You are on page 1of 38

Geotechnical and Geological Engineering (2005) 23: 365–402  Springer 2005

DOI 10.1007/s10706-004-1607-3

Geotechnical characterisation of pyroclastic soils


involved in huge flowslides

EDUARDO BILOTTA, LEONARDO CASCINI, VITO FORESTA and


GIUSEPPE SORBINOw
Department of Civil Engineering, Faculty of Engineering, University of Salerno, via Ponte don
Melillo, 84084 Fisciano (Salerno), Italy
(Received 4 August 2002; revised 4 August 2003; accepted 24 February 2004)

Abstract. Landslides of the flow type involving granular geo-materials frequently result in
casualties and damage to property because of the long travel distance and the high velocities
that these may attain. This was true for the events that took place in Campania Region
(Southern Italy) in May 1998, involving pyroclastic soils originating from explosive activities
of the Somma-Vesuvius volcano. Although these phenomena have frequently affected various
areas of the Campania region over the last few centuries, there were no useful geological and
geotechnical references available in the aftermath of the May 1998 events. For this reason
Salerno University, which was involved in the scientific management of the emergency, ad-
dressed the issue of acquiring data on the geological, geomorphological and hydrogeological
features of the slopes where the landslides had taken place. The information acquired made it
possible to set up a slope evolution model that is able to interpret, from a geological point of
view, past and more recent landslides that had occurred in the same area. As preliminary
geotechnical analyses had already validated the above model, more detailed investigations
were performed both on the pore pressure regimen of the covers still in place as well as on the
physical and mechanical properties of pyroclastic soils, in saturated and unsaturated condi-
tions. The present paper begins by discussing the data acquired during the first phase of the
studies and then goes on to illustrate the laboratory results so far obtained with the aid of
approximate procedures. These help advance our knowledge of pyroclastic soils within a
reasonable time frame, thus improving landslide triggering analysis.

Key words. flowslides; pyroclastic soils; shear strength; soil water characteristic curve;
unsaturated soils.

1. Introduction
Rapid flow type movements can certainly be considered one of the most insidious
landslide phenomena because of their potential for causing casualties and huge
economic damage (Sassa, 1998). They involve different kind of soils (Costa and
Wieczoreck, 1987; Hutchinson, 1988; Hungr et al., 2001) and are widespread in
many countries (Jones, 1973; Costa and Baker, 1981; Ellen and Fleming, 1987;
Sassa, 1988; Takahashi, 1991; O.U., 1998a; Picarelli, 1999). Significant examples are
w
Corresponding author: Department of Civil Engineering, Faculty of Engineering, University of
Salerno, via Ponte don Melillo, 84084 Fisciano (Salerno), Italy; tel.: +39-089-964329, fax: +39-089-
964045, (e-mail: g.sorbino@unisa.it).
366 EDUARDO BILOTTA ET AL.

Figure 1. Location Map.

those periodically occurring in the Campania Region (Southern Italy) triggered by


critical rainfall events (Rossi and Chirico, 1998; Versace, 2001). They involve
unsaturated pyroclastic soils – originating from the explosive phases of the Vesuvius
volcano (Figure 1) – which mantle the limestone and tuffaceous slopes over an area
of about 3000 square kilometres.
In the upper area, there are more than 200 towns frequently suffering these flow
type phenomena, as pointed out by historical data acquired over a time interval from
the 16th century up to the present day (O.U. 2.38, 1998b; Cascini, 2002; Cascini
et al., 2002). The most recent and calamitous events occurred in May 1998, causing
160 casualties and seriously damaging four small towns (Bracigliano, Quindici,
Sarno and Siano) located at the toe of the Pizzo d’Alvano relief (Figure 2).
After the latest events, the National Italian Department of Civil Defence Pro-
tection entrusted the task of scientific management for flowslide risk mitigation to
the University of Salerno. Among the many activities developed (O.U., 1998a;
Cascini et al., 2000), a considerable effort was channelled into systematically fur-
thering knowledge of geological and triggering factors responsible for landslide
occurrence.
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 367

Figure 2. Geomorphological features of Pizzo d’Alvano Massif with landslides occurred in May 1998.

In particular, thorough geological, geomorphological and hydrogeological inves-


tigations were conducted throughout the area shown in Figure 2 and at different
scales (1:25,000 and 1:5000), while accurate geotechnical investigations were carried
out inside sample areas at a more detailed scale (1:2000). The geotechnical investi-
gations aimed to acquire data on the in situ soil suction regimen, as well as on the
mechanical properties of pyroclastic soils in both saturated and unsaturated con-
ditions. After a brief description of the landslide events of May 1998 and the pre-
liminary efforts devoted to modelling them, the present paper summarises the results
acquired through laboratory investigations and discusses their usefulness in the
triggering analysis of such landslides.

2. Geological and Geotechnical Problems


Within an interval of about 10 h on 5th–6th May 1998, landslide phenomena
occurred in almost all the basins of the Pizzo d’Alvano relief (Figure 2), with
368 EDUARDO BILOTTA ET AL.

triggering areas located primarily in the upper parts of the basins where the average
slope angle varied from 35 to 41. Pyroclastic soil thickness in the triggering areas
ranged from 0.5 to 5.0 m.
In the majority of the triggering areas, there were multiple failures which pro-
gressively involved ample portions of the slopes, following time sequences that are
not easy to reconstruct. The mobilised soil masses rapidly travelled down the slope
and increased their initial volume through the mobilisation and/or erosion of further
in-place soil. The total volume of the landmass which invaded the downstream areas
was estimated at about 3 million cubic meters. According to a back analysis of the
damage sustained by buildings (Faella and Nigro, 2001), the estimated velocities in
the urbanised areas ranged from 1 to 20 m/s.
In general, the May 1998 landslides along the Pizzo d’Alvano slopes can be defined
as complex landslides. According to the classification proposed by Hungr et al.
(2001), most of these landslides showed characteristics that embrace, at least, three
rapid flow type movements: flowslides, debris flows and debris avalanches. The
flowslide characteristic can certainly be detected when considering the early stages of
post-failure movements in the landslide triggering areas: here, the loose, metastable
structure of pyroclastic soils (Olivares and Picarelli, 2001; Sorbino and Foresta,
2002) collapsed just after the initial failure stages, thus inducing the build up of high
pore water pressures and the generation of liquefaction phenomena (Eckersley, 1990;
Olivares and Picarelli, 2001; Wang and Sassa, 2001). The debris flow and/or debris
avalanche characteristics emerged when examining the post-failure evolution stages.
At these stages movements developed either along open slope (debris avalanche type
movement), inside drainage channels of various orders, as well as along established
gullies that controlled the direction of the flow (debris flow movement type) downhill
to urbanised areas. Although the above three flow type movements can be detected
for the May 1998 landslides, it must be noted that the presence of liquefaction
phenomena in the triggering areas can be considered a common factor affecting the
initial failure stages of such landslides. For this reason, and also according to Hungr
et al. (2001), landslides of the Pizzo d’Alvano massif will be considered as flowslides
from now on in this paper, in order to stress the importance of the liquefaction
processes in the landslide occurrences.
With reference to the geological features of the Pizzo d’Alvano massif, as well as
the geotechnical properties of pyroclastic soils, it must be underlined that only very
poor quality data was available at the time of last flowslide events, even though these
phenomena had occurred frequently over the centuries. Indeed, the only geological
data was reported in the ‘Italian Geological Map’ (scale 1:100,000), while geomor-
phological and hydrogeological studies for the area shown in Figure 2 were only
related to the water-supply potentiality assessment of the limestone bedrock (Civita
et al., 1975; Celico and De Riso, 1978; Cinque et al., 1987; Cascini and Cascini,
1994).
As for the geotechnical aspects, no references on flowslide characterization were
available and only a few studies were dedicated to the mechanical properties of soils
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 369

