Data assimilation for the calibration of flume tests with different granular mixtures

The flow-type landslide (mud-flow, debris-flow, etc.) could often cause large economic and social damages. Besides the mechanisms that rule the triggering phenomena, the comprehension of their propagation behaviour represents an important aspect to be addressed, because from this issue it depends the reliability of different risk scenarios and the evaluation of mitigation strategies. The kinematic of these phenomena as well as the shape and the heights of the final deposit is related to the viscous character of the involved materials that in turn depends on the mutual concentration of the component phases: the water, the fine matrix and the coarser fraction. In order to understand the composition effects on the kinematic of flow-type landslides several flume tests were carried out using different kaolin-sand mixtures, and then simulated with a particle-based SPH code. The constitutive parameters derived from a laboratory characterization of two components were then fine-tunes by mean of a data assimilation numerical procedure in order to individuate the most suitable values for describing the observed propagation.