LARGE-SCALE IMPACT TESTS ON ROCK FALL GALLERIES

K Schellenberg, ETH Zurich, Switzerland A Volkwein, WSL, Switzerland

A Roth, Geobrugg Protection Systems, Switzerland T Vogel, ETH Zurich, Switzerland

Abstract Protection galleries are important structures to reduce the risk caused by falling rocks. In Switzerland about 350 galleries are in use for that purpose. The aim of the tests described in this paper is to gain experimental data to improve the design methods for rock fall galleries. A series of rock fall impact tests on six reinforced concrete slabs with a cushion layer consisting of compacted gravel was performed. Concrete cubes of 800 kg and 4000 kg, respectively, were dropped with different falling heights. Special cushion systems consisting of high-tensile steel wire mesh and cellular glass were tested. The relevant data to describe the dynamic processes in the structure are the measured reaction forces at the supports, the accelerations in the boulder and in the slab as well as the strains at the upper slab surface and in the reinforcement. In addition the impacts were recorded by high speed cameras. This paper describes the conducted impact tests in an outdoor rock fall testing area. Keywords: rock fall impact, large-scale tests, cushion system, cellular glass, reinforced concrete slabs 1. Introduction The risk of rock fall events increases due to global warming and due to the population growing into alpine regions. Meanwhile the risk acceptance in our society decreases according to the state of our proper economical situation. Considering the high mobility requirements also in alpine regions, professionals need to improve the protection against rock fall hazards. Rock fall galleries are an efficient measure to protect roads and railways, mainly if the danger is locally concentrated. A study on the Swiss rock fall galleries has shown that most of the existing galleries consist of reinforced concrete slabs and are covered with a cushion layer [1]. The cushion layer distributes the contact stresses, reduces the accelerations in the striking body and increases the impact time. Normally, granular soil from the surroundings or gravel is used as cushion layer. Protection galleries typically span 9 m with a slab thickness of approximately 0.70 m. The back side of the galleries is clamed supported at the retaining wall, the valley side is supported on columns (see Figure 1). A typical column spacing is 7 meters.

The impact load capacity of the existing galleries is of great interest e.g. to decide on the necessity of renovation or strengthening. The Swiss design guideline for rock fall galleries was published in 1998 [2]. Older galleries are mostly designed based on oversimplifications by local engineers. The guideline is based on impact tests carried out in 1996 [3] that focused on the influence of the cushion layer. The test results were extrapolated by using finite element simulations [4]. Further research was performed on the dissipation capacity of different cushion materials [5]. To understand the response of the structure and the interaction between the impacting rock, the cushion layer and the reinforced concrete slab are the main focus of the actual study. Therefore, large-scale field tests on reinforced concrete slabs are performed in an old quarry close to Walenstadt in the Swiss Alps. 2. Test Setup Six reinforced concrete slabs covered with a cushion layer were subjected to falling weight impacts. The slabs correspond to an average Swiss rock fall protection gallery in a scale of 1:2. With three different slab types the influence of the slab thickness and reinforcement on the impact behavior could be explored. Additional to the standard gravel cover, two special cushion systems were tested. The impact was defined by different artificial boulders dropped from varying heights. For each slab, the impact energy was increased until the slab failed. The kinematics of the striking body was analyzed and the dynamic response of the reinforced concrete slab was investigated by measuring accelerations, reaction forces and strains. The test unit consists of two HEM 360 beams supported by four reinforced concrete footings of 0.5 x 0.5 m and 0.6 m height. Both HEM beams are 4.5 m long. One of them serves as line support for the slabs and is placed on two load cells (shown on the left hand side in Figure 2). A mortar layer with a thickness of about 1 cm is placed between the line support and the reinforced concrete slab. It insures a continuous and regularly distributed support and compensates geometrical irregularities and also restrains the horizontal movements. The remaining two corners of the slabs are simply supported on the load cells. They model the column supported points of the galleries. On this side the load cells are placed on top of the second HEM beam (shown on the right hand side in Figure 2). The four load cells have a capacity of 1000 kN each. The reaction forces at all supports provide the most important data to study the impact response of the structure.

