Engineering Geology 196 (2015) 88–98

Contents lists available at ScienceDirect

Engineering Geology

journal homepage: www.elsevier.com/locate/enggeo

An integrated procedure to evaluate rheological parameters to model

debris flows
Anna Maria Pellegrino a,⁎, Anna Scotto di Santolo b, Leonardo Schippa a
a
Department of Engineering, University of Ferrara, Via Saragat 1, 44122 Ferrara, Italy
b
Pegaso University, Piazza Trieste e Trento, Napoli, Italy

a r t i c l e i n f o a b s t r a c t

Article history: In the present paper, the problem of modeling the propagation of debris flow using suitable rheological parameters
Received 21 October 2014 is considered. A new procedure is proposed based on field observations, laboratory investigations and numerical
Received in revised form 29 June 2015 analysis. The back analysis of a debris flow event that occurred in the Campania region (southern Italy) was
Accepted 2 July 2015
used to investigate the reliability of the assumed rheological parameters. The well-known Bingham constitutive
Available online 6 July 2015
equation was chosen to model the behavior of the equivalent flow mixture. Numerical analysis was conducted
Keywords:
using DAN-W, a code based on a single-phase, depth-averaged continuum mechanics approach. The results
Debris flow show that extrapolating model parameters only from laboratory experiments may lead to inaccurate predictions.
Runout Instead, when the rheological parameters are assumed according to stress field observations and experimental
Numerical modeling data, the model satisfactorily replicates the case studies. Such conclusion suggests that the calibration of model pa-
Bingham model rameters using an integrated approach may be applicable for field-scale debris flow simulations and can yield use-
Pyroclastic soil mixtures ful information about the dynamics of flows to be investigated. The accuracy of this approach was demonstrated by
Hazards the numerical prediction of other fifty events that occurred in the same area.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction Mangeney et al., 2003; McDougall and Hungr, 2004; Pudasaini and
Hutter, 2007; Hungr and McDougall, 2009; Christen et al., 2010) and re-
Rapid, long runout landslides jeopardize areas situated at a consider- cently applied to hazard evaluation (Van Westen et al., 2006; Scotto di
able distance from the source and represent a difficult challenge in haz- Santolo and Evangelista, 2009). These models are based on physical
ard studies. During these catastrophic events, debris flows move data and numerically solved, simulating the motion of flow using consti-
suddenly and swiftly down-valley, invading the plains at the mountain tutive equations of fluid mechanics in one or two dimensions. Most
foothills and affecting towns, roads, and factories (Cruden and Varnes, models are based on a “continuum approach” that considers the multi-
1996). The prediction of the runout distance, flow velocity, and depth phase moving mass of a debris flow as a continuum. This way, the dy-
(i.e., dynamic parameters) is necessary in order to plan and design pro- namics of debris flow are modeled using an equivalent fluid, whose
tective measures and it is a key requirement to design hazard zones. rheological properties can approximate the expected bulk behavior of
Models used to predict flows at field scale may be classified either as the real mixture. In the continuum mechanical approach, the flow is
empirical or analytical. Empirical models are based on limiting criteria modeled by a De Saint Venant-type system (e.g., shallow water equa-
(e.g. Fannin and Rollerson, 1993) or statistical relations (Sheidegger, tions) derived in a reference frame linked to an infinitely inclined plane.
1973; Cannon, 1993; Corominas, 1996; Fannin and Wise, 2001; This allows numerical simulation of rapid mass movements over complex
Rickenmann, 2005; Zhang and Yin, 2013). Although empirical models topographies. Many of the analytical models are based on single-phase
are easy to use, their application is limited to conditions matching those solutions of depth-averaged equations of motion and relatively simple
of the original model. However these models were utilized for hazard one-phase rheological relationships, which define the frictional force act-
mapping (e.g., Calcaterra et al., 2003). Several analytical runout models ing at the interface between the flow and the bed path. However, the
were developed (Savage and Hutter, 1989; Chen and Lee, 2000; choice of a rheological law and the calibration of its peculiar parameters
still remain open problems. The most common rheologies used in the dy-
namic models are the Coulomb model, the Voellmy model, the Bingham
generalized (or Herschel & Bulkley) model, and the Quadratic model
(Naef et al., 2006). Currently, the rheological parameters of constitutive
⁎ Corresponding author.
E-mail addresses: [email protected] (A.M. Pellegrino),
equations can be estimated following either a measurements-based ap-
[email protected] (A. Scotto di Santolo), [email protected] proach or a calibration-based approach. Using a measurements-based ap-
(L. Schippa). proach, the input parameters can be obtained through controlled

http://dx.doi.org/10.1016/j.enggeo.2015.07.002
0013-7952/© 2015 Elsevier B.V. All rights reserved.
 A.M. Pellegrino et al. / Engineering Geology 196 (2015) 88–98 89

laboratory experiments on reconstructed samples of the involved materi- 2. Case histories

