Summer Research Fellowship Programme of India's Science Academies

Parameterization of porous structures using Computer aided design

Ankita Sheoran

Department of Mechanical Engineering, National Institute of Technology, Delhi, 110040

Professor B. Gurumoorthy

Center for Product Design & Manufacturing, Indian Institute of Science, Bengaluru, 560012


Building models of complex porous structures in Computer-aided design environment is an arduous job because of high nonuniformity, unevenness and intricacy in the shape and size. Modeling of both regular pore shaped structures like honeycomb structure and irregular pore shaped structures based on voronoi tessellation is done. Grasshopper, a visual programming language and environment that runs within Rhinoceros 3D CAD application has been utilized to model the same. This approach is tedious as the construction of the model is not related to either some of the geometric attributes of interest or to the physical properties desired in the porous structure. Consequently, the process involves several iterations of `generate and test' before obtaining the desired model. The overall idea is to develop a toolbox for porous structures, which can return geometrical properties on being supplied with Physical properties as an input. The geometrical properties so obtained are used to finally get an optimized CAD model as an output maintaining the user defined physical properties. This report presents techniques for the extraction of Geometric properties from a given CAD model of a Porous structure. This is done to aid in second stage of project that is generating an optimized CAD model from geometric properties obtained from deep learning. Scripts have been written in Python using libraries in Grasshopper to generate basic geometric primitives like cube with a single open pore, cube with multiple open pores and cube with a closed pore, followed by scripts written to extract geometric properties like Porosity, internal surface area, tortuosity and strut thickness from a given porous structure. For better understanding the geometric properties of a porous structure, algorithms have been developed and implemented as scripts to facilitate the identification of pores or void in a porous structure.

Keywords: porosity, geometric properties, Grasshopper 3D, Rhinoceros 3D CAD application, python scripting, deep learning

Written, reviewed, revised, proofed and published with