having an analogous origin, but located in the areas around the city of Naples
(Pellegrino, 1967; Ciollaro and Romano, 1995; Esposito and Guadagno, 1998;
Nicotera, 1998).
For the above reasons, preliminary investigations were addressed to acquiring
information on the geological and geomorphological features of the Pizzo d’Alvano
massif. With reference to the geological characteristics, the investigations and studies
carried out, together with the information available in the literature, showed that the
massif consists of a sequence – up to several hundred meters thick – of limestones,
dolomitic limestones and, subordinately, marly limestone rocks, dating back from
the Lower to the Upper Cretaceous Age.
Pyroclastic soils from Somma-Vesuvius volcanic activities diffusely cover the
limestone slopes, both as primary air-fall deposits and re-worked deposits (vulcan-
oclastic deposits). Air-fall deposits generally have a small thickness, ranging from
several decimetres up to a maximum of 6–7 m. Stratigraphic conditions are almost
everywhere characterized by alternated layers of pumiceous soils and ashy soils,
sometimes with the presence of paleosoil horizons. On average the re-worked
deposits have a greater thickness (up to 20 m) than air-fall deposits, and they consist
of debris and colluvium mainly located in the morphological concavities, in the
karstic depressions and at the toe of the valleys (Del Prete et al., 1998; Cascini et al.,
2000; Fiorillo et al., 2001).
As for geomorphological features, the Pizzo d’Alvano massif is a NW–SE oriented
morphological structure, with gently sloping summit plains and relatively smooth
summit ridges. The border slopes are the product of long-lasting modelling events
related to displacements of border faults caused by parallel linear recession mech-
anisms and subsequently modified by linear erosion (Cinque et al., 1987).
The slopes’ outline is influenced by structural factors (minor faults and jointing) as
well as by morphological frames derived from limestone layers more resistant to
erosion.
Above the morphological frames, the paleo-drainage network of the limestone
slopes are mainly filled with air-fall deposits or with debris colluvial material from
upstream slopes, thus forming the so-called ‘Zero Order Basins’ (ZOB) (Cascini
et al., 2000), having a slope angle varying from 35 to 41. Alluvial fans of various
ages are located at the toe of the valleys, highlighting the systematic occurrence of
depositional events from flow type movements, superimposed onto older debris
deposits.
By comparing the maps produced according to the acquired geological and geo-
morphological data with slope portions affected by the May 1998 flowslides (Fig-
ure 2), it was observed that the majority of triggering zones are located inside the
ZOB areas and that the final destination of the mobilised soil masses was in the
alluvial fans areas described above. On the basis of these results, additional in situ
investigations and studies were devoted to the hydrogeological aspects of the Pizzo
d’Alvano massif, with the aim of setting up a slope evolution model for the pyro-
clastic covers.
370 EDUARDO BILOTTA ET AL.

Figure 3. Frame of main structural and hydrogeological elements (from Cascini et al., 2000, modified).

In this respect, previous studies (Civita et al., 1975; Celico et al., 1987; Del Prete
et al., 1998) indicated the presence of perched water tables, which characterize the
subsurface flow system in the upper part of the massif. Although detailed in situ
investigations confirmed this hydrogeological framework, they also pointed out
some additional features of the subsurface flow regimen, which is strictly linked to
the arrangement of the main and secondary structural lineaments, representing
relative impermeable layers (Cascini et al., 2000). The structural, wedge-shaped
lineaments are closely associated to the seasonal and temporary springs (or outlets)
located in the upper parts of the slopes (Figure 3). Temporary springs are mainly
located inside the ZOB, and at the top of the filled main catchment basins
(Figure 4(a) and (b)). Seasonal springs can even be observed at the toe of
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 371

Figure 4. (a) Part of hydrogeological map, (b) outlets flow rate versus time, (c) preliminary hydrogeo-
morphological model and (d) typical location of bedrock outlets (from Cascini and Sorbino, 2002,
modified).

morphological discontinuities (Figure 4(d)), in correspondence of similar, although


upper ordered, hydrogeological structures.
The acquisition of all the above information made it possible to set up a slope
evolution model (Cascini et al., 2000) which can provide a geological interpretation
of the past and more recent flowslides that have occurred on the Pizzo d’Alvano
massif. This model also stresses the important role played by the temporary springs
in triggering flowslides (Cascini et al., 2000) (Figure 4(c)).
In order to evaluate the model’s reliability, a preliminary back-analysis was per-
formed for the failure conditions of a huge flowslide that occurred in May 1998
(Cascini et al., 2000) (Figure 5), with the support of poor laboratory data on the
mechanical properties of pyroclastic soils.
372 EDUARDO BILOTTA ET AL.

Figure 5. Daily rainfall, stratigraphic section, initial and boundary conditions in the flow transient
analysis and modelled time sequel of the slope failure (from Cascini et al., 2000, modified).

Despite the above limitations, the results obtained were particularly encouraging
as they were able to simulate, for a given set of mechanical properties, a time
sequence of flowslide triggering which was in a good agreement with that observed
by eye-witnesses at the time of failure (Figure 5). Moreover, the performed analysis
confirmed the relevant role played by bedrock temporary springs and highlighted the
presence of negative pore pressures in many zones of the pyroclastic covers just
before the failure occurrence.
On the basis of the above results, it was considered worthwhile to perform further
in situ and laboratory investigations in order to acquire detailed data on: strati-
graphic conditions inside and outside the ZOB areas; negative pore pressure regimen
during dry and wet seasons; and the mechanical properties of pyroclastic soils, both
in saturated and unsaturated conditions.
With reference to the first topic, seismic refraction prospects were performed along
with more than 300 shallow pits. The overall collected data revealed that the
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 373

Figure 6. Typical stratigraphic conditions in (a) slope sector and (b) alluvial fans sector of the Pizzo
d’Alvano Massif (from Cascini et al., 2000, modified).

thickness and stratigraphic conditions of pyroclastic covers are highly variable due
to their arrangement (air-fall or re-worked deposits), the different exposure of the
slopes towards the Somma-Vesuvius volcano, and the different ripeness of the
geomorphological slopes; however, on average, pyroclastic covers can be considered
as being composed of the following lithological items (Figure 6):
– layers of undisturbed or pedogenetisized ashy soils, constituted by a silty sand
with a negligible fraction of small pumices and volcanic scoriae;
– pumice layers related to the several eruptive ages of the Somma-Vesuvius vol-
cano (8000–3800 years ago);
374 EDUARDO BILOTTA ET AL.

Figure 7. Daily rainfall and suction data collected at the investigation sites.

– layers constituted by ashy soils, pumices and volcanic scoriae, re-worked and
mixed by means of colluvial processes or landsliding occurrences;
– ashy soils highly pedogenetisized, with the presence of small roots and organic
materials.
With reference to negative pore pressure regimen inside the pyroclastic covers,
measurements were performed in the areas shown in Figure 2, by using portable
tensiometers (Quick-Draw tensiometers) and Jetfill in-place tensiometers (Cascini
and Sorbino, 2002) which make it possible to measure soil matric suction. Mea-
surements started on November 1999 and a total number of about 2900 have been
taken so far. The depths investigated by means of the Quick Draw tensiometers
ranged between 0.2 and 1.6 m, while the tip installation depths for Jetfill tensiom-
eters varied between 1.5 and 4.0 m from the ground surface.
In Figure 7 the overall suction data are plotted against time for the period
November 1999 – April 2002, together with the daily rainfall recorded in the same
period. As can be seen, the suction data show significant scattering which is, of
course, related to the different sites and depths where measurements were performed.
In spite of the dispersion and the different rainfall pattern characterizing the entire
measurement period, the data highlight that suction values range from a maximum
of 65 kPa to a minimum of 1.0–2.0 kPa in each site and at any depth. The data
scattering almost disappears if the monthly average suction values collected at the
same depth are considered. In this case (Figure 8), monthly average suctions seem to
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 375

Figure 8. Daily rainfall and monthly average suction values (a) at depths lower than 1.0 m and (b) at
depths ranging from 1.0 to 1.8 m.

show the same behaviour on a yearly scale, regardless of the measurement site. In
particular, at depths not exceeding 1.0 m from the ground surface, they are char-
acterized by a similar time trend strictly related to rainfall regimen, with maximum
values (35 kPa) attained in summers and minimum values reached (5 kPa) in win-
ters. At higher depths, suction values present an analogous time trend but show two
drying periods with a maximum attained in summer (37 kPa) and at the end of
autumn (30 kPa).
In conclusion, suction measurements reveal persistent unsaturated conditions in-
side the pyroclastic covers throughout the measurement period of about three years.
Therefore, it cannot be excluded that such conditions did characterize large portions
of the up-slope covers during the May 1998 events. This aspect is highlighted by the
preliminary geotechnical model (Figure 5), as well as by further analysis based on
more detailed stratigraphic conditions deriving from further in situ investigations.
376 EDUARDO BILOTTA ET AL.

Figure 9. Location map of the three investigated sites.

In order to improve the modelling of the rainfall–pore pressure relationships and


consequently to perform more accurate stability analyses, laboratory investigations
on mechanical properties were carried out so as to characterise pyroclastic soils in
both saturated and unsaturated conditions.
The experimental programme and the interpretation of the obtained results, per-
formed with the aid of models available in the literature, will be discussed in the next
section.

