Novel time-efficient approach to calibrate VARANS-VOF models for simulation of wave interaction with porous structures using Artificial Neural Networks

Abstract

Numerical models are valuable tools to provide information on wave-structure interaction processes that are difficult to measure in a physical model. The current level of accuracy of numerical models is relatively high but an adequate validation to establish the models' empirical parameters is required to ensure a correct representation of the phenomena to be investigated. To this end, a "trial and error" approach is typically adopted, potentially resulting in a large number of simulations. In this study, a methodology based on Artificial Neural Networks (ANN) is presented, to obtain the optimal combination of values for the empirical coefficients that characterize the porous media in a VARANS-VOF model for predicting mean overtopping discharges. From an initial reduced set of simulations, input and output data are obtained for the training of the ANN. The ANN is then used to estimate the best values for the combination of coefficients that describe the three breakwater layers (armour, filter, and core), which are subsequently applied in the numerical model. The method was successfully applied to a set of laboratory experiments aimed at obtaining overtopping discharges for a Single-Layer Cube armoured breakwater. It resulted in a large reduction of computational effort when compared to the simulation of all possible combinations of values.