al. However, a mere laboratory analysis may not be adequate to repro-
duce the properties of a complex debris flow because such properties, In 1997 numerous landslides occurred in Campania due to a storm
even if measurable, may be scale-dependent (Blanco et al., 2009; event occurred on January 10th. Among those, the first to be considered
Pellegrino et al., 2014). Using a calibration-based approach, the rheologi- was a fatal debris flow event occurred at Pozzano, which is located on
cal parameters are evaluated with systematic adjustment during trial- the Sorrentina Peninsula (Naples, Campania region, see Fig. 1). The
and-error back analysis of full-scale events. Although this approach Pozzano debris flow involved the foothills of Castellammare di Stabia,
makes simpler the study of past debris flow phenomena, the effective- whose slopes are covered by pyroclastic soils. The event occurred at
ness of the calculated parameters is strictly related to the degree of sim- 20:15 on January 10th, as a result of an intense and prolonged rainfall
ilarity between the characteristics of historical and potential events. This in the area. The debris flow destroyed a civil building and flooded the
limitation restricts the use of rheological parameters measured with a 145 State Route, causing four deaths, twenty-two injuries and the inter-
calibration-based approach to the study of a particular area of interest ruption of the State Route for about two months. The case history of the
(Scotto di Santolo and Evangelista, 2009). Currently, methods for deter- Pozzano event is well-documented in the literature. Several pieces of in-
mining the dynamic characteristics of a debris flow may be based on ex- formation are available: the topography of the area before and after the
perimental data as well as on data generated by physical models. By event, the geotechnical characteristics of the involved materials (Scotto
integrating these two sources of data, a promising approach could di Santolo et al., 2000), the geometry of the slope, the shape and the size
emerge to refine the methods for the assessment of the flowing mass dy- of the triggering zone, the mobilized volume and the propagation path
namics (Luna et al., 2011). (Calcaterra et al., 2003; Calcaterra and Santo, 2004; Iadanza et al.,
The present paper suggests an original procedure that allows to esti- 2009). The landslide started as a translational slide and continued
mate the rheological model parameters of the flowing mass through an along the State Route, filling an old quarry before reaching the sea.
integrated approach based on laboratory and field investigations. For The details of the morphology and the slope profile are shown in
this purpose, the dynamic characteristics of well-documented events Fig. 2. The crown of the landslide is located at approximately 465 m
(e.g., runout distance, flow velocity and deposition depth) were cali- above sea level. The flowing mass traveled for 1300 m, with a deposition
brated by numerical back analysis. length of approximately 665 m. The medium measured depth of the de-
Some debris flow events that periodically occurred in the Campa- posited material was 2.5 m, and the inclination of the deposition area
nia region in southern Italy were considered in order to assess the was 7°. The depth of the flow was approximately 1.2 m, and the channel
method (e.g. Scotto di Santolo, 2002; Calcaterra et al., 2003). These slope was 35° (Calcaterra and Santo, 2004; Iadanza et al., 2009). The
events were back-calculated using the 2D numerical code DAN-W Pozzano's debris flow is a mixed type event. After an initial J-path, in
(Hungr, 1995) based on the continuum mechanics approach. The the middle portion of the event, the landslide is canalized into a
classic, one-phase Bingham model was chosen. In order to calibrate preexisting gully (see Fig. 2).
the rheological model parameters, the Pozzano landslide (Naples) In this case, the displaced mass started to flow and the mass move-
was chosen as the case study. The resulting calibrated model param- ment transformed into an extremely rapid debris flow. A first part of
eters were used for the prediction of the second case study occurred this debris stopped in the landslide channel. Another part of the debris
in the same area (Corbara, Salerno). Based on the acquired experi- (about 10,000 m3) stopped on State Road 145. The remaining part of
ence (Scotto di Santolo and Evangelista, 2009; Schippa and Pavan, the mass reached the sea (see Fig. 2), about 50 m beyond the shoreline.
2011) and considering the limited reliability of the forecast — from As highlighted by Calcaterra and Santo (2004), the Pozzano landslide
the triggering to the ceasing of the flow — , the proposed method can be defined as a complex movement: the first movement was a
could be a useful tool to model the runout and the velocity of a des- small rotational slide affecting a thin slab of deeply altered ashy pyroclas-
ignated debris flow. tic; a translational slide then occurred; the event reached a preexisting
The proposed procedure, even if utilized on Italian debris flows, has gully, and so was filled by some meters of altered and reworked pyro-
general validity and can be applied to other events when the rheological clastic deposits. A canalized debris flow was initiated here, facilitated
behavior of the involved materials can be described with the Bingham by the high water content. The involved soils show a wide grain-size dis-
generalized model (e.g., fluid-like behavior). tribution. They comprise a slightly gravelly silty sand (Fig. 3). The

Fig. 1. Inventory map of the January 1997 landslides in the Sorrento Peninsula: (1) Debris or earth fall; (2) rock fall; (3) rotational slide; (4) translational slide; (5) flow; (6) complex land-
slide. The circle refers to the area of interest (modified from Calcaterra and Santo, 2004).
90 A.M. Pellegrino et al. / Engineering Geology 196 (2015) 88–98

Fig. 2. Pozzano's debris flow (Naples, 1997). Event landform (on the right) and elevation profile (on the left) with stratigraphy (modified from Calcaterra and Santo, 2004).

volcanoclastic deposits offered low bulk and dry densities and did not di Santolo, 2002). The landslide started as open, then became channel-
show plastic behavior. The shear strength show a substantial depen- ized and stopped near an old quarry. The details of the slope profile
dence on the water content: when the soils were sheared with their are shown in Fig. 4. The depth of the flow was approximately 1 m, and
“natural” water content, they showed a distinct but small cohesion the channel slope was 35° (Scotto di Santolo, 2002). The crown of the
(about 20 kPa) and a friction angle of 27–30°. After saturation, the landslide is located at approximately 710 m above sea level. The flowing
same materials behaved as cohesionless soils because the matric mass traveled for 1145 m, with a deposition length of approximately
sunction was annulled (Scotto di Santolo et al., 2000). 720 m. The depth of the deposited material was 2 m, and the medium
The second event analyzed in this study was a debris flow occurred inclination of the deposition area was 10–15°. The soil involved shows
on a slope with a low hierarchized drainage basin at Corbara, a town lo- a wide size distribution and it comprise a slightly gravelly silty sand,
cated near Salerno (Campania region, see Fig. 1), Fig. 4. The flow was a as shown in Fig. 3.
mixed type (Calcaterra et al., 2003; Di Crescenzo and Santo, 2005;
Scotto di Santolo and Evangelista, 2009). On the east side of the hill, 3. An overview on DAN-W code
the flow started as a triangular shape and then became channeled on
the drainage basin; on the west side of the hill, the flow was directly The DAN code, developed by Hungr in 1995, is based on a continuum
channeled. In the present work, the debris flow that occurred on the mechanics approach and it was especially developed to simulate the
east part of the hill has been considered (Fig. 4) because it involved a motion of flows, flow-like slides and avalanches. It reduces a complex
considerable initial volume of material (8547 m3). As shown in Fig. 4, and heterogeneous three dimensional problem into an extremely sim-
the flow was stopped completely at a quarry. Several information are ple formulation. The simplicity of the model and the possibility of choice
available: the topography of the area before and after the event, the geo- among different rheologies, some of which are particularly simple,
technical characteristics of the involved materials, the geometry of the make DAN an interesting tool to be employed. DAN is a Windows-
slope, the shape and the size of the triggering zone, the mobilized vol- based program and it implements a Lagrangian solution for the equa-
ume and the propagation path (Di Crescenzo and Santo, 2005a; Scotto tions of motion for a mass of earth material which starts from a pre-
scribed static configuration and flows according to one of the several
rheological relationships. The model treats the slide mass as an “equiv-
alent fluid”, a hypothetical material governed by simple internal and
basal rheological relationships (see Fig. 5). DAN-W is based on shallow
flow equations and on the Lagrangian solution of St. Venant's equation.
This equation can be derived by applying a momentum balance equa-
tion to the thin slices of flowing mass that are perpendicular to the
base of the flow. These “boundary blocks” divide the slide mass into n
“mass elements” of constant volume.
The resulting formula for the net driving force acting on each bound-
ary block i is:

F ¼ ρ  g  Hi  ds  Bi  sinϑ−T−P: ð1Þ

Hi is the depth of a considered i-element in which the mass can be

discretized, F is the net driving force equal to F = ρ ⋅ H ⋅ ds ⋅ B ⋅ (dv/dt),
where ρ is the bulk density of the flow material, H is the flow depth nor-
Fig. 3. Main grain size distribution of soil samples taken from the pyroclastic cover of
mal to the base, B is the channel width, ds is the infinitesimal length of the
Pozzano (continuous line with fill black squares) and Corbara (continuous line with fill boundary block, θ is the angle of the base from horizontal, T is the
black circles). resisting shear force acting on the base, P is the tangential internal
 A.M. Pellegrino et al. / Engineering Geology 196 (2015) 88–98 91

