above was tested for the three pyroclastic deposits under examination. The examples given in the Appendix are from samples collected from a quarry some 10 km south of Rome.
The classification exercise indicates that Pozzolanelle are a welded pyroclastic rock with light reddish brown clasts in a pinkish white matrix when dry and reddish brown clasts in a pinkish grey matrix when wet. Their structure is massive and quite homogeneous at the scale of the deposit, presenting limited local variations in grading. However, there was a 1 m thick horizon with a significant increase of the cinder fraction near the top of the deposit which was not considered in the description.
The pyroclasts are more than 90% juvenile, with 60% matrix and 30% clasts (20% scoria, 5% pumices and 5% crystals). The texture is intermediate with few contacts between randomly oriented clasts (isotropic).
The material falls in the lapilli-tuff field of Fischer’s chart, consisting of lapilli (50%) coarse ashes (40%) and medium to fine ashes (10%), while blocks and bombs are totally absent. Clast shape is sub angular; vesiculation is low to absent.
Porosity is medium to high: laboratory determination of voids ratio yielded a value of n = 0.64 ± 0.01. True bonding is medium and related to the formation of secondary minerals; apparent cohesion is revealed by shear tests carried out in dry and saturated conditions.
Pozzolana Nera is a welded pyroclastic rock with light grey clasts in a greenish grey matrix when dry, or dark greenish grey matrix when wet. Its structure is massive and relatively homogeneous at the scale of the deposit, presenting only limited local variations in grading.
The pyroclasts, in all size classes, are mainly juvenile (95%), with 65% matrix and 35% (30% scoria clasts, no pumice) and 5% crystals. The texture of the material is intermediate between granular- and matrix-sustained with randomly oriented clasts (isotropic).
The material consists of lapilli (>50%), coarse ashes (30%) and medium to fine ashes (<20%) while blocks and bombs are totally absent. Clast shape varies from sub-angular to angular, although clast agglomeration tends to increase the angularity. Vesiculation is low to absent.
Porosity is medium to low: laboratory determination of voids ratio yielded a value of n = 0.45 ± 0.01. True bonding is medium to high, probably due to the mechanism of deposition rather than diagenetic factors.
Conglomerato Giallo is a lithoid pyroclastic rock with clasts of various colours (dark bluish grey, pinkish brown, black) in a light yellowish brown matrix when dry and brown matrix, when wet. Its structure is massive and quite homogeneous at the scale of the deposit.
The pyroclasts are 80% juvenile, with 50% matrix and 30% clasts, of which 20% are scoria, 5% pumices, and 5% crystals, 15% secondary, with only 5% rock fragments of volcanic origin or accidental. The texture is matrix sustained.
The material falls in the lapilli-tuff field of Fischer’s chart, consisting of lapilli (30%), coarse ashes (35%), and medium to fine ashes (35%); blocks and bombs are totally absent. Clast shape is sub angular; vesiculation is low to absent.
Porosity is low: laboratory determination of voids ratio yielded a value of n = 0.29 ± 0.01. True bonding is high, associated with diagenetic factors.
Microstructural features and physical properties
Published data (Cecconi and Viggiani 2001) on thin sections of Pozzolana Nera in plane-polarised light show that the material has a clastic texture, diffused in a scoriaceous matrix containing frequent crystals of leucite, pyroxene and biotite as well as lithic fragments of lava. Scanning electron microscopy (SEM) of a small titanium-coated dried sample of Pozzolana Nera also shows that the microstructure of the material consists of sub-angular grains of very variable size with a rough and pitted surface, at times bonded by physical bridges of the same mineralogical nature as the grains, and evenly coated by microspherules with a minimum diameter of about 0.5 μm.
In thin section, the Pozzolanelle appear to consist of a compact matrix containing darker clasts, scoria with various degrees of vesiculation and frequent crystals, recognizable from their straight edges. The Conglomerato Giallo has a better defined clastic texture, diffused in a yellowish brown scoriaceous matrix containing frequent crystals as well as lithic fragments of lava, probably derived from the volcanic chimney. The mineralogical nature of the crystals would have been better defined in plane polarised light but is likely to consist mainly of leucite, biotite and clinopyroxene. In thin section, pores appear as white areas with no defined edges. Inter-granular pore features, such as size, shape, and orientation are very variable, particularly in the case of the Conglomerato Giallo; for both materials, inter-granular pores appear to be partially filled by altered material and secondary minerals. Intra-granular pores are generally isometric. The closed porosity, within the clasts, is very small, as confirmed by measurements of specific gravity carried out on increasingly finer, powdered material.
