Predictive maintenance

We build predictive maintenance in three stages: reliable data and event records, verified diagnostic rules, and prediction proven in the field.

Approach

Predictive maintenance in three stages

Stage 1

Reliable data & event records

Fault codes, operating status, protection events and suitable temperatures are recorded continuously. If a brand doesn't expose a value, it reads "not assessable" — missing data is never counted as "healthy".

Stage 2

Verified diagnostic rules

Specific degradation signatures become cases through rules. A hot day, the morning start-up or a switched-off unit must not create false cases. Rules improve as cases are verified in the field.

Stage 3

Prediction proven in the field

We will call it prediction when field data shows specific fault types flagged early enough. We're collecting that evidence together through pilot projects.

The maintenance case

What does a maintenance case contain?

"Protection trips have increased" achieves nothing on its own. A case answers four questions together.

What was observed? The measured change and the comparison period.
What might cause it? Probable causes, in order of likelihood.
What should be checked? The first things to inspect on site.
How do we know it's fixed? The value to re-measure under similar conditions.

An outdoor unit fault doesn't open separate cases on every connected indoor unit; the case attaches to the outdoor unit and lists the affected zones.

hvac · zone 3B · pull-down time Demo data

This zone now takes longer to reach setpoint under similar outdoor and load conditions.

Pull-down time: 74 min
Previous similar day: 38 min
Outdoor band: 31–34 °C

Probable causes

Dirty indoor filter or coil
Obstructed air distribution
Refrigerant circuit performance loss

Recommended check

Clean the filter and coil; measure supply/return air temperature difference.

How do we know it's fixed?

Pull-down time returns to its previous level within the same outdoor temperature band.

Open
Acknowledged
Actioned
Verified

Sample case text. Real cases are generated from field data and the points your brand exposes.

Case types

HVAC maintenance cases defined in FiotX

Whether a case works on your site depends on the equipment exposing the relevant signal. Which cases apply to your system is settled at the survey.

Case	Equipment	Signals used
Repeated protection / short cycling	VRF outdoor unit	Protection state, compressor start count and runtime, operating mode
Slow pull-down	VRF zone / indoor unit	Setpoint and room temperature, outdoor air, operating status
Filter due	Fan coil	Filter alarm or runtime threshold
Condensate pump	Fan coil, AHU	Continuous pump running, level warning
Filter clogging	AHU	Filter differential pressure
Low delta-T	FCU / AHU coil	Flow and return water temperatures

No automatic intervention

The system never changes setpoints or resets protections by itself. The technical team decides.

Condition-matched verification

Pre- and post-intervention periods are compared under similar outdoor and usage conditions.

One issue, one case

A recurring finding doesn't open a new record; the existing case's last-seen time is updated.

Worth knowing

Around AC and VRF, we also do all of this.

Central control & scheduling After-hours shutdown, setpoint limits, bulk commands. Air quality & monitoring CO₂, temperature, humidity; battery-powered wireless sensors. Maintenance & mobile work orders Checklists, photo-backed completion, entitlement tracking. Energy metering & BMS Circuit-level consumption; Modbus and BACnet. Residential blocks Per-apartment control, central monitoring.

Let's set up predictive maintenance on your system.

In a pilot we agree in writing which cases will run on your system and how they'll be verified.

Request a pilot conversation