Fig. 4. Corbara's debris flow (Salerno, 1997). Event landform (on the right) and elevation profile (on the left) with stratigraphy.

pressure and v is the mean flow velocity. Eq. (1) can be solved for succes- to the appropriate equations for T. More details may be found in Hungr,
sive time steps after the initial, at-rest condition of the slide mass. 1995.
Referring to finite difference scheme, for a single time step of Δt and One of the most critical aspects in using a physically based debris-
a unit length of ds, the change in velocity of each boundary block is: flow model is the choice of a representative and adequate constitutive
relationship for the mixture. Among the rheological models available
Δv ¼ gð FΔt−MÞ=ðγ  Hi  Bi  dsÞ ð2Þ in the DAN -W code, the Bingham model was used because it was pre-
viously employed to describe the experimental data from tests on
where γ is the unit weight of the boundary block, and F is the net force reconstituted samples of pyroclastic soils from debris flows in the Cam-
as defined in Eq. (1). M is a term for momentum flux which is caused by pania region (Scotto di Santolo and Evangelista, 2009; Scotto di Santolo
erosion or entrainment of material (Hungr, 1995). The velocity change et al., 2012). The Bingham model also seems adequate to represent real
can be represented by the difference between the new velocity at the debris-flow in a variety of situations (Luna et al., 2011; Bertolo and
end of time step Δt and the old velocity. Now, integrating Eq. (2) over Wieczorek, 2005; Nigussie, 2013; Jeong, 2013; Jeong, 2010; Bisantino
the same time step Δt gives the curvilinear displacement S of each et al., 2009). According to the Bingham model, the fluid initially assumes
boundary block: a solid-like behavior (shape- and volume-defined) and then reveals its
liquid characteristics during the flow when the shear stress is higher


Si ¼ Si;old þ Δt= 2 vi þ vi;old : ð3Þ than the yield shear stress τc. The resisting shear stress is assumed to de-
pend on a constant strength. The viscous term depends on the velocity
and on the inverse of the debris sheet thickness. The rheological equa-
This result can be used in combination with the constant-volume
tion depends upon two constants: the yield stress (τc) and the Bingham
constraint to find the new height of each boundary block, defined by
viscosity (μB). Accounting for zero yield stress, the flow becomes Newto-
the mean depth h of adjacent mass elements j–1 and i:
nian. The mean flow velocity is derived from the assumption of a linear
  increase in shear stress with depth. In the Bingham model, the basal
Hi;new ¼ h j−1 þ hj j =2 ð4Þ flow resistance term, T, is a function of flow depth, flow velocity, con-
stant yield stress and Bingham viscosity as follows:
where hj is equal to (2 Vj/[(Si + 1 − Si) (Bi + 1 + Bi)]), and Vj is the con-  
stant volume of boundary element j. The basal flow resistance term, T, in vi ¼ 1=6Hi =μ B 2τc =Ai −3τc þ τ3c Ai =T2 ð5Þ
Eq. (1) is governed by the rheology of the material. Several different
rheological models are available in the DAN-W model. They are related where Ai is the boundary block base area equal to (ds · Bi).

Fig. 5. Prototype of a heterogeneous and complex moving mass and model of equivalent homogeneous “apparent fluid” (from Hungr, 1995).
92 A.M. Pellegrino et al. / Engineering Geology 196 (2015) 88–98

Fig. 6. Profile of a debris flow. H is the difference in altitude between the crown and the toe
of the debris. L is the traveled distance of the debris. α is the “fahrboschung” (“travel Fig. 8. Main grain size distribution curve of samples taken from the pyroclastic deposit
angle”) as the slope of a line connecting the crest of the source area with the distal tip of analyzed.
deposits, measured on a straightened profile of the path (modified from Heim, 1932).

• velocity of the front vf;

4. Calibration procedure • height of the deposit.

On the basis of an extensive experimental study on the rheological

behavior of natural pyroclastic soils with different processes remixed In the area of interest, flow velocities were estimated through the
with water, a homogeneous fluid-like behavior is obtained only within back analysis of the damages caused by the flows that took place in
a very narrow range of concentrations. Below this range the material the area of Sarno–Quindici (Faella and Nigro, 2003) and through dy-
rapidly settles; above this range it behaves like a solid. In the fluid-like namic analysis related to geometrical characteristics of the flows
range the rheological behavior is well reproduced by the generalized (Revellino et al., 2004, 2006). For this reason, other indications available
Bingham model with static yield stress larger than its dynamic ones from the back-analysis of debris flow events that occurred worldwide
(Scotto di Santolo et al., 2012). For this reason the parameters of the are considered too (Luna et al., 2011; Faella and Nigro, 2003; Bertolo
Bingham model (e.g., yield stress τc and Bingham viscosity μB) used and Wieczorek, 2005; Prochaska et al., 2008). The range of velocity be-
for numerical simulations were defined in three ways: tween 16 up to 35 m/s was detected.

• firstly based on field observations (define later Model A, Section 4.1); 4.1. Model A
• secondly using experimental results of rheometer tests on soil–water
mixtures prepared using field-collected soil samples (define later This method is a direct application of theories concerning the
Model B, Section 4.2); flowing and stopping processes of yield stress fluids in idealized steady
• thirdly the original procedure proposed by the authors (defined later uniform flow on an infinitely wide inclined plane (Johnson, 1970;
Model C, Section 4.3). Coussot, 1997), Fig. 7.
The main assumptions are as follows: the free surface is necessarily
parallel to the inclined plane and the shear stress at the base is equal
The models, discussed afterwards, were calibrated with reference to:
to the yield stress τc, evaluated for x = 0 as:
• runout distance L, horizontally between the edge of the toe and the
crown (Fig. 6); τc ¼ ρ  g  hD  sinðiD Þ ð6Þ

Fig. 7. Model A: free surface flow on an infinitely wide inclined plane: simplified scheme of stress and velocity distribution in a flowing yield stress fluid/viscous debris flow and distinction
of plug and shear zone.
 A.M. Pellegrino et al. / Engineering Geology 196 (2015) 88–98 93

1000 Based on the calculated plug thickness hplug, the thickness of the
shear zone hshear, is calculated as:

hshear ¼ h−hplug ð9Þ

shear stress (Pa)

100 where h is the observed flow depth. For a simplified linear velocity dis-

tribution, the apparent shear rate γ relative to the observed velocity vs,
is:
Experimental data  vs
Bingham Model γ¼ : ð10Þ
hshear
10
0,1 1 10 100 1000 Assuming an identical density ρ over the entire flow depth h, the bed
shear rate (1\s) shear stress τbed is:

Fig. 9. Laboratory flow curve (symbols, diamonds) and Bingham model (continuous line) τbed ¼ ρ  g  h  sinðiÞ: ð11Þ
of a mixture of water and pyroclastic soil at solid volumetric concentration of Φ = 42%.

