A hybrid investigation of geometrical effects on landslide-tsunamis: Generic hazard assessment and numerical benchmark test cases

本研究旨在探讨水体几何形状对滑坡海啸的影响。要预测此类事件，至少可以采用三种方法：(i)物理模型测试，（ii）数值模拟和（iii）经验方程。传统的经验方程是基于物理模型测试。不幸的是，这种模型测试主要是在狭窄的波浪通道（2D）中进行的，并且没有系统地解决水体几何形状的影响。这种剧烈的自由表面流动的数值模拟直到最近才进入实际应用。光滑粒子流体力学SPH是解决这些问题的一种很有前途的计算方法。建议的工作采用所有三种方法，使用物理模型测试，验证和先进的SPH代码，并开发迄今为止最通用的经验方程。这项工作直接建立在PI之前通过其危险评估手册在物理滑坡-海啸建模和实际应用方面的成功基础上。海啸不仅是由地震活动引起的，比如2011年的日本海啸和2004年的印度洋海啸灾难，还包括山体滑坡、岩石坠落或冰崩解等其他大规模运动。例如，这种山体滑坡海啸（或湖泊或水库中的冲激波）发生在公元前6100年影响苏格兰的挪威海岸，1963年导致约2000人死亡的瓦昂特灾难，以及1998年导致2100人死亡的巴布亚新几内亚。冲激波在奥地利、加拿大、中国、意大利、挪威、瑞士和土耳其等多山国家也经常发生，因此需要在水库规划阶段进行危害评估。英国也经历过类似的情况，例如在苏格兰，它可能受到来自加那利群岛拉帕尔马的潜在山体滑坡海啸的影响，并且在冰崩解的背景下，它与航运和石油和天然气工业的相关性越来越大。与地震海啸相比，在理想条件下，用经验方程可以相对较好地预测滑坡海啸。然而，这些方程主要是在波浪通道（2D）和波浪盆地（理想3D）中发展起来的，其中基于2D和理想3D的方程的波高预测由于相同的滑动特征而相差高达一个数量级。目前，由相同的滑坡特征产生的二维海啸和理想的三维海啸之间的差异还知之甚少。基于2D或理想化的3D数据的公式是否更适用于特定的原型形状，以及对中间几何形状的一般预测（例如大坝侧面的滑动冲击）也完全未知。迄今为止，水体几何形状的重要影响尚未得到系统的解决，尽管众所周知，它是导致滑坡-海啸（脉冲波）预测中相当大的低估或高估的原因。本文将采用混合（实验-数值）方法系统地研究水体几何形状的影响。物理模型实验将在室内波浪水槽和波浪盆中进行，从而排除显著的尺度效应。测量的独特组合将包括整个滑动的运动学，在滑动前的压力记录和波的特征。这将产生高质量的基准测试用例，对处理流固耦合应用和剧烈自由表面流动的数值模拟界有价值。最近发布的功能，包括复杂的几何形状，通过Blender在spphysics将有可能进行额外的测试在2D，理想的3D和中间几何纯粹的数值。这种混合方法的数据集将用于开发迄今为止预测滑坡海啸（脉冲波）的最通用经验方程，从而大大提高对湖泊、水库、峡湾和海洋的预测。

The proposed research concerns the investigation of the effect of the water body geometry on landslide-tsunamis. To predicting such events, at least three approaches may be adopted: (i) physical model testing, (ii) numerical simulations and (iii) empirical equations. Traditionally empirical equations are based on physical model testing. Unfortunately, this model testing has mostly been conducted in narrow wave channels (2D), and did not systematically address the effects of the water body geometry. Numerical simulations of such violent free surface flows have only recently come into practical reach. A promising computational method for solving these problems is Smoothed Particle Hydrodynamics SPH. The proposed work adopts all three methodologies, using physical model testing, to validate and advanced SPH code, and develop the most generic empirical equations to date. The work directly builds upon the PI's previous success in physical landslide-tsunami modelling and real-world applications through his hazard assessment manual.Tsunamis are not only caused by seismic activities such as in the 2011 Japan and 2004 Indian Ocean tsunami catastrophes, but also by other mass movements such as landslides, rock falls or ice calving. Such landslide-tsunamis (or impulse waves in a lake or reservoir) occurred, for instance, off the coast of Norway affecting Scotland in 6100 BC, in the 1963 Vaiont catastrophe with a death toll of about 2,000 or in Papua New Guinea in 1998 with 2,100 fatalities. Impulse waves occur also quite frequently in mountainous countries such as Austria, Canada, China, Italy, Norway, Switzerland and Turkey such that a hazard assessment is required in the planning phase of reservoirs. The UK also experiences similar cases e.g. in Scotland, it may be affected by a potential landslide-tsunami originating at La Palma on the Canary Islands and it is of increased relevance for the shipping and oil and gas industries in the context of ice calving.In contrast to seismic-tsunamis, landslide-tsunamis may be relatively well predicted with empirical equations under idealised conditions. However, these equations were mainly developed in wave channels (2D) and wave basins (idealised 3D) where the wave height predictions with equations based on 2D and idealised 3D differ up to an order of magnitude by identical slide features. At present the differences between 2D and idealised 3D tsunamis generated by identical slide features is poorly understood. It is also not well understood whether formulae based on 2D or idealised 3D data apply better to a specific prototype shape and generic predictions of intermediate geometries (e.g. slide impact at dam flank) are unknown at all. The important effect of the water body geometry was not systematically addressed to date even though it is widely known to be responsible for considerable under- or overestimations in landslide-tsunami (impulse wave) predictions.The proposed work will systematically investigate the effect of the water body geometry with a hybrid (experimental-numerical) approach. The physical model experiments will be conducted in the in-house wave flume and a wave basin thereby excluding significant scale effects. The unique combination of measurements will include the entire slide kinematics, pressure recordings at slide front and wave features. This will result in high quality benchmark test cases valuable for the numerical modelling community dealing with fluid-structure interaction applications and violent free surface flows. The recently released feature to include complex geometries via Blender in SPHysics will make it possible to conduct additional tests in 2D, idealised 3D and intermediate geometries purely numerically. The data set from this hybrid approach will be used to develop the most generic empirical equation to predict landslide-tsunamis (impulse waves) to date resulting in considerably improved predictions in lakes, reservoirs, fiords and the sea.