Alpine countries like Austria are especially vulnerable to natural phenomena like landslides which can cause severe damages (Fig. 1). Today, the understanding of the sequence of accelerations and decelerations of natural slope movements is limited and reliable prediction models do not exist. Therefore, high-sensitivity early warning systems are required which detect changes in the deformation behavior at an early stage. Sufficient warning time allows counter actions or at least can reduce the number of lost human lives.
Another challenge in alpine areas is the construction of roads or railway tracks in steep terrain. Today, reinforced earth structures are more and more used instead of conventional retaining walls. However, failures of such structures are known (Fig. 2). One key element of an early warning system for reinforced earth structures is the high-sensitivity monitoring of the internal strain distribution of the embedded geogrids.
In this paper, we report the development and implementation of monitoring systems for both applications using internal fiber optic measurements.
High-sensitivity landslide monitoring
The Gradenbach Observatory
The Gradenbach landslide (Fig. 3) is a deep-seated mass movement in the south of Austria. Its active deformation zone covers an area of approximately 800 m × 1800 m. The main scarp is located slightly below the mountain ridge with a height of 2268 m above sea level. The Gradenbach landslide has been monitored using epoch-wise measurements for more than 50 years. The landslide is constantly moving with a typical velocity of about 12 cm/year. The steady movement is interrupted by sudden acceleration and deceleration phases (Brückl et al., 2006).
Since 1999, monitoring activities increased within the IDNDR (International Decade for Natural Disaster Reduction) and ISDR (International Strategy for Disaster Reduction) research programs of the Austrian Academy of Sciences (OeAW). As a result, the Gradenbach Observatory was installed (Brückl et al., 2013).
This observatory consists of a geodetic component, a hydro-metrological component (precipitation, temperature, snow cover) and a seismic component. An overview of all sensors installed at the Gradenbach Observatory is given in Fig. 4. The Institute of Engineering Geodesy and Measurement Systems (IGMS) of Graz University of Technology is responsible for the GPS measurements, the terrestrial surveys and the local strain measurements. The measurements are carried out epoch-wise or continuously and the results are accessible on the homepage of the Gradenbach Observatory.
Fiber optic strain rosette
Development and installation
Absolute displacements of the landslide can be determined with the continuous GPS monitoring system and the implemented height correction models with an accuracy of less than 1 cm (Gassner et al., 2002). This accuracy is well suited to determining the long-term behavior. For the determination of local deformations, measurement systems with higher precisions and higher resolutions are required. Fiber optic sensors fulfill these requirements and are robust enough to be embedded in landslides. One successful application of fiber optic measurements for the determination of the location of the boundary between stable and sliding areas of a landslide in Switzerland was reported in Iten et al. (2009). In case of the Gradenbach Observatory, the focus was placed on the early detection of an acceleration of the landslide. Therefore, a large fiber optic strain (LFOS) rosette was embedded in the central landslide area to detect local compression and decompression.
The LFOS rosette is composed of three SOFO sensors with a length of 5 m. Alternatively fiber Bragg grating (FBG) sensors can also be used. The SOFO sensors are based on an interferometric measurement principle. Each SOFO sensor consists of a stretched fiber and a loose fiber. A length change between the two anchor points of the sensor only affects the strained fiber. On the contrary, a temperature change has an influence on both fibers. Since the measurement result is the length difference of both fibers, the temperature influence is eliminated. In our application, the sensors were separated to each other by an angle of about 120° to form a rosette in analogy to strain rosettes used in classic mechanical stress analysis (Fig. 5).
The sensors were installed in 2007, parallel to the surface below the frost penetration depth at a depth of about 2 m (Fig. 6). One sensor (sensor A) was oriented in the direction of movement of the landslide. More details about the development of the fiber optic strain rosette can be found in Wöllner et al. (2011) and Woschitz and Brunner (2008).
