Using a Data Driven Approach to Predict Waves Generated by Gravity Driven Mass Flows

When colossal gravity-driven mass flows enter a body of water, they may generate waves which can have destructive consequences on coastal areas. A number of empirical equations in the form of power functions of several dimensionless groups have been developed to predict wave characteristics. However, in some complex cases (for instance, when the mass striking the water is made up of varied slide materials), fitting an empirical equation with a fixed form to the experimental data may be problematic. In contrast to previous empirical equations that specified the mathematical operators in advance, we developed a purely data-driven approach which relies on datasets and does not need any assumptions about functional form or physical constraints. Experiments were carried out using Carbopol Ultrez 10 (a viscoplastic polymeric gel) and polymer water balls. We selected an artificial neural network model as an example of a data-driven approach to predicting wave characteristics. We first validated the model by comparing it with best-fit empirical equations. Then, we applied the proposed model to two scenarios which run into difficulty when modeled using those empirical equations: (i) predicting wave features from subaerial landslide parameters at their initial stage (with the mass beginning to move down the slope) rather than from the parameters at impact; and (ii) predicting waves generated by different slide materials, specifically, viscoplastic slides, granular slides, and viscoplastic granular mixtures. The method proposed here can easily be updated when new parameters or constraints are introduced into the model.

Experimental study of impulse waves generated by viscoplastic fluid

Zhenzhu Meng

Landslide-generated waves, also called impulse waves, occur as a result of the intrusion of landslides (such as rock falls, debris flows, and avalanches) into bodies of water (such as lakes, reservoirs, and seas). The objective of this thesis was to study the momentum transfer from the slide material to the body of water, in order to develop a better understanding of how the slide material's properties affect the wave generation and to discover alternative modeling approaches to existing empirical equations. Previous experimental studies have usually used blocks and granular materials to mimic natural landslides. However, many landslides in real worlds have been idealized as viscoplastic fluids in theoretical and numerical studies. No studies have used viscoplastic material in experimental studies of landslide-generated waves. The originality of this thesis lies in the use of a viscoplastic material called Carbopol Ultrez 10, an artificial aqueous micro-gel whose rheological behavior can be described using the Herschel-Bulkley model. Carbopol's cohesive and deformable properties are different to both block and granular slides. Further, Carbopol is transparent and can easily be seeded with micro-seeding particles, so its velocity field can be measured using particle image velocimetry (PIV). As a comparison of Carbopol, I also used a granular material named polymer-water balls whose density is close to that of Carbopol. The investigations of this thesis are as follows: I conducted two series of experiments. First, I observed waves generated by Carbopol, water balls, and mixtures of them using high-speed cameras, to investigate the role of slide material's properties in wave prediction. Second, I conducted PIV experiments with Carbopol to investigate the internal dynamic of slide-water interaction. (1) I developed a theoretical model that combined the momentum conservation of twophase flow in a control volume (Zitti et al., 2016) and viscoplastic theory (Ancey et al., 2012). With the experimental results obtained from PIV measurements, I analyzed the drag force and hydrostatic force that act on stopping the sliding mass, and validated the theoretical model. (2) I developed empirical equations using the dimensionless groups that emerged from the governing equations to quantify the wave characteristics (for example, maximum wave amplitude and height) as functions of the slide parameters. Using empirical equations, I compared the characteristics of waves generated by cohesive Carbopol and cohesionless water balls, and discussed the effect of slide cohesion on wave generation. (3) Taking advantage of a purely data-driven approach that strictly relies on the dataset and does not need any physical constraints in advance, I applied an artificial neural network method to predict wave characteristics under complex configurations, such as dealing with an experimental dataset with several different slide materials (Carbopol, water balls, and mixtures of them). (4) Using a panel data model called random coefficient model, I predicted the time series data of wave characteristics from the time series data of slide parameters on impact. Given the slide parameters on impact by the viscoplastic theory, the temporal wave characteristics were quantified from the parameters of the slide material at the initial stage (at rest on the slope and then starting to move).

EPFL2020

Experimental study of impulse waves generated by viscoplastic fluid

Zhenzhu Meng

EPFL2020