How to fault-find a BMS without guessing

Two engineers can go to the same fault and come back with completely different results. One changes a controller, resets everything, and the fault’s back on Thursday. The other spends twenty minutes looking before touching anything, finds a sensor that’s been knocked, and writes a report that gets the client to approve two other repairs. The difference isn’t talent. It’s method.

1. Understand the actual complaint

“The AHU’s not working” isn’t a fault description. Before you go near the plant, find out:

• what exactly is happening (too hot, too cold, noisy, alarming, not running?)
• where, and since when
• whether it’s constant or intermittent, and if intermittent, when it happens
• whether anything changed recently, such as works, a power cut, a software change or a new tenant

Ten minutes with the person who reported it, or with the helpdesk notes, can save hours.

2. Look before you touch

Check the BMS before you start changing things:

• Active and recent alarms. What’s actually alarming, and since when?
• Overrides. Is anything in hand, overridden, or forced to a value?
• Trends. What were the setpoint, the measured value and the output doing when it went wrong? If there are no trends, set some up now.
• Schedules. Is the plant supposed to be running?

A good look at the trends often tells you where the fault is before you’ve opened a panel.

3. Split the problem

Every BMS fault sits in one of a few places. Work out which:

Where	Typical signs
Field device	Sensor reading wrong, actuator not moving, switch stuck
Wiring	Readings that jump about, open circuit, shorted outputs
Controller	Outputs not following commands, controller offline, failed I/O
Network	Devices offline, values not updating, comms errors
Software and logic	Wrong setpoint, wrong schedule, a bug in the strategy
The plant itself	Valve moves but temperature doesn’t change, pump runs with no flow

The last one matters. A lot of “BMS faults” are mechanical or hydraulic; see hydraulic problems that get blamed on the BMS.

4. Prove it, one step at a time

Work from the controller outwards, measuring as you go:

1. Is the controller doing what you’d expect? Check the output value in the software.
2. Is that signal leaving the controller? Measure it at the terminals: 0–10 V, a relay contact, whatever it should be.
3. Is it arriving at the device? Measure at the actuator or relay.
4. Is the device responding? Watch it move, or listen for the relay.
5. Is the plant responding? Check the temperature or pressure change.

For inputs, go the other way. Measure what the sensor is actually producing, check it arrives at the controller, then compare the software value with a reference reading.

At each step you either find the break or rule that part out. That’s what stops you swapping parts on a hunch.

5. Use the right tools, safely

A decent multimeter covers most BMS fault-finding: volts, resistance, milliamps and continuity. Add a calibrated reference thermometer, a manometer for pressures, and a laptop with the right software and drivers. For comms, a USB-to-RS-485 adapter and a protocol tool are worth their weight in gold.

Panels have 230/400 V in them, and the plant you’re testing can start without warning. Follow the site’s permit and isolation procedures, prove dead before working, and tell the building team before you put plant in hand.

6. Fix the cause, not the symptom

Resetting an alarm, applying an offset or overriding a valve can get the building comfortable today. That’s fine as a temporary measure, as long as you record it and come back to fix the real cause. Temporary fixes that become permanent are how sites end up full of mysterious overrides.

7. Write it up properly

A good fault report is short and specific:

• what