The Bingham viscosity parameter μB can be calculated as:

τbed −τc
μB ¼  : ð12Þ
where ρ is the mixture density, g is the gravitational force, hD is the de- γ
posit depth, and iD is the slope of the deposition plane.
Owing to the presence of the yield stress τc, it seems important to This method allows to estimate the Bingham parameters τc and μB of
anticipate that there would be a plug-like region near the free surface, the debris flow, which are required when the post-failure behavior is
as shown schematically in Fig. 7. If this region extends up to x = x0, modeled with a numerical code based on equivalent homogenous
the velocity of the plug will be constant in the region 0 b x b x0. Beyond fluid-flow, on the basis of field measurements of the deposit depth hD
x N x0, the velocity will progressively decrease from the plug velocity to and the slope of the deposition plane iD.
a zero value at the wall due to the no-slip condition. It has to be noted
that such method can also be applied for numerical codes which are 4.2. Model B
based on a Herschel–Bulkley model. In this case the more sophisticated
Herschel–Bulkley fluid is reduced to the simplified Bingham fluid. The rheological parameters assigned to Model B are determined using
The Bingham viscosity parameter μB is estimated accounting for the the experimental data from rheometrical tests performed on mixtures of
yield stresses τc, as well as the observed flow depth h and surface veloc- water and sediment. In our back analysis the experimental results are re-
ity vs of the debris flow. The calculation of the Bingham viscosity param- lated to the mixture of pyroclastic soil involved in the debris flow oc-
eter μB, is given as follows. The flow depth h of a yield stress fluid is curred in South Italy. The materials tested were collected from the
roughly divided into a plug zone with the depth hplug and a shear source area. The top soil, in a thickness of about a meter, is characterized
zone, Fig. 7. The velocity distribution within the shear zone is simplified by pyroclastic deposits deriving from the volcanic activity of Mount
to a linear distribution (Coussot, 1997). With the knowledge of the yield Somma/Vesuvius. The grain size distributions of the collected samples
stress τc, in the case of no lateral boundaries effects, the thickness of the are reported in Fig. 8. The soil is a sandy silt with a small clay fraction.
plug zone hplug can be estimated as: The bedrock underlying the soil is a limestone. The main physical proper-
ties of the considered soil are the specific gravity of soil particles Gs equal
to 2.61, the dry weight of soil per unit volume γd equal to 9.08 kN/m3, the
τc
hplug ¼ ð7Þ total weight of soil per unit volume γ equal to 11.35 kN/m3, and the po-
ρ  g  sinðiÞ
rosity n equal to 66%. The complete description of the laboratory activity
carried out on the soil is illustrated in Scotto di Santolo et al. (2012). The
where i is the bed slope. In the case of a confined channel, the lateral rheometrical tests were performed on the soil fraction having a particle
boundaries effects must be considered. For the simplified case of a semi- diameter less than 0.5 mm, which represent almost the 70% of the
circular channel (Johnson, 1970), it is: whole grain size distribution (see Fig. 8). Since the rest of the particles
which are contained in the whole material are not colloidal, when they
are mixed with the paste to get the complete mixture they can be expect-
ed to essentially increase the values of the rheological parameters but
2τc they do not affect the behavior type. Thus, the idea is that working with
hplug ¼ : ð8Þ
ρ  g  sinðiÞ the whole mixtures would have got similar qualitative trends as

Table 1
Data available in literature on rheological modeling of debris flow materials. (LS-rheo = large scale rheometer; BMS = ball measuring system; Rheo = conventional rheometer; DF = debris
flow material; DF Campania = debris flow material from Campania region).

Author/s Year Set up Materials dmax (mm) Φ (%) Model τc (Pa) μ (Pa · s) K (Pa · sn)

Major and Pierson 1992 LS-rheo DF 2 52–66 Bingham generalized 12–405 – 0.4–16
Schatzmann et al. 2009 BMS DF 10 35–64 Bingham generalized 2.7–265 – 0.41–47
Kaitna et al. 2007 Rotating flume DF 5 47–62 Bingham 10–80 0.1–2 –
O'Brien and Julien 1988 Rheo DF 2 15–45 Bingham 0.2–100 0.4–10 –
Coussot and Piau 1995 Rheo DF 0.4 35–40 Bingham generalized 29–220 – 18–242
Martino 2003 Rheo DF Campania 0.4 25–53 Bingham generalized 1.1–39 – 0.02–0,3
Schatzmann et al. 2009 Rheo and BMS DF material 0.4 22–30 Bingham generalized 30–98 – 1.18–14
Pellegrino et al. 2010 Rheo DF Campania 0.5 30–42 Bingham generalized 1.2–90 0.2–2 0.34–4,7
94 A.M. Pellegrino et al. / Engineering Geology 196 (2015) 88–98

described below. For the present work, the parameters of the Bingham Table 2
model obtained were used to describe the experimental data for a mix- Rheological parameters assumed for Models A, B, and C (see Section 4).