3. Laboratory Investigations
In order to perform laboratory tests, soil samples were collected at several dates after
May 1998 events, inside the triggering areas of three huge flowslides. The three
investigation sites (hereafter referred to as sites 1, 2 and 3) belong to Pizzo d’Alvano
slopes facing Sarno town (Figure 9). In ashy layers, undisturbed samples were ob-
tained from excavated pits by using cylindrical samplers 0.1 m wide and 0.3 m long,
driven by hand or by hydraulic jack. As it regards pumice layers, only partially
disturbed soil samples could be collected.
Physical properties were determined for all the samples while mechanical properties
were investigated only for the ashy soils (Table 1). As far as the physical properties
are concerned, tests for grain size distribution, Atterberg limits and index properties
were carried out. Mechanical characterization has dealt with hydraulic properties and
Table 1. Number of the performed laboratory tests
Physical properties Strength properties Hydraulic properties
Grain size Atterberg limit Direct shear NCTX VE PP SCO
distribution

Undisturbed Re- Undis- Undisturbed


moul- turbed
ded

Site Date Sieve Sediment WL wp Gs na w Sr c cd wim wn wpp wim

2 Jun 1998 42 9 10 1 16 29 29 29 29 29 15 3 14
Jan 2000 11 6 11 11 11 11 11 5

Jul 1998 51 13 11 7 11 44 44 44 44 44 19 8 8
Apr 1999 70 2 12 44 44 44 44 44 14 18 8 21
3 Oct 1999 3 3 3 3 3 3 3
Apr 2000 29 13 6 16 28 28 28 28 28 6 3 3 4
Mar 2001 2 12 12 12 12 12 12
Mar 2002 12 2 12 12 12 12 12 12 3
1 May 1999 40 6 18 16 40 40 40 40 40 22 18
Sep 1999 32 6 20 20 20 20 20 17 3
Oct 1999 29 2 6 29 29 29 29 29 21 8
Nov 1999 11 2 3 11 11 11 11 11 11
Dec 2001 14 4 14 14 14 14 14 2 6 6 3
Jan 2001 3 1 3 3 3 3 3 3 2 1
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS

Feb 2001 8 4 8 8 8 8 8 5 2 1
Feb 2002 15 4 15 15 15 15 15 6 6 3

wim – submerged in distilled water; VE – volumetric extractor; SCO – suction controlled oedometer; wn – natural water content; NCTX – suction controlled
triaxial; PP – pressure plate; wpp – imposed water content.
377
378 EDUARDO BILOTTA ET AL.

shear strength, in both saturated and unsaturated conditions. Hydraulic properties in


saturated conditions were investigated by means of permeameter tests and using three
different laboratory instruments, namely suction controlled oedometer (SCO), vol-
umetric pressure plate extractor (VE) and Richards’ pressure plate (PP) for unsatu-
rated conditions. Finally, shear strength has been evaluated by means of conventional
direct shear tests and suction controlled triaxial tests.
For all these tests, the experimental procedures and the obtained results will be
presented in the following sections.

3.1. PHYSICAL PROPERTIES

With reference to plasticity properties, it must be noted that all the tested samples
can be essentially classified as non-plastic or slightly plastic soils (Bilotta and For-
esta, 2001), according to Picarelli and Olivares (2001) who examined pyroclastic soils
of analogous origin, collected in other sites of the Campania Region. Low plastic
properties have been highlighted by the particularly laborious determination of the
liquid limit, for which results could only be achieved roughly by using Casagrande’s
cup (Table 1); determination of the plastic limit could also be performed in only a
very limited number of cases and only for those specimens having organic material
or a small clay content.
Grain size distributions for ashy soils were mainly performed on undisturbed
samples used for mechanical tests. The overall grain size distributions of both ashy
soils and pumices are reported in Figure 10, which shows a high variability of the
grading curves for all the soils located in the investigated sites. Although of minor
extent, the above variability can also be detected with reference to ashy soil layers
(Figure 11), which have a sandy component ranging between 40% and 60% and a
silty component varying from 20% to 60%; ashy soil composition also includes some
gravel, due to the presence of small isolated pumices, but always below 22%.
Further elements assessing the variability of the investigated soils are provided by
other physical properties (Table 2), i.e. the specific gravity Gs, the dry unit weight cd,

Figure 10. Ranges of the grain size distribution for ashy soils and pumices.
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 379

Figure 11. Ashy soils: range of the grain size distribution.

Table 2. Range of main physical properties


GS na c (kN/m3) cd (kN/m3)
Soil class Min Max Min Max Min Max Min Max

A 2.33 2.59 0.63 0.74 8.88 14.40 6.58 9.50


B 2.45 2.70 0.53 0.69 9.20 16.59 5.71 12.32

the total unit weight c, and the porosity na, where na refers to inter-particle porosity
and represents the free voids between particles (Pellegrino, 1967; Wesley, 2001).
An analysis performed on the above physical properties enables a close compar-
ison of the values obtained for homogeneous ashy soil layers (named as a, … , f)
recovered in the three investigation sites. In this respect, Figure 12 reports the per-
centage of ashy soil samples having values of na, cd and Gs lower than those reported
in abscissa.
By comparing the ‘physical properties distributions’ thus obtained and the grading
curves, it was possible to classify ashy soils into two homogeneous classes, named ‘A’
and ‘B’ (Bilotta and Foresta, 2001). As can be noted, class ‘A’ includes ashy soils
with a finer grain size distribution (Figure 11), higher porosity values and lower
specific gravity values than those belonging to class ‘B’ (Figure 12).
With reference to the average values of physical properties (Table 3), it can be
observed that specific gravity is always quite small because of the presence of small
internal voids within the single soil particles (Esposito and Guadagno, 1998; Wesley,
2001). This characteristic, together with the high porosity values and the low degree
of saturation values, makes this soil very light, as testified by the values of the total
unit weight (Table 3).
The overall results obtained for the physical properties allow some considerations
to be made both on the triggering phase and on the post-failure stage, when the
mobilised soil evolves into a flowslide.
As far as the triggering stage is concerned, some stability analyses reveal a minor
influence on the safety factor induced by the increase in total unit weight during
380 EDUARDO BILOTTA ET AL.

Figure 12. Percentage of ashy specimens whose physical properties have values lower than those reported
in abscissa.

Table 3. Average values of the physical properties


Soil class Gs na c (kN/m3) cd (kN/m3) W (%) Sr (%) WL (%) WP (%)
A 2.51 0.70 11.4 7.5 51.9 54.7 53.8 49.3
B 2.63 0.64 12.7 9.4 36.8 52.1 40.6 –

rainfall, if compared to the role of suction reduction and bedrock outlets. On the
contrary, the grain size and the high porosity of the ashy soils could induce positive
generation of pore pressures in the post-failure stages, with a progressive loss of
strength. This behaviour has been quoted by some Authors (Olivares and Picarelli,
2001) when discussing the results obtained in undrained triaxial tests aiming to assess
the occurrence of static liquefaction (Sassa, 2000) as a relevant mechanism for the
modelling of post-failure movements.
3.2. HYDRAULIC PROPERTIES

The hydraulic properties of ashy soils were investigated by testing undisturbed


specimens (Sorbino and Foresta, 2002) collected from sites 1 and 3 (Figure 9).
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 381

In saturated conditions the hydraulic conductivity, estimated by means of per-


meameter tests, was found to range from a minimum of 5.0 · 10)6 m/s to a maxi-
mum of 4.8 · 10)5 m/s.
As for the unsaturated conditions, determination of the hydraulic properties of
course requires knowledge of the relationships between stress state variables and
both the hydraulic conductivity k and the volumetric water content h (i.e. the
product of porosity na and the degree of saturation Sr). The stress state variables are
identified by the net stress (r ) ua) and the matric suction (ua ) uw), where r is the
total stress, ua the pore air pressure and uw the pore water pressure (Matyas and
Radhakrishna, 1968; Fredlund and Rahardjo, 1993). However, it must be noted that
unsaturated soil in the field, undergoing processes of drying and wetting due to
climatic changes, is often subjected to more frequent and significant changes in
suction than in net stress. In this case, an appropriate estimate of the volumetric
water content and hydraulic conductivity can be obtained by considering both soil-
water retention and hydraulic conductivity relationships for a constant value of the
net stress applied (Vanapalli et al., 1996).
Preliminary results on these relationships were obtained by Sorbino and Foresta
(2002) by means of SCO tests, VE tests and PP tests performed on specimens whose
physical properties are shown in Table 4. In particular, SCO tests (Table 4) were
carried out at vertical net stresses ranging from 10 to 100 kPa, with the lowest value
corresponding to the minimum allowed by the adopted apparatus and the highest
representing about twice the maximum in situ vertical stress.
Sorbino and Foresta (2002) did not indicate any influence of net vertical stress on
the unsaturated hydraulic properties inside the 0–20 kPa range. In addition, within
this interval the relationships between suction and saturation degree did not show
any significant hysteretic behaviour, in agreement both with the absence of collapse
phenomena and the incompressible behaviour shown by the soil skeleton. Further-
more, for these net vertical stress values the authors underlined the importance of

