What is it about?

This research is about developing a faster, more efficient way to evaluate how wind affects buildings, particularly at the pedestrian level. Traditionally, wind environment investigation uses something called Computational Fluid Dynamics simulations, which are complex and take a long time to conduct. This research aims to create a new system, a 'hybrid framework', that combines several tools - parametric design (which uses computer programming to generate design variants), image processing, and machine learning - to predict wind behaviour swiftly and accurately. This system is trained with a large amount of data from 300 simulated building cases, allowing it to predict Low-Velocity Areas (LVAs) - places where the wind speed is decreased - around rectangular buildings. The model's predictions were tested against new building cases and the results were close to those obtained from traditional methods, proving its efficiency and accuracy. This research is about a new tool that can quickly and accurately predict how wind behaves around buildings, which can be useful in designing sustainable, comfortable outdoor spaces. Highlights include • A proposed hybrid framework for rapidly evaluating built wind environment. • Parametric model, CFD, image process, and machine learning jointly formulated the framework. • Low-velocity regions around buildings could be predicted in a very short time. • Benchmark investigations have demonstrated the efficiency and accuracy of this framework. • Rapid evaluation of the urban environment could benefit the development of green urban design.

Featured Image

Why is it important?

Firstly, understanding wind behaviour around buildings is crucial for sustainable architecture and urban design. Wind can impact a building's energy efficiency, the comfort of pedestrian spaces, and even the structural integrity of the building. Secondly, the traditional method of using Computational Fluid Dynamics (CFD) simulations to study wind behaviour can be complex and time-consuming. The proposed hybrid framework provides a quicker, more efficient alternative. This can speed up the design process and allow for more design variations to be tested and compared. Thirdly, this tool can be used in the early stages of design, providing valuable information for design optimization. It can help architects and urban planners make informed decisions about their designs earlier, potentially saving time and resources down the line. Overall, this tool could innovate the field of architecture and urban planning by making wind analysis more accessible and efficient, leading to better, more sustainable design outcomes.


My research is truly cutting-edge and could have significant implications across several fields. It matches my overall research undertakings in computational architecture and urban design. From a technology and engineering standpoint, our endeavours represent a remarkable fusion of several advanced tools - parametric design, Computational Fluid Dynamics (CFD) simulation, image processing, and machine learning. This multidisciplinary approach allows for a more holistic and efficient wind analysis method, pushing the boundaries of what is currently possible. From an architectural perspective, the potential to rapidly evaluate building performance relative to wind is incredibly valuable. It allows architects to design buildings that optimize wind flow to increase energy efficiency and improve pedestrian comfort levels, leading to more sustainable and livable urban environments. From an urban planning perspective, understanding wind behaviour can help guide strategic decisions about building placement and city layout. This can help minimize wind-related issues such as urban heat islands and pedestrian wind discomfort. From a climate perspective, reducing energy consumption in buildings is a key strategy for mitigating climate change. This hybrid framework could be important in designing energy-efficient buildings and promoting sustainable architecture. In essence, my research signifies an important stride toward more sustainable and optimized urban design practices, leveraging the power of advanced computational tools and machine learning to enhance our understanding of wind behaviour around buildings.

Professor Marc Aurel Schnabel
Xi'an Jiaotong Liverpool University

Read the Original

This page is a summary of: Hybrid framework for rapid evaluation of wind environment around buildings through parametric design, CFD simulation, image processing and machine learning, Sustainable Cities and Society, October 2021, Elsevier,
DOI: 10.1016/j.scs.2021.103092.
You can read the full text:



The following have contributed to this page