Inspiration
Extreme heat is becoming a serious challenge for cities, affecting indoor comfort, energy use, and the daily lives of residents. Traditional approaches to reducing urban heat may treat large areas uniformly, even though different buildings have different roof materials, structures, and cooling needs.
We created RoofRx to explore a smarter approach: using data and AI-assisted optimization to identify which rooftops could be prioritized for cool-roof interventions when resources are limited.
What it does
RoofRx is a climate-tech prototype that models roof-level cooling benefits and helps rank buildings for potential cool-roof coating interventions.
The project includes:
- Thermal modelling: Estimates potential indoor temperature changes under modelled roof-coating scenarios.
- Weather data handling: Supports weather inputs and a synthetic-data fallback for demonstrations.
- Budget-aware optimization: Ranks candidate roofs to explore how a limited intervention budget could be allocated.
- Strategy comparison: Compares roof-level prioritization with ward-level and random-selection baselines.
- Interactive dashboard: Uses Streamlit to present results and make the prototype easier to explore.
The current demonstration uses synthetic data. Its results are modelled estimates, not verified real-world cooling measurements.
How we built it
We built RoofRx using Python and a modular architecture that separates thermal modelling, weather handling, building data generation, optimization, and the application interface.
NumPy and Pandas support numerical calculations and data processing, while SciPy supports optimization-related functionality. Matplotlib is used for visualizations, and Streamlit provides the interactive dashboard.
We also created automated tests to check important model and optimization behaviours. The prototype is designed to support further development with real building and weather datasets.
Challenges we ran into
One of our main challenges was modelling a complex real-world problem with limited validated data. Building-level thermal performance depends on several factors, including roof material, building structure, weather, and other environmental conditions.
Another challenge was ensuring that budget allocation remained consistent while comparing different intervention strategies. We also had to distinguish between synthetic demonstration results and conclusions that would require real-world validation.
These challenges helped us recognize the importance of transparent assumptions, reproducible experiments, and careful testing.
Accomplishments that we're proud of
We developed a modular prototype that connects thermal modelling, building data, budget-aware optimization, and visual comparison in one project.
We are particularly proud of exploring how targeted roof-level interventions could be compared with broader allocation strategies under a constrained budget.
The project also provides a foundation for future work on real building datasets, more realistic thermal models, and practical climate adaptation tools.
What we learned
Through RoofRx, we learned how to connect data processing, modelling, optimization, and an interactive interface into a single climate-tech application.
We also learned that AI-assisted recommendations must be supported by reliable data, transparent assumptions, and validation. A promising simulation is a starting point for investigation, not proof of real-world impact.
What's next for RoofRx
Our next steps are to validate the thermal model against measured building temperatures, integrate reliable local weather and building data, and evaluate the prototype across different climate conditions.
We also plan to improve uncertainty estimates, strengthen comparisons between intervention strategies, and make the dashboard more useful for urban planners, sustainability teams, and communities exploring heat-mitigation solutions.
Our long-term goal is to help decision-makers investigate where cool-roof interventions may provide the greatest benefit while making effective use of limited resources.
Built With
- ai
- climate
- coolroofs
- data
- involver-engagement-optimization
- matplotlib
- numpy
- pandas
- python
- science
- scipy
- streamlit
- tech
Log in or sign up for Devpost to join the conversation.