Table 4. Testing conditions, initial physical properties and saturated hydraulic conductivity (SCO and VE
tests) (from Sorbino and Foresta, 2002, modified)
Site Soil class Specimen rv ) ua (kPa) Wi (%) cdi (kN/m3) nai Sri (%) GS kS (m/s)
3 A SCO_4[10] 10 52.2 7.1 0.72 52.8
SCO_2[20] 20 47.5 7.3 0.70 50.7 2.50 5.0E)06
SCO_3[50] 50 55.4 6.4 0.72 48.7
SCO_4[100] 100 85.9 7.3 0.71 89.8
VE_1 0 47.8 7.3 0.70 50.7
VE_2 0 56.2 6.8 0.72 53.9 2.50 5.0E)06
VE_3 0 56.1 6.9 0.72 54.8

1 B SCO_1[50] 50 15.9 9.5 0.63 24.3 2.62 4.8E)05


SCO_5[20] 20 42.9 8.4 0.68 52.9 2.66
VE_4 0 34.5 8.1 0.68 41.7 2.62 4.8E)05
VE_5 0 31.1 9.4 0.65 46.3
VE_8 0 33.3 9.3 0.65 48.9 2.62 1.7E)05
382 EDUARDO BILOTTA ET AL.

Table 5. Initial physical properties of specimens in VE tests


Site Soil Specimen rv ) ua (kPa) Wi (%) cdi (kN/m3) nai Sri (%) Gs
class
3 A VE_14 0 35.5 7.1 0.71 35.4 2.47
1 B VE_9 0 28.1 9.6 0.64 42.7 2.65
VE_10 0 29.8 10.5 0.62 50.6 2.76
VE_11 0 42.7 8.4 0.68 52.8 2.65
VE_12 0 41.3 8.6 0.68 52.6 2.65
VE_13 0 27.8 12.5 0.51 68.3 2.56
VE_15 0 22.3 10.6 0.60 39.5 2.65
VE_16 0 48.9 7.2 0.73 48.8 2.64

considering the soil–water characteristic curves, instead of the degree of saturation–


suction relationships, resorting to a single soil–water characteristic curve regardless
of the particular wetting and drying path followed.
With reference to net vertical stress values greater than 20 kPa, experimental re-
sults for class ‘A’ soil pointed out the influence of the latter variable, with a pro-
gressive flattening of the soil–water characteristic curve and the presence of
hysteretic phenomena, associated to a collapse behaviour during the wetting path.
On the basis of the above preliminary results and in order to acquire further
information on the unsaturated hydraulic properties for both classes of soils, the
experimental programme has been integrated with a series of VE tests mainly per-
formed on class ‘B’ specimens, whose physical properties are listed in Table 5.
Table 6 shows the suction paths followed in the tests, together with the maximum
and minimum suction variations applied.
According to Sorbino and Foresta (2002) and to other researchers dealing with
pyroclastic soils of analogous origin (Ciollaro and Romano, 1995; Nicotera, 1998),
the experimental volumetric water content versus suction data was modelled by the
non-hysteretic equation suggested by van Genuchten (1980), which can be written:
hs  hr
hðsÞ ¼ hr þ m ð1Þ
½1 þ ðasÞn 
where hs and hr are the saturated and residual volumetric water content respectively,
a is an empirical parameter whose reciprocal value can be assumed as the soil’s air
entry value (aev), s ¼ (ua)uw) is the suction, n is a fitting constant reflecting the slope
of the retention curve, and m is a parameter linked to n by means of the equation

Table 6. Suction paths in VE tests


Soil class No. of specimens Suction paths (kPa) |D (ua ) uw)min|kPa) |D (ua ) uw)max|(kPa)
A 1 0.1 fi 160 2 60
B 2 0.1 fi 160 fi 0.1 2 60
5 0.1 fi 160 2 60
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 383

Figure 13. Soil water characteristic curves.

m ¼ 1)1/n. The parameters were estimated by using the RETC code (van Genuchten
et al., 1991).
The best fit curves, obtained by means of van Genuchten’s model, are plotted in
Figure 13 in terms of the degree of saturation versus suction, while Table 7 shows
the values of the estimated model parameters (i.e. hs, hr, a and n). As can be noted,
specimens belonging to class ‘A’ are characterized by air entry values and a volu-
metric water content higher than those belonging to class ‘B’, thus reflecting the
different grain size distribution and different pore shapes of the two soil classes
(Corey, 1994).

Table 7. Estimated van Genuchten’s model parameters


Site Soil class Specimen hws hwr a n m R2 a.e.v. k0 (m/s)
(kPa)1) (kPa)
3 A SCO_4[10] 0.68 5.6E)07 0.31 1.23 0.19 0.981 3.22 1.0E)05
SCO_2[20] 0.66 4.3E)07 0.19 1.28 0.22 0.984 5.33 5.4E)06
SCO_3D[50] 0.70 1.0E)01 0.18 1.34 0.25 0.987 5.44 9.8E)06
SCO_3W[50] 0.65 1.2E)01 0.21 1.36 0.26 0.999 4.76 2.8E)06
SCO_4[100] 0.63 2.8E)01 0.07 1.79 0.44 0.988 13.87 6.9E)07
VE_1 0.68 0.00 0.23 1.22 0.18 0.991 4.39
VE_2 0.68 0.00 0.51 1.18 0.15 0.968 1.96
VE_3 0.67 0.00 0.87 1.18 0.15 0.975 1.16
VE_14 0.69 0.00 1.02 1.19 0.16 0.991 0.98
1 B SCO_1[50] 0.63 5.0E)02 1.27 1.31 0.24 0.998 0.79 7.6E)04
SCO_5[20] 0.68 4.8E)03 2.27 1.17 0.15 0.999 0.44
VE_4 0.68 0.00 24.58 2.16 0.07 0.995 0.04
VE_5 0.65 0.00 7.14 1.19 0.16 0.997 0.14
VE_8 0.65 1.2E)01 1.31 1.39 0.28 0.999 0.76
VE_9 0.64 0.00 23.84 1.16 0.14 0.999 0.04
VE_10 0.61 0.00 14.75 1.15 0.13 0.989 0.07
VE_11 0.66 0.00 3.79 1.19 0.16 0.991 0.26
VE_12 0.67 6.7E)02 1.42 1.29 0.23 0.999 0.70
VE_13 0.51 0.00 1.54 1.19 0.16 0.979 0.65
VE_15 0.59 0.00 6.23 1.19 0.16 0.993 0.16
VE_16 0.72 0.00 1.58 1.20 0.17 0.982 0.63
384 EDUARDO BILOTTA ET AL.

The three pieces of equipment used do not make it possible to measure any direct
hydraulic conductivity of the specimens at the various suction values applied. To this
end, the unsaturated hydraulic conductivity was obtained by following the meth-
odology proposed by Sorbino (1994, 1995), which entails the numerical integration
of the equations governing the transient unsaturated water flow triggered by a
variation in applied suction.
With reference to the influence of the soil–water characteristic curves on the
rainfall–pore water pressure modelling, it must be observed that net vertical stress
values ranging from 0 to 20 kPa are of particular interest because they represent the
values frequently attained in the triggering zone of flowslides, where the sliding
surface is usually located at depths not exceeding 3.0 m. For these net vertical stress
values, it seems possible to resort to a single soil water characteristic curve regardless
of the particular wetting and drying path followed. If this behaviour is confirmed by
further experimental results, then hydraulic unsaturated properties within the above
net vertical stress range may be adequately determined by using equipment for VE
tests and PP tests which are also rapid means of testing.

