Most plant faults never cross a threshold.
A stuck economizer damper does not set off an alarm. Neither does a valve passing 4%, a sensor drifting half a degree a month, or an air handler heating and cooling at the same time. Every one of them is invisible to a limit check and expensive for years.
A fault is a departure from the sequence, not a number out of range
Limit alarms ask whether a point is outside its band right now. That catches failures and misses degradation, which is where the money is. Indagate evaluates equipment against the sequence it is supposed to be running — the ASHRAE Guideline 36 shape for that equipment type, adapted to how your site actually operates — so the question becomes whether this air handler is behaving like an air handler in this mode, at this outside condition.
That is a question about several points at once over a window of time, which is exactly the question a single-point threshold cannot express.
The faults that quietly cost the most
Simultaneous heating and cooling. The coil and the reheat both working, each doing its job perfectly, together burning two utilities to hold one setpoint.
Economizer stuck. Free cooling available outside and a damper that has not moved since someone commanded it to hand. Nothing reads as broken; the chiller simply runs on days it should not.
Valve or damper leak-by. A few percent that never closes. Small enough to look like noise, constant enough to be a line item.
Sensor drift. Half a degree a month, and every control decision downstream is made against a number that is no longer true.
Schedule drift. An override from a plant visit in October that nobody released, running occupied hours through every weekend since.
Diagnosis, not just detection
Detection tells you the air handler is misbehaving. Diagnosis tells you which of the seven things that produce that signature it is, and rules out the other six with evidence. Because the fault is raised against a modelled asset, the diagnosis arrives with the trend that produced it, the mode the equipment was in, the outside condition at the time, and what closed the last three occurrences.
Ranked by cost, not by severity label
Severity labels are a scheme someone chose during commissioning. What an operator with a fixed number of hours needs is the order in which faults are worth fixing — energy at stake, comfort at stake, and whether this one is degrading or stable. A fault costing real money every day outranks an alarm that has been amber for two years, and the list says so.
Specifics
- Evaluated against
- Equipment sequences by type, in the Guideline 36 shape, adapted to site operation
- Window
- Faults evaluated across time and across points, not on instantaneous values
- Delivered as
- A fault on the asset record, with trend, mode, outside condition and prior resolutions
- Ranked by
- Estimated cost and whether the fault is degrading, not by a severity label
Common questions
Do we need to write the fault rules ourselves?
No. Rules come with the equipment type, because a fault in an air handler is a fault in an air handler. Site-specific behaviour is adapted during the survey, and anything unusual about how your plant is operated is captured there rather than assumed away.
How is this different from the alarms we already have?
Your alarms answer whether a point is out of range. These answer whether the equipment is running the sequence it should be. The two overlap less than people expect — most of the expensive faults never cross a limit at all.
Will it flood us with findings on day one?
The first run of any FDD on a real estate finds a lot, because the faults were always there. They arrive ranked by what they cost, so the list is a work plan rather than a queue. The alternative is not fewer faults — it is the same faults, unfound.
Bring your point list.
The first conversation is a survey: what you run today, and what it would look like modelled.