Design of a Tiny House Generator with Location Parameterisation Function

Abstract

In light of growing housing shortages and rising rental prices, alternative housing forms such as Tiny Houses are becoming increasingly popular. This housing form is characterised by compact floor plans, with sizes typically under 45 m² (480 sqft) per person. As the trend originated in the USA, much of the literature and design proposals refer to the prevalent climate conditions found there. This research aims to bridge this gap by developing a script that generates a proposal for a Tiny House for any given location, using construction strategies adapted to the local climate. First, an analysis was conducted to determine which building components of a Tiny House are particularly susceptible to climatic influences and how specific weather conditions affect these components. Based on four case studies and relevant literature, parametric construction principles were developed. These principles were incorporated into a script that used weather and climate data to generate a 3D model of the Tiny House. The script was implemented within the Grasshopper environment of the 3D modelling software Rhino 3D. To provide a user-friendly interface, it was integrated into a web application. This allows users to select locations and various input parameters, to visualize the generated model, as well as to access detailed information about the construction decisions and how they are influenced by the local climate. To exemplify the output generated by the tool, three models for different locations were selected and slightly modified to show how these buildings might be built and look in reality. The thesis was successful in developing a fully parametric building generator, which can further be expanded to include features such as complete indoor climate simulations. The script and implementation are fully documented. However, given the general complexity of architecture and construction, the question arises as to whether a future approach based on artificial intelligence might be more effective than the algorithmic approach taken here.