a)

b)

Figure 1: Typical rock fall galleries in Switzerland a) Galeria Val Funtana, 1971, b) Avalanche gallery Buggital from 1982

Horizontal movements of the slabs are possible at the simple supports enabled by Teflon laminated sliding plates (Figure 3). Rotations are enabled by spherical calottes placed between the slide plates and the load cells. Horizontal displacements of all components of the test unit are restrained by bolts. The second HEM beam allows the test unit for being adjusted to other support conditions. For further tests, it could be used for two line supports, even along the short side of the testing unit. The reinforced concrete slabs are covered by the cushion layer. The boulder is dropped on the slab from a predefined height. The slabs, the cushion layers and the boulder are described in the next sections. The tests are recorded by a digital video camera with a recording rate of 250 frames per second. A posterior analysis of the recorded trajectory can be done using tracking software. The camera and the instrumentation are triggered manually.

Figure 2: Test setup with gravel cushion and 4000 kg boulder

a) b)

slide plate

UNP 220350

calotte

load cell

60

120

250

HEM 360

Figure 3: Simple support a) load cell with sliding plates and calotte, b) technical drawing of cross section

2.1 Reinforced Concrete Slabs Six reinforced concrete slabs with the dimensions 3.5 x 4.5 m were tested. Table 1 shows an overview of the three different slabs types, Figure 4 the cross section along the short side. Slabs 1 and 2 have a thickness of 0.25 m. Slabs 3 to 6 have a thickness of 0.35 m. Slabs 5 and 6 have additional continuously distributed shear reinforcement. To avoid shear failure of the simple supports, two UNP 160 or 220 joists welded to an elbow are cast in at 250 mm distance from the corner. Each side of the UNP profiles has a length of 1.20 m. Within the concrete slabs, acceleration and strains are measured with a sampling rate of 3200 Hz. The accelerations are measured in two places: First in the center of the slab and second with 1.5 m distance of the center along the diagonals towards the simple supported corners. The measurement range of the acceleration sensor is ± 1000 m/s2.

In the center of the slab the strains in the bending reinforcement and at the slab’s upper surface are measured (Figure 5). The reinforcement strains are measured in two additional rebar of 2 m length located orthogonal to each other in the center of the slab. There, also the strains at the slab surface are measured. These strain gages have to be protected by silicon and by a 0.5 cm thick mortar layer. For a better observation of the crack pattern the soffit of the slabs is painted in white.

Slabs Thickness Bending reinforcement

Shear reinforcement

1 & 2 0.25 m d = 18 mm s = 155 mm

no

3 & 4 0.35 m d = 22 mm s = 155 mm

no

5 & 6 0.35 m d = 22 mm s = 155 mm

d = 10 mm, s = 155 mm

Table 1: Slabs thickness and reinforcement

a)

b)

c)

Figure 4: Reinforcement layout of a) slabs 1 & 2, b) slabs 3 & 4 and c) slabs 5 & 6

a) b)

