railway tunnel

Interaction of a railway tunnel with a deep slow landslide in clay shales


Varco d’Izzo landslide system develops in an urbanized area at the eastern suburbs of Potenza, the capital of Basilicata Region. It is located in a slope facing the valley of Basento river that the geotechnical research group of University of Basilicata has been monitoring and studying for a long time. Many results have already been found and published for the adjacent Costa della Gaveta landslide 1,2,3. More recently, attention has been focused on Varco d’Izzo landslide system and, in particular, on its interaction with man-made works. The highway and the railway, whose tunnel crosses the accumulation zone of an earthflow of the landslide system (Fig. 1), need frequent maintenance and continuous monitoring. The railway line is important for the whole region since it connects Potenza to the Tyrrhenian and Ionian coasts. The earthflow displacements, slow but continuous, are in the order of several cm/year upslope from the highway and of one cm/year in the accumulation. They have led, in the last years, to the eviction of a house, the dismantling of a pedestrian bridge, damages to roads and other structures.
The geology of the earthflow, which develops in Varicoloured Clays, has been studied by several authors among whom 4; Di Maio et al.5 reported results of a geotechnical investigation carried out in 2005 in the same zone, with six boreholes equipped with inclinometers. The authors analyzed data from other previous geotechnical investigations together with those of the 2005 investigation. All the experimental data were used to define the main geometrical and kinematic features of the landslide system. Calcaterra et al.6 reported displacement time trends evaluated by inclinometers and by a network of permanent and non-permanent GPS stations still operative nowadays in the area under study.
The railway tunnel “Calabrese”, 200 m long, crosses the accumulation. Its state of fissuring has significantly advanced in the last years, as will be described in the section 4. The tunnel can actually be considered an artificial tunnel, having maximum overburden of about 5 m. It was completely re-built in 1992 between two strutted sheet pile walls, by the cut and-cover method, after the previous tunnel had suffered severe damage due to the landslide.
The interaction landslide-structure is thus characterized by the fact that the latter is constituted by different elements (piled walls and tunnel box structure) that were constructed in sequence and that also interact with one another. Most research studies available in the literature proposing methods for the evaluation of the stabilizing force that is provided in a landslide by a single row of piles (e.g. 7,8,9,10) are thus not specifically useful for the studied case. Moreover, unfortunately, the literature is not very rich of comprehensive studies of landslide-tunnel interactions (see 11 and references therein) and most of them consider the case of circular, deep tunnels not reinforced by piles (e.g. 12, 13).
In this paper, a preliminary analysis of the landslide-tunnel interaction in Varco D’Izzo earthflow is carried out by considering the results of inclinometer and GPS measurements, in-situ tunnel inspections, laser scanner surveys, simplified FEM calculations.

Fig. 1. a) Map with location of study area and railway tunnel, and with indication of main damages due to the landslide; b) Front view of railway
tunnel “Calabrese”.

Brief description of the earthflow and its kinematics

The landslide occurs in a clay shale formation of Upper Cretaceous – Oligocene, locally known as Varicoloured Clays, constituted by a succession of tectonized, chaotic, heterogeneous scaly clays. There are marly clays in the lower part of the formation and calcareous marls, calcilutites and mcalcarenites in the upper part. The accumulation is very heterogeneous, with a fine matrix including rock fragments and blocks whose size varies in a wide range. The matrix is characterized by a clay fraction in the range 25% – 70%, the liquid limit wL ranges between 40% and 80%.
In these soils the residual friction angle too varies in a wide range, and can be even lower than 5°. Due to the difficulty of undisturbed sampling, the peak values of shear strength parameters could not be reliably evaluated5. Section 5 will thus consider parametric values for earth pressure estimation.
The landslide deformations and displacements were evaluated by means of inclinometer, GPS and topographic measurements. All the measurements were carried out after the reconstruction of the tunnel. The inclinometer profiles allow to reconstruct the geometry of possible slip surfaces or, at least, of parts of them. Some inclinometer profiles are sketched in Fig. 2 in two vertical sections of the landslide accumulation. The figure gives an idea of the complexity of the displacement field. In particular, the hypothesis of the existence of two slip surfaces, one at about 20 m and the other at about 40 m depth in the tunnel section, sounds reasonable. However, the profiles refer to different times (reported between brackets in the figure) and different time intervals, besides different experimental techniques. Furthermore, the deepest “slip surface” in some verticals coincides with the boundary between the clay formation and the ancient Basento river bed14, where the relative movements can be due to processes different from those currently influencing the landslide.
Actually, further systematic investigation is required for a consistent description of all the phenomena occurring in the considered part of the landslide. At the moment, the available inclinometer profiles can be used with confidence only to put in evidence: i) the depths of failure corresponding to the shallower slip surface, and ii) the deformations of the accumulation above it. As for i), it is worth noting that in 2001 and 2004 the inclinometers F6 and FSB, close to the upslope wall of the tunnel in correspondence to the axis of the landslide, underwent failure beneath the piles and not on a surface which could cross them. As for ii), displacement rates underwent limited

