Computer-aided design (CAD) and analysis (structural, computational fluid dynamics CFD, analysis) provide simplified and discrete solutions to various engineering design problems. Iterative optimization techniques based on these numerical methods for design improvement is not generally used in the industry when dealing with complex physical phenomena, due to high computational costs. To optimize an intended performance, the complete design space has to be searched by continuous variation of the design parameters. This requires enormous amount of time in a user-interactive and an interfacing CAD-based system as the user has to perform many numerical experiments and transfer information from one system to another. The paper describes a feature-based parametric design to model and evaluate based on CFD, and design a fan impeller used for dehusking grains. The project was part of an endeavour to facilitate technological support to grassroots innovations and indigenous knowledge systems in North East India. Optimization of a special double layer multi-bladed bamboo fan impeller is described as an application study. The effects of critical geometric features on the airflow characteristics, mainly the turbulence of the flow field in front of the fan, are examined using the feature-based geometric modelling tool.