Building Automation Services
Most building faults are already visible in data nobody is looking at. A damper stuck at minimum position. A control valve leaking by. A zone heating and cooling at the same time. A fan running hard against a closed damper. Each of these leaves a clear signature in trend data, often for weeks, before anyone raises a ticket. Fault detection and diagnostics reads those signatures automatically, every day, across every piece of plant — and tells you what is wrong, how confident it is, and what it is costing you to leave it alone.
Ask about a fault detection assessmentEvery BMS already has alarms, and in most buildings they have stopped meaning anything. Hundreds of standing alarms, most of them nuisance, all of them acknowledged out of habit. Alarming works on thresholds: a value went above or below a number. It is blind to everything that is wrong while every value stays inside its limits.
Fault detection works on relationships. It compares what the plant is doing against what it should be doing given the conditions — the outside air temperature, the mode, the setpoint, the command, the position feedback, the time of day. A chilled water valve commanded to 40 per cent while the supply air temperature refuses to move is not an alarm condition. It is a fault, and it will show up on your energy bill long before it shows up on a screen.
The failure mode of most FDD deployments is not technical. It is that the system produces four hundred faults a week, the facilities team looks at it twice, and it is never opened again. We design against that from the start.
Every fault carries a probable cause, not just a detection. “AHU-3 economiser not economising” is a symptom; “outside air damper actuator not responding to command” is something a technician can act on. Faults are ranked by consequence — energy cost, comfort impact, compliance risk or equipment damage — so the list is worked in the order that matters.
Where the data supports it, each fault carries an estimated cost of inaction, expressed in kWh or currency. This is what gets a work order approved. Faults are deduplicated and grouped. One stuck damper should produce one item, not forty daily instances. Rules are tuned during a commissioning period specifically to drive down false positives, because adoption dies on the first week of noise.
Most planned preventive maintenance is calendar-driven: every filter changed quarterly, every belt inspected half-yearly, whether or not the equipment needs it. That is simultaneously wasteful on the assets that are fine and too slow for the assets that are degrading.
Where the instrumentation supports it, we move maintenance onto the condition of the asset: filter replacement triggered by actual differential pressure trend and rate of change, servicing driven by accumulated runtime and start counts rather than elapsed time, degradation trending on chiller approach temperature, plant kW/TR, pump specific power and heat exchanger effectiveness, each measured against the commissioned baseline.
Escalation paths agreed, dashboards and reports configured, CMMS integration tested, and your team trained on how to interpret and act on a fault. Rule documentation handed over in full.
A recurring review cycle — typically monthly or quarterly — where we assess which faults recurred, which were closed, which rules need retuning, and what the programme has saved. This is where FDD and continuous commissioning become the same activity.
FDD does not fix anything. It tells you what is wrong. Someone still has to hold a spanner, and if nobody is resourced to act on the output, the deployment will fail no matter how good the rules are. We would rather say this before you buy than after.
It also cannot detect what your instrumentation cannot see. It will not compensate for a sensor you do not have, and it will not manufacture history that was never trended. Where we find those gaps in the readiness assessment, we will tell you what it would cost to close them and let you decide whether it is worth it.
Finally, no rule set is perfect on day one. Expect a tuning period. Any supplier who tells you their FDD works out of the box on your estate has not looked at your estate.
No. FDD runs on the data your existing system already produces. If anything, it is most valuable on older estates, because those are the ones that have drifted furthest. What matters is trend coverage and history retention, not the age of the platform.
They overlap deliberately. Continuous commissioning is an engineering programme delivered by people: review, rewrite, tune, verify. FDD is the automated layer that watches between those visits and tells you when something has drifted again. Most clients who take both find the commissioning visits get shorter, because the faults have already been identified.
Wherever you want it to. Rules can run entirely on-premise with nothing leaving the building, at the edge, or in your own cloud tenancy. We do not require you to host data on our infrastructure, and if you do choose a hosted option it is your tenancy and your account.
The readiness assessment usually returns findings immediately — dead sensors, missing history and standing overrides tend to surface on the first pass, and those are fixable straight away. A tuned rule set typically takes 4-8 weeks from data access to operational handover, depending on estate size and how much modelling is required.
That is expected in the first weeks, which is why the commissioning stage exists. Rules are tuned against confirmed and dismissed faults until the output is something a team can actually work through. We treat a persistently noisy rule as our defect, not your problem.
Send us the specification, the point list, or just a description of what is not working. You will get a reply from an engineer within one working day.