Introduction

Mitigation of large, deep, and fast-moving landslides is a difficult engineering task, which requires a good understanding of the causes and mechanisms that lead to slope failure. According to the International Union of Geological Sciences (IUGS) Working Group on Landslides, the landslide mitigation process includes the choice of one or a combination of the following four principle measures: modification of slope geometry, drainage, retaining structures, and internal slope reinforcement (Popescu 2001). The key elements for the successful use of one or a combination of these measures are field investigations and measurements that must provide all information necessary to investigate the stability of slopes.

At the time when several large deep-seated landslides with volumes of the order of 1 million m3 were triggered during heavy rainfall in Slovenia (central Europe), there was little local experience with such large landslides. It was obvious that the mitigation approach should follow the basic principles combined with experience gained in the rehabilitation of smaller in size but phenomenologically similar landslides. In the case of the Macesnik and the Slano blato landslides, the general approach to mitigation was based on the frequently used measures in the Alpine region. These measures include the construction of surface drainage system and subsurface drainage system by means of interceptor trench drains or drainage wells in order to slow down the landslide movements to the extent that allows the construction of retaining structures in the form of stabilizing piles, retaining walls, or large diameter cast-in-place piles/shafts (Corsini et al. 2006; Marcato et al. 2012). For the structural mitigation of the two landslides, deep reinforced concrete (RC) shafts/wells were selected in order to combine the restraining effect of stabilizing piles (Hong and Han 1996; Popescu and Schaefer 2008; Kang et al. 2009; Song et al. 2012) and the dewatering effect of deep bore wells (Marschallinger et al. 2009) in a single type of construction.

Once detailed field data are collected, the design process of structural mitigation of the landslide is governed by slope stability analysis and retaining structure design calculations by means of various analytical and numerical methods. Although some analytical methods enable the calculation of lateral forces on piles/shafts in soil undergoing lateral movement (Ito and Matsui 1975; Ito et al. 1982; Firat 2009) and can be used for the design, they usually require a separate uncoupled slope stability analysis based on the limit equilibrium method in order to incorporate the reaction exerted on the unstable soil by the piles/shafts. Due to the three-dimensional (3D) nature of the problem and the complicated interaction behavior between the shaft and the moving soil mass, some of these methods are reported to be unreliable (Ito et al. 1982; Firat 2009). This is one of the reasons why numerical methods such as two-dimensional (2D) or 3D Finite Element Method (FEM) are becoming a viable, but not widely applied, alternative (Ng et al. 2001; Liang and Zeng 2002; Liang and Yamin 2010) for the structural design and stability evaluation.

In this paper, a brief history of the Macesnik and the Slano blato landslides is presented, with the emphasis on the mitigation approach and structural design of RC shafts/wells by means of 3D finite element analysis, together with construction experiences and performance evaluation after the completion of mitigation works.

Field conditions and brief history of the Macesnik and the Slano blato landslides

The Macesnik landslide is located at 46°26′ N, 14°41′ E above the village of Solčava in N Slovenia near the border with Austria (Figs. 1 and 2). The landslide was triggered in 1989 and is more than 2,500 m long, up to 100 m wide, and, on average, 10 to 15 m deep with an estimated volume of the sliding mass about 2 million m3. The active landslide lies within the fossil landslide that is up to 350 m wide and 50 m deep with the total volume estimated at 8 to 10 million m3. The crown of the landslide is at an altitude of 1,360 m and the toe, where the landslide was stopped by a large rock outcrop, at 840 m asl. In the period between 2000 and 2004, the landslide movements in the most critical upper part of the landslide reached up to 50 cm/day with an average value of 15 cm/day. The detailed description of the landslide occurrence, together with the explanation of geological–geotechnical conditions and measures taken in the period between 2001 and 2004 for a stepwise mitigation of the Macesnik landslide was previously reported by Mikoš et al. (2005). The mitigation works, which consisted of subsurface drainage works and retaining works with deep RC shafts/wells, were completed in 2006.

