Hello OEM Community,
I’ve been developing a Building Energy Management System (BEMS) within Home Assistant to monitor the thermal performance of a 2,440 sq ft residence in Connecticut (Zone 5).
While most HA setups stop at tracking runtime, I’m trying to quantify the “Digital Twin” of the building envelope—specifically tracking UA (Heat Loss Coefficient) and EUI daily to identify performance drift and wind infiltration penalties.
The Methodology:
-
HDD Normalization: I’m calculating daily HDD (Base 65) using 10-minute sampling to ensure high-fidelity weather data.
-
Overcoming the Setback Proxy Error: One issue I encountered was the inaccuracy of using a 2-point proxy (Start/End temps) for HDD during overnight setbacks. I have moved to a 15-minute Riemann Sum integration during “the coast” to compute a true integrated outdoor average. This has significantly smoothed my UA estimates during recovery periods.
-
Efficiency vs. Norms: I cross-reference daily performance against an 18-year historical climate dataset (P10/P90) using a Python-driven command-line sensor. This helps differentiate between a “cold day” and a “system efficiency drop.”
Data Integrity & Persistence:
Because I don’t trust the standard HA Recorder for long-term building science, I’ve implemented a shell-command pipeline that exports daily, monthly, and yearly summaries to rotated CSVs. This gives me a permanent, portable record for regression analysis in Excel/Python outside of the HA ecosystem.
System Specs:
-
Heating: 60k BTU/hr Gas Furnace (2-zone).
-
Metrics Tracked: UA (Btu/hr-°F), Cycle Efficiency Ratio, Design-Day Utilization, and Recovery Rate vs. Setback HDD.
Seeking Peer Review on:
-
Thermal Decay Rate (D_s): I am planning to implement a “Passive Decay” sensor to measure the slope of temperature drop during idle periods. Has anyone here had success correlating this with wind speed to calculate a “Wind Leakage Index”?
-
Dynamic Baselines: My current UA baseline is static. How are others accounting for the “Slab Effect” or ground-coupling changes as the deep soil temperature shifts throughout the winter?
-
Sensor Placement: I’m considering adding ESP32-S3 based probes for a true Delta-T (Supply/Return) calculation. Is it worth the complexity over my current runtime-based BTU estimation?
I’ve documented my YAML logic and the underlying building physics in my repo here: [Link to your GitHub]
Looking forward to any critiques on the math or the sampling methodology.