A concrete-framed office building with a curtain wall facade stores roughly the same thermal energy as 5,000 gallons of water. When you adjust the HVAC setpoint, you're not changing the temperature of the air in the room — you're attempting to change the temperature of the room's structural mass, which is connected to the air through convection and radiation, and responds on a timescale of hours, not minutes.
This is thermal inertia. Every building has it. And almost no BMS schedule accounts for it correctly.
What Thermal Inertia Actually Is
Thermal inertia — sometimes called thermal mass — refers to the resistance of a material to changes in temperature. High thermal mass materials (concrete, brick, masonry, stone) absorb and release heat slowly. Low thermal mass materials (glass, thin metal cladding, light-frame wood construction) respond much faster.
In a commercial building, thermal inertia manifests as lag: the delay between when you change an HVAC setpoint and when the space temperature reaches the new target. This lag is not a fixed number. It varies based on:
Outdoor temperature. On a 95°F summer day, the building envelope is absorbing heat from outside faster than HVAC can remove it. The effective cooling rate is slower — the thermal lag is longer. On a 68°F day, the building is near equilibrium with outside, and the same setpoint change takes significantly less time to achieve.
Solar gain. South- and west-facing zones with significant glazing receive direct solar radiation that adds thermal load irrespective of outdoor air temperature. A southwest corner office suite on a clear afternoon in July may have a cooling lag 40% longer than a north-facing interior zone in the same building.
Construction type. An older 1980s concrete-frame office building with deep floor plates and limited exterior glazing has very high thermal mass — it's slow to heat and slow to cool. A newer curtain-wall construction with 50% glazing ratio has lower mass but much higher solar gain sensitivity. A light-frame retail structure behaves almost differently again.
Zone size and geometry. Large open floor plates equilibrate differently than cellular office zones with many interior walls. The HVAC distribution system (VAV box positions, ductwork layout, diffuser placement) also affects how quickly conditioned air reaches the space temperature sensor versus the thermal mass.
Why Fixed Schedules Fail Under Variable Thermal Lag
A fixed BMS schedule says "start cooling at 6:00 AM." That instruction has an implicit assumption: 3 hours is enough to condition the building for an 8 AM occupancy start under all conditions.
On a 65°F November morning in Portland, 3 hours is more than enough — the building is already close to target temperature, the thermal lag is minimal, and the schedule is running 2 hours more HVAC than necessary. On a 92°F August morning with clear skies, 3 hours may not be enough — the southwest zones are fighting solar gain, the thermal mass is saturated from the previous day's heat, and the spaces that matter most to occupant comfort (perimeter offices, conference rooms with western exposure) are still 3–4°F above target at 8 AM.
The fixed schedule simultaneously over-conditions on mild days and under-conditions on the days when comfort matters most. And crucially, on hot days it is running full HVAC capacity in the 6–8 AM pre-peak window when utility rates are lower, then running continued full capacity into the peak tariff window when it can't quite achieve target temperatures anyway.
The Per-Zone Thermal Fingerprint
Every zone in a building has a distinct thermal response profile. When we start observing a new building, we collect zone temperature data at 5-minute intervals alongside outdoor temperature, solar irradiance proxy (cloud cover data from NOAA), and HVAC operation state. After 7 days of passive observation, we can construct a rough per-zone model that answers a specific question: given outdoor temperature X and starting zone temperature Y, how many minutes of HVAC operation are required to reach target temperature Z?
For a building we connected earlier this year — a 78,000 sq ft concrete-frame office in Portland's Lloyd District with a Siemens Desigo CC BMS — the thermal fingerprint for the southwest conference suite (Zone 11B, 4,200 sq ft, 60% glazed south/west exposure) looked roughly like this:
- Outdoor temp below 70°F: 45 min pre-cooling needed to reach 71°F from 76°F
- Outdoor temp 70–80°F: 90 min pre-cooling, same transition
- Outdoor temp above 85°F with afternoon cooling load: 140–160 min pre-cooling needed, and the zone will drift back up 1–2°F during peak hours without active cooling
The existing BMS schedule was running pre-cooling from 6 AM to 8 AM regardless of outdoor conditions. On mild days: 75 minutes of unnecessary conditioning. On very hot days: still undershooting by 30–60 minutes relative to the zone's actual thermal response requirement.
