This repository contains the dissertation project of Abhishek Palit submitted in partial fulfillment of the requirements for the Master of Architecture at Indian Institute of Technology Roorkee.
The dissertation explores Generative Design techniques to optimize the spatial layout of urban neighborhoods. Using computational methods, the study aims to develop an algorithm that considers multiple urban design goals such as profit, open spaces, proximity, and shading to generate and evaluate design solutions. This algorithm has been applied to the Salt Lake City, Kolkata neighborhood as a case study.
The growing global population and urbanization have placed immense pressure on cities, making it more challenging for planners and architects to create sustainable, livable environments. With increasing demand for land, cities are expanding their boundaries and developing new neighborhoods that require optimized layouts for the well-being of their residents.
Urban design is a multi-faceted process involving several stakeholders, goals, and constraints, often making it difficult for planners and architects to achieve a balance. This project aims to address these challenges through a Generative Algorithm that outputs multiple scenario-based designs, helping stakeholders make informed decisions.
This dissertation focuses on creating a Generative Design Algorithm for two neighborhoods—Block BD and Block CD—in Salt Lake City, Kolkata, driven by multiple design goals. The algorithm explores layouts that optimize the following key metrics:
- 🏢 Profit – Maximize the number of plots while minimizing roads.
- 🌳 Urban Quality – Maximize open spaces.
- 🚶 Proximity – Minimize distance to schools and open spaces.
- 🏖️ Urban Comfort – Maximize shading in outdoor spaces for comfort.
-
Input Parameters:
- Goals specified by stakeholders.
- Environmental and spatial constraints.
-
Generative Design Output:
- Multiple design solutions that align with the specified goals.
-
Performance Evaluation:
- A normalized metric to compare the solutions.
- Selection of the top 10 high-performing layouts based on their scores.
-
Comparison with Existing Design:
- Evaluation of the two best solutions against the current neighborhood layout.
- The results show that the Generative Algorithm's designs outperform the existing layout in various metrics.
- The algorithm generated optimal designs that balance profit, urban quality, proximity, and comfort.
- The top solutions provide better outcomes than the existing neighborhood design in Salt Lake City, demonstrating the value of generative algorithms in urban planning.
- Optimization Study: The generative design algorithm generated multiple solutions based on specified goals such as maximizing plots, minimizing road areas, and ensuring proximity to amenities.
Below are some key slides from the final dissertation presentation, summarizing the study and its outcomes:
