Application of physical models to improve the mitigation structures of debris flow in a case study

. Debris flows are a phenomenon in mountains areas, where a large amount of mobile rock material, large amounts of water and steep slopes meet together. They are characterized by high capacity of sediment transport, high concentration of sediment with wide range of grain sizes and high front-velocity. To improve the protection measures against debris flow, it is necessary to know the processes of debris flows and its interaction with structures of the measures. By investigation of the mitigation structures in experimental scaled physical models it is possible to understand the functionality of the structures under different scenarios as well as optimization of the structures. For the protection of the counties Agarn und Leuk in Valais in Switzerland a multi-level debris flow basin system has been designed. By means the physical model in scale of 1:55 the functionality and the dimensions as well as the details of the structures could be optimized.


Introduction
Debris flow is a flow of sediment and water mixture in a manner as if it was a flow of continuous fluid driven by gravity, and it attains large mobility from the enlarged void space saturated with water or slurry [1]. Debris flows are mainly subjected to the inertial and gravitational forces. Therefore, Froude's law of similarity can be applied, for physical models of debris flow [2].
Physical models of debris flow have of course limitations and simplifications. For example, the terrain surface in most of the models is non-erodible. The sediment entrainment from torrent bed can't be investigated in such models. The choice of the model scale depends not only on the influence of scale effects but also, among other things, on the space conditions in the laboratory and handling of material logistics for the required debris volumes. Based on our experiences, the appropriate choice of the model scale is usually between 1:40 and 1:60. This is accompanied by the fact that the fine components can no longer be reproduced in models. The natural mixture of debris flow is strongly variable and can't be reproduced 100% in the laboratory. However, in order to obtain meaningful results, the debris flow mixtures are calibrated to target values. The target values include the volume of surge , the maximum discharge , and the maximum front velocity of the debris flow. In this way the flow behaviour of debris flow can be reproduced, although the model mixture is not exactly the same as the mixture in the nature. However, the large grain sizes are considered in the model mixture, which are very * Corresponding author: davood.farshi@ost.ch important for clogging of the outlet structures of sediment traps.

The multi-level debris flow basin system Meretschi
In the catchment area Meretschibach, between Agarn and Leuk in Canton Valais, smaller debris flows occurred repeatedly in the past years. Based on the new hazard assessment map, a large debris flow potential up to 130'000 m 3 with a probability of 0.003 in a year can occur. The residential settlements and traffic routes located in the valley are at considerable risk. In order to mitigate the risk a multi-level debris flow basin system has been designed on the slope above the valley (Fig. 1).
The five deposition areas have different tasks. For example, the sediment trap 1 (Geschiebesammler 1) is intended to ensure the deposition of very frequent events (probability over 0.1 in a year) with a volume of up to 20'000 m 3 . When the spillway is activated (Überlastfall 1), all subsequent surges have to be safely transferred to the sediment trap 2 (Geschiebesammler 2). The sediment trap 2 provides a deposition volume of about 70'000 m 3 and transfers the subsequent debris flow surges in case of a full filling via the spillway (Überlastfall 2A) further towards the sediment trap 3 (Geschiebesammler 3) with a capacity is about 20'000 m 3 . The excessive debris flow load will be transferred over the spillway 2B and 3 (Überlastfall 2B and 3) to the corridors 2 and 3 (Überlastkorridor 2 and 3) These complete the deposition capacity up to 130'000 m 3 [3].

Physical model of Meretschi
Due to the complexity of the debris flow and also the protection structures functionality, the whole multi-level debris flow basin system has been tested by using a physical model.
The physical model of the system represents a real area of about 600 x 180 m with a height 100 m at a scale of 1:55. The entire structure and detailed individual structures are depicted in Fig. 2.
The following objectives were defined for the model tests: • The protective functionality of the system has to be checked and documented depending on the existing debris flow scenarios for the events with a probability of 0.033, 0.01 and 0.0033 in a year. • The functionality of spillways and corridors (overload structures) hast to be checked. • The model structure has to be optimized on the basis of the knowledge gained from the tests in order to achieve the two objectives mentioned above.
The tests are divided into three phases: preliminary tests, main tests for the spillways structures (phase 1) and main tests on the entire system (phase 2). The preliminary tests have been used to calibrate the model mixture of the debris flow. The calibration is preformed that the model mixtures show the same flow condition as in the nature although not all components are the same as in the nature. The maximum front velocity ( ) and die maximum discharge ( ) are used as reference values for the calibration of the model mixture. The model mixture shows the same flow characteristics as in the nature.

Debris flow mixture
In order to calibrate the model mixture, the reference values of the debris flow (Table 1) have been estimated based on empirical methods under natural condition [5].
Additionally, the grain size distribution is necessary for the mixture. Especially the max. size of the grains is very important in this project, because based on this value the grids at the outlet of the sediment traps will be plugged. The grain size distribution has been derived from depositions of last debris flow occurrences [3]. However, the grain size distribution in physical model doesn't have to be exactly the same as in nature, primarily it is important to achieve the target values (Fig. 3). For the fine and viscose part of the mixture we use special ingredient from food industry.  The model mixture is calibrated for each 5 surge volumes differently during preliminary tests. More than 70 surge tests are performed for the calibration. The water content has a strong influence on the flow conditions of the debris flow [4]. Therefore, in first step the calibration of the mixture is done by changing the water content. For measuring the maximum front velocity ( ) at the upper part of the model at least 3 ultrasonic sensors are used (Fig. 4). The maximum discharge ( ) can be controlled through adjusting the opening of the debris container (Fig. 5). If adjustment of the water content does not lead to the target values, the viscose part of the model mixture can be changed.

Main tests
After calibrating the model mixture of the debris flow surges, the main tests are performed in two phases. In the first phase the functionality of the spillways is checked. This happens with and without filled sediments traps with fresh and dried debris mixture (Fig. 6). The tests show that the system has to be modified in 6 positions. In Fig. 7 the modifications at the spillway 2 (Überlastfall 2A/B) are depicted.
In the second phase the entire system is tested through 6 successive surges with total volume varies between 56'000 to 130'000 m 3 (in natural scale). The tests show that at least 5 modifications are needed to have the correct functionality. In Fig. 8 the modifications at the spillway 2 (Überlastfall 2A/B) are depicted.
The tests show also that after the modifications the corridors work well and control the overrun volumes. It shows also that part of the corridors can be neglected in final design (Fig. 9). Totally more than 80 tests are performed for the main part.         Fig. 10 shows the volume of events and the total volume as well as the functionality of each part after all modifications. It shows for which event volume, which part will be activated.

Conclusions
By means of physical model tests, the robustness and functionality of the individual structures are to be checked in order to achieve the protection goals. If necessary, the structure will be further optimized during the model tests. For this purpose, the project perimeter is reconstructed in physical model at a scale of 1:55. The test show that the mitigation structures have to be optimized in different parts to achieve the functionality. It shows also that debris flow close to structures has a complex flow properties, which is still can't be simulated inaccurate way. Therefore, physical models are very important to understand the flow behaviour close to such structures.