Main physical properties of the materials under investigation
13.0 ± 1.7
16.46 ± 0.06
14.64 ± 0.03
0.45 ± 0.01
43.5 ± 3.5
16.3 ± 1.9
11.92 ± 0.06
9.93 ± 0.03
0.64 ± 0.01
25.0 ± 5.2
26.8 ± 0.6
17.40 ± 0.06
17.40 ± 0.03
0.29 ± 0.01
88.0 ± 1.5
The first technique was only effective for the Pozzolanelle, while the second technique was adopted for both Pozzolanelle and Conglomerato Giallo. While the grain size distribution of Pozzolana Nera and Conglomerato Giallo is not affected by the sieving time or number of sieving cycles, the grain size distribution obtained for the Pozzolanelle depended on the number of sieving cycles, with a tendency for the fine fraction to increase with increasing number of cycles, indicating imperfect separation of small aggregates. The grain size distribution curve obtained with this procedure stabilizes after about 5 sieving cycles. A much more pronounced shift of the grain size distribution is observed for cycles of freezing and thawing, showing how the second procedure is much more effective in breaking the original structure of the material. Again, after three cycles of freezing and thawing the grain size distribution of Pozzolanelle tends to stabilise.
The core of the experimental work consisted of direct shear tests on intact samples of Pozzolanelle. A smaller number of one-dimensional compression, uniaxial compression and indirect tension tests on intact and reconstituted samples of Pozzolanelle and Conglomerato Giallo were also carried out, to determine the compressibility of the two materials and the strength of the intact Conglomerato Giallo.
The direct shear tests were carried out in standard 60 × 60 mm shear boxes. However, samples of 40 mm were used to avoid the rupture surface crossing the boundaries of the shear box. The shear force was developed by an electric motor, providing a variable speed ranging from 5 × 10−4 to 2 mm/min. All tests were performed at a displacement rate of 0.1 mm/ min. Micrometer dial gauges with a resolution of 0.001 and 0.01 mm were used to measure vertical and horizontal displacements respectively. One dimensional compression tests were carried out in standard oedometer cells with a diameter of 71.4 mm, to a maximum vertical stress of about 4,000 kPa, or in reduced diameter (35.1 mm) cells to a maximum vertical stress of 12,800 kPa.
For the direct shear and oedometer tests, intact samples of Pozzolanelle were obtained manually from partially saturated natural blocks by pushing a square or circular thin-walled soil sampler into the block and trimming the material around it with a knife. To ensure good contact between the sample and the loading system any voids or irregularities were filled with a mix of 85% crushed Pozzolanelle passing sieve N40 and 15% alabastrine gypsum. Direct shear tests were carried out on both vacuum dried and saturated natural samples; the latter were placed into the shear box, submerged in water and saturated under vacuum. Reconstituted samples of Pozzolanelle and Conglomerato Giallo for oedometer testing were formed using material passing ASTM sieve 10 (2 mm) which was oven-dried, mixed with de-aired distilled water at a water content ≅20% and compacted in the oedometer in thin layers.
Intact samples of Conglomerato Giallo for uniaxial compression and indirect tension tests were obtained by rotary coring partially saturated natural blocks. Before coring, the blocks were immersed in water, which increased the saturation degree up to Sr ≅ 0.7, and then frozen at −20°C; the ends were then squared on a band saw up to a tolerance of 0.2 mm which, for the grain size distribution of the material, is the smallest achievable.
The effects of freezing on the microstructure of the material were partially investigated by means of compression wave velocity measurements as the velocity with which stress waves are transmitted through a soil depends on the elastic properties and density of the material. The velocity of propagation of P-waves has been traditionally related to the degree of fissuring and micro-cracking of rock masses (Goodman 1989). In this case a significant reduction of the velocity of propagation of P-waves would indicate a reduction of the initial stiffness of the material, probably due to the de-structuring induced by freezing. Measurements of P-wave velocity were first made on vacuum dried natural samples of approximate dimensions of 0.15 × 0.15 × 0.15 m. P- and S-waves were transmitted and received across unconfined samples using electro-mechanical 15 mm diameter transducers with a characteristic frequency of vibration of 1 MHz under an axial load of 1.0 kN for the Comglomerato Giallo and of 0.8 kN for the Pozzolanelle.
For the Conglomerato Giallo the measurements were repeated on the same samples after they had been immersed in water, frozen at −20°C, thawed and vacuum dried; the procedures and the times for saturation and freezing were the same as those employed for the preparation of laboratory samples. The results showed the effects of freezing on the stiffness properties of the Conglomerato Giallo to be very minor, with a reduction of the P-wave velocity of about 3%. For Pozzolanelle one cycle of freezing and thawing causes the complete loss of cohesion of the material so it was not possible to prepare undisturbed samples by this technique nor to repeat the measurements after freezing and thawing.
Like other Italian pyroclastic soils and weak rocks (Croce 1954; Evangelista and Aversa 1994) the materials under investigation may exhibit significant long term deformation due to creep. This was investigated by performing conventional oedometer tests on both natural and reconstituted samples. Some results for the secondary one-dimensional compression of Pozzolana Nera (Cecconi et al. 2005) are compared with the results from the other two deposits investigated.