ture of water and sediment having a solid volumetric concentration Φ Model Model parameters
equal to 42%. In Scotto di Santolo et al. (2012), a consistent study was τC μB
done in order to have an overview of the behavior of pyroclastic material
(Pa) (Pa s)
as a function of the solid concentration and to identify the range of con-
centrations in which it may be considered as fluids and thus character- A 3112 484
B 144 1.8
ized with usual rheological tools. Three different possible states
C 3112 1.8
appeared. For sufficiently low solid volume fractions (less than 32%),
the particles rapidly (within a few seconds) settle down, leading to an ap-
parent phase separation. It was obviously impossible to carry out material flows and behaves like a fluid has been clearly defined accord-
rheometrical tests because of when the particles have settled, no longer ing to the natural water content, the viscosity of the flowing mass can be
dealing with a homogeneous material, and nothing can be said about evaluated carrying out rheometrical experiments on reconstituted sam-
its viscosity. For high volume fractions (up to 42%), the suspension ob- ples of the mixture. In this way some definitions result: the integrated
tained is a kind of paste of high strength, which easily breaks like a Model C which assumes the yield stress according to field data using
solid when it is deformed. the state of stress related to the natural deposition depth (materials
Such material cannot be considered as a fluid able to undergo revers- on the right scale); and the viscosity obtained from laboratory using
ible large deformations without changing its basic properties and it was the solid volumetric concentration at which the material behavior is
impossible to carry out accurate rheometrical tests. For intermediate fluid-like on the base of natural porosity.
volume fractions (32% b Φ b 42%), the material thus remains homoge- From the data available in the literature on the rheological parameters
neous over a reasonable time of observation and can flow like a liquid. of debris flows occurred worldwide, some of these reported in Table 1
This is in its “fluid-like” intervals. (e.g., Major and Pierson, 1992; Coussot and Piau, 1995; Martino, 2003;
The experimental flow curve obtained for Φ = 42% is reported Schatzmann et al., 2009; Kaitna et al., 2007; Pellegrino et al., 2010;
in Fig. 9. The best-fitting Bingham parameters are τc = 144 Pa and Bisantino et al., 2009; Jeong, 2010), the viscosity values do not exceed
μB = 1.8 Pa · s. The theoretical curve, used after for model B, is 10 Pa · s while the yield stress does not exceed 400 Pa. For the considered
shown in Fig. 9. These values were compared also with the results soil–water mixture (Scotto di Santolo et al., 2012) the values of the mea-
obtained with inclined plane. For the laboratory details and fitting sured viscosity and yield stress (1.8 Pa · s and 144 Pa) are in the range of
procedures, see Scotto di Santolo et al. (2012). those reported in the literature.
Regarding the shear stress of the soil before flowing, several works
4.3. Model C on the mechanical properties of pyroclastic cover in the source area of
debris flow report that this value has to be about a tens of kPa for uncon-
Several studies have already put in evidence that one-phase fluid- fined stress in order not to fail (Scotto di Santolo et al., 2000). This value
dynamic approach is suitable to describe mass movement during the of shear strength, known as apparent cohesion, is related to the matric
run-out process, even though significant difficulties remain in terms of suction present into the soil due to the unsaturated state of the cover
quantification of the model characteristic parameters (e.g., Major and as demonstrated on the basis of laboratory tests and in situ suction mea-
Pierson, 1992; Coussot and Piau, 1995; Martino, 2003; Schatzmann surements (e.g. Calcaterra et al., 2003; Cascini et al., 2008). Assuming
et al., 2009; Kaitna et al., 2007; Pellegrino et al., 2010; Bisantino et al., that the yield stress coincides with this apparent cohesion ca, the yield
2009; Jeong, 2010; Scotto di Santolo et al., 2012). On the basis of the shear stress measured during laboratory test into conventional rheom-
in situ and laboratory experience, it has been found that the value of eter is too low and in our case is about one order of magnitude less
the yield stress evaluated only from rheometrical experiments is gener- (0.1 kPa vs 10 kPa).
ally underestimated of about an order of magnitude (Coussot, 1997). Making use of dimensional analysis, for the factor of safety of an infi-
This happens because the yield stress (static and dynamic) of the mate- nite slope reported in Fig. 7 assuming a purely cohesive soil, the only di-
rial does not only depend on the characteristic of the mixtures during mensionless group is (ca / γ · h). In order to maintain the same margin of
the flow but also on the stress history in situ before failure and during safety in a model and in a prototype, not only the geometry (i) should be
depositions. On the contrary, the value of viscosity is a property of the kept constant but also the dimensionless group (ca / γ · h). If the slope of
mixture in the fluid-regime and governs the behavior during the prop- the model is the same as the prototype, so as the unit weight of the soil
agation. So, once the solid volumetric concentration at which the (γ), the ratio ca/h should be a constant. For this reason the yield stress
measured in the laboratory with an inclined plane on sample having
the same relative density, can be considered representative of the yield
stress value of the prototype if the scaling law was applied. In this as-
sumption there are a lot of simplification but with clear physical
meaning.
In the proposed model C, the viscosity is assumed equal to the one
estimated with laboratory activity, whereas the yield stress τc is as-
sumed equal to the value evaluated on the field observation according
to the procedure described for Model A (e.g., using the Eq. (6)) or by ap-
plying the scaling law to the laboratory results.

Table 3
Pozzano's debris flow. Results of the parameter calculations for Model A.

τC hplug hshear γ

τbed μB
−1
(Pa) (m) (m) (s ) (Pa) (Pa s)
Fig. 10. Longitudinal and transversal profile of Pozzano's event reconstructed using
3112 0.21 0.987 8.10 7029 484
DAN-W.
 A.M. Pellegrino et al. / Engineering Geology 196 (2015) 88–98 95

3000 Table 4
Pozzano's debris flow. Results of the numerical analysis using Models A, B, and C.
2500
Model Runout Thickness of Maximum Slope
DAN Runout distance (m)

distance deposit velocity inclination

2000
(m) (m) (m/s) (°)

A 385 0,3 4,47 29,84

1500 B 2763 0,09 87 13,63
C 1267 0,2 67 24,36
1000
Model A
500 Model B the runout distance measured in situ was 1300 m. This value is most
Model C close to the value predicted by Model C.
0 Fig. 12 shows the results of the numerical analyses for Models A, B,
0 500 1000 1500 2000 2500 3000 and C in terms of maximum flow velocity reached by the flowing
Field Runout distance (m) mass during propagation. In the graph, the maximum values of velocity
for debris flow type events from literature are reported as a straight line.
Fig. 11. Pozzano's debris flow: comparison between the measured runout distance and the Due to the lack of in situ measurements, the following values of velocity
runout distance calculated with DAN-W for the three different models proposed. are contemplated:

5. Results and discussion • the debris flow velocity calculated by Faella and Nigro (2003) through
the back analysis of the damage caused by the flows that took place in
5.1. The Pozzano case study the area of Sarno-Quindici (Campania region). The value reported by
the authors is 21 m/s;
The Pozzano debris flow has been considered as a benchmark be- • the debris flow velocity calculated by Revellino et al. (2006) through
cause of the large amount of in-situ and field data available and reliable. dynamic analysis related to geometrical characteristics of some flow
The numerical simulations performed with DAN-W on the Pozzano de- events occurred in Campania region. The value reported by the au-
bris flow was carried out considering a soil profile reconstructed from in thors is 20 m/s;
situ survey and geometrical data taken from literature (Calcaterra and • the debris flow velocity calculated by Bertolo and Wieczorek (2005)
Santo, 2004; Iadanza et al., 2009). The geometrical characteristics of through the back analysis of some debris flows that took place in
the debris flow, in terms of channel slope and channel width, are repre- California (United States). The value reported by the authors is
sented in Fig. 10. The values of the rheological parameters used for the 35 m/s;
back-analyses of this case are reported in Table 2. They were estimated • the debris flow velocity calculated by Prochaska et al. (2008) through
according to the proposed procedures, explained in Section 4, for the back analysis of some debris flows occurred in Colorado (United
models A, B, and C. The parameters obtained from field data and used States). The value reported by the authors is 21 m/s;
for Model A are shown in detail in Table 3. • the debris flow velocity calculated by Luna et al. (2011) through the
The results of the numerical analyses for models are showed in back analysis of some debris flows occurred at Faucon torrent
Fig. 11 in terms of runout distance and in Fig. 12 in terms of velocity as- (France). The value reported by the authors is 16 m/s;
sumed during the travel path. Table 4 reports all the results of the nu-
merical analysis in terms of runout distance, deposit thickness,
maximum velocity reached during propagation and angle of extension The results from models A and B do not satisfactorily match the in
(i.e. Fahrböschung). The predicted runout distances for Models A, B, situ measurement of maximum flow velocity. The results from model
and C were 385 m, 2736 m, and 1267 m, respectively. For a comparison, C rather well match the in situ measurement in terms of maximum
flow velocity, although the model overestimates the plausible values re-
ported in the literature (Luna et al., 2011; Faella and Nigro, 2003;
Bertolo and Wieczorek, 2005; Prochaska et al., 2008).
From the numerical results obtained the Model C seems to be the
most accurate in terms of total runout, debris spread, spatial distribution,
and flow velocity. The results show that past events can be modeled with
a reasonable accuracy using a combination of field and laboratory results
for the calibration of the run-out models. It as to be recalled that Model C
evaluates the yield stress τc from the geometric characteristic of the de-
bris flow deposit and the viscosity from the conventional rheometer
tests. The model C proposed by the authors is based on the observation

