Characterizing the Impact of Temperature on Clay-Water Contact Angle in Geomaterials during Extreme Events by Deep Learning Enhanced Method

For unsaturated soils, the reduction of matric suction caused by climate change (global warming) or extreme events (rainfall and wildfire) can potentially trigger landslides. It is well known that temperature can impact the strength of unsaturated soils by changing capillary pressure. A key parameter to determine capillary pressure in unsaturated soils is the contact angle between clay and water (e.g., through the Young-Laplace equation). In this paper, we conduct a series of molecular dynamics simulations of a clay-water model to investigate the impact of temperature on the clay-water contact angle. The spatial location of individual water molecular is derived from the coordinate of the molecule’s center-of-mass. A numerical approach is adopted to determine the contact angle by treating the water droplet as a three-dimensional point cloud. For each molecule, we fit a tangent plane to the surface spanned by its k-nearest neighbors using the covariance matrix method. The normal vector of the planar surface is used to derive the contact angle. For implementation, we enhance the covariance matrix method by a deep learning algorithm based on the Hough transform, a shape feature extraction technique by a voting procedure. Through the Hough transform, all the normal vectors are mapped into a spherical accumulator space (Hough space) divided by many bins of nearly the same area. Each bin corresponds to a small range of normal. The voted sphere accumulator is then transformed to a square image-accumulator, as the input of a convolutional neural network. The output of the convolutional neural network is two angles representing the normal direction in spherical coordinate system. The numerical results have demonstrated that the deep learning enhanced method is robust in characterizing variations of the contact angle between clay water at elevated temperature. It is expected the results can shed some light on interpreting unsaturated soil slope failures triggered by environmental temperature increase during extreme events.