3.3. SHEAR STRENGTH

3.3.1. Conventional direct shear tests


As observed in the previous sections, ashy soils in the triggering areas are charac-
terized by a marked variability in both physical and hydraulic properties. The pre-
liminary stages of the research programme aiming to characterize shear strength in
saturated and unsaturated conditions were therefore devoted to the conventional
direct shear tests (DS tests). These are less time consuming than the other techniques
usually employed for determining soil shear strength. This made it possible to ac-
quire a considerable data set within a reasonable time frame, and these data laid
down guidelines for the subsequent experimental programme on the unsaturated
shear strength, to be performed using non conventional equipment.
DS tests were carried out on undisturbed ashy soil specimens from the three sites in
Figure 9 and belonging to classes ‘A’ and ‘B’, as previously defined. In order to
estimate the influence of the in situ colluvial processes on soil shear strength, DS tests
on remoulded specimens were also performed. In particular, remoulded specimens
were obtained from a soil sample mechanically mixed with distilled water, in order to
obtain a water content of about 1.5 times the liquid limit (wL); the sample was then
consolidated in a large oedometer (250 mm diameter), under an effective vertical
stress of 10 kPa.
All the tests were performed at a strain rate of 28 mm/day, on specimens mea-
suring 60 · 60 · 20 mm3, for net vertical stresses ranging from 10 to 106 kPa. The
former stress value corresponds to the minimum allowed by the adopted direct shear
apparatus, while the latter represents about twice the maximum in situ vertical stress.
As shown in Table 1, three different initial conditions were considered, i.e. satu-
rated condition (wim), natural water content (wn) and water content (wpp) obtained in
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 385

the Richard’s pressure plate apparatus for suction values of 50, 100 and 200 kPa.
Saturated conditions were obtained by submerging specimens in distilled water at the
beginning of the consolidation stage and throughout the shearing stage. As far as wn
and wpp tests are concerned, specimens were tested without supplying any water and
packing the shear box with a cellophane film in order to avoid evaporation phe-
nomena (Bilotta and Foresta, 2001).
Typical results of DS tests performed on specimens belonging to both classes ‘A’
and ‘B’ are shown in Figures 14–16 in terms of shear stress–displacement (s–dh) and

Figure 14. Typical results of DS tests, at wim condition, for undisturbed specimens.
386 EDUARDO BILOTTA ET AL.

Figure 15. Typical results of DS tests, at wn condition, for undisturbed specimens.

volume change–displacement (dv–dh). A comparison of results from wim tests (Fig-


ure 14) and wn tests (Figure 15) shows a dilatant behaviour for both ‘A’ and ‘B’
classes only for wn tests carried out at low vertical stress levels, i.e. not exceed-
ing 30 kPa. As for the shear stress-horizontal displacement curves, a strain–hard-
ening behaviour was systematically detected for vertical stress values higher than
30 kPa.
Remoulded specimens (Figure 16), on the contrary, show a strain-softening
behaviour, almost regardless of the vertical stress applied, resulting in peak shear
strength values higher than those attained by the undisturbed specimens.
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 387

Figure 16. Typical results of DS tests, at wim condition, for remoulded specimens belonging to class ‘A.

With reference to the shear strength parameters, Figure 17(a) shows the Mohr–
Coulomb peak envelope for saturated specimens (wim), which clearly points out the
differences between the ‘A’ and ‘B’ classes. The first class has an effective friction
angle u¢ ranging between 30 and 35; the second is characterized by higher strengths
(36 £ u¢ £ 41). Both classes always have a very low effective cohesion intercept,
with values not exceeding 3 kPa.
In order to investigate the relationships between saturated shear strength and
physical soil properties, Bilotta and Foresta (2002) compared the effective friction
angles with the mean values attained by the porosity and the dry unit weight. This
showed (Figure 17(b)) that the experimental data are mainly located in a narrow
band, with friction angles increasing as cd increases and na decreases. The location of
the points in Figure 17(b) not belonging to the above band can probably be ex-
plained by considering that the lowest point (1 in Figure 17(b)) is associated with a
specimen having the highest clay fraction percentage, while points 8 and 9 have the
highest gravel fraction percentage.
Another factor clearly influencing the saturated shear strength is shown in
Figure 18, which compares the Mohr–Coulomb failure envelopes of undisturbed and
388 EDUARDO BILOTTA ET AL.

Figure 17. (a) Saturated shear strength envelopes and (b) effective friction angle versus physical properties
(from Bilotta and Foresta, 2002, modified).

remoulded specimens belonging to class ‘A’. As can be noted, the shear strength of
remoulded specimens is higher than that observed for undisturbed ones, with
effective friction angles of about 41–42 and an intercept effective cohesion of 4 –
5 kPa. The considerable differences in shear strength are certainly related to the soil
porosity decrease induced by remoulding.
As for DS tests at constant water content (wn and wpp tests), it must be noted that
the above mentioned experimental procedure was particularly satisfactory since the
maximum percentage difference between water contents measured at the beginning
and at the end of the tests was always lower than 1%.
As far as the experimental results are concerned, Figure 19 shows the best linear
envelopes of the peak shear strengths (sf) for some values of the saturation degree
(Srf) attained at the peak shear strength. For specimens not showing a well-defined
peak value in the experimental (s–dh) curve (strain-hardening behaviour), peak shear
strength values were estimated according to the procedure suggested by Fredlund
and Rahardjo (1993) and also adopted by de Campos and Carrillo (1995).
With reference to class ‘A’ soils for all applied net vertical stresses, Figure 19(a)
shows a clear increase in shear strength as the degree of saturation decreases up to
values of 50–60%.
Whereas, for class ‘B’ soils (Figure 19(b)) the available data do not show an easily
recognisable dependence on the degree of saturation, probably because of the high
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 389

Figure 18. Comparison among the shear strength envelopes obtained from DS tests on remoulded and
undisturbed specimens for soils of class ‘A’ (from Bilotta and Foresta, 2002, modified).

heterogeneity of these soils. In any case, the contribution of the degree of saturation
to the shear strength is minor for soils belonging to class ‘A’, especially at high values
of net vertical stress.
For the latter soils, the well-defined trend shown in Figure 19(a) allows the rela-
tionships between shear stress, net vertical stress and matric suction to be investi-
gated. In this respect, it was considered appropriate to first investigate specimens
obtained from site 3 which, for class ‘A’ soils, are those characterised by the smallest
variability in physical properties. For these specimens, the peak shear strength
against net vertical stress, at various saturation degrees, is shown in Figure 20. The
effective friction angle and the effective cohesion are, respectively, equal to 30 and to
3 kPa.
In order to estimate the values of matric suction for the data plotted in Figure 20,
the following procedure was adopted: for each value of net vertical stress applied, the
corresponding soil water characteristic curve was selected; the curve was then used to
estimate matric suction on the basis of the measured values of the degree of satu-
ration at peak strength (Srf). The obtained data for net vertical stresses of 10, 20, 50
and 100 kPa, were then plotted in the shear stress–suction plane of Figure 21. An
analysis of such data will be presented later, after a brief description of some
interpretative models available in the literature.
390 EDUARDO BILOTTA ET AL.

Figure 19. Peak shear strength versus net vertical stress for some values range of the saturation degree
(Srf) for (a) soils of class ‘A’ and (b) soils of class ‘B’.

3.3.2. Interpretative models for unsaturated shear strength


The data in Figure 21 suggest a non-linear failure envelope with respect to matric
suction. In literature, the non-linearity of such an envelope is testified by several
researchers (Escario and Saez, 1986; Gan et al., 1988; de Campos and Carrillo, 1995;
Oloo and Fredlund, 1996; Vanapalli et al., 1996; Oberg and Sallfors, 1997; Rassam
and Williams, 1999; Vanapalli et al., 1999).
The current explanation of such non-linear behaviour is normally attributed to a
fall in matric suction contributing to shear strength, as residual soil water content is
approached (Oloo and Fredlund, 1996; Vanapalli et al., 1999). Consequently, for a
given soil, the angle of shearing resistance with respect to matric suction (ub) is
practically equal to the effective friction angle u¢ for matric suction values varying
from zero up to the air entry value.
Experimental results published in the literature seem to confirm this general
behaviour with reference to the suction interval (0–500 kPa) currently encountered
in soil engineering practice (Fredlund and Rahardjo, 1993). An almost linear
envelope of unsaturated shear strength for soils characterised by high values of a.e.v.
(e.g. clayey soils) (Rahardjo et al., 1995) and a clearly defined non-linear envelope for
coarser grained soils (Gan and Fredlund, 1996) have also been highlighted.
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 391

Figure 20. Class ‘A’ soil coming from site 3: (a) Peak shear strength versus net vertical stress for some
values range of the saturation degree (Srf) and (b) grain size distribution.

