17th September 2023

Innovation through technology: genetic algorithms

Alibaba Xuhui Binjiang Park Zone Y

In recent years, genetic algorithms have been incorporated into Foster + Partners’ design processes. Based on the principles of evolution, these computational techniques have positively shaped a number of the practice’s projects, including Alibaba Xuhui Binjiang Park Zone Y in Shanghai. Genetic algorithms are used to address complex optimisation challenges. They are one of the most popular forms of machine learning, at the core of leading AI research.

The new Alibaba offices showcase the Chinese ecommerce giant’s unique working culture to the public. The innovative design process stemmed from a comprehensive understanding of the company’s structure and their ethos of collective success. The building reflects the ways in which Alibaba functions, their advanced technological vision for the future and the relationship between customers and employees.

The dynamic form of the building was guided by a genetic algorithm, written by in-house specialists within the design team. A workable, three-dimensional grid of cells was used to produce massing options and advance the building concept. The genetic algorithm was set up to create the massing geometry using a three-dimensional reductive process, similar to carving a block of stone. Both divisions of the design team, based in London and Beijing, were able to review and interact with the various massing options in real time. The script was adapted over time to further develop the concept.  

Ryan Mehanna, Associate Partner, outlines this process and describes the human interventions that contributed to the final solution. ‘It involved an intuitive and careful combination of digital processes and human judgment, to create a building that meets the client’s needs and vision. We wrote the genetic algorithm within a few weeks, which quickly yielded a range of meaningful massing options. This gave us the advantage of locking in the concept early on, thereby allowing ample time to develop and refine the design.'  

Initially, the algorithm combined three performance-driven criteria, selected by the architects in response to the project brief: responsiveness to environmental conditions, maximised outside views, and functional area requirements, including a central public space to enhance social cohesion. The criteria were assigned relative values to produce optimal massing options which would provide year-round user comfort, protect the building from harsh weather conditions and create tailored workspace solutions for the different departments.

Once the final massing solution was selected by the architects, the cells were manually modified to improve the building’s structural, functional and aesthetic qualities. A cascading rooftop, controlled rooflights and winter gardens were introduced to enhance the quality of the space. Core supports were added to the four corners of the building to improve structural efficiency and meet fire safety requirements. The desire to enhance visual and physical relationships with the surrounding area also resulted in further modifications to the massing. In order to connect the inner public space with the urban realm, voxels were carved out to create efficient routes through the building.  

Following these manual adjustments, the design team re-analysed the model digitally, to test its environmental performance and make sure it was achieving the desired results.

The Alibaba Xuhui Binjiang Park Zone Y project demonstrates how Foster + Partners is using technology as a tool to radically push forward the way the practice designs, enhancing sustainability, functionality and form.