Figure 5: Strain gages on a) the reinforcement and b) at the upper slab surface

2.2 Cushion Systems For the tests 0.40 m gravel is used as a cushion layer. In the scale of 1:2 the thickness of the cushion corresponds to the average layer thickness placed on typical Swiss rock fall galleries. The cushion layer is kept together by a steel mesh cylinder with a diameter of 3 m. Considering a load spread angle of about 45°, the influence of the steel mesh on the behavior of the cushion layer should be negligible. In order to avoid the cushion materials to penetrate the mesh, a geo-textile is applied. The grain size distribution of the gravel is shown in Table 2. In order to quantify the energy dissipated by the gravel, rounded material is selected. After the impacts probes of the cushion layer are taken from the contact area and examined with respect to the amount of broken grains. The compaction of the cushion layer is measured before and after each impact. The compaction measurement is carried out by a light drop-weight device. It consists of a 10 kg weight dropping from 1 m on a steel plate, where accelerations are measured. From the accelerations, a compaction modulus of the soil can be defined. Two special cushion systems with high-tensile steel wire mesh (TECCO) and cellular glass (MISAPOR) were tested and compared to the gravel layer. The main purpose of these tests is to analyze the suitability and behaviour of these systems as alternative cushion material. The high-tensile steel wire mesh is a chain-link mesh with a wire diameter of 3 mm and a tensile strength of 1770 N/mm. The ultimate load of the mesh is about 170 kN/m. The cellular glass is produced from recycled glass and has a cube compressive strength of 6 N/mm2. The granulation of the cellular glass amounts ranges from 10 to 50 mm and its density is only 2.5 kN/m3. The main advantage of the cellular glass is the low dead load acting on the concrete slab. Thus, it offers cost savings in transportation and installation. Furthermore, it was evaluated how much the impact forces can be reduced by this special cushion system compared to gravel due to the thicker but still lighter cushion layer and also due to the absorption capacity of the cellular glass and mesh system. The high-tensile mesh was so far used for slope stabilization and rock fall barriers and has a proven energy absorption capacity. The first setup (called Geobrugg A) consists of three layers of 40 cm of cellular glass, which means three times thicker than the gravel layer but still has only half the weight. Between the layers of the cellular glass and also on top of the cushion, a layer of high-tensile mesh is installed in order to get an improved load distribution and to activate more cellular glass. The second setup (called Geobrugg B) is made of modular cylinders made from high-tensile mesh and filled with cellular glass. The cylinders have a diameter of 1 m and a height of 0.6 m. Seven of them are placed and then covered with a layer of high-tensile mesh, again 7 cylinders and again a top layer of the high-tensile mesh. The cylinders restrict the lateral displacement of the material and create modular energy dissipating systems.

Portion Mass % 0 – 4 mm 33 4 – 8 mm 11

8 – 16 mm 20 16 – 32 mm 36

Table 2: Grain size distribution of cushion gravel

a)

b)

c)

Figure 7: Cushion layers a) 0.40 m gravel layer, b) 1.20 m Geobrugg A and c) Geobrugg B system

2.3 Boulder Artificial concrete boulders are dropped from different heights ranging from 2 to 15 m released by a remote-controlled hook. The boulders consist of two hexahedron obtuse (Fig. 8) out of fiber-reinforced high performance concrete with a weight of 800 kg and 4000 kg, respectively. The boulders are equipped with six acceleration sensors measuring the vertical acceleration. The range of the sensors is ±500 m/s2, the data is sampled with 10 kHz. The data of 0.5 s before triggering and 2.5 s afterwards is saved. The vertical rock velocities and movements are obtained through numerical integration of the acceleration data. Further steps are taken to increase the preciseness of this analysis: The large number of sensors allows for averaging the acceleration and the known start position and final position allow the consideration of necessary correction factors. A tailor-made analysis software [6] eases the procedure. The trajectory of the boulder is additionally recorded by the high-speed video system. This redundancy allows an additional check or - if necessary - a further correction. Previous work [7] has shown that the acceleration and the time-integrated velocities taken from the acceleration sensors are most precise. However, the best displacement curve is taken from the video recordings. By differentiation velocities and accelerations are obtained from the displacement curve. 3. Test program The magnitude of rock fall impacts is difficult to predict and their values are governed by large uncertainties. Compared to that the statistical spread of the test results is of minor importance and each impact test

a)

b)

Figure 8: Boulder a) geometry, b) 4000 kg boulder with target points

Name

Slab Cushion layer

Falling Weight [kg]

Falling height [m]

Impact Energy [kJ]