The most popular relationship describing the shear strength of an unsaturated soil
is the one proposed by Fredlund et al. (1978), who extends the Mohr–Coulomb
failure criteria for saturated soils with the equation:

s ¼ c0 þ ðrn  ua Þ tan u0 þ ðua  uw Þ tan ub ð2Þ

Figure 21. DS tests: experimental peak shear strength versus matric suction for soils of site 3 belonging to
class ‘A’.
392 EDUARDO BILOTTA ET AL.

where s is the shear strength, c¢ is the effective cohesion, (rn)ua) is the net normal
stress at failure on the plane of failure, (ua ) uw) is the matric suction and ub is the
above mentioned angle of shearing resistance with respect to matric suction.
Equation 2 can be used for both the linear failure envelope (i.e. ub equal to a
constant value) and the non-linear envelope, assuming ub as a function of matric
suction.
In order to assess the variation of ub with respect to matric suction, Vanapalli et al.
(1996) presented a model in which this variation is linked to the soil water charac-
teristic curve. In particular, they assumed:

tan ub ¼ ðHÞk tan u0 ð3Þ


where k is a fitting parameter and Q is the normalized water content, given by:
hðua  uw Þ
H¼ ð4Þ
hs
which coincides with the degree of saturation–matric suction relationship in the case
of a rigid soil skeleton.
The basic idea underlying Equation 3 relies on the hypothesis that the rate at which
suction contributes to shear strength is a function of the normalized area of water (i.e.
the ratio of the area of water at any saturation degree to the area of water in saturated
condition). The same concept has also been used by Oberg and Sallfors (1997).
Vanapalli et al. (1996) also proposed an alternative form of Equation 3 without
using the fitting parameter k:
 
h  hr
tan ub ¼ tan u0 ð5Þ
hs  hr
The above expressions (4)–(5) have been particularly satisfactory in the analysis of
experimental shear strength–matric suction data obtained for till specimens prepared
at different initial conditions (Vanapalli et al., 1996).
Rassam and Williams (1999) presented an alternative model to interpret experi-
mental data coming from triaxial tests performed on two tailings samples with dif-
ferent grain size distribution. In particular, they used an unsaturated shear strength
envelope formed of two distinct surfaces. The first is a planar surface describing the
unsaturated shear strength at the various net normal stresses for suction values less
than or equal to the aev. The second is a non-linear surface, which is described by the
product of a power function of the matric suction and a linear function of the net
normal stress.
The global unsaturated shear strength envelope is then defined by the following
expressions:
8
< F1 ¼ aev1 þ aevs ðr3  ua Þ
>
F2 ¼ a þ tan u0 ½ðr3  ua Þ þ ðua  uw Þ ð6Þ
>
:
F3 ¼ ½ðua  uw Þ  F1b ½c þ kðr3  ua Þ
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 393

where a is a parameter corresponding to cohesion c¢ and b, c and k are fitting


parameters.
3.3.3. Unsaturated shear strength envelope and validation
The data in Figure 21 were interpreted by using the Vanapalli et al. and Rassam and
Williams models. In particular, for the Vanapalli et al. models, the soil water
characteristic curves in expressions (4) and (5) were assumed as those derived from
the van Genuchten model by using the parameters found in the previous sections
(Table 7) for each net stress considered. The parameter k was then estimated by
means of a simple least squares method. For the Rassam and Williams model, a
preliminary estimate of the relation between the air entry value and the net vertical
stress was made with reference to the investigated soils. In this respect (Figure 22),
the air entry values estimated by means of van Genuchten’s model were plotted
against net vertical stress; as can be seen, the data can be usefully interpolated by a
linear regression utilised for the calculations. In order to achieve an estimate of the
four Rassam and Williams’ parameters, a least squares method was also utilised.
Figure 23 shows the best-fit curves obtained by adopting the two models described
above, while in Table 8 the corresponding values of the parameter involved are
listed. As can be seen, the best-fit envelope is represented by the Rassam and Wil-
liams model (Figure 23b), which attains (Table 8) correlation coefficients always
higher than 0.95; the companion values provided by Vanapalli et al. models (Fig-
ure 23a) are lower and result in k parameter values ranging from 0.8 to 2.2 (Table 8).
In order to validate the procedure to relate the matric suction values and the shear
strength data (Figure 21) and, hence, to validate the interpretative models of Fig-
ure 23, non conventional triaxial tests (NCTX) were performed on specimens of
68 mm in diameter and 136 mm in height from samples collected in site 3. Tests were
carried out using the Bishop modified apparatus shown in Figure 24 (Rampino,
1997; Nicotera, 1998). In this apparatus, suction values are imposed via the axis
translation technique (Richards, 1941; Hilf, 1956), while volumetric strains of the
specimen and its water volume variations are measured by means of two differential

Figure 22. Air Entry Value versus net vertical stress for soils of site 3 belonging to class ‘A’.
394 EDUARDO BILOTTA ET AL.

Figure 23. Experimental and predicted shear strength envelopes: (a) Vanapalli et al. (1996) model and (b)
Rassam and Williams (1999) model.

pressure transducers (DPT), both linked to two different reference burettes. In


particular, for the volumetric strains, one DPT is connected to a water chamber
surrounding the specimen, while the other is connected to a burette located along the
drainage line.
On the basis of the expected results provided by shear strength–suction relation-
ships (Figure 23), triaxial tests were performed at two different matric suction values,

Table 8. Fitting parameters and correlation coefficients for Rassam & Williams and Vanapalli models
Rassam and Williams (1999) Vanapalli et al. (1996)
rv ) ua (kPa) a b c k R 2
k(a) R2(a) R2(b)

10 0.980 0.82 0.963 0.936


20 0.958 0.85 0.920 0.890
3.345 1.364 0.067 1.13E)04
50 0.969 1.28 0.958 0.833
100 0.984 2.18 0.908 0.704
(a)
First approach.
(b)
Second approach.
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 395

Figure 24. Cross-sectional view of suction controlled triaxial apparatus (from Rampino, 1997, modified.

namely 20 and 50 kPa, under constant values of net minor principal stress (r3 ) ua)
respectively equal to 10, 30 and 50 kPa (Table 9). The same table also reports the
physical properties of the tested specimens together with initial suctions of speci-
mens, estimated on the basis of the methodology suggested by Nicotera (1998). All
the tests were carried out at vertical strain rate of 3.3 mm/day.

Table 9. Testing conditions and initial physical properties (suction controlled triaxial tests)
Specimen r3 ) ua Wi (%) cdi nai Sri (%) GS [ua ) uw]i [ua ) uw]im
(kPa) (kN/m3) (kPa) (kPa)
USP1 50 62.5 7.9 0.68 73.0 2.45 4.1 50
USP3 10 62.6 7.9 0.68 73.1 2.45 8.1 50
USP4 30 67.5 7.4 0.70 71.3 2.45 9.7 50
USP5 50 59.3 8.1 0.66 72.4 2.38 12.9 20
USP6 30 62.0 7.4 0.69 66.3 2.38 14.5 20
USP8 10 63.3 7.6 0.68 70.2 2.38 13.8 20

[ua ) uw]i = Initial matrix suction; [ua ) uw]im = imposed matrix suction.
396 EDUARDO BILOTTA ET AL.

Figure 25. Deviatoric stress and volumetric strain versus axial strain curves (suction controlled triaxial
tests).

Assuming that r1 is the major principal stress, the measured values of deviatoric
stress [q ¼ (r1 ) r3)/2] and volumetric strain ev are plotted against vertical strain ey
in Figure 25. As observed, tests performed at net minor principal stress of 30 kPa
and 50 kPa show no strain softening behaviour (Figure 25(a)) or a reduction of
volume specimens during the shearing (Figure 25(b)), in agreement with the exper-
imental results of the DS tests, carried out at similar level of net stress (Figures 14–
16). As for DS tests, a slightly strain-softening behaviour (Figure 25) was revealed by
tests performed at a low stress level [(r3 ) ua) ¼ 10 kPa].
In order to make a comparison between the results obtained by DS and NCTX
tests, the procedure suggested by Fredlund and Rahardjo (1993) and Rahardjo et al.
(1995) was adopted for the NCTX test evaluation of shear strength (sf) and net
normal stress (rn ) ua)f acting on the failure plane. To this end, by assuming a
constant effective friction angle u¢ of 30 over the range of applied suction, the points
[sf, (rn ) ua)f] were determined by drawing the u¢ tangent line to each failure Mohr
circle at a specific suction value (Figure 26).
The points thus obtained, together with the DS data and the best fit curve of the
Rassam and Williams model, are shown in Figure 26, which highlights a good
agreement between the NCTX data and those predicted using only the DS data.
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 397

Figure 26. Predicted shear strength envelopes with Rassam and Williams model and experimental data
from DS and suction controlled triaxial tests.