Long-term strain measurements with the static SOFO reading unit can be performed with a precision of 2 μm according to the specifications of the manufacturer (Inaudi, 2004). This was also confirmed by our own investigations (Lienhart, 2005). This corresponds to a precision of strain measurements of 0.4 μm/m taking into account the length of the sensors (5 m). The principal strain values ε1 and ε2 as well as their orientation φ can be derived from the strain measurements εA, εB and εC of the individual sensor using the following equations:
The LFOS measurements at the Gradenbach Observatory are usually performed automatically every 15 min from spring to autumn. In this paper, we focus on three measurement epochs where additional geodetic measurements are available. Fig. 7 displays the resulting strain ellipses with respect to the reference epoch June 2007. It can be seen that almost no strain changes occurred between June 2007 (epoch #1) and July 2008 (epoch #2). However, a significant compression occurred between July 2008 and July 2009 (epoch #3).
This is a typical result for the Gradenbach landslide. During the period of constant movements (2007–2008), no local strain changes occurred. However, in case of acceleration phases, the upper parts of the landslides are faster than the lower parts (2008–2009) and therefore local compression occurred.
Comparison with geodetic measurements
A local terrestrial geodetic network was established to check the results of the fiber optic strain rosette for plausibility since such a LFOS rossette was installed in a landslide for the first time. This network was measured three times (June 2007, July 2008, and July 2009) with a total station. To compare the data of the geodetic network and the fiber optic strain rosette, the movements determined by the terrestrial measurements were interpolated to the end points of the strain rosette.
Fig. 8 shows a comparison of the strain ellipses derived from the two different measurement techniques. The principal strain determined by the LFOS rossette is (−261 ± 0.6) μm/m, and (−245 ± 28) μm/m from the geodetic network. The difference is statistically not significant. However, geodetic measurements only provide a qualitative control of the fiber optic measurements due to the 50 times lower precision of the geodetic measurements.
The importance of the fiber optic measurements can also be seen when compared to the GPS measurements. The distance changes between the GPS points MA and MC (Fig. 4) can be converted into strain changes. Since the orientation of the line between points MA and MC corresponds to the orientation of sensor A, the strain values can directly be compared. It can be seen in Fig. 9 that the GPS measurements depict the overall trend of the strain changes, however, detailed strain analysis is impossible due to the high noise of the GPS data. This confirms that the high resolution fiber optic strain measurements are crucial for depicting the early phase of a change in moving behavior of the landslide.
High-sensitivity monitoring of reinforced earth structures
The Longsgraben disposal site
The New Semmering Base Tunnel is a key project of the upgrade of the Austrian North-South railway connection. Prior to tunnel excavation, preparatory works had to be completed including the construction of the Longsgraben disposal site. This disposal site covers an area of approximately 2 × 105 m2 at depths of 1.05–1.25 m above sea level. During operation of the site, the valley will be filled up to a height of 50 m. Afterwards, the site will be covered with earth and afforested.
Prior to establishing the disposal site, the small mountain stream running at the valley floor had to be relocated to a new bed at the edge of the site, up to 50 m above the current valley bottom (Fig. 10). Therefore, reinforced earth structures with a length of 1.22 m were constructed on the steep valley side.
Additionally, a maintenance road was constructed with a reinforced earth ramp (Fig. 11) with a maximum height of 25 m and a length of 80 m.
Distributed fiber optic measurement system
Development and installation
The reinforced earth structures were monitored during construction based on the observation method of Eurocode 7. Therefore, prism targets were mounted on the surface of the structure. The absolute movements of the prisms were measured epoch-wise twice per week with total stations. Additionally, a distributed fiber optic measurement system was installed to measure the current operation grade of the geogrids within the earth structure.
Up to now, several different types of geotextiles with integrated fiber optic sensing cables are available (Artieres et al., 2010, Krebber et al., 2012). However, these geotextiles were not suitable for the Longsgraben disposal site. The challenge in this application was to determine the operation grade of the specific geogrids which were used to build the structure. Therefore, we developed a monitoring system with adapters that connect fiber optic sensing cables to standard geogrids. In total 2 km of sensing cables were installed in the project area. The BRUstrain V4 from Brugg cables was selected to monitor three cross sections of the reinforced earth structure. In each cross section, up to five geogrid levels were equipped with the sensing cables. A cross section of one measurement profile is shown in Fig. 12. The end point of each measurement level was directly connected to a geodetic observation point to relate the internal strain measurements to displacements measurements of the surface of the structure. The sensing cables were connected to the geogrids with adapters developed at IGMS in distances of more than 1 m.