Fig. 1
figure 1

Locations of the five large landslides in Slovenia

Full size image
Fig. 2
figure 2

Aerial view of the Macesnik landslide in 2005

Full size image

The same approach using drainage works and RC shafts, acting also as drainage wells, was applied on the Slano blato landslide, which is located at 45°54′ N 13°51′ E (Figs. 1 and 3) in the Eocene flysch region of western Slovenia. The landslide was first mentioned in 1887 with the occurrence of an earth flow destroying the main road in the valley about 2 km away. The landslide was thereafter remediated with a series of torrential check dams along the Grajšček stream. The landslide remained stable till November 2000, when during heavy rainfall, a large landslide of mud and debris was triggered, moved along the Grajšček stream at a speed of up to 100 m/day and threatened the village of Lokavec situated in the valley below. The history of the landslide and geological, hydrogeological, hydrological, and geotechnical conditions for the occurrence of the landslide have been reported by several authors (Kočevar and Ribičič 2002; Majes et al. 2002; Logar et al. 2005; Placer et al. 2008). The landslide is about 1,600 m long, 60 to 250 m wide, with a total sliding mass of about 900,000 m3 located between 270 and 650 m asl. In 2001 and 2002, approximately 230,000 m3 of accumulated debris material were removed from the area of a rockfill dam that was built about 300 m above the village as a part of protection measures, followed by surface and subsurface drainage works and a construction of three RC shafts/wells in 2004, with gradual construction of additional eight RC shafts that were completed in 2007.

Fig. 3
figure 3

Aerial view of the Slano blato landslide in 2006

Full size image

Although the conditions for the occurrence of the two landslides are geologically different—the Slano blato landslide is conditioned by weathering of the flysch at the border of Triassic limestone and Eocene flysch and the Macesnik landslide by complex geological conditions at the junction of Carboniferous, Triassic, and Oligocene rocks—they have a lot in common. Both landslides can be categorized as rainfall-induced landslides, which can also behave as earth flows consisting of clayey/silty gravels with high percentage of fines (>30 %) of low plasticity (PI > 30) (Logar et al. 2005; Fifer and Zupančič 2009). Both landslides are long and relatively shallow with variable slope inclination and cannot be stopped by a single retaining structure. Thus, a similar design approach was adopted in the selection and design of mitigation measures for both landslides.

Evaluation of mitigation measures

In cases when rainfall-induced landslides are active, any execution of restraining work is a difficult, if not impossible, engineering task. It is of utmost importance that the restraining measures are designed not only to efficiently restrain the sliding mass but also to allow safe construction, with the likelihood of damage during the construction process reduced to an acceptable level. There were at least three additional problems that confronted the design of retaining constructions for the Macesnik and the Slano blato landslides. The first is that under heavy rainfall, the soil above the sliding plane can turn into a viscous earth flow, where the use of conventional geotechnical methods and material models can hardly be implemented. The second problem was associated with the length of both landslides, irregularity of the geological profile in the longitudinal and transverse directions, and finally, in the change of the terrain slope along the landslides. The third problem is the length of both landslides as compared to the depth of the sliding surface. In such conditions, it is usually impossible to stabilize the landslide with one retaining structure at a single location. Therefore, in order to achieve the best effect on stabilization, the selection of retaining structure locations was extremely important.

In the case of the Macesnik landslide, it was evident that the narrowing of the landslide in the area of a pontoon bridge is an ideal place for the first retaining structure in order to assure stability of the upper part of the landslide (Fig. 2). For the stabilization of the middle part of the landslide, the second location was selected in the area where the landslide is twice crossed by the ruined panoramic road. The third possible location was considered in the lower part of the landslide above the rock outcrop as shown in Fig. 4.

Fig. 4
figure 4

Map of the Macesnik landslide with marked areas for the construction of retaining structures (from Mikoš et al. 2005)

Full size image

For the Slano blato landslide, the shaft location (see Fig. 3) was selected on the basis of observed sliding mechanism and professional judgment that is necessary to retain the sliding mass in the upper part of the landslide and prevent slippage of large quantities of weathered material along the Grajšček stream and to reduce progressive spreading of the landslide in the uphill direction. Considering the geology, size, and dynamics of the two landslides, the surface and subsurface drainage works (surface drainage and trench drains) were indispensable and were able to slow down the two landslides and enable the execution of restraining works.

