Research / R.01

HVAC Rightsizing

2025–26

A data pipeline estimating peak heating and cooling loads for roughly 130 million U.S. residential building records, built by aligning property data with the ResStock housing-stock energy model from NREL.

Role
Dataset review, harmonization, stratification
Stack
Databricks · SQL · Python
Scale
~130M records · EnergyPlus-derived loads
HVAC rightsizing load-transfer pipeline Property records and the ResStock energy model feed into harmonization, then stratification, then load transfer, producing peak heating and cooling loads per property. A dashed imputation step feeds into load transfer to fill missing attributes. PROPERTY RECORDS ~130M U.S. RESIDENTIAL RESSTOCK ENERGY MODEL ENERGYPLUS PEAK LOADS HARMONIZATION SQL / DATABRICKS STRATIFICATION COVERAGE-VERIFIED STRATA LOAD TRANSFER MATCH + PYTHON MODEL PEAK LOADS (kW) PER PROPERTY RECORD IMPUTATION CONDITIONAL SAMPLING
Fig. 01 — Load-transfer pipeline. Dashed: fallback path for missing attributes.
01 / Method
Harmonization

Variables overlapping between the two datasets are identified and inventoried. Units are standardized and categorical values are harmonized to ensure structural consistency across the database.

Stratification

A key stratification scheme is defined for downstream estimation. Sample coverage is verified to ensure each stratum has a sufficient number of ResStock samples.

Load Transfer

Multiple load transfer methods are evaluated, utilizing a hybrid approach. Records are first matched within key strata, then a predictive Python model is deployed within each stratum to improve accuracy and prevent unrealistic extrapolation. Missing information is addressed using the laboratory's conditional distributions, sampled from each property's observed characteristics.

02 / Contribution
My Role

I reviewed the datasets, harmonized the two sets and stratified the combined set, scripted in SQL and Python on Databricks with AI-assisted development.

Next Research AEM Surveying for Water Security →