A1 1 Gravel 800 2 16 A2 1 Gravel 800 5 39 A3 1 Gravel 800 5 39 A4 1 Gravel 800 5 39 A5 1 Gravel 800 7.5 59 A6 1 Gravel 800 10 78 A7 1 Gravel 800 12.5 98 A8 1 Gravel 800 15 118 B1 3 Gravel 800 5 39 B2 3 Gravel 800 7.5 59 B3 3 Gravel 800 10 78 B4 3 Gravel 800 12.5 98 B5 3 Gravel 800 15 118 B6 3 Gravel 4000 2 79 B7 3 Gravel 4000 5 197 C1 2 Geobrugg A 800 15 118 C2 2 Geobrugg A 800 5 39 C3 2 Geobrugg A 800 5 39 C4 2 Geobrugg A 800 5 39 C5 2 Geobrugg A 800 10 78 C6 2 Geobrugg A 800 15 118 C7 2 Geobrugg A 4000 2 78 C8 2 Geobrugg A 4000 5 196 D1 4 Gravel 4000 2 78 D2 4 Gravel 4000 5 196 E1 5 Gravel 800 15 118 E2 5 Gravel 4000 5 196 E3 5 Gravel 4000 7.5 294 F1 6 Geobrugg A 4000 7.5 294 F2 6 Geobrugg A 4000 7.5 294 F3 6 Geobrugg B 4000 7.5 294 F4 6 Geobrugg B 4000 7.5 294 F5 6 Gravel1) 800 10 79 F6 6 Gravel2) 800 15 118 F7 6 Gravel 800 15 118 F8 6 Gravel 800 15 118 F9 6 Gravel 4000 5 198 F10 6 Gravel 4000 7.5 295 Table 3: Program of the 38 large scale tests 1) thickness: 0.20 m 2) incompact

is carried out only once. Taking into account the high costs of large-scale tests this decision is advisable. The test program was set up to obtain the best comparisons possible between the single impact tests. To avoid damage to the test unit and to the instrumentation the load capacity of the slabs should not reach the ultimate load level, although it would be interesting since information about the governing failure mode and the dynamic material properties can be gained. The tests have been performed by increasing the falling height until plastic strains in the bending reinforcement reach a certain level or shear failure occured. Using the six available slabs the test program with 38 impacts as shown in Table 3 has been executed. The procedure for the impact testing includes the following ten steps: 1) Placing the slab on the testing unit and mounting the instrumentation, 2) installing the cushion layer on top of the slab, 3) compacting the cushion layer and measuring the compaction, 4) placing the boulder into initial position and measuring the position, 5) lifting the boulder to the predefined falling height and releasing it, 6) measuring the final position after impact and removing the boulder, 7) measuring the compaction and taking samples from the cushion layer, 8) tracking the cracks of the slab, 9) loosing of the compacted cushion layer at the impact location and returning to step 3) for the next impact test. Finally, 10) removing the slab from the testing unit and taking drilling core samples. For each slab (except the last one) the tests lasted one day. The influence of the slab thickness on the impact response is the major goal of the experimental study. Additionally, the following comparisons could be obtained from the test program:

1) impulse variation: large mass with low impact velocity compared to small mass with high impact velocity

2) slab response with different degradation states, 3) load capacity of the slab with and without shear reinforcement 4) conventional and special cushion layer

4. Discussion It could be observed that the boulder’s impact surface was not completely horizontal at the moment of contact. This lead to a small rotation of the boulder during the impact and produced a horizontal load acting on the structure. The slabs received a small lateral displacement. The bolts ensuring the horizontal restrain of the test units had to be replaced several times. Neither the influence of the boulder shape nor the compaction of the cushion layer has been studied during theses large-scale test. The compaction of the cushion layer has been studied parametrically in previous studies [3]. Before every test the gravel layer has been compacted with a compacting machine to Mv-values around 45 MN/m2. This value corresponds to the upper range of compaction that can be expected on top of a rock fall gallery. After the tests, the cushion layer was less compacted than before. The separation of cushion layer and slab was clearly observable with the high-speed video recording. In difference to real rock fall galleries, the slabs are not retrained from lifting off the supports. This was also observed in the high-speed videos. During the latest phase of the slab response, the slab, the cushion system and the boulder are in

a) b)