The RC shafts (Fig. 5) were designed to provide stabilizing and dewatering effect on the sliding mass. The primary lining of the shaft consists of up to 30 cm thick and 1 m high reinforced cast in place concrete rings with an internal diameter of 5 m, which were connected in the vertical direction and separated from the sliding mass with an outer layer of drainage geotextile. With gradual deepening of the well, the rings were adequately perforated to allow the infiltration of the water and, if necessary, reinforced with steel wire mesh/ribs and up to 10 cm of shotcrete in order to sustain the loads of the sliding mass during the construction. In the stable base, additional polyethylene high-density (PEHD) drainage tubes (Φ125 mm) were installed at vertical distances of 3 m with a discharge into the interior of the well.

Fig. 5
figure 5

Vertical cross-section of the reinforced concrete shaft

Full size image

After reaching the final depth of the shaft/well, up to 4 m thick RC base plate was constructed, followed by the execution of the RC secondary lining with a thickness of 80 cm, which was designed to sustain the full force of sliding mass with prescribed factor of safety according to Eurocode standards. Finally, the shafts were interconnected with drainage pipes and a sewer pipe with a diameter of 160 mm was drilled by using horizontal directional drilling technology from the deepest well to the terrain surface and connected to the main drainage channel. In the case of the Slano blato landslide, additional vertical drainage pipes were installed through the base plate of the shaft to collect water from the flysch layers and acting as pressure relief pipes. The details dealing with the design and execution of mitigation works on both landslides are given below for each landslide separately.

Design and execution of mitigation works

Macesnik landslide

The stability analysis of the Macesnik landslide and shaft design calculations were performed using 3D FEM with the adoption of elastoplastic constitutive model for the displaced material and for the underlying layers and bedrock. Commercial FEM software Plaxis 3D Foundation (Brinkgreve and Broere 2004) was used for the analysis.

Elastoplastic model with isotropic hardening, also known as hardening soil (HS) model (Schanz et al. 1999), was used to model the soils and bedrock. The HS model is an advanced soil model with stress-dependent stiffness and allows for plastic straining due to primary compression and deviatoric loading and distinguishes between primary loading and elastic unloading/reloading. The failure is defined according to the Mohr–Coulomb criterion. Soil stiffness and strength parameters for the numerical analysis were determined on the basis of laboratory tests and are shown in Table 1.

Table 1 Basic soil/rock stiffness and strength parameters
Full size table

For computational reasons, it was not possible to model the entire landslide. Thus, only appropriate sections with the proposed shafts were considered in the analysis. The design approach and method of analysis for all three locations were the same as described below for the case of shafts at the uppermost location of the landslide (Fig. 2). For this section of the landslide, the total length of numerical model in the slope direction was determined according to the expected effect of the shafts on the landslide stability and set to 200 m, extending 70 m above and 130 m below the planned shafts as shown in Fig. 6a, where the finite element mesh of the model with activated shaft elements is depicted. The width of the computational model (30 m) was consistent with the landslide width at the proposed shaft location. Other geometric characteristics of the computational model were set according to the transverse and longitudinal profiles of the terrain, geological data, and geometry/symmetry of the planned constructions.

Fig. 6
figure 6

Finite element mesh with activated structural elements (a), failure mechanism at the state of nonconvergence (back analysis) with deactivated structural elements (b) and calculated displacements for the final state plotted as contour shadings of displacements with a detailed view of the calculated displacements around shafts (c)

Full size image

Two RC shafts, both 22 m deep and 5 m in diameter, were positioned 10 m apart and modeled as concrete elastic shells with primary lining thickness of 30 cm and final wall thickness of 110 cm. Young modulus of the concrete (class C25/30) was set according to Eurocode 2 (EN 1992-1-1:2004) E c  = 31 GPa.

The design calculation was carried out on the basis of back analysis of the landslide, which is commonly used to estimate the shear strength of the sliding surface and reproduce the actual displacement of the sliding mass (Ishii et al. 2001). Numerical modeling was conducted in the following steps:

  • Calculation of the initial stress state;

  • Back analysis of the landslide;

  • Calculation of loads on the primary shaft lining at various construction stages; and

  • Calculation of total loads on the primary and secondary shaft linings for the final state of construction and for the critical state under the worst possible conditions.

It should be noted that the dynamic effects due to the speed of the landslide movement were not considered in the analyses because of the successful completion of drainage measures, which reduced the displacement rates to an acceptable level.