Table 5
Corbara's debris flow. Values assigned to the rheological parameters for cases 1a, 1b, 1c, 2,
and Model B.

Cases considered Geometric features Rheological parameters

hD iD τC μB

(m) (°) (Pa) (Pa s)

Model C c1 2 10 4434 1.8

c2 1 10 1862 1.8
c3 1.5 10 2793 1.8
Fig. 12. Back analysis of Pozzano's debris flow. Velocity profiles along the path for Models
c4 4 15 15,779 1.8
A, B, and C. The horizontal continuous lines represent the velocity limits for debris flow
Model B – – 144 1.8
type event reported in the literature.
96 A.M. Pellegrino et al. / Engineering Geology 196 (2015) 88–98

900 80
Case c1
800 70 Case c2
Model C
Case c3
700 60 Case c4

Front velocity (m/s)

Model B
600 50
Altitude (m)

500 40
Bertolo & Wieczorek, 2005

400 30
Faella & Nigro, 2003; Revellino et al, 2006
300 20
Luna et al., 2011
200 10

100 0
0 500 1000 1500 2000 2500 3000
0
Runout distance measured (m)
0 200 400 600 800 1000 1200
Runout distance (m) Fig. 15. Back calculation results for Corbara's debris flow. Velocity profiles obtained with
DAN-W for the five cases considered. The horizontal continuous lines represent the veloc-
Fig. 13. Longitudinal profile of Corbara's debris flow reconstructed using DAN-W. ity limits for a typical debris flow event deducted from the literature.

that the evaluation of τc from the field is a very robust straightforward

method using the natural materials on the right scale. This τc value can Table 5. For a comparison, a numerical simulation was performed with
predict very well the run-out distance. Instead the viscosity values are Model B, as well.
less important on the estimation of the travel distance but they assume The numerical simulations with DAN-W on the one of Corbara debris
a great importance to estimate the velocity. Moreover the validation on flows were performed considering a flow profile reconstructed from the
the velocity was very weak because of the poor information on real veloc- geometric characteristics of the event taken from literature and field
ity data available in this studied area. survey (Di Crescenzo and Santo, 2005; Scotto di Santolo, 2002).
The longitudinal slope profile is schematically represented in Fig. 13.
In Fig. 14, the results of the numerical analysis are shown for the cases
5.2. The Corbara case study considered in terms of runout distance. The predicted runout distances
for the Model C cases c1, c2, and c3 are 699 m, 1363 m and 1157 m, re-
The back analyses conducted for the Pozzano case, which was cho- spectively. No numerical results were obtained for Model A because the
sen as a benchmark, have shown that Model C is the most accurate pro- model did not predict a motion. For a comparison, the runout distance
cedure for calibrating the rheological parameters of the constitutive measured on field was 1145 m. The runout distance was most closely
model in terms of runout distance and maximum velocity of the front. predicted by case c3. Fig. 15 shows the results of the numerical analysis
The Corbara debris flow event was used to validate the proposed proce- in terms of maximum front velocity reached by the flowing mass during
dure, Model C, and its parameters. The values of the parameters of the propagation. For each case, the results rather match well in terms of
geometric fan for the Corbara debris flow are not uniquely defined in maximum front velocity. The back-calculated values for maximum
the literature. In order to investigate the influence of deposit shape on flow velocity are compatible with the reasonable range of values report-
runout, four combinations of geometric features were considered. ed in literature as explained in Section 5.2. Model B overestimated the
These scenarios (c1, c2, c3, and c4) were obtained on the basis of the maximum flow velocity. The results of the numerical simulations for
values of the deposit characteristics reported in the literature the Corbara case are summarized in Table 6.
(Calcaterra and Santo, 2004; Di Crescenzo and Santo, 2005; Scotto di
Santolo, 2002). These scenarios arise from the dependence of the yield
stress τc on the deposit's shape when the hypothesis of the uniform 6. Conclusion
flow was assumed and the rheological parameter considered are calcu-
lated using Eq. (6) (see Model A in Section 4.1). In order to define the In the present work, two well-documented debris flow case histories
parameters of the four scenarios, a parametric analysis was conducted. suitable for back analysis were selected. The case histories of Pozzano
The values of the rheological parameters obtained are reported in and Corbara were analyzed in 2D using DAN-W (Hungr, 1995). The clas-
sic, one-phase Bingham rheological model was chosen. For the Pozzano
case study, three approaches were adopted to set the characteristics pa-
rameters of the model (i.e., the yield stress, τc, and the Bingham viscos-
ity, μB). Using the first approach, Model A, the rheological parameters
were determined based on the geometric characteristics of the debris
flow event measured in situ. For Model B, the rheological parameters

Table 6
Corbara's debris flow. Results of the numerical analysis for Model C and Model B.

Cases Runout distance Thickness of deposit Maximum velocity

considered
(m) (m) (m/s)

