Numerical Simulation Of Landslide Generated Waves In Reservoir

Landslide generated waves hazard originates from a certain volume of landslide quickly impacting a large water body, producing initial waves and propagating far away, threatening lives and facilities in and around the water. Fast but small volume landslide impacting water is not able to push water to wave hazard. Giant but slow landslide into water will not cause high impulsive waves, large and fast landslide hitting a small stream is incapable of making wave hazards neither. Therefore, a certain landslide volume, an impact speed and a large water volume are three necessary conditions for composing landslide generated waves hazards.Many landslides distribute along reservoirs, rivers, lakes and coastal in China and may be triggered by earthquakes, volcanoes, rainfalls and water level fluctuates, followed by river or lake tsunami. The influence region of wave hazards is much larger and the consequence is often more severer than landslide itself. The worst combination comes from a large fast landslide impacting a deep and wide water body and exposed elements in the region such as lives, structures, ships, docks and vegetation will suffer a huge disaster.China has get achievements on landslide generated waves research both on laboratory experiment and computational simulation by now. However almost all empirical formulas have their own limitations in application. Influence factors such as landslide type, failure model, volume and velocity beyond the scope of application will result in incorrect wave height. Computational simulation is becoming more and more popular for recent decades, with the widely using of hydrodynamics theory. The advantage of modeling is very obvious compared with lab texting and mathematical analysis. While many problems still exist in the present simulation methods such as inadequately dealing with complex river boundary conditions, insufficient computational accuracy, blur in the water and landslide interacting during impacting, and unable to coupling the landslide modeling and wave’s generation.This thesis proposes a new modeling method named’Tsunami Squares’ for landslide generated waves. It masterly avoids all the traditionally complex problems such as mesh, solve differential equations, deal with wet and dry boundaries, and seamlessly couple models the landslide and waves generation. The thesis has detailed described the theory and characteristics of Tsunami Squares method, and validated the method by three ways:analytic method, lab testing and case study. The results show a high feasibility and precision of TS method and it will play a significant role in landslide generated wave hazard research. Main conclusions for the thesis are listed as follow:(1) Summarize typical landslide generated waves methodsThe influence factors for landslide generated initial wave height contains still water depth, impacting speed, width, length, thickness, volume, impacting angle, density, underwater moving time, landslide shape, front facing water area. Two factors landslide impacting speed and water depth are considered by all researchers in lab experiments, and the next is volume. Run out height depends on the wave height at bank toe and influenced by bank slope angle, platform, roughness, and coming wave angle. All formulas have their own application scope, and people should be careful in application.(2) New method for landslide-generated waves simulation-----Tsunami SquaresModeling objects in Tsunami Squares are composed by multiple small and uniform sized squares that carry all physical quantities and can be further divided. TS method can model a variety of moving materials with flow features, such as landside, debris flow, lava, avalanches, waves, floods, tsunami, glacier, etc. It has many advantages compared to traditional ways:All square grids are getting from DEM directly and no need for complexes meshes; Conservation equations are transformed according to regularity of squares that obviates the differential calculations, so that the computing efficiency has greatly enhanced and the application scope is extended; the divisible feature makes the results accuracy can be higher than the single squares size; Care little about wet and dry squares leads to the seamlessly coupling the landslide and wave generation modeling.(3) Variation landslide moving modesLandslide property closely relates to the moving velocity. It belongs to solid motion during initiation and it transfers to flow like sliding when increasing velocity and losing cohesion. The upper surface of landslide provides the acceleration now. When the movement is close to stoppage, a low dynamic basal friction will transfer to a high static value to stop slide, and the landslide comes back to solid again. Stopped material thus becomes part of an evolving topography that influences the paths of later passing material.Landslide movement is traditionally treated as solid motion and the acceleration comes from basal surface gradients. Here we have a new sight of variation landslide moving property that treats landslide as flow like material and it is closer to the real world nature. Here we focus on the variation landslide moving property and treats landslide as flow like material. This is a new sight in landslide modeling.The parameter system contains basal friction, dynamic friction and repose angle. The friction force is assumed to be composed of a basal friction independent of velocity and a dynamic friction proportional to the velocity square. The two frictions control the moving property variations. Repose angle control stoppage and deposits. The combined parameter effects make landslide modeling close to the real moving nature.