Creep parameters α and β for the three materials under investigation
1.67 × 10−4
1.67 × 10−4
1.77 × 10−4
This paper describes the results of an experimental investigation of the engineering geological properties of three pyroclastic deposits from the Colli Albani volcanic complex, typical of the subsoil of Rome (Italy). In their natural state, these materials are coarse-grained weak rocks, generally unsaturated in situ.
A technical sheet for the description and classification of these deposits was developed during the experimental work and is presented in the paper. The sheet provides a useful index to follow for the standard description of pyroclastic soils and a support for the engineering geologist to define qualitatively and semi-quantitatively some preliminary properties of the material.
The experimental investigation consisted mainly of identification and classification tests, one-dimensional compression and direct shear tests on saturated and dry samples of Pozzolanelle and Conglomerato Giallo. An investigation of the micro-structural features of the materials was also undertaken by means of optical and electron scanning microscopy. Unlike other pyroclastic materials, pyroclastic flow deposits from the Colli Albani exhibit a very small closed porosity within the clasts.
The measurement of the physical properties and mechanical characteristics of the materials was carried out using conventional laboratory techniques with the goal of identifying some of the problems that may be encountered in the laboratory characterisation and to assist in the choice of suitable testing programmes. Due to the nature of the soils, however, non-standard techniques were developed for sample preparation and testing.
The procedures adopted to prepare the material for sieving significantly affect the experimental grain size distribution curve. For welded pyroclastic rocks, such as Pozzolanelle and Pozzolana Nera, the material can be broken by a weak mechanical action such as that applied by human hands, which is sufficient to separate aggregates but not strong enough to break individual particles. However, at the lower bound of welded pyroclastic rocks (Pozzolanelle), the grain size distribution of the material obtained by hand separation is often still significantly affected by the number of sieving cycles, testifying imperfect separation of aggregates. Subjecting the material to cycles of freezing and thawing before hand separation was found to be very effective in separating larger aggregates; only a small number of thermal cycles (three in the case of Pozzolanelle) being sufficient for the grading curve to stabilise. Moreover, for lithoid or semi-lithoid pyroclastic rocks, such as the Conglomerato Giallo, the material can only be broken by hand after the original structure of the rock has been weakened by cycles of freezing and thawing.
Sample preparation for mechanical testing is a very delicate issue. Cylindrical samples of lithoid and semi-lithoid pyroclastic rocks, such as the Conglomerato Giallo, may be obtained by rotary coring partially saturated frozen natural blocks. The technique was originally developed and successfully used for the Pozzolana Nera, at the upper bound of welded pyroclastic rocks. In both cases, the effects of freezing on the microstructure, as revealed by measured values of compression wave velocity through the samples, are negligible. The technique, however, proved completely ineffective for other welded pyroclastic rocks, such as the Pozzolanelle, whose structure is completely destroyed by even one cycle of freezing and thawing. In this case, intact samples can only be prepared by pushing circular or square (for direct shear) thin-walled samplers into the natural blocks and trimming the material around it with a knife. This is a very wearisome and time consuming procedure; an alternative reported in the literature is that of double-barrel coring of natural samples without freezing (Diano 2005) although this was not undertaken in the present study.
From the point of view of grading, all the tested materials were gravel with sand. However, quite often larger lithic fragments were present in the matrix, causing small jerks and jolts in the observed response; note for example the sharp increase of stiffness during unconfined compression of sample 2 at about 0.5% axial strain. The shape of the shear surfaces of direct shear tests is also strongly affected by the occurrence of larger fragments. In some cases it can be so uneven as to intersect the sides of the box, e.g. sample 8, tested at 150 kPa vertical stress. For all these reasons it is recommended that larger samples are used.
Direct shear tests on natural samples of Pozzolanelle indicate that the strength of the saturated material is affected very significantly by dilation, both at peak stress and at the end of test as the horizontal displacements imposed in the shear box are far too small to reach constant volume shearing. On the other hand, for the material at the natural water content, the large values of measured cohesion may be attributed to a combination of increased dilation (due to the roughness of the shear surface) and suction.
The results of the oedometer tests confirm that the initial voids ratio is the main factor affecting the compressibility of these materials, with an almost linear increase of Cc with e0. The trends obtained for the three materials under investigation are consistent with literature data for other natural soils and weak rocks. Like many pyroclastic soils and weak rocks all three materials exhibit significant long term deformations due to creep.
The authors are grateful to Tatiana Rotonda of Roma La Sapienza and Marco Marinelli of Roma Tor Vergata, who made available the experimental set-up and their expertise on how to carry out elastic wave measurements and SEM analyses respectively. Francesca Bozzano and her group were the source of much valuable information on the geology of the Colli Albani. The authors also wish to thank the technical staff at Geoplanning S.r.l. for their help with the experimental work and Mr Tozzi, owner of I.MA.TER. quarry.