The strongest arguments are built on evidence, structured by analysis, and sharpened by design. My research bridges data science and policy — turning spatial analysis and predictive modeling into actionable recommendations for a more circular, equitable built environment.
Thesis · 2025–2026
A year-long research project investigating how circular economy principles could reduce embodied carbon in New York City's construction sector — combining quantitative modeling, policy analysis, stakeholder interviews, and design communication.
Construction is responsible for a massive share of global carbon emissions — not just from building operations, but from the embodied carbon locked into materials themselves. Concrete, steel, and glass carry enormous carbon footprints before a building ever opens its doors. Yet most climate policy focuses on operational energy, leaving embodied carbon largely unaddressed.
NYC, with its relentless pace of construction, is both a major contributor to this problem and a potential laboratory for solutions. What if the city adopted circular economy principles — reusing materials, substituting lower-carbon alternatives, diverting construction waste from landfills?
The research combined three analytical streams: quantitative modeling of embodied carbon across building types using OLS regression and Random Forest; scenario analysis projecting cumulative carbon savings under four policy pathways over 20 years; and qualitative interviews with construction industry practitioners and policymakers to ground the models in real-world feasibility.
The data pipeline processed building-level records from NYC DOB permits, material specifications from EPDs (Environmental Product Declarations), and spatial data from city planning datasets — all cleaned, joined, and analyzed in Python using pandas, GeoPandas, scikit-learn, and statsmodels.
The modeling revealed significant variation in embodied carbon intensity across building types, structural systems, and material choices. Click any chart to expand and read the findings.
Embodied carbon varies dramatically by material. Steel and concrete dominate emissions, but the specific mix matters — low-carbon concrete substitutes can reduce material-level emissions by 30-40% without compromising structural performance.
Random Forest analysis reveals which building characteristics most influence embodied carbon intensity. Floor area and structural system dominate, suggesting that design-phase decisions have the greatest impact on lifecycle emissions.
Structural system choice is one of the strongest predictors of total embodied carbon. Steel-frame buildings consistently show higher ECI than concrete or hybrid systems, moderated by building height and floor area.
Small changes in cement content produce outsized carbon reductions. Blends using fly ash, slag, and supplementary cementitious materials represent the single highest-leverage intervention identified in the analysis.
Embodied carbon intensity increases non-linearly with building height. Taller buildings require more structural material per square foot, but the relationship plateaus above ~30 stories.
OLS regression coefficients quantify the marginal effect of each building characteristic on embodied carbon. The model identifies which variables have statistically significant relationships with carbon intensity, controlling for all other factors.
The analysis modeled four policy scenarios — from procurement reform to subsidy programs to hybrid regulation — projecting their impact on market transformation and cumulative carbon savings over 20 years.
Current policy (EO23) covers only a fraction of NYC's construction emissions. The vast majority — 32.8 million metric tons from private sector construction — falls outside existing regulation. Closing this gap requires extending embodied carbon requirements beyond public projects.
The hybrid approach — combining subsidies with regulation — achieves 40% market adoption of circular practices within 20 years, compared to just 10% under baseline conditions. Subsidies alone reach 25%, while procurement reform reaches 15%.
The hybrid scenario could save 57 million tCO₂e over 20 years — nearly 4x the baseline trajectory. Even the more modest procurement-only approach yields 19 million tCO₂e in savings, demonstrating that any policy action significantly outperforms inaction.
The ~20% viability threshold — where circular practices become self-sustaining — is reached within 5 years under the hybrid scenario but never under baseline conditions. This suggests that early policy intervention can create lasting market transformation, while delay locks in decades of avoidable emissions.
Circular economy principles are not just aspirational — they are quantifiable, modelable, and policy-ready. By connecting material science, spatial analysis, and policy design, this research offers a framework for cities to address embodied carbon with the same rigor they apply to operational energy.
This research draws on every practice: predictive models quantify impact, maps reveal spatial patterns, and sensing grounds it in lived experience.