In the same figure, the peak shear strength from Figure 20 is also plotted at
vertical stress of 40 and 80 kPa. For each of the above points, suction values were
estimated as follows. First of all, for each of the above net vertical stresses, the two
experimental characteristic curves close to that of interest were selected (i.e. for
40 kPa the curves at net vertical stress of 20 and 50 kPa; for 80 kPa the curves at net
vertical stress of 50 and 100 kPa). Then, the suction value at a given degree of
saturation was calculated as the weighted average suctions detected on the pair of
two close curves. As can be seen (Figure 26), the points thus determined can also be
well interpreted by adopting the Rassam and Williams model.
In order to better quantify the dependence of the unsaturated strength envelope
both on net normal stress and suction, the points of coordinates [sf, (ua ) uw)]
obtained from the best-fit curves in Figure 26 were interpreted by means of Equation
2. The resulting ub values are plotted in Figure 27. The figure shows a rapid decrease
of ub in quite a low suction values range, from 0 kPa to about 30 kPa, regardless of
the net normal stress; beyond this range the contribution of soil matric suction to
shear strength progressively decreases, with ub values reaching about 4 at matric
suctions approaching 100 kPa. From the chart depicted in Figure 27, it is also
interesting to observe that, at matric suction values higher than 65–70 kPa, the ub
angle seems independent of the net normal stress since the observed differences
amount to only one degree.
The above results allow some considerations to be made with reference to the
unsaturated shear strength of the investigated soil.
First of all, the experimental procedure used to estimate the unsaturated shear
strength envelope – by coupling the results from DS tests at constant water content
and the soil water characteristic curve – seems to be particularly satisfactory, as
testified by the suction controlled triaxial tests. This is of particular interest when
considering the area extension of pyroclastic soils subjected to potential flowslide
phenomena, which can be characterized with equipment and experimental proce-
dures allowing significant information to be collected in a reasonably short time.
398 EDUARDO BILOTTA ET AL.

Figure 27. Variation of friction angle /b versus matric suction.

Moreover, the experimental programme has allowed a good definition of the


unsaturated shear strength envelope, for a group of class ‘A’ soils. The results ob-
tained reveal a non-linear trend for this envelope, its dependence on the net normal
stress applied and a constant value of the ub angle (equal to the effective friction
angle u¢), in the range of matric suctions not exceeding the soil air entry values.
Finally, for the investigated soils Figures 26 and 27 show that the contribution of
matric suction to shear strength is significant for the entire range of in situ suction
values [0 £ (ua ) uw) £ 65 kPa] so far collected.

4. Concluding remarks
With reference to the ashy soils involved in the flowslides that recently occurred in
Campania Region, the experimental laboratory programme has made it possible to
highlight some features of these soils which can make a significant contribution to
improve the modelling of the above phenomena.
First of all, it must be stressed that the experimental results obtained for samples
collected at different triggering areas and in different layers of the Pizzo d’Alvano
massif can be classified into homogeneous classes, with respect to their physical and
hydraulic properties as well as the shear strength characteristics.
With reference to the hydraulic properties, the results so far obtained have shown
the absence of any hysteresis effects affecting the soil-water characteristic curves in
the 0–20 kPa range of net total stress values. As these net total stress values are
attained in almost all the triggering zones of the flowslides, the transient seepage
regimen induced by rainfall can be modelled using the same characteristic curve
during both the wetting and the drying processes.
As for saturated shear strength, the experimental results obtained for undisturbed
samples have clearly shown that the differences observed between the soil classes can
be usefully interpreted referring to their physical properties and, in particular, to
their unit dry weight and porosity which control the effective friction angle values.
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 399

In the unsaturated conditions, the definition of shear strength envelopes for the
different classes appears more complex because of the intrinsic heterogeneity of the
soils. In this respect, it seems particularly useful to resort to conventional direct shear
tests performed on specimens with water contents below the saturated state, and
interpret the results by a simple conceptual procedure involving the soil water
characteristic curves. The unsaturated non linear shear strength envelopes thus ob-
tained have also been validated with the aid of suction triaxial controlled tests, and
well modelled by the Rassam and Williams model.
If this procedure is confirmed by the experimental programme currently in pro-
gress, a considerable set of data could be obtained within a reasonable time frame.
This might then give an advantageous spin-off in the back analysis of the phenomena
as well as a better estimation of the stability conditions of the covers still located in
the ZOB areas.

References
Bilotta, E. and Foresta, V.: Sulla resistenza a taglio di alcune piroclastiti dei monti di Sarno,
Report no. 20, Department of Civil Engineering, University of Salerno, 2001, 26 pp.
Bilotta, E. and Foresta, V.: On the measured shear strength of some pyroclastic soils of Sarno
mountains, In: Proc. of the 3rd International Conference on Unsaturated Soils, Recife,
Balkema, Rotterdam, Vol. 2, 2002, pp. 495–500.
Cascini, L.: Il rischio da frana in aree urbane dell’Appennino Centro-meridionale, In: Proc. of
the XXI Convegno Nazionale di Geotecnica;11–13 settembre 2002, L’Aquila, 2002, pp. 127–
134.
Cascini, E. and Cascini, L.: Forecasting spring flow time series, J. Italian Statist. Soc., 3,
(1994), 1–23.
Cascini, L. and Sorbino, G.: Soil suction measurement over large areas: a case study, In: Proc.
of the 3rd Int. Conf. on Unsaturated Soils, Recife, Balkema, Rotterdam, Vol. 2, 2002, pp.
829–834.
Cascini, L. Guida, D. Nocera, N. Romanzi, G. and Sorbino, G.: A preliminary model for the
landslides of May 1998 in Campania Region, In: Special Lecture in Proc. 2nd Int. Symp. on
Geotechnics of Hard Soil-Soft Rock; Naples Vol. 3, 2000, pp. 1623–1649.
Cascini, L., Ferlisi, S. and Tagliafierro, G.: Il contributo delle indagini storiche nella defi-
nizione del rischio da frana: un caso di studio, In: Proc. of the XXI Convegno Nazionale di
Geotecnica, 11–13 settembre 2002, LÕAquila, 2002, pp. 135–142.
Celico, P. and De Riso, R.: Il ruolo idrogeologico della Valle Caudina nella idrogeologia del
Casertano e del Sarnese (Campania), Boll. Soc. Nat., 88, (1978), 1–26.
Celico, P., Guadagno, F.M. and Vallario, A.: Proposta di un modello interpretativo per lo
studio delle frane nei terreni piroclastici. Geol. Appl. e Ldrog., 21, (1987), 173–194.
Ciollaro, G. and Romano, N.: Spatial variability of the hydraulic properties of a volcanic soil,
Geoderma, 65, (1995), 263–282.
Cinque, A., Hossein, H., Laureti L. and Russo F.: Osservazioni preliminari sull’evoluzione
geomorfologica della Piana del Sarno (Campania, Appennino Meridionale), Geogr. Fisica
Din. Quat., 11, (1987), 38–44.
Civita, M., De Riso, R. and Lucini, P.: Studio delle condizioni di stabilità dei terreni della
Penisola Sorrentina, Mem. e Note Istituto di Geologia Applicata, Napoli (1975), 22 pp.
Corey, A.T.: Mechanics of Immiscible Fluids in Porous Media, 3rd edn. Water Resources
Publications, Highlands Ranch, Colo, 1994.
400 EDUARDO BILOTTA ET AL.

