Engineering & Technology · Building Systems, Services & Environmental Performance
Building controls, monitoring and performance gaps
A new control system is followed by a lower electricity bill, but the building also had fewer occupied days. The change may be real while its cause remains uncertain. Building-performance evidence needs both a working physical sequence and a comparison that accounts for the service and conditions delivered.
Use heat transfer, civil systems, electrical quantities and the preceding instrumentation/control chapters. Elementary energy and material balances support the worked examples; all numerical building inputs are supplied teaching assumptions.
Back to the chapterAfter reading, you can
Connect control sequences, measurements and comparable baselines to a defensible building-performance claim.
An automation record needs physical meaning
A building controller receives measurements, applies a sequence and sends requests to equipment. The actual temperature, airflow or delivered heat can differ from the requested value. Sensor location, calibration, filtering and data age affect the interpretation. A communication acknowledgement establishes only the event defined by its contract; it does not prove a valve moved or a space received the intended service.
Schedules, occupancy modes and manual overrides can change how systems interact. Heating and cooling can compensate for each other, concealing a fault in an aggregate comfort reading while consuming additional energy. Keep the active mode, commands, relevant physical outputs and intervals on a compatible timeline so that an unexpected pattern can be investigated.
The baseline is part of the claim
A performance gap is a difference between expected and observed behaviour under a stated comparison. Weather, occupancy, operating hours, plug loads and required indoor conditions can change energy use. A design model’s assumptions and a measured building’s conditions may differ. Comparing two bill totals alone cannot assign the difference to one control change.
Use three original matched-condition teaching pairs: baseline 12, 14, 16 kWh and later 10, 12, 14 kWh. Each difference is 2 kWh. Mean baseline is 14 kWh and later mean 12 kWh; relative difference is 2/14≈14.29%. Matching is an explicit supplied assumption here, rather than something proved by the numbers.
A detected anomaly still needs investigation
A model can help identify an unusual load profile or a persistent deviation from expected performance. Its threshold, input quality and operating range matter. A flagged pattern is not necessarily a uniquely diagnosed fault. Missing data and values estimated by a model must stay distinguishable from actual observations, especially when the same model is used to judge a later change.
Commissioning and operational verification connect intended requirements to documented installed behaviour. They include relevant modes, interfaces, records and responsibilities rather than merely a successful software connection. After a correction, retain the original evidence, change record and follow-up observations. Maintenance staff need an actionable explanation and ownership, because a dashboard alone does not repair a physical cause.
Compare a matched account with an unmatched bill
The supplied pairs total 42 kWh baseline and 36 kWh later, a difference of 6 kWh over the three cases. Their relative difference 6/42 equals the mean comparison 2/14. The arithmetic is clear, but three constructed pairs do not establish uncertainty, seasonal persistence or independent causation. Another unrecorded change could still explain an actual observational comparison.
An unmatched example of 100 kWh predicted versus 130 kWh measured likewise does not establish 30 kWh of waste. First examine the required service, weather, occupancy, model scope and meter boundary. A separate sensor check can matter too: a display of 21 °C with an independent reference of 23 °C would challenge the interpretation of the loop before any energy-control adjustment is evaluated.
Try the reasoning
Why adjust a baseline for changed weather or occupancy?
Show the answer and reasoning
Those conditions can change the required service and energy use independently of the control modification.
Do the three paired differences prove seasonal savings in an actual building?
Show the answer and reasoning
No. The cases and matching are supplied illustrations; uncertainty, confounding changes and persistence still need evidence.
Should reconstructed meter values be presented as measured data?
Show the answer and reasoning
No. Preserve their estimated status and the model used to produce them.
What to carry forward
Link control changes to observed physical service and a comparable, traceable energy baseline.
References
Teaching resources and research supporting this lesson’s explanations and examples.