(4) Concepts of wave generationThree ways of wave generation to different sources:Complete Momentum Transfer (CMT), No Momentum Transfer (NMT) and Drag Along (DA). CMT combined with DA is used for landslide generated waves, corresponding to volume waves and impulsive waves. Analyzing the water and sliding interaction, landslide displacement, velocity, thickness and water velocity play a role during wave generation(5) Lab testing and TS simulation results comparisonComparing the Lab testing and TS simulation results we can find that when DA coefficient cd=0.1, the simulated initial wave height is the closest to the measured value. Landslide volume and velocity are most important factors for wave generation, while the drag along force from slide to water plays subordinate effects. The character of wave generation process in modeling is similar to the lab testing and the wave attenuation trends are the same. Both wave heights decay fast when wave propagates in lkm and change little beyond 1 km. The wave train frequency of modeling and lab testing has the correct similarity 1:(?). The comparison proves TS simulation method is feasible and reliable.(6) Verify the accuracy of TS method by two case studiesTwo cases are the 2008 Gongjiafang landslide event in China Three Gorges Reservoir and 2007 Chehalis lake rockslide event in Canada. Clear out the failure reason and initiation pattern; build a landslide model in TS program and coding the progressive failures. Choose appropriate parameters and simulate the landslide and wave generation, propagation and run out. Conclusions are list as follow:1. Evolving topography and impacting velocity are two important factors during wave generation. The’topography’is changing every time step and it relates to landslide thickness (volume), moving path and final deposition, so accurate landslide modeling is a key precondition for wave heights and shapes. Before a correct inverse modeling, one should clear out the landslide geo-conditions, material, failure mode, deposited shape etc.2. Four parameters basal friction μb, dynamic friction μd, repose angle θrepose, and critic velocity vcrit need to settle in landslide modeling. In Gongjiafang case, μb=0.2,μd=0.05 (in the air), μd=0.2 (in water),θreposee=30°; in Chehalis case,μb=0,μd=0.003 (in the air),μd=0.02 (in water), θrepose=20.5°; If lansldie surface angle is smaller than the repose angle, the value of μb increase gradually.3. The two cases are not fail as whole. Gongjiafang landslide collapse from lower part to upper parts influenced by water level, while Chehalis rockslide starts from upper parts influenced by snow and rainy weather. Different iniation codes are embeded in TS program and it tracks velocities in multipe positions. The simulated velocities and deposition states agree with the real case.4. Both cases use DA combined with NMT to produce waves. Drag Along parameter cd=0.1 in Gongjiafang case while it equals to 0.11 in Chehalis case. The simulated wave heights are highly consistent with measured values, which indicate the TS method can well handle the landslide generated wave simulation.5. Micro topography leads to board variations with distance for wave decay curve. Mutual relation of wave propagation direction and bank slope aspect affects run up heights. Oblique crossing of two directions makes higher waves especially when bank slope surface perpendicular to the wave direction. Large scale bends in the propagation path do not reduce wave height, but small scale bends do.6. Simulation results show that initial wave height largely depends on landslide impacting velocity and spatial scales. Wave energy dissipation and run up height are influenced by distance, water depth, propagation direction and band slope aspects.(7) study landslide volume, river width, and water level effects on wave heights by TS simulationEstablishing four volume scales of landslide in TS program, small scale, middle scale, large scale and giant scale and impacting a same river with a same speed respectively. Searching the relationship between volume scale and wave height and it shows that:landslide volume plays a significant role in wave generation and initial wave height is proportional to log volume, the larger the higher; Initial waves in small and middle scale landslide is higher than cross river run up height, while it’s the same or even smaller in large and giant landslide; Large and giant landslide waves decay fast in 3 km upper and downstream and then keep the value for kilometers; Middle landslide waves reduced to a small value beyond 2 km upper and downstream and small landslide makes low waves all the way. The large and giant landslide generated wave in reservoir should be paid a high attention, middle landslide’s speed should be concerned, and small landslide would not cause wave hazards.Establishing a straight river with four widths (500m,750m,1000m and 1200m) and 30° of bank slope in TS program. Impacting them by a same landslide model and searching the relationship between river width and wave height. The results show that:the initial wave heights are the same in four river widths; the landslide in this experiment is middle scale volume with speed of 20 m/s, all initial waves’heights reaches to 25m and decay fast to 5 m in 0.5 km upper and downstream. In 0.5 km to 2 km away from landslide point, wave decay relatively slower in narrower river and faster in wider river. Waves in narrower river propagates further, influence more area, and cause severer consequences.Establishing a straight river with an exponent cross section and put water to 130m,145m, 160m and 175m in river depth in TS program. Impacting them by a same landslide model and searching the relationship between river depth and wave height. The results show that:wave heights attenuate to a same rule in four depths that they all decay fast in 1 km and change little beyond.0 to 0.5km decays fastest and 0.5 to 1km followed. Initial wave height positively correlates to water level while cross river wave height negatively correlates to the river width; initial wave height is close to the cross river wave height in a low water level but it is much higher in a high water level.