What Is Fault Detection and Diagnostics (FDD), and How Does It Differ from Predictive Maintenance?

Fault detection and diagnostics, usually shortened to FDD, is software that continuously checks HVAC operating data against rules to flag faults that already exist: a stuck damper, a unit heating and cooling at the same time, a sensor reading nonsense. Predictive maintenance answers a different question. FDD tells you something is wrong now. Predictive maintenance tells you something is about to go wrong. If you're evaluating monitoring for your building or your kitchen, the difference matters less than the vendors make it sound, but it's worth understanding what each term actually means.

What does FDD actually do?

FDD watches the data your HVAC equipment already produces and compares it against rules for how the equipment should behave. When reality and the rule disagree for long enough, it raises a fault. The classic catches are operational waste and control problems: an economizer that never opens, a rooftop unit fighting itself with simultaneous heating and cooling, a schedule that runs the building all weekend, a temperature sensor that drifted out of plausibility.

The "diagnostics" half is the ranking and explanation: good FDD tools don't just list faults, they point at the likely cause and sort by impact, because a big building can generate hundreds of faults and someone has to decide which ten matter.

Where does FDD come from?

FDD grew up in large commercial buildings, on top of building automation systems. Offices, campuses, and hospitals already had thousands of data points flowing through a BAS, and FDD emerged as the analytics layer that made sense of them. That heritage shapes the tools: most FDD products assume a BAS exists, integrate through it, and price for buildings that have one.

That's also why the term barely appears in the restaurant world. A standalone rooftop unit and a walk-in cooler don't have a BAS to plug into, so classic FDD tooling has mostly never reached them.

How is predictive maintenance different?

Predictive maintenance looks forward instead of sideways: rather than checking today's operation against rules, it tracks how equipment condition trends over time and flags degradation before it becomes a fault. The two answer different questions about the same machine.

In practice the boundary is soft. A fault FDD catches early, like a defrost cycle misbehaving, is often the same signal a predictive system reads as a failure developing. Same data, different lens.

Which one does a restaurant actually need?

Neither term should drive the decision, because what a restaurant needs is simpler than either category: continuous monitoring of the handful of units that can shut you down, with warnings early enough to schedule the fix. That system will do FDD-style checks (is defrost running, is the schedule sane) and predictive-style trend watching (is runtime creeping up week over week) without you ever caring which is which.

What matters more is fit. Tools built for a BAS-equipped office tower won't map onto a kitchen with two rooftop AC units and three cold boxes. Our guide to what predictive maintenance looks like in practice covers what the restaurant version actually involves, and what it can and can't catch.

We're building that restaurant-shaped version at SmartHVAC. If you want to follow the failure library as it grows, sign up below.