Fault Detection and Diagnostics (FDD) Services for Building Systems
Fault Detection and Diagnostics services apply automated analytical methods to building system data streams to identify equipment malfunctions, degraded performance, and control sequence errors — often before occupants or operators notice any symptom. This page covers the definition and scope of FDD as applied to commercial and institutional buildings, the technical mechanism by which faults are detected and classified, the operational scenarios where FDD delivers measurable impact, and the decision boundaries that determine when FDD is appropriate and which approach fits a given facility. Understanding FDD is increasingly central to any building energy management technology services strategy, given the direct link between undetected faults and energy waste.
Definition and scope
Fault Detection and Diagnostics is a class of analytical services that continuously or periodically evaluate building system performance data against expected operating conditions, rule sets, or learned baselines to flag deviations (detection) and identify probable root causes (diagnostics). The distinction matters: detection alone tells an operator that something is wrong; diagnostics narrows the cause to a specific component, sequence, or configuration error.
The scope of FDD in buildings typically spans HVAC systems (air handling units, chillers, boilers, variable air volume boxes), building automation control loops, electrical distribution, and — increasingly — integrated systems monitored through IoT integration services for smart buildings. The U.S. Department of Energy's Building Technologies Office has published FDD as a priority technology under its Commercial Buildings Initiative, citing studies where simultaneous heating and cooling faults alone account for 5–20% of HVAC energy waste in affected systems (DOE Building Technologies Office, "Fault Detection and Diagnostics").
ASHRAE Guideline 36-2021, High-Performance Sequences of Operation for HVAC Systems, establishes standardized control sequences that serve as the normative baseline against which many FDD engines evaluate performance, making it a foundational reference document for scoping FDD services (ASHRAE Guideline 36-2021).
How it works
FDD services operate through a structured analytical pipeline. The following breakdown describes the five primary phases common to commercially deployed FDD platforms:
-
Data acquisition — Sensor readings, meter pulses, and controller outputs are collected from the building automation system or directly from field devices. Data quality checks flag missing points, stuck sensors, and out-of-range values before analysis begins.
-
Normalization and contextualization — Raw data is aligned to a common time base and enriched with contextual variables (outdoor air temperature, occupancy schedule, equipment mode) that govern expected operating ranges.
-
Fault detection — Analytical rules, physics-based models, or statistical algorithms compare actual performance against expected performance. Three detection approaches are recognized in the literature: rule-based (threshold and expert rules), model-based (first-principles energy or mass balance), and data-driven (machine learning trained on historical operation). Each carries distinct trade-offs detailed in the decision boundaries section below.
-
Diagnostics — When a fault is detected, diagnostic logic isolates probable causes by evaluating correlated signals. For example, a simultaneous heating and cooling fault in an air handling unit triggers diagnostics that examine economizer damper position, heating coil valve command, and cooling coil valve command together.
-
Prioritization and work order generation — Detected faults are ranked by estimated energy impact, comfort risk, or equipment damage potential and pushed to a predictive maintenance technology services workflow or a CMMS platform for technician dispatch.
NIST has contributed foundational metrological work on sensor performance standards that underpin step one; NIST SP 1169 documents performance metrics specifically for HVAC FDD tools (NIST SP 1169).
Common scenarios
FDD services address a defined set of high-frequency failure patterns in commercial buildings. The four most operationally significant scenarios are:
-
Simultaneous heating and cooling — Heating and cooling coils active in the same air stream, typically caused by a failed economizer sequence or misconfigured control logic. DOE-cited studies place this fault among the top contributors to HVAC energy waste.
-
Air handling unit economizer faults — Damper stuck open, stuck closed, or failing to modulate. A stuck-closed economizer prevents free cooling, increasing mechanical cooling load. A stuck-open damper in humid climates imposes uncontrolled latent load. Title 24 in California and ASHRAE Standard 90.1 both mandate economizer function, making these faults compliance-relevant as well as operational (ASHRAE 90.1-2022).
-
Chiller and boiler plant inefficiency — Staging sequences that run excess capacity, condenser water temperature setpoints that are not reset with ambient conditions, or failed isolation valves that allow bypass flow all produce measurable delta between actual and optimal plant efficiency. These scenarios are prime candidates for integration with smart building data analytics services.
-
Variable air volume (VAV) box faults — Actuator failures, sensor offsets, and controls that fail to achieve airflow setpoints generate both comfort complaints and energy penalties. ASHRAE Guideline 36-2021 defines the expected control sequences for VAV boxes, providing the normative baseline FDD rules reference.
Decision boundaries
Selecting an FDD approach requires matching the analytical method to facility type, data availability, and operational capacity. Three method classes define the primary decision space:
| Method | Data requirement | Interpretability | Commissioning effort |
|---|---|---|---|
| Rule-based | Low (BAS points only) | High | Low |
| Model-based | Medium (physics inputs) | High | Medium |
| Data-driven (ML) | High (historical data) | Lower | High |
Rule-based FDD is appropriate for facilities with limited historical data or those undergoing initial smart building commissioning services — the rules encode expert knowledge directly and require no training period. Model-based FDD fits facilities where physics-based energy models already exist, such as buildings with a validated digital twin. Data-driven approaches require 12–24 months of clean operational data to train reliably and are better suited to large portfolios where patterns generalize across buildings.
A second boundary separates continuous FDD from periodic FDD. Continuous FDD ingests streaming BAS data and surfaces faults in near-real-time, typically integrated with remote monitoring and management services. Periodic FDD runs batch analysis on logged data at defined intervals (weekly or monthly) and suits smaller facilities where real-time alerting infrastructure is not cost-justified.
Facility size is a practical threshold: DOE guidance and practitioner consensus generally support continuous FDD investment for buildings above 50,000 square feet of conditioned area, where the fault-driven energy savings reliably offset service costs. Below that threshold, periodic FDD or integration into a smart building managed services contract is a more proportionate approach.
References
- U.S. Department of Energy, Building Technologies Office — Fault Detection and Diagnostics
- NIST SP 1169 — Framework for the Evaluation of Intelligent Building Environments/FDD Tools for Commercial HVAC Systems
- ASHRAE Guideline 36-2021 — High-Performance Sequences of Operation for HVAC Systems
- ASHRAE Standard 90.1-2022 — Energy Standard for Sites and Buildings Except Low-Rise Residential Buildings
- California Energy Commission — Title 24, Part 6 (California Energy Code)