Real buildings. Demand charge data, not projections.
Commercial office, retail, and healthcare administrative buildings across Oregon. Results measured from 15-minute interval meter data against a 60-day pre-deployment baseline — the demand charge line item, not blended kWh. Synthetic company names; identities anonymous by request.
Northgate had a consistent pattern of peak demand spikes on Monday mornings between 8am and 10am. The BMS was configured with a 6am warm-up schedule set during commissioning in 2019 and never adjusted for seasonal occupancy variation. During summer months, the air handler staged up all zones simultaneously — creating a demand spike three times the building's midday load.
After a 7-day passive learning period, Celaxis identified the Monday spike pattern and implemented a staggered pre-conditioning schedule — pre-warming occupied zones (floors 1–4) starting Sunday at 11pm, then staging remaining zones with 45-minute offsets. Peak demand on the Monday morning interval dropped by 41% in the first measured week. Full-month demand charge reduction stabilized at 34%.
Retail properties with food service tenants have complex thermal adjacency: food court exhaust heat rejection affects adjacent non-food HVAC zones. The existing Siemens scheduling treated the building as uniform zones, causing overcooling in zones adjacent to food courts during lunch hours — the exact period when demand charges are highest.
Celaxis's thermal model identified the food court thermal signature and calibrated adjacent zone setpoints to account for expected heat contribution during service hours. Pre-cooling those zones by 1.5°F before the 11am lunch period eliminated the overcooling correction cycle that was creating the midday demand spike. Average demand reduction across the three properties was 27%, with the largest property at 31%.
Medical office buildings with clinic tenants have irregular occupancy — many clinics close at 3pm or have half-day Fridays. The existing schedule ran full HVAC through 6pm regardless. Floors 2–4 (outpatient clinic) were conditioning empty space for 3 hours every afternoon, including during PGE's demand peak window.
Celaxis's occupancy inference model detected the 3pm departure pattern across clinic floors within the first two weeks and began implementing graduated setback from 2:45pm. Friday half-day detection reduced Friday afternoon HVAC load by 52%. The approach was deliberately conservative — maintaining a 1.5°F buffer from setback limits to stay clearly within the medical office comfort standard required by the building's commercial leases.
How we measure results
All results are measured from 15-minute interval meter data. Baseline is the 60 days immediately before Celaxis deployment. Comparison is the 60 days following the first full optimization period. Savings percentages represent demand charge line items only — not total energy usage. We do not report blended kWh reduction because that number is less meaningful than the actual line item on your bill that Celaxis targets.
Interval meter data
15-minute interval readings from the building's utility meter or sub-metering system. No self-reported data.
60-day baseline
Baseline captures the same 60-day period prior to deployment to control for seasonal variation in occupancy and weather.
Demand charge line item
Savings measured against the demand charge line item on the utility bill, not blended kWh cost. That's where the lever is.