The initial stress state in the model was generated for the prefailure state with a low ground water level approximately 5 m below the terrain surface, as measured during the dry periods after the surface drainage system was executed. Numerically, it was done in the initial phase of the finite element calculation in which soil weight and water pressures were applied by means of gravity loading.

In the next phase, the back analysis of the landslide was performed. In order to numerically simulate on-site conditions and behavior of the landslide, the ground water level was gradually “raised” from the initial state (−5 m). Through iterative computation, the model was brought to a state of nonconvergence when the water level was just below the surface of the terrain (−0.1 m). As the numerical procedure fails due to soil plastification, it generates large displacements, which indicate the sliding mechanism (Fig. 6b). The results of back analysis were consistent with the sliding mechanism, depth of the sliding zone, and field conditions as observed on site.

As it was realized that in dry periods, the surface drainage works are sufficient to stabilize the landslide to the extent that it allows the construction of shafts, the primary shaft lining, composed of segmental cast-in-place RC rings, was designed to sustain the loads during the construction when the landslide is dormant. The negative experience at the Slano blato landslide showed that such an approach was too risky, leading to a change in the design and construction method as described in the next section.

Although the shafts were designed also as drainage wells, the drainage effect was not considered in the critical analysis in order to get the characteristic shaft loads for the worst-case scenario (WCS), e.g., failure of the subsurface sewer pipe and drainage system. Therefore, the loads on the shafts (primary and secondary linings) were calculated with the numerical simulation of groundwater level rise to the model surface. Figure 6c shows the calculated displacements plotted as contour shadings for the WCS, with clearly visible arching effect behind the shafts and sliding of the soil mass along the shaft wall with final displacement reaching up to 0.5 m, which is sufficient to achieve full load transfer of the active landslide upon the shafts. The calculated shaft displacements with maximum value reaching up to 56 mm are shown in Fig 7. Figure 8 shows the most important internal axial forces, bending moments, and shear forces in the shaft wall plotted as contour shadings, with the notation according to the scheme of internal forces as shown in Fig. 9.

Fig. 7
figure 7

Contour shadings of maximum shaft displacements

Full size image
Fig. 8
figure 8

Contour shadings of the characteristic internal axial forces, bending moments, and shear forces in the shaft wall

Full size image
Fig. 9
figure 9

Scheme and notation of internal forces

Full size image

The calculation of safety factor for the WCS by using shear strength reduction method (Zienkiewicz et al. 1975; Dawson et al. 1999) showed that the impact of shafts on the stability of the slope below the shafts was marginal, which did not allow the calculation of the design loads at safety factor of 1.25 required by Eurocode 7 (EN 1997–1:2004). Therefore, the design values of shaft internal forces (axial forces (N 1;d, N 2;d), bending moments (M 11;d, M 22;d, M 12;d) and shear forces (Q 12;d, Q 13;d, Q 23;d) for the WCS were calculated according to Eurocode 7 (EN 1997–1:2004), which allows the multiplication of the effects of actions shown in Fig. 8 (characteristic values of shaft loads), with the partial safety factor of 1.35.

Additional analyses showed that the downslope stability could only be ensured, if the ground water level is kept −3 to −3.5 m below the terrain surface. Therefore, the construction of 3 to 4 m deep drainage trench along the slope was proposed and executed after the completion of the shafts (Fig. 12). The analysis taking into account the lowering of ground water level and a safety factor of 1.25(case herein named LWL) yielded comparable distribution but, in general, less extreme design values of shaft internal forces than the analysis for the WCS. Extreme design values of shaft internal forces for WCS and LWL analyses are shown in bold text in Table 2 together with accompanying values of internal forces in the same cross-section.

Table 2 Extreme and accompanying design internal forces for WCS and LWL case (LWL values in brackets)
Full size table

Although the internal forces were calculated by taking into account the shaft wall thickness of 110 cm, the 80 cm thick secondary lining was designed to sustain the whole (100 %) load induced by the sliding mass. The primary lining was designed to sustain only the loads during the construction. Figure 10 shows the ultimate resistance surface of the secondary lining and ultimate design loads for WCS at the N 1 − M 11 diagram according to Eurocode 2 standard (EN 1992-1-1:2004).