Model C c1 699 1 12.50

c2 1363 0.5 44.69
c3 1157 0.5 34.98
c4 443 – 0
Fig. 14. Corbara's debris flow: comparison between the measured and calculated runout
Model B 2677 0.3 70.65
distance for the different models and scenarios proposed.
 A.M. Pellegrino et al. / Engineering Geology 196 (2015) 88–98 97

were derived using experimental data obtained through rheometer Cruden, D.M., Varnes, D.J., 1996. Landslides: investigation and mitigation, landslides types
and processes. In: Turner, A.K., Schuster, R.L. (Eds.), Transportation Research Board
tests and inclined plane performed on mixtures of water and sediment Special Report 247. National Academy Press, WA, pp. 36–75.
using a pyroclastic soil similar to that involved in the debris flow event Di Crescenzo, G., Santo, A., 2005. Debris slides-rapid earth flows in the carbonate massifs
being considered (Scotto di Santolo et al., 2012). An integrated proce- Campania region (Southern Italy): morphological and morphometric data for evalu-
ating triggering susceptibility. Geomorphology 66, 255–276. http://dx.doi.org/10.
dure to fit model C, in which the yield stress parameter τc was calculated 1016/j.geomorph.2004.09.015.
from the geometric characteristic of the considered debris flow event Faella, C., Nigro, E., 2003. Dynamic impact of the debris flow on the constructions during
measured in situ, and the Bingham viscosity μB was calculated from the hydrogeological disaster in Campania-1998: failure mechanical models and eval-
uation of the impact velocity. Fast Slope Movements — Prediction and Prevention for
the experimental data obtained from the rheometer test. The numerical Risk Mitigation. Naples 1, pp. 179–186.
results for each model were assessed by matching the total horizontal Fannin, R.J., Rollerson, T.P., 1993. Debris flow: some physical characteristics and behav-
runout and the flow velocities to the values measured in situ. Model C iour. Can. Geotech. J. 30, 71–81. http://dx.doi.org/10.1139/t93-007.
Fannin, R.J., Wise, M.P., 2001. An empirical-statistical model for debris flow travel dis-
is the most accurate in terms of total runout, debris spread, spatial dis-
tance. Can. Geotech. J. 38, 982–994. http://dx.doi.org/10.1139/t01-030.
tribution, and flow velocity. The results show that past events can be Heim, H., 1932. Landslides and Human Lives. In: Skermer, N. (Ed.), Bi-Tech Publishers,
modeled with reasonable accuracy using yield stress τc calculated Vancouver, p. 196.
from geometric characteristics of the debris flow event measured in Hungr, O., 1995. A model for the runout analysis of rapid flow slides, debris flows, and av-
alanches. Can. Geotech. J. 32 (4), 610–623. http://dx.doi.org/10.1139/t95-063.
situ and Bingham viscosity that has been experimentally measured. To Hungr, O., McDougall, S., 2009. Two numerical models for landslide dynamic analysis.
investigate the influence of deposit shape on runout, four combinations Comput. Geosci. 35, 978–992. http://dx.doi.org/10.1016/j.cageo.2007.12.003.
of geometric features were considered. These results have highlighted Iadanza, C., Triglia, A., Vittori, E., Serva, L., 2009. Landslides in coastal areas. In: Violante, C.
(Ed.), Geohazard in Rocky Coastal Areas. Geological Society Special Publication 322,
significant considerations regarding the flow dynamics for the events pp. 121–142.
analyzed. In particular, when estimating resistance stress (i.e. yield Jeong, S.W., 2010. Grain size dependent rheology on the mobility of debris flows. J. Geosci.
stress), accurate measurement of deposits leads to very good predic- 14 (4), 359–369. http://dx.doi.org/10.1007/s12303-010-0036-y.
Jeong, S.W., 2013. The viscosity of fine-grained sediments: a comparison of low-to
tions of travel distance, velocity, and flow depth. The calibration of the medium-activity and high-activity clays. Eng. Geol. 154, 1–5. http://dx.doi.org/10.
rheological parameters is clearly scale-dependent. However, the appli- 1016/j.enggeo.2012.12.006.
cation of a combined approach for fitting rheological parameters of a Johnson, A.M., 1970. Physical Processes in Geology: a Method for Interpretation of Natural
Phenomena Intrusions in Igneous Rocks, Fractures and Folds, Flow of Debris and Ice.
model supports us to identify a procedure suitable for assessing the dy- Freeman, Cooper and Co., San Francisco, California.
namic characteristics of debris flow. Kaitna, R., Rickenmann, D., Schatzmann, M., 2007. Experimental study on the rheologic
Integrated with back analysis of historical cases, the use of both in behaviour of debris flow material. Acta Geotech. 2, 71–85. http://dx.doi.org/10.
1007/s11440-007-0026-z.
situ observations and laboratory experience can lead to accurate
Luna, B.Q., Remaître, A., van Asch, Th.W.J., Malet, J.P., Van Westen, C.J., 2011. Analysis of
methods for the prediction of the dynamic parameters of potential de- debris flow behaviour with a one dimensional run-out model incorporating entrain-
bris flow events. The simulated, order-of-magnitude effects of the ana- ment. Eng. Geol. 128, 63–75. http://dx.doi.org/10.1016/j.enggeo.2011.04.007.
lyzed case studies could be helpful for predicting future events not Major, J.J., Pierson, T.C., 1992. Debris flow rheology: experimental analysis of fine-grained
slurries. Water Resour. Res. 28 (3), 841–857. http://dx.doi.org/10.1029/91WR02834.
only in the considered area. Although it is still difficult to make accurate Mangeney, A., Vilotte, J.P., Bristeau, M.O., Perthame, B., Bouchut, F., Simeoni, C., Yerneni, S.,
predictions about the most likely runout, the simple analysis conducted 2003. Numerical modelling of avalanche based on Saint Venant equations using ki-
in this study provides useful hints about dynamic behavior of possible netic scheme. J. Geophys. Res. 108, 2527. http://dx.doi.org/10.1029/2002JB002024.
Martino, R., 2003. Experimental analysis on the rheological properties of debris-flows de-
debris flows regardless the geographical contest. It is worth noting that posit. In: Rickenmann, D., Chen, C.L. (Eds.), Proceeding 3rd International Conference
the properties of the granular-fluid mixture as a whole change as the mix- Debris Flow Hazard Mitigation: Mechanics, Prediction and Assessment. Millpress,
ture flows downhill, and they affect the reliability of the model prediction. Davos, Rottedradam.
McDougall, S., Hungr, O., 2004. A model for the analysis of a rapid landslide runout motion
Further research, including experimental tests, should address this topic. across three-dimensional terrain. Can. Geotech. J. 41 (6), 1084–1097. http://dx.doi.
org/10.1139/t04-052.
Naef, D., Rickenmann, D., Rutschmann, P., McArdell, B.W., 2006. Comparison of flow resis-
References tance relations for a debris flows using a one-dimensional finite element simulation
model. Nat. Hazards Earth Syst. Sci. 6, 155–165.
Bertolo, P., Wieczorek, F., 2005. Calibration of numerical models for small debris flows in Nigussie, D.G., 2013. Numerical Modelling of Run-Out of Sensitive Clay Slide Debris (Mas-
Yosemite Valley, California, USA. Nat. Hazards Earth Syst. Sci. 5, 993–1001. http://dx. ter Thesis) p. 95.
doi.org/10.5194/nhess-5-993-2005. Pellegrino, A.M., Scotto di Santolo, A., Evangelista, A., Coussot, P., 2010. Rheological behav-
Bisantino, T., Fischer, P., Gentile, F., 2009. Rheological characteristics of debris-flow mate- iour of pyroclastic debris flow. Third International Conference on Monitoring, Simu-
rial in South-Gargano watersheds. Nat. Hazards 54, 209–223. http://dx.doi.org/10. lation, Prevention and Remediation of Dense and Debris Flow “Debris Flow III”,
1007/s11069-009-9462-4. 2010, 24–26 May, Milano, Italy.
Blanco, A., Rodriguez, C., Garcìa, R., 2009. Can we accurately use laboratory-scale rheolog- Pellegrino, A.M., Scotto di Santolo, A., Schippa, L., 2014. Numerical Modeling of a Debris
ical parameters to model field-scale debris flow? Rev. Fac. Ing. 1 (24), 45–55. Flow Event Occurred in Campania Region, Southern Italy: Consideration on the Rhe-
Calcaterra, D., Santo, A., 2004. The January 10, 1997 Pozzano landslide, Sorrento Peninsu- ological Model Parameters on the Run-Out, Latest Trends in Engineering Mechanics,
la, Italy. Eng. Geol. 75, 181–200. http://dx.doi.org/10.1016/j.enggeo.2004.05.009. Structures, Engineering Geology, 2014, 3–5 June, Salerno (Italy), pp. 62–71.
Calcaterra, D., de Riso, R., Evangelista, A., Nicotera, M.V., Santo, A., Scotto di Santolo, A., Prochaska, A.B., Santi, P.M., Higgins, Cannon, S.H., 2008. A study of methods to estimate de-
2003. Slope instabilities in the pyroclastic deposits of the Phlegraean district and bris flow velocity. Landslides 5, 431–444. http://dx.doi.org/10.1007/S10346-008-0137-0.
the carbonate Apennine (Campania, Italy). Proceeding of the International Workshop Pudasaini, S.P., Hutter, K., 2007. Avalanche Dynamics–Dynamics of Rapid Flows of Dense
on Occurrence and Mechanisms of Flows in Natural Slopes and Earthfills, IW- Granular Avalanches. Springer-Verlan, Berlin.
FLOWS2003, Pàtron Ed., Bologna, Sorrento, May 14–16, 2003, pp. 61–75. Revellino, P., Hungr, O., Guadagno, F.M., Evans, S.G., 2004. Velocity and runout prediction of
Cannon, S.H., 1993. An empirical model for the volume-change behaviour of debris flows. destructive debris flows and debris avalanches in pyroclastic deposits, Campania Re-
Hydraulic Engineering Conference, American Society of Civil Engineering, 93, 1993, gion, Italy. Environ. Geol. 45, 295–311. http://dx.doi.org/10.1007/s00254-003-0885-z.
25–30 July, San Francisco, pp. 1768–1993. Revellino, P., Guadagno, F.M., Hungr, O., 2006. Criteri morfologici e modellazione
Cascini, L., Cuomo, S., Guida, D., 2008. Typical Source Areas of May 1998 Flow-Like dinamica nella valutazione della suscettibilità di frana lungo versanti carbonatici
Mass Movements in the Campania Region, Southern Italy. Eng. Geol. 96, 107–125. dell'Appennino campano. G. Geol. Appl. 3. http://dx.doi.org/10.1474/GGA.2006-03.
http://dx.doi.org/10.1016/j.enggeo.2007.10.003. 0-21.0114 (159–156).
Chen, H., Lee, C.F., 2000. Numerical simulation of debris flows. Can. Geotech. J. 37, Rickenmann, D., 2005. Runout prediction methods. In: Jakob, M., Hungr, O. (Eds.), Debris-
146–160. http://dx.doi.org/10.1139/t99-089. Flow Hazards and Related Phenomena, pp. 305–324.
Christen, M., Barlett, P., Kowalski, J., Stoffel, L., 2010. Calculation of dense snow avalanches Savage, S.B., Hutter, K., 1989. The motion of a finite mass of a granular material down a rough
in three-dimensional terrain. Cold Reg. Sci. Technol. J. 63 (1–2), 1–14. http://dx.doi. incline. J. Fluid Mech. 199, 177–215. http://dx.doi.org/10.1017/S0022112089000340.
org/10.1016/j.coldregions.2010.04.005. Schatzmann, M., Bezzola, G.R., Minor, H.E., Windhab, E.J., Fischer, P., 2009. Rheometry for
Corominas, J., 1996. The angle of reach as a mobility index for small and large landslides. large particulated fluids: analysis of the ball measuring system and comparison to de-
Can. Geotech. J. 33, 260–271. http://dx.doi.org/10.1139/t96-005. bris flow rheometry. Rheol. Acta 48, 715–733. http://dx.doi.org/10.1007/s00397-009-
Coussot, P., 1997. Mudflow rheology and dynamics. IAHR Monograph Series. Balkema, 0364-x.
Rotterdam. Schippa, L., Pavan, S., 2011. Numerical modeling of catastrophic events produced by mud
Coussot, P., Piau, J.M., 1995. A large-scaled field concentric cylinder rheometer for or debris flows. Int. J. Saf. Secur. Eng. 1 (4), 403–422. http://dx.doi.org/10.2495/SAFE-
the study of the rheology of natural coarse suspensions. J. Rheol. 39, 105–124. V1-N4-403-422.
http://dx.doi.org/10.1122/1.550693. Scotto di Santolo, A., 2002. Le colate rapide, Hevelius.
98 A.M. Pellegrino et al. / Engineering Geology 196 (2015) 88–98

