Thomas Wortmann

Home / People / PhD Students / Thomas Wortmann

PhD Graduate

Email: 
Pillar / Cluster: Architecture and Sustainable Design

Biography

Thomas currently is a lecturer in architectural design at Xi’an Jiaotong-Liverpool university. His research and teaching interests are the integration of computation into architectural design processes, focusing on performance-informed design. He leads the development of Opossum, an award-winning, machine-learning-related optimization tool freely available from food4rhino.com. Currently, he is developing multi-objective, as well as visual and interactive, versions of Opossum.

Thomas is a registered architect and experienced computational designer, having used computer programming for his own projects, the pioneering digital architecture practice of NOX / Lars Spuybroek, and Web Structures, a multi-disciplinary engineering practice in Singapore. In 2015 and as a member of the Advanced Architecture Laboratory at SUTD, he automated the generation and documentation of the Bayfront Pavilion’s envelope, a forty-meter span gridshell with around 10.000 individually-sized and -perforated cladding panels in Gardens by the Bay.

For his publications on computational design, architectural design optimization and related topics, please visit Google Scholar and ResearchGate. To keep up-to-date with his research, please follow Thomas on LinkedIn.

PhD Summary

The PhD thesis (advised by Prof. Thomas Schroepfer) presents optimization as a co-design method for architects and engineers that is grounded in computer science (CS) and applied mathematics. The thesis utilizes CS methods such as benchmarking and surrogate modelling, test problems related to structure, building energy, and daylight as well as design methods including prototyping, visualization, and user-testing.

A key element of the PhD research was the development of Opossum, an optimization tool (plugin) for Grasshopper. Opossum offers an implementation of RBFOpt (Radial Basis Function Optimization), a novel machine learning (ML)-based algorithm that accelerates optimization processes by approximating fitness landscapes.

The PhD and related publications demonstrate that RBFOpt is the most efficient optimization algorithm in Grasshopper when only hundreds of design candidates can be evaluated, e.g. simulated. This fast convergence is relevant especially for “expensive” optimization problems that require time-intensive simulations or physical experiments. Opossum allows architects and engineers to, for example, optimize daylight or aerodynamics, which typically are time-consuming to simulate.

This research has received international recognition. It is applied by numerous users worldwide and generates impact in peer-reviewed international journals. Opossum and RBFOpt won the international COIN-OR Cup 2016 from the operations research community and the SG Mark award 2017 from the design community. The PhD received the best dissertation and best design practice awards SUTD. Opossum is freely available on www.food4rhino.com and has been downloaded more than 3,500 times. The SUTD-MIT International Design Center partially supported the development of Opossum with two research grants.

2020-02-28T11:56:38+00:00February 28th, 2020|PhD Students, PhD Students ASD|