Fig. 10
figure 10

Ultimate resistance surface of the secondary lining and ultimate design loads (N 1−M 11 diagram) with a scheme of longitudinal reinforcement in the secondary lining of the shafts

Full size image

The construction of the first two shafts, each about 22 m deep and 5.4 m in external diameter, began in autumn 2004, after three drainage trenches (250 m long, up to 8 m deep) had been executed in order to reduce unacceptably high displacement rates up to 50 cm/day in the 800 m long uppermost section of the landslide. The shafts are situated in the area just above the pontoon bridge and were completed in spring 2005, followed by the construction of the drainage trench on the steeper slope below (Figs. 11 and 12). In 2006, the mitigation works were completed with the construction of two additional deep RC shafts in the middle area of the landslide, where it is twice crossed by the panoramic road. No further measures were taken in the lower part of the landslides as planned previously (Mikoš et al. 2005).

Fig. 11
figure 11

Interior of the shaft during the execution of the secondary lining

Full size image
Fig. 12
figure 12

Construction of the drainage trench in the middle section of the Macesnik landslide

Full size image

Slano blato landslide

A similar approach as for the Macesnik landslide was used to design shafts for the Slano blato landslide, taking into account the relevant geological, hydrological, and geotechnical data (Majes et al. 2002; Fifer and Zupančič 2009). Initially, five shafts at mutual distances of 15 m, each 5.4 m in diameter and 20 m deep, were planned to be executed in the central part of the landslide, with gradual construction of additional shafts towards the edges of the landslide, if it proved to be necessary. Since the back analysis with the numerical consideration of water level on the ground surface was insufficient to create sliding mechanism as observed on site, additional artesian water pressure (50 kPa) at the contact between the sliding mass and flysch base was considered in the analysis (Fig. 13), confirming the field observations that an important influx of water to the sliding mass also came from a lower flysch layer (Logar et al. 2005; Placer et al. 2008). In order to minimize the negative effect of these pressures, additional vertical drainage pipes were planned to be installed through the concrete shaft base to work as pressure relief pipes.

Fig. 13
figure 13

Cross-section of the Slano blato landslide

Full size image

The construction of three central shafts started in autumn 2004. After intense rainfall in October 2004 (monthly rainfall accumulation 377 mm, of which 118 mm between October 29 and 31), a large amount of water and strong inflow of ground water from the flysch rock into the area of shafts, which were under construction and had not yet penetrated into stable bedrock, caused excessive earth movements between November 1 and 2. In just 2 days, the shafts were displaced by nearly 20 m (Figs. 13 and 14). The primary lining of one shaft, which was at that time excavated to the depth of 6 m, was damaged beyond repair, while two others suffered only minor damage and were later completed at the displaced location. Although not reaching the stable bedrock and severely displaced, the shafts were able to collect a large amount of water, which was pumped out and discharged from the area. In the period between October 31 and November 6, each shaft was able to collect about 200 to 240 m3 of water per day (2.3–2.8 l/s). The decreased displacement that took place in the next few days was mainly due to successful drainage of the sliding mass. This event confirmed the hydrological forecast that during intense rainfall each well should be able to collect about 2.8 l/s of water from the most permeable top 4 m of the sliding mass and additional 1 l/s from the soil below (Prestor et al. 2004).

Fig. 14
figure 14

Primary lining of the shaft at the Slano blato landslide during construction. The photo shows the drainage ability of the shafts/wells

Full size image

To speed up the construction and reduce the risk of potential damage to shafts during their construction, the design of the shaft was changed. The segmental concrete rings (primary lining) were replaced with eight bored RC piles with diameters of 150 cm, which were then on their tops connected with a 1.5 m high RC ring beam. The centers of piles are on circles with a diameter of 6.5 m to allow the final construction of the secondary RC lining with the inner diameter of 4.2 m. The spatial scheme of the shaft with its cross-section is shown in Fig. 15. The damaged shaft was replaced in 2005 and four additional shafts were constructed on the right-hand side of the landslide. By the end of 2007, when mitigation works on the upper part of the landslide were completed, four additional deep shafts were executed on the left-hand side of the landslide. The shafts were positioned in semicircular form to act as an arch and transfer the loads from the retained soil laterally into to the stable ground.