Fig. 2. Vertical sections across the landslide foot with inclinometer profiles: a) plan view; b) section 1-1′; c) section 2-2′.

variations in the monitoring period5 and thus it is possible to hypothesize a general displacement rate field. This latter (Fig. 3a) shows that the displacement rate decreases from the highway to the tunnel (i.e. from inclinometer S1f to FS6), suggesting an important role of the tunnel in contrasting soil movements. However, Fig. 3b shows that GPS benchmark CS01, very close to the tunnel entrance, moves by about 1 cm/year 6. Displacement rates measured by GPS have kept fundamentally the same annual rate also in the last years 15. Summing up, the available data suggest that: a) until 2004 the tunnel with its sheet pile wall was able to contrast the landslide movements, at least in correspondence to the axis of the landslide; b) the western entrance of the tunnel moved by about 9 cm in the last 9 years.

Fig. 3. Distribution of average yearly rates of a) relative displacements on the upper slip surface; b) displacements at ground level.

Description of railway tunnel “Calabrese”

The artificial tunnel “Calabrese” was re-built in 1992 by the ‘cut-and-cover’, bottom-up technique. It was constructed by excavating to the maximum depth of about 15 m with the support of strutted piled walls (Fig. 4), casting a concrete box structure, and covering it with soil. A schematic section is provided in Fig. 5a. The concrete box has internal dimensions 7.5 x 7.5 m. Along its length, it is constituted by eight segments separated by construction joints. The overburden reaches a maximum of 5 m at the middle length of the tunnel. The two rows of sheet piled walls are made of contiguous piles, 1 m in diameter and 20 m in length, connected with a top capping beam. Two double rows of piles were constructed at the entry and exit (Fig. 6). Characteristics of tunnel lining and pile sections, obtained on the basis of available design drawings and documents16 provided by the railway company, are summarized in Table 1.

Fig 4. Construction of railway tunnel Calabrese by the cut and cover method: a) schematic section; b) photo of construction stage before casting
reinforced concrete box lining.

The design documents of the 1990s contain no specific consideration of the effects of landslide displacements, suggesting that the pile walls were designed and constructed only as temporary support to the excavation. In particular, the tunnel lining was designed by the structural scheme and loads reported in Figs. 5b and 5c considering i) as vertical loads self weight of concrete members and 5 m soil surcharge on the top slab, and ii) as horizontal load, the effect of the increment of horizontal earth pressure due to = 5 m backfill of the excavation. Such increment was calculated by assuming at rest (K0) conditions and, thus, a pressure equal to Ko· ·Δ, with being soil bulk unit weight. The state of stress in reinforced concrete lining was obtained by linear elastic analysis.

Fig. 5. a) Schematic section of the tunnel at its middle length; b) structural scheme used to simulate pile-tunnel interaction; c) design loads used
in 1990’s design of the tunnel concrete box.
Fig. 6. Plan view of the tunnel showing different typologies of piles used: A (diameter D=1 m, length L=20 m); B (diameter D=1 m, length L=21
m); C (diameter D=1 m, length L=16 m).