Costa, J.E. and Baker, V.R.: Surficial Geology: Building with the Earth, John Wiley & Sons,
Inc., New York, 1981, 498 pp.
Costa, J.E. and Wieczoreck, G.F.: Debris flows/Avalanches: Process, Recognition and
Mitigation, Reviews in Engineering Geology, Vol. 7, Geol. Soc. Of America, Boulder,
Colorado, 1987, 239 pp.
de Campos, T.M.P. and Carrillo, C.W.: Direct shear testing on an unsaturated soil from Rio
de Janeiro, In: Proc. of 1st International Conference on Unsaturated Soils, Vol. 1, 1995, pp.
31–38.
Del Prete, M., Guadagno, F.M. and Hawkins, A.B.: Preliminary report on the landslides of 5
May 1998, Campania, southern Italy, Bull. Eng. Geol. Env., 57, (1998), 113–129.
Eckersley, J.D.: Instrumented laboratory flowslides, Ge´otechnique 40(3), (1990), 489–502.
Ellen, S.D. and Fleming R.W.: Mobilization of debris flows from soil slips, S. Francisco Bay
Region; California, In: Debris Flows/Avalanches: Process; Recognition and Mitigation:
Reviews in Engineering Geology, Vol. 7, Geol. Soc. of America, Boulder, Colorado, 1987,
31–40.
Escario, V. and Saez, J.: The shear strength of partially saturated soils, Ge´otechnique, 36(3),
(1986), 453–456.
Esposito, L. and Guadagno, F.M.: Some special geotechnical properties of pumice deposits,
Bull. Eng. Geol. Env., 57, (1998), 41–50.
Faella, C. and Nigro, E.: Effetti delle colate rapide sulle costruzioni; Parte II: Valutazione della
velocità di impatto. Forum per il Rischio Idrogeologico in Campania ‘‘Fenomeni di colata
rapida di fango nel maggio ‘98’’, (2001), 113–125.
Fiorillo, F., Guadagno, F.M., Aquino, S. and De Blasio, A.: The December 1999 Cervinara
landslides: Further debris flows in the pyroclastic deposits of Campania (southern Italy),
Bull. Eng. Geol. Env., 60, (2001), 171–184.
Fredlund, D.G., Morgenstern, N.R. and Widger, R.A.: The shear strength of unsaturated
soils, Canadian Geotechnical Journal, 15, (1978), 313–321.
Fredlund, D.J. and Rahardjo, H.: Soil Mechanics for Unsaturated Soils, John Wiley & Sons
Inc, New York, 1993.
Gan, J.K-M. and Fredlund, D.G.: Shear strength characteristics of two saprolitic soils.
Canadian Geotech. J., 33, (1996), 595–609.
Gan, J.K-M. Fredlund, D.G. and Rahardjo, H.: Determination of the shear strength
parameters of an unsaturated soil using the direct shear test, Can. Geotech. J., 25, (1988),
500–510.
Hilf, J.W.: An Investigation of Pore-water Pressure in Compacted Cohesive Soil, Ph.D. Dis-
sertation, Tech. Mem. Dept. of the Interior, Bureau of Reclamation, Design and Con-
struction Div., Denver, CO., US, 1956.
Hungr, O., Evans, S.G., Bovis, M.J. and Hutchinson, J.N.: A review of the classification of
landslides of the flow type, Environ. & Eng. Geosci., VII(3), (2001), 221–238.
Hutchinson, J.N.: Morphological and Geotechnical parameters of Landslides in relation to
Geology and Hydrogeology. State of the art Report, In: Proc. V Intl Symposium on
Landslides; Lausanne, Vol. 1, 1988, 3–35.
Jones, F.O.: Landslides of Rio de Janeiro and the Serra das Araras Escarpment, Brazil, U.S.
Geological Survey, Professional paper 697, (1973), 42pp.
Matyas, E.L. and Radhakrishna, H.S.: Volume change characteristics of partially saturated
soils, Ge´otechnique, 18(4), (1968), 432–448.
Nicotera, M.V.: Effetti del grado di saturazione sul comportamento meccanico di una pozzolana
del napoletano, Ph.D. Thesis, University of Naples, 1998.
GEOTECHNICAL CHARACTERISATION OF PYROCLASTIC SOILS 401

Oberg, A.-L. and Sallfors, G.: Determination of shear strength parameters of unsaturated silts
and sands based on the water retention curve, Geotech. Testing J., 20(1), (1997), 40–
48.
Olivares, L. and Picarelli, L.: Susceptibility of loose pyroclastic soil to static liquefaction –
some preliminary data, In: Intl. Conference on Landslides – Causes, Impacts and Coun-
termeasures, Davos, Switzerland, 2001, pp. 75–85.
Oloo, S.Y. and Fredlund, D.G.: A method for determination of ub for statically compacted
soils, Can. Geotech. J., 33, (1996), 272–280.
O.U. 2.38.: Relazione sull’attività svolta dall’U.O. dell’Università di Salerno dal 21/05/1998 al
05/07/1998. G.N.D.C.I. – Dept. of Civ. Eng., University of Salerno, Italy, 1998a.
O.U. 2.38.: Ricerca storica sulle colate di fango in terreni piroclastici della Campania.
G.N.D.C.I. – Dept. of Civ. Eng., University of Salerno, Italy, 1998b.
Pellegrino, A.: Proprietà fisico-meccaniche dei terreni vulcanici del Napoletano, In: Proc. of
VIII Convegno Italiano di Geotecnica, Cagliari, Vol. 3, (1967), pp. 113–146.
Picarelli, L.: Alcune considerazione sui meccanismi di innesco e di propagazione delle colate in
terreni sciolti e detritici, In: Proc. on Previsione e Prevenzione di movimenti franosi rapidi,
Trento (Italy), 1999, pp. 163–179.
Picarelli, L. and Olivares, L.: Innesco e Formazione di colate di Fango in terreni sciolti di
origine piroclastica, Forum per il Rischio Idrogeologico in Campania, Napoli 22 Giugno
2001, 2001, pp. 26–38.
Rahardjo, H., Lim, T.T., Chang, M.F. and Fredlund, D.G.: Shear-strength characteristics of a
residual soil, Can. Geotech. J., 32, (1995), 60–77.
Rampino, C.: Comportamento meccanico di una sabbia limosa ed argillosa costipata parzial-
mente satura, Ph.D. Thesis, University of Naples, 1997.
Rassam, D.W. and Williams, D.J.: A relationship describing the shear strength of unsaturated
soils, Can. Geotech. J., 36, (1999), 363–368.
Richards, L.A.: A pressure membrane extraction apparatus for soil suction, Soil Sci. 51,
(1941), 377–386.
Rossi, F. and Chirico, G.B.: Definizione delle soglie pluviometriche d’allarme, G.N.D.C.I. –
Dept. of Civ. Eng., University of Salerno, Italy, 1998.
Sassa, K.: Special Lecture: Geotechnical model for the motion of Landslides, In: Proc. ISL,
Lausanne, Vol. 3, 1988.
Sassa, K.: Recent urban landslide disasters in Japan and their mechanisms, In: Proc. 2nd Int.
Conf. Environmental Management, Vol. 1, (1998), pp. 45–78.
Sassa, K.: Mechanism of flows in granular soils. Intl. Conference Geotechnical and Geological
Engineering, GEOENG 2000, 2000, pp. 1671–1702.
Sorbino, G.: Il regime delle acque sotterranee nelle rocce metamorfiche alterate, Ph.D. Thesis,
University of Naples, 1994.
Sorbino, G.: Unsaturated hydraulic characteristics of weathered gneissic soils, Proc. X Pan
American Conference, Guadalajara (Mexico), 1995.
Sorbino, G. and Foresta, V.: Unsaturated hydraulic characteristics of pyroclastic soils, In:
Proc. of the 3rd International Conference on Unsaturated Soils, Recife, Balkema, Rotter-
dam, Vol. 1, 2002, pp. 405–410.
Takahashi T.: Debris Flow, IAHR Monograph, A.A. Balkema, Rotterdam, 1991.
van Genuchten, M.Th.: A closed-form equation for predicting the hydraulic conductivity of
unsaturated soil, Soil Sci. Soc. Am. J. 44(5), (1980), 892–898.
van Genuchten, M. Th., Leij, F.J. and Yates, S.R.: The RETC code for quantifying the
hydraulic functions of unsaturated soils, EPA/600/2-91/065, 1991.
Vanapalli, S.K., Fredlund, D.G., Pufahl, D.E. and Clifton, A.W.: Model for the prediction of
shear strength with respect to soil suction, Can. Geotech. J., 33, (1996), 379–392.
402 EDUARDO BILOTTA ET AL.

Vanapalli, S.K., Fredlund, D.G. and Pufahl, D.E.: The influence of soil structure and stress
history on the soil–water characteristics of a compacted till, Ge´otechnique, 49(2), (1999),
143–159.
Versace, P.: La riduzione del Rischio idrogeologico nei Comuni colpiti dagli eventi del Maggio
1998 in Campania, In: Forum per il Rischio Idrogeologico in Campania; Napoli 22 Giugno
2001, 2001, pp. 11–20.
Wang, G. and Sassa, K.: Factors affecting rainfall-induced flowslides in laboratory flume tests,
Ge´otechnique 51(7), (2001), 587–599.
Wesley, L.D.: Determination of Specific Gravity and Void Ratio of Pumice Materials, Geo-
tech. Test. J., GTJODJ, 24(4), (2001), 418–422.

You might also like