Fig. 15
figure 15

Spatial scheme with vertical and horizontal cross-section of the modified design

Full size image

Monitoring and performance evaluation

The Macesnik and the Slano blato landslides are typical examples where conventional methods for monitoring displacement or pore water pressures (e.g., inclinometers and piezometers) can hardly be implemented due to excessive and rapid movements of the sliding mass. In order to evaluate the performance of the applied mitigation measures, one had to rely on other methods of monitoring, which contained conventional geodetic measurements of displacements, Global Positioning System (GPS) based displacement measurements, advanced geodetic measurements including ground and air Light Detection And Ranging (LIDAR) measurements, weather monitoring (rainfall, temperature, wind, and sunshine duration), monitoring of ground water levels in the soil/wells, and soil matric suction measurements.

As both landslides are rainfall induced, the amount of precipitation has a major influence on the landslide activity. Thus, it is of great importance (Hong et al. 2005) to assess the influence of rainfall on landslide movements and, consequently, to assess the impact of the measures implemented.

Macesnik landslide

As shown in Fig. 16, where local daily, monthly, and cumulative yearly precipitation measured in the rainfall gauging station in the village of Solčava are depicted (1 km away from the toe of the Macesnik landslide), the precipitation in 2006 was slightly lower than in previous years but at the same level in the subsequent years with a series of rainfall events that would normally trigger the landslide.

Fig. 16
figure 16

Rainfall data measured in the village of Solčava—Macesnik landslide (2000–2012)

Full size image

The intense movements up to 50 cm/day at the Macesnik landslide in the period between 2000 and 2004 have rapidly declined after the completion of mitigation works in the upper part of the landslide in September 2004. Figure 17 shows the absolute displacements and displacement rates in measuring cross-sections on the Macesnik landslide for the period from 8 September 2004 to 29 September 2005. The displacement rates were further reduced when drainage works and retaining works had been completed in the middle section of the landslide in 2006. The maximum recorded displacement of the upper shafts was 36 mm, which is about 63 % of the predicted displacement for the WCS (56 mm). From 2006 to date, no landslide activity has been reported in the upper and middle sections of the landslide where mitigation measures were executed as planned.

Fig. 17
figure 17

Absolute displacements and displacement rates in measuring cross-sections on the Macesnik landslide (see Fig. 4 for the position of cross-sections)

Full size image

Slano blato landslide

The precipitation data for the Slano blato landslide (Fig. 18) were taken at a rainfall gauging station in the village of Otlica situated approximately 3.8 km east of the landslide. The landslide was triggered in November 2000 with a monthly rainfall amount of 723 mm and daily maximum 86 mm. A series of events causing serious advancements of the sliding mass in the subsequent years showed good correlation with the amount of precipitation. Both peaks observed in March and September 2001 (Fig. 18) with monthly rainfall over 400 mm and daily maximum over 70 mm were followed by significant earth flow advancements. The next triggering event was in November 2004 (377 mm/month and 83.6 mm/day), when the shafts were severely displaced. Based on these events, it was estimated that monthly rainfall over 350 mm accompanied with the daily maximum of about 70 mm of rainfall is sufficient to trigger the landslide.

Fig. 18
figure 18

Rainfall data measured in the village of Otlica—Slano blato landslide (2000–2012)

Full size image

As seen from the rainfall data measured in the rainfall gauging station in the village of Otlica (3.8 km away from the Slano blato landslide) in Fig. 18, 2005, and especially 2006 and 2007 were rather dry years. However, after 2007, when mitigation works at the Slano blato landslide were completed, a number of critical rainfall events have happened, causing some local earth movements at the main scarp but have not been able to trigger the landslide. The final test of the executed measures was conducted during heavy rainfall in September 2010, when the record with 498 mm of rainfall in 48 h between 17 and 19 September was recorded and large landslide Stogovce located nearby was triggered (Petkovšek et al. 2011). The impact on the Slano blato landslide was marginal, as it remained stable and no damage at the 11 executed RC shafts and drainage channels was observed, as can be seen from the aerial view of the upper part of the remediated landslide (Fig. 19). In 2007, after the completion of the shafts, ground LIDAR measurements were taken in order to get better insight into the behavior of the upper part of the landslide, where occasional local earth movements were still observed during heavy rainfall.