The Pre-Conditioning Calculation
Knowing the thermal fingerprint allows a specific, calculable optimization: given tomorrow's weather forecast and the peak tariff window start time, when is the latest point at which pre-conditioning should begin to achieve target temperature before both the occupancy start and the tariff peak window?
This is different from the question most BMS schedules implicitly answer ("when should HVAC start?"). The BMS schedule starts at a fixed time because that's the only parameter available. A thermal model starts at a calculated time because it knows how long the building actually needs.
On a day with a forecast high of 88°F and a peak tariff window starting at 3 PM, the optimal pre-cooling strategy for Zone 11B might look like: begin at 10:45 AM (not 6 AM), run at full capacity until 1:30 PM, then set the zone to 69°F (2 degrees below target) to use thermal mass as a buffer for the 3–6 PM period where HVAC runs at reduced load to avoid peak demand contribution.
That 69°F pre-cool target — often called a "thermal battery" or "pre-cool to storage temperature" strategy — is where the real demand charge reduction happens. The building is deliberately overcooled slightly before the peak window so it can coast on stored cooling during the expensive period. Occupants experience 70–72°F during peak hours without the chiller running at full capacity. The demand contribution during the peak window drops 40–60% compared to reactive cooling.
Building Types and Their Thermal Response Ranges
Not all buildings are equally amenable to thermal pre-conditioning. Here is a general framework for evaluating thermal lag in your buildings:
High inertia, best pre-conditioning candidates: Concrete-frame construction from 1960s–1990s, masonry exteriors, limited glazing ratio (below 30%), deep floor plates. These buildings hold pre-conditioning for 3–5 hours. The pre-cool thermal battery strategy is highly effective — you can cool aggressively 3–4 hours before peak and have the building absorb heat slowly enough to maintain comfort through a 2–3 hour peak window without active cooling. Portland has a significant stock of this building type in its commercial core.
Medium inertia, moderate candidates: Steel-frame with composite deck floors, 30–50% glazing, curtain wall construction from the 1990s–2000s. Thermal lag is 1.5–2.5 hours depending on orientation. Pre-conditioning works but the thermal storage period is shorter. Pre-cool scheduling 90–120 minutes before peak is effective; extended coasting is limited to 90 minutes or so without noticeable comfort drift.
Low inertia, limited candidates: Light-frame construction, high-glazing ratio (above 60%), very thin floor plates, large open atrium spaces. Thermal lag is short (30–60 minutes in some zones). Pre-conditioning can still reduce demand peaks but the window is narrow. More of the cooling work has to happen closer to occupancy and peak window, limiting the degree of demand avoidance.
What BMS Doesn't Know and Can't Know From Inside
The thermal fingerprint can be measured and modeled. The piece a BMS cannot supply from within the building is what it needs to know to act on that fingerprint: tomorrow's weather forecast, which determines how the thermal lag will behave tomorrow specifically.
We're not saying BMS platforms are deficient in design — they were built to control a building, not to consume external data feeds and run predictive models. The NOAA hourly weather API, utility tariff schedule data, and a dynamic thermal inertia model are simply outside the scope of what BMS platforms were designed to do. Adding a predictive layer on top of the BMS — one that issues setpoint commands based on the thermal model and external data — is an architectural complement, not a replacement.
The thermal fingerprint of your building exists whether or not you've measured it. The question is whether your control system knows what it is and uses it to decide when to start the HVAC — or whether it starts at 6 AM because that's what was entered during commissioning.
Measuring Your Building's Thermal Response
You can do a rough thermal lag measurement yourself without any additional sensors. Here is the procedure: on a weekend or holiday when the building is unoccupied, set a zone to full cooling from setback to target temperature (say, from 80°F to 70°F) and record how long it takes. Do this at two different outdoor temperatures — a mild day and a hot day — and note the difference.
The ratio between those two durations gives you a rough sense of the temperature sensitivity of that zone's thermal lag. A zone that takes 60 minutes on a 70°F day and 150 minutes on a 90°F day has a 2.5× thermal lag scaling factor with outdoor temperature — meaning the pre-conditioning start time needs to move roughly 90 minutes earlier for every 20°F increase in outdoor temperature. That is the kind of knowledge a fixed BMS schedule cannot incorporate but a dynamic control system uses on every cycle.