Scotto di Santolo, A., Evangelista, A., 2009. Some observations on the prediction of the dy- Sheidegger, A.E., 1973. On the prediction of the reach and velocity of catastrophic land-
namic parameters of debris flows in pyroclastic deposits in the Campania region of slides. Rock Mech. 5 (4), 231–236. http://dx.doi.org/10.1007/BF01301796.
Italy. Nat. Hazards 50, 605–622. http://dx.doi.org/10.1007/s11069-008-9334-3. Van Westen, C.J., van Asch, Th.W.J., Soeters, R., 2006. Landslides hazards and risk
Scotto di Santolo, A., Nicotera, M.V., Evangelista, A., Pellegrino, A., Ramondini, M., Urciuoli, zonation—why is it still do difficult? Bull. Eng. Geol. Environ. 65 (2), 167–184.
G., 2000. Some remarks on the shear strength of neapolitan pyroclastic deposits. In- http://dx.doi.org/10.1007/s10064-005-0023-0.
ternational Conference on Geotechnical Engineering, GeoEng (Melbourne, novembre Zhang, M., Yin, Y., 2013. Dynamics, mobility-controlling factors and transport
2000, ISBN 1-58716-068-4). mechanisms of rapid long-runout rock avalanches in China. Eng. Geol. 167, 37–58.
Scotto di Santolo, A., Pellegrino, A.M., Evangelista, A., Coussot, P., 2012. Rheological behav- http://dx.doi.org/10.1016/j.enggeo.2013.10.010.
iour of reconstituted pyroclastic debris flow. Géotech. J. 62 (1), 19–27. http://dx.doi.
org/10.1680/geot.10.P.005.