Energy Use Analysis
Build an energy balance, then explain it
Energy use analysis answers two questions: where does the energy go, and what drives it? Start with 12–36 months of utility data, build a monthly profile, and separate base load (schedule- and occupancy-driven) from weather-dependent load (heating and cooling). Then disaggregate end uses using sub-meters, run-time data, or engineering estimates, and present it as a Sankey-style balance that closes to the meter.
Key tools: EUI and benchmarking against ENERGY STAR or CBECS-type stock data; load factor (average kW ÷ peak kW) to reveal how peaky a facility is; change-point regression of consumption on heating and cooling degree days to obtain a slope (weather sensitivity) and intercept (base load); and normalization for occupancy, production, or floor area so year-to-year comparisons mean something.
Traps. Regressing on degree days when the load is really schedule-driven (a small slope with a huge intercept); comparing billed demand to an instantaneous interval peak; treating a single billing month as the annual peak; mixing site and source energy when comparing fuels; counting a submetered load twice in the balance; and forgetting that gas may serve process or DHW, not just space heating.
Rule of thumb
If the intercept explains most of the variation, fix operations first; if the slope is large, envelope and plant efficiency dominate.
Worked example. Monthly electricity regressed on cooling degree days gives an intercept of 180,000 kWh/month and a slope of 12,000 kWh per CDD. Annual CDD total is 1,500. Annual use ≈ 180,000 × 12 + 12,000 × 1,500 = 2,160,000 + 18,000,000 = 20,160,000 kWh, of which only about 11% is weather-independent base load - this is a cooling-dominated building where chiller plant work pays.