Infrared - AI-Powered Environmental Simulations
Background

Case studies

Explore real-world examples of how organizations are using infrared.city to make data-driven climate decisions

author
Oana
 · 5 min read

SvN Architects + Planners: Optimizing Urban Wind Comfort through AI-Driven Design

Overview

Firm: SvN Architects + Planners

Project: Wind Comfort Optimization of Mixed use Towers scheme

Location: Mississauga, Ontario, Canada

Completed: 2025

image Fig.1: Optimization process studied position, proportion and corner chamfer variables in relation to distribution of potentially uncomfortable wind zones image Fig.2: Optimization results showing total reduction in “uncomfortable wind”(above 5m/s) of 11.6% and a reduction of “very uncomfortable wind”(above 8m/s) of 9.6% of a designated focus zone marked in red. The image shows the importance of positioning, alignment and spacing of tall buildings as well as the added impact of the building footprint proportions.

The Challenge

Urban wind comfort significantly impacts how people experience outdoor spaces. As cities grow taller and denser, the wind tunneling effects between buildings can create uncomfortable or even dangerous conditions for pedestrians. This is particularly sensitive in cold climate locations where wind acceleration exacerbates cold stress, reducing comfort throughout seasons. Traditional computational fluid dynamics (CFD) simulations, while accurate, are computationally intensive and time-consuming, limiting the application to high budget projects, and further limiting the number of design iterations architects can feasibly test.

Once the low cost, high impact early stage decisions are set, design teams are left with costly and aesthetic altering mitigation solutions in order to achieve necessary comfort and safety criteria.

SvN Architects + Planners needed a solution that would allow them to:

  • Rapidly evaluate multiple building configurations
  • Optimize building massing for pedestrian wind comfort
  • Work within their existing design strategies built on extensive project experience
  • Generate data-backed design options to support decisions for stakeholders

The Solution

Looking to capture their extensive project experience and knowhow into a repeatable early stage design testing process, SvN’s computational design team gathered recurrent design strategies into a parametric algorithm capable of iterating through a large design option space given a starting condition. The team wanted to implement a genetic optimization workflow powered with infrared.city’s AI models and real-time simulation to find optimal building forms in relation to year round wind comfort.

Key Implementation Features

  • AI-Accelerated Analysis: infrared.city’s AI model predicted simulation results at speeds 100-1000x faster than traditional CFD
  • Seamless Integration: Direct connections to both Rhino and Revit design platforms enables real-time data transfer and simulation runs directly from the team’s design environment
  • Genetic Algorithm Approach: Utilizing our RawData transfer feature, the team can access detailed numerical results through Grasshopper. These are used to define key performance indicators and support the optimal solution search.
  • Targeted Solution Space: The team has run extensive testing to identify the key design parameters most critical to pedestrian comfort

image Fig.3: Base test conditions indicating the overall design scheme, the position and height of the geometries this study is focussed on, and the predominant wind direction, extracted from the local wind rose and considered for the study.

Design Parameters Tested

For this trial, SvN specifically optimized tower positioning, rotation (orthogonal only), depth, and chamfering in relation to the most challenging westerly winter wind (270° at 15 m/s) for a site in Mississauga, Canada.

The team developed a performance metric (fitness criteria) by measuring percentage of site affected by uncomfortable or dangerous wind speeds, under normal wind conditions. Through a weighted fitness formula, areas experiencing winds above 5 m/s (uncomfortable) and above 8 m/s (very uncomfortable), were quantified and minimized during the optimization process.

image _Fig.4: Design parameters. Setting up an optimization process requires the definition of a design space. This represents the totality of “acceptable” designs, and is defined through design parameters, their ranges and step size. _

The Outcome

The optimization generated significant improvements:

  • Reduced uncomfortable wind speeds by up to 11.9% in critical focus areas
  • Decreased dangerous wind speeds by up to 9.7%
  • Maintained consistent design criteria while optimizing for wind performance
  • Validated the results of the optimal solution by benchmarking against other simulation solutions
  • Validated the improvement on Annual Pedestrian Wind Comfort achieved by the Optimal solution against the same analysis on the Benchmark solution

The genetic algorithm iterated through multiple tower configurations to arrive at optimal solutions that balanced building form considerations, established both through strict rules in the parametric design strategy, and as elements of the Fitness Criteria, with maximizing pedestrian comfort. image Fig.5: Parameter selection strategy. The team decided to run a multi-step optimization process, with critical parameters being explored first, and detail parameters explored in a second step. This ensured faster and more targeted optimal results.

Design Insights

The study revealed critical insights:

  • Tower depth had significant impacts on wind performance

  • Narrower towers (18-19m) performed better than wider options

  • Tower positioning with regards to corners and other massing on the site, and the distance between multiple towers have dramatic impact

  • Solution space needs to be carefully controlled to reduce computational time for unfeasible solutions, and ensure quick convergence to optimal results

  • Testing specific parameters independently was an effective strategy to understand causal relationship. The goal should be to optimize only necessary variables simultaneously

Beyond This Project

SvN intends to expand this approach across their design practice:

  1. Early-stage design optimization: Using the tool to rapidly test massing scenarios before detailed design
  2. Multi-criteria optimization: Expanding to include thermal comfort, solar radiation, and daylight analysis
  3. Environmental process standardization: Through application on project types and location, create and implement a cross practice environmental design standard starting from understanding the local climate, setting coherent sustainability goals and iterating to achieve desirable KPI’s for each project
  4. Client communication: Leveraging data visualization to communicate performance benefits to stakeholders. Explore the automated report generation tools as a way to deliver performant design options to client and stakeholders in an interactive, accessible way.
  5. Design research: Contributing to broader understanding of building performance in relation to urban microclimates by continuing to run studies and inform further design work

image

Fig.6: Base test axonometric view with initial simulation results

In Their Words

This workflow is envisioned as a tailored genetic algorithm that speeds up the ideation and assessment of designs based on a narrow solution space. The smaller the solution space, the more likely we are to hone in on a solution that contributes to good design and comfortable public spaces.

  • SvN Architects + Planners

Technologies Used

  • infrared.city: AI-powered wind speed simulation
  • infrared.city: AI-powered Annual Pedestrian Wind Comfort simulation using General Lawsons Criteria
  • Rhino/Grasshopper: Parametric 3D modeling and genetic solver plugin
  • External Wind Speed & Wind Comfort simulation for comparison

Images courtesy of SvN Architects + Planners and Infrared City

  • Case study

Recent Articles