Energy Analysis and Modeling
Simulation fundamentals
Whole-building energy models solve hourly heat and mass balances. Know the inputs that most affect results: weather file (TMY3 vs. actual year), infiltration/ventilation schedules, internal gain schedules, HVAC part-load curves, and envelope thermal bridging. Calibration compares modeled to metered data using NMBE (normalized mean bias error) and CVRMSE (coefficient of variation of RMSE); ASHRAE Guideline 14 gives typical thresholds (roughly ±5% NMBE and 15% CVRMSE for monthly data).
Parametric and uncertainty work
Understand baseline vs. proposed modeling, parametric runs, sensitivity analysis, and life-cycle cost analysis (LCCA) including discount rate and escalation. Recognize degree-day methods as simplified screening tools, not substitutes for hourly simulation.
Traps
- Comparing models built on different weather files or vintages.
- Forgetting thermal bridging and fan/pump energy in savings claims.
- Using CVRMSE alone without NMBE; both matter.
- Treating a calibrated model as predictive truth - calibration can mask compensating errors.
- Mixing energy and demand savings in one percentage.
Worked example
Monthly metered use averages 80,000 kWh; model predicts 84,000 kWh. NMBE = (Σ(model−metered)/Σmetered)/n × 100. Sum difference = 4,000 kWh over, say, 12 months; Σmetered = 960,000. NMBE ≈ (4,000/960,000)/12 × 100 ≈ 0.035% - well within tolerance. A large NMBE signals bias; a large CVRMSE signals scatter. Report both.