Tunnel surveys

A new field investigation of the tunnel w as carried out in September 2015. Many cracks, broken glassmeters, joints open along the entire height of the tunnel were detected (Fig. 7). In comparison with previous inspections, a more extensive pattern of cracks was found. The cracks are mostly vertical, and more severe ones are located along the upslope wall facing the landslide.
A survey was performed by means of a pacometer to verify the location and spacing of the actual as-built arrangement of steel reinforcement, showing their consistency with data provided in the original design drawings.
Indirect determination of concrete properties was also carried out by performing rebound hammer tests from the inside of the tunnel, on the upslope wall. The main objective was to check the uniformity of concrete. In fact, rebound test, as other non destructive test methods, should not be used to directly estimate in-situ concrete strength, while it can be used effectively as a means of determining the uniformity of concrete properties in structures17. Test results show a good uniformity with a min-max range of the rebound measurements equal to 38-43. Based on the mean value of all the measurements (equal to 41), the cylindrical strength value fc estimated through the correlation curves provided in some references (e.g.18) should be around 35 MPa, larger than the value provided in the original design specifications, thus showing the development of carbonation effects, typical of old concretes.
A high definition survey of the current internal geometry was done on October 2015 by 3D laserscanning19. The scan provided the geometrical anomalies throughout the length of the tunnel thus contributing to understand the deformation pattern due to landslide movements. In particular, it showed that in some portions of the tunnel, mostly in the western part, the deformed shape of vertical cross sections is the one that would be expected from unbalanced actions acting on the upslope and downslope wall, with a maximum relative displacement of about 5 cm between tunnel roof and base. In other portions, the shape is more symmetrical (Fig. 8).

Fig. 7. a) 3D representation of tunnel lining damages in the central sectors (more severe cracks in red); b) open joint in the tunnel.

Preliminary structural analyses of landslide-tunnel interaction

This section reports the results of some analyses carried out to preliminarily study the stress and deformation state of the tunnel as a result of the soil pressure exerted by the landslide.
As mentioned above, the tunnel lining is made up of eight segments separated by construction joints and is protected by sheet pile walls with a top beam connecting all the piles. This connection is effective only in the upper part of the structure allowing the piles to behave and deform independently in the deeper one. Therefore, different segments are generally subjected to different actions and, beyond an internal stress state, they experience relative displacements. For these reasons, the study of the landslide-tunnel interaction is really complex and requires sophisticated FEM analyses in order to adequately take into account all the interaction effects and the real 3D nature of the problem. At the current state of the study, simplified analytical developments along with 2D FEM structural analyses have been performed in order to get a first interpretation of the observed behavior and field data.

Fig 8. Schematic plan view (a), and cross sections of the tunnel at 50 m (b), and 125 m (c) from entry. Laser scanner survey and restitution by Dr. M. Limongiello (University of Salerno).


The Calabrese railway tunnel crosses the accumulation of a deep and complex earthflow in clay shales.
Displacement data collected since the re-construction of the tunnel in 1992 seem to suggest that in correspondence to the axis of the landslide, until 2004 the upslope sheet pile wall had not undergone significant deformations due to landslide movements. However, the monitoring carried out since 2006 shows that the western entrance of the tunnel moved by about 1 cm per year in the last 9 years. A detailed study, both experimental and theoretical, is currently being carried out to evaluate the stress-strain condition of the structure and clarify if the piles are still able to hinder landslide movements.
Preliminary structural analyses were performed by a 2D FEM non-linear model of a tunnel cross section which was possible to elaborate thanks to the availability of the original design drawings. The structural part was based on a lumped plasticity model. Earth pressure acting from upslope at the tunnel level was considered with triangular distribution and varying between at rest conditions and passive conditions. Due to the significant heterogeneity of the accumulation zone and to the difficulty of obtaining undisturbed samples, the peak values of soil shear strength parameters could not be reliably evaluated. The analyses thus had to be parametric, with the friction angle varying in a reasonable range of values. Even considering that a quota of the observed deformation should be ascribed to concrete creep effects and construction defects, the calculated deformations compare satisfactorily with the real measured deformations of the tunnel lining. In addition to the FEM model, simplified calculations allowed to confirm that the tunnel lining is in the post-elastic branch of its moment curvature relationship although the chord rotation demand is only a fraction of the rotation capacity.
The shape of the accumulation, in-situ investigation of the state of damage, laser scan surveys of the current internal geometry of the tunnel suggest that the landslide-tunnel interaction is strongly influenced by 3D effects.
Thus, future study, which will benefit from new field investigations, will be focused on the 3D behaviour of the system, also by using the appropriate geotechnical and structural software. With the collaboration of the National Railway Company, the results of new inclinometer and earth pressure measurements will play a fundamental role in ascertaining the current state of health of the tunnel and its possible evolution with time.

Source: Interaction of a railway tunnel with a deep slow landslide in clay shales

Authors: Roberto Vassallo, Mayank Mishra, Giuseppe Santarsiero, Angelo Masi

One thought on “Interaction of a railway tunnel with a deep slow landslide in clay shales

Leave a Reply

Your email address will not be published. Required fields are marked *