Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture

Save
Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture

Information

  • Project Name: Flow
  • Student Name: Alex Li
  • Awards: Outstanding Design Award (Best in Class) | BDP Quadrangle Excellence in Digital Fabrication Award
  • Softwares/Plugins: Rhinoceros 3D , Grasshopper , Visual Studio Code , LM Studio
More Info Less Info

Excerpt: Flow’ is a Bachelors Design Project by Alex Li from the ‘School of Architecture – University of Waterloo.’ The project aims to develop robust yet flexible design frameworks that enable designers to generate and fabricate kinetic geometries across multiple scales. By connecting digital simulation with material-driven fabrication, the project emphasizes adaptability, performance, and designer autonomy, allowing the systems to be applied to a wide range of design contexts.

Introduction: How can AI be leveraged to rapidly produce functional geometry? 

FL◇W is a project that utilizes the computational strength of AI to create advanced geometric systems. At its core is a robust yet flexible framework that enables generation, iteration, validation, and physical testing in a continuous loop. The designer acts as a validation checkpoint, maintaining full agency and control.

The project consists of two complex forms: a Bistable Auxetic Surface, and a Kinetic Transformation Frame. Each can function independently or combine to create semi-autonomous multi-state geometries, scalable from tabletop objects to architectural form. 

AI democratizes the design process by augmenting human creativity, enabling designers of all skill levels to rapidly produce advanced, functional geometry, which historically demanded high levels of expertise, skill, and time.

Save

Site Context

FLOW is a project that aims to provide robust yet flexible frameworks for designers to generate kinetic geometries meant to be applied to a wide range of scales and uses. The two geometries demonstrated in the project are a bistable auxetic surface and a kinetic geometry frame. Both geometries may act independently of one another or as a combined system. On a wide scale, the combination may form an architectural canopy. On small scales, the geometries have high potential to be applied in performance materials. The emphasis on framework allows for the designer to have autonomy over the outputs and apply it to any design scenario.

Design Process

Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Step 1 | Step 2
Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Step 3 | Step 4
Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Bistable Auxetic Animation

The design process centers around an iterative open-system loop between the designer and a large language model. Through different AI applications such as RAG analysis, generative code, and image generation, design iterations can occur rapidly. Each output is evaluated through specific criteria within a matrix and scored in comparison to each iteration to determine optimal solutions. 

Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Kinetic Frame Assembly
Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Digital Deformation Simulation
Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Workflow

Prompt engineering becomes critical to the project as the designer learns to speak the AI’s language. The kinetic designs are first digitally simulated in Grasshopper and measured as pass/fail before being finalized through digital fabrication. This ensures that designs that do make it to the fabrication phase have a high chance of success.

Final Outcome

Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Final Diagram
Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Bistable Digital Animation
Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Architectural Example

The final outcome is to translate digital simulation into fabrication. The bistable auxetic pattern demonstrated optimal performance when 3D printed in TPU, while the kinetic frame achieved the best results through 3D printing that embedded and fused TPU joints within PLA members.

Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Combined Deformation 1 | Combined Deformation 2

Although the geometries can also be produced using laser cutting and other precision-cutting techniques, material selection becomes critical. Repeated deformation demands resilient materials with minimal structural weaknesses, which are more difficult to maintain with subtractive methods, as cutting exposes vulnerabilities more readily than additive manufacturing.

Flow: Artificial Intelligence as a Framework for Generating Functional Geometry | Bachelor’s Design Project on Computational Architecture
Combined Geometries

Conclusion: Ultimately, FLOW demonstrates how flexible design frameworks can translate digital simulations into effective kinetic systems, revealing the potential of adaptable geometries across scales, materials, and applications.

[This Academic Project has been published with text and images submitted by the student]

​​To submit your academic project for publication at ArchiDiaries, please visit the following link >> Submit

Leave a Reply