Practical analysis on demand charge mechanics, forecast accuracy benchmarks, building envelope physics, BACnet/Modbus integration, and HVAC staging schedules. Written by Tobias Schulz and the Heatvelo team — for facilities managers and energy engineers who want to understand their buildings, not just receive recommendations about them.
Forecasting
Static setback schedules were designed for predictable occupancy. When real-world building use diverges from the schedule — as it always does — HVAC systems overcycle at exactly the wrong tariff moment.
Energy Economics
Most utility bills bury demand charges in technical line items. We break down exactly how the 15-minute peak interval gets captured and why a single HVAC spike can inflate your bill for the entire billing month.
Building Science
Every commercial building contains tonnes of concrete, steel, and glass that absorb and release heat over hours. A good forecast lets you use thermal mass as a free demand-response buffer.
Data & Integration
Most energy models treat occupancy as a fixed schedule. Real-world occupancy varies by 40–60% from nominal. We look at how occupancy sensor feeds transform forecast accuracy.
Case Studies
Buildings in the Upper Midwest face extreme outdoor temperature swings that compress the pre-cooling window and amplify the consequence of a missed forecast.
Data & Integration
Most commercial building automation systems already speak BACnet or Modbus. Pushing a staging schedule from an external forecast engine requires only a network-accessible endpoint.
Energy Economics
Commercial utility rate schedules are dense documents, but the part that matters most to HVAC operators fits on one page: the time-of-use window structure.
Energy Economics
Utility demand response programs offer real bill credits, but they require your building to curtail load within minutes of a dispatch signal. Thermal forecasting lets you pre-position HVAC before the signal arrives.
Forecasting
Mean Absolute Percentage Error is the standard benchmark for load forecasting accuracy. We explain what MAPE < 8% on a 24-hour horizon actually means for your peak demand charge exposure.
Use Cases
Retail buildings have one advantage over offices: foot-traffic data is inherently available from POS systems and entry counters. Integrating transaction volume as an occupancy proxy dramatically tightens the forecast.
Building Science
Solar gain through glazing is the largest uncontrolled heat input in most commercial buildings. A building with 50% WWR in Minneapolis faces radically different cooling demand dynamics than one with 30% WWR.
Data & Integration
A forecast engine can push its output as a discrete staging schedule or as a continuous setpoint offset stream. The right choice depends on how your BMS controller handles external overrides.
Case Studies
After our first full pilot cohort, we share the honest numbers: peak demand reduction range, forecast accuracy across building types, and the one variable that consistently surprised us in cold-climate pilots.