Fig. 19
figure 19

Aerial view of the upper part of the Slano blato landslide

Full size image

Figure 20 shows a 3D view of the upper part of the landslide as recorded by ground LIDAR measurements in 2007 and Fig. 21, the difference in the terrain height recorded in February and October 2007. The colder colors (blue) show the areas where the terrain was lower in October 2007 than in February 2007, while warmer colors (red) show the opposite. From the data (Fig. 21), it is evident that some local sliding of the soil occurred in the upper part of the landslide, which leads to the change of terrain level within −3 m and +4 m. The terrain around the shafts remained stable, although some ground subsidence (blue areas) can be detected above the shafts (due to the drainage) and in the area below the shafts, where some erosion and shallow earth slips were detected. The net volume of the soil mass, obtained as the difference in the terrain height, was found to be positive, which means that the terrain measured in October was, on average, higher than that measured in February 2007. This is most probably due to the accumulation of the material from the periphery of the observed area and flysch rock deterioration, although it could also be due to the horizontal movements. Nevertheless, the calculated net volume confirms the ability of the applied stabilizing measures to successfully retain the earth mass in the upper area of the landslide.

Fig. 20
figure 20

3D view of the upper part of the Slano blato landslide (February 2007)

Full size image
Fig. 21
figure 21

Difference in the terrain height in the period from February 2007 to October 2007

Full size image

Drainage ability of the shafts was observed by real-time measurements of the water levels in one of the wells and piezometer, which was placed about 10 m above the shafts in the stabilized ground. The data compared to the amount of daily precipitation are shown in Fig. 22. The groundwater level in the area above the shafts, which used to be at the terrain surface, has dropped by about 4 m and remains at that level regardless of the daily precipitation. The water level in the well shows a rapid increase immediately after heavy rainfall and rapid decrease thereafter. This indicates high drainage capacity of wells that exceeds the capacity of subsurface sewer pipes with a diameter of 160 mm. Consequently, shafts/wells act as temporary water detention storage. The data also show a successful reduction of the groundwater level in the shaft hinterland, which was not taken into account during the design and significantly contributes to the stability.

Fig. 22
figure 22

Groundwater levels and precipitation data for the period between 1 May and 21 September 2009—Slano blato landslide

Full size image

Conclusions

This study presents a case history of successful remediation of the Macesnik and the Slano blato landslides with a combination of drainage and retaining works with an emphasis on the design and execution of the RC shafts/wells. The simultaneous use of shafts and deep drainage works has proved to be an effective measure to increase the stability and reduce the landslide risk. Several years after the completion of RC shafts at the two landslides, the following conclusions can be drawn:

  1. 1.

    The study shows that initial geological and geotechnical surveys, together with an efficient monitoring system, have a decisive influence on the choice of mitigation measures. The locations for retaining constructions should be selected according to the size and speed of landslide movements, length of the landslide, irregularities in the geological profile in the longitudinal and transverse directions, and change of the slope along the landslide.

  2. 2.

    In case of fast-moving landslides, the execution of drainage works is indispensable in order to slow down the landslide and to enable the execution of retaining works. In order to design appropriately a retaining construction, its execution must be taken into account in order to minimize the construction risk and to avoid damage during the construction.

  3. 3.

    The FEM has been applied for the design of deep RC shafts. The numerical analysis should be based on the back analysis of the landslide and consider all construction phases during and after the execution of works. Numerical simplifications are inevitable and therefore, the impact on the final results should be carefully examined in the design process.

  4. 4.

    RC shafts with twofold function of restraining and draining have proved to be an effective measure for the structural mitigation of large and fast-moving landslides such as the Macesnik and the Slano blato landslides. Both landslides were efficiently slowed down. The shaft behavior was found consistent and in good agreement with the design expectations. The measured shaft displacements do not exceed the predicted values and no structural damage was observed during the extreme rainfall events after the completion of works.

  5. 5.

    The measurements of the groundwater level in the vicinity of the shafts have proved the high drainage capacity and successful reduction of the groundwater level, which adds considerably to the stability of both landslides.

  6. 6.

    According to the monitoring data, the mitigation measures at both landslides can be assessed as effective in preventing the reoccurrence of large-scale sliding under heavy rainfall, although the emergence of minor earth movements is still possible.