Figure 9: Slabs after failure in test D2 a) front view b) chipping of concrete in the compaction zone

free oscillation. From the oscillation period, the actual stiffness of the slab can be deduced. The failure mode that could be observed in all slabs was a combined bending shear failure close to the simple supported corner (Figure 9). According to the design of the slabs, a bending failure along the middle of the slab was expected. Punching resistance of the slab was close to the bending failure. For the structural analysis and the design of the structure, the supposed failure mode plays an important role [8]. Also for the assumptions of dynamic material characteristics, the structures response is important. 5. Conclusions Large-scale tests have been presented that simulate the impact of a falling rock onto a rock fall protection gallery. An instrumented boulder has been dropped onto six concrete slabs that were covered by a conventional and by a special cushion system consisting of high-tensile steel wire mesh filled with a layer of light-weight cellular glass (Misapor). With this system it was possible to reduce the support forces substantially. It could be shown that the test setup produces reliable results. The obtained data is very detailed and allows for an extensive analysis describing the rock impact, the behavior of the cushion system and the interaction between impacting boulder and concrete slab. 6. Outlook The test results, the evaluation methods and the comparison with analytical models will be published next. Studies of the structural performance of concrete slabs subjected to rock fall impacts will lead to a design concept for rock fall galleries. The teamwork between the different disciplines in rock fall studies (detachment of blocks from cliffs, trajectory analysis, geotechnical studies of the cushion layer and structural response of the gallery) will improve the handling of rock fall problems, i.e. mitigate the damage of infrastructure or humans lives due to rock fall. Acknowledgements These large-scale tests have only been possible to perform thanks to the help of many involved. The authors wish to thank Werner Gerber, Andreas Müller, Jorge H. Schellenberg, Markus Baumann, Christoph Gisler, Matthias Denk, Bruno Fritschi, Daniel Caduff, Sara Ghadimi, Stephan Fricker, Thomas Jaggi, Heinz Richner, Jan Laue, Reto Hess, Heinz Gubser and Hans Kienast. Special thanks are also addressed to the highway administrations of the cantons of Grison and Uri for their indispensable financial support. References [1] Schellenberg, K., Vogel, T. (2005). Swiss Rockfall Galleries - Impact Load, Proceedings IABSE Symposium Lisbon 2005, 'Structures and Extreme Events', IABSE Zurich, 2005, pp. 302/303 and CD-ROM file LIS099.PDF, pp. 1-8. [2] ASTRA, SBB (1998). Einwirkungen auf Steinschlagschutzgalerien, Richtlinie, Bundesamt für Strassen, Baudirektion SBB, Eidgenössische Drucksachen- und Materialzentrale, Bern. [3] Montani, S. (1998). Sollicitation dynamique de la couverture des galeries de protection lors de chutes de blocs, Dissertation, No. 1899, École Polytechnique Fédérale de Lausanne, 190 pp. [4] Bucher, K. (1997). Dynamische Berechnung von Steinschlageinwirkungen, Proceedings, Schweizerische Gesellschaft für Boden und Felsmechanik, Conference Paper Montreux, [5] Chikatamarla, R. (2005). Rockfalls on slopes and structures, Dissertation, No. 16315, Swiss Federal Institute of Technology (ETH), Zurich. [6] Volkwein, A., Schädler, S., Fritschi, B., Grassl, H. (2005). Spatial tracking of a falling rock using internal acceleration sensors, EGU, Vienna. [7] Schaedler, S. (2004). Ermittlung dreidimensionaler Starrkörperbewegungen anhand von Beschleunigungsdaten, Diplomarbeit Fachhochschule Weingarten. [8] Schellenberg, K., Vogel, T. (2007). Tests and analytical model of rockfall impacts on galleries, Proceedings of Protect2007, Structures under Extreme Loading, Aug. 20-22, Whistler, p. 27 and CD-ROM file SWO04_Schellenberg.pdf, pp. 1-10.