A loop configuration with a stretched and a loose fiber (Fig. 12) was implemented in each measurement level. The stretched fiber is sensitive to length changes between the adapters and to temperature changes, whereas the loose fiber is only sensitive to temperature changes. By combining both measurement signals, the temperature influence can be eliminated. The cables were locally protected between the adapters in order to guarantee constant strain of the stretched fiber and to avoid trans-axial pressure to the cable. All sensing cables were connected to one single 2 km loop to measure all cross sections simultaneously with one distributed fiber optic sensing unit.
Fiber optic measurements systems provide new insights into the behavior of geotechnical structures (Iten, 2011), and distributed sensing is especially suited for large structures. Distributed fiber optic sensing can be based on Rayleigh, Raman or Brillouin scattering (Bao and Chen, 2012). Brillouin scattering occurs when light traveling in a single-mode optical fiber is reflected by the refractive index modulations produced by acoustic waves (Horiguchi et al., 1989). BOTDA (Brillouin optical time domain analyzer) and BOFDA (Brillouin optical frequency domain analyzer) systems use a loop configuration where one end of the fiber is connected to a continuous wave pump laser and the other end to a laser which emits light pulses (BOTDA) or amplitude modulated light waves (BOFDA). In BOFDA, the light pulse is replaced by sinusoidal waves of a tunable frequency and the response to these frequencies is measured (Galindez-Jamioy and Lopez-Higuera, 2012, Nöther and v.d. Mark, 2012). In theory, a Fourier transform of the measurement data of a BOFDA system yields the same strain and temperature profiles along the sensing fiber as is measured with a BOTDA system (Nöther, 2010). In our monitoring setup, a FibrisTerre BOFDA reading unit was placed into a central measurement container (Fig. 13).
The measured Brillouin frequencies along the 2 km sensing cable are shown in Fig. 14. The zoom window displays the measured Brillouin frequencies of level 2 in cross section CS2. This level consists of four measurement segments (s1–s4, with s1 closest to the surface). Clearly visible is an increase of the strain with increasing distance from the surface of the structure.
During the construction phase, internal monitoring measurements were carried out weekly and reported to the geotechnical engineer. After the construction was finished, the measurement interval was extended to monthly measurements. The strain development of one selected monitoring level (level 2 in cross section CS2) is shown in Fig. 15. It can be seen that the strain increased up to 1% during the construction phase. As already discussed before, the strain increased with increasing distance from the surface of the structure. After the construction phase has finished, the strain increase slowed down and stabilized.
The new internal strain measurements provided important insights into the behavior of the reinforced earth structure. The nonlinear strain distribution inside the structure cannot be detected using conventional geodetic measurements on the surface. Currently, the measurements are used to calibrate the numerical results of finite element method of the structure. One year after installation, the fiber optic monitoring system is still fully operational and will be used to monitor the internal strain development until the disposal site is completely covered with the tunnel excavation material.
We presented two applications of high-sensitivity fiber optic monitoring installations for natural and engineered slopes. The great advantage of fiber optic sensors compared to traditional geodetic sensors is that the fiber optic sensors can be embedded in the structure and therefore deliver deformation information at locations that are otherwise not accessible. Furthermore, in many cases deformation changes at an early stage can only be detected with fiber optic sensors due to their high resolution and high precision. Conventional geodetic sensors and fiber optic sensors complement each other. Geodetic sensors are well suited to depicting position changes of objects whereas fiber optic sensors are better suited to determining internal strain changes.
Source: Case studies of high-sensitivity monitoring of natural and engineered slopes
Author: Werner Lienhart