Site Optimization
2022 - Efficiency Study
A computational optimization study commissioned by MOOKAMBOO, on behalf of a real estate developer in Northeast Brazil. The project's starting point was a financial viability model: given the developer's target return on investment, cost per square meter, and available land, I built an algorithm to determine the minimum square meters required to achieve the projected profit, working backwards from economics to built area to site occupancy.
The site comprised a total area of approximately 330,000 square meters, divided across multiple plots facing a beachfront. The constructive unit was a 15x15m module arranged in a grid pattern, representing the buildable footprint for each unit.
I developed the optimization in two stages. In the first, I applied a dedicated evolutionary algorithm independently to each plot, iterating over the rotational angle of the module grid to find the orientation that maximized land occupancy in 2D, accounting for the irregular geometry of each individual terrain.
In the second stage, I extended the optimized 2D layouts into the third dimension. Each positioned module could receive between one and four floors, and I ran a second evolutionary algorithm to determine the optimal vertical distribution across the site. The governing rule was that units deeper into the plot could receive more floors, preserving unobstructed beachfront views for all units, balancing density with livability.
The result was a fully generative site layout that pushed occupancy to its mathematical limit. The combined optimization I developed reduced the gap between available land area and the developer's target built area to less than 1%, an outcome that would be practically unachievable through conventional manual planning at this scale.