Wind Flow

Applying Component Micro-Wind-Behaviour to Macro-Structures

A 2011 VEIV EngD Group Project by David Andreen, Gwyneth Bradbury and Samuel Wilkinson

The use of Computational Fluid Dynamics (CFD) for wind simulation and subsequent façade component optimisation can be a computationally inefficient process, requiring lengthy simulations of many iterations of the entire building under various wind conditions. We propose a method for learning the behaviour of individual components at a small scale to then apply to large scale structures. Bentley Generative Components is used to model the component parametrically, NextLimit XFlow is used to simulate the wind environment and Matlab is used to correlate the training set and implement a k-nearest neighbour search for the application test. A number of application tests are conducted in order to explore the inherent issues and possibilities of using such an approach.

One of the potential benefits of this method is reduced batch-simulation time for a later generative design stage detailed optimisation. The possibility of learning local component behaviour and mapping this to larger structures is shown to have more effective success for certain applications over others and shows a strong need for further research