Prediction-of-the-roughness-equivalent-sandgrain -height

Description

This repository is prepared to predict the roughness equivalent sandgrain height, k_s, using Machine Learning (ML) techniques.

Deep Neural Network (DNN) and Gaussian Process Regression (GPR) are employed to predict k_s as a function of roughness geometrical parameters.

For more details on the methods, please refer to Aghaei Jouybari, M. Yuan, J. Brereton, G.J. and Murillo, M.S. (2021) Data-driven prediction of the equivalent sand-grain height in rough-wall turbulent flows. J. Fluid Mech. 912, A8.

Folder 1

Please consult with this folder if you are interested to use the already trained version of ML networks and predict k_s for your surface.

Folder 2

Please consult with this folder if you are interested to train your own version of ML networks. This folder contains some helper codes that can receive arbitrary geometrical parameters and train DNN and GPR networks to predict k_s.

Please consult with README.md in each sub-folder.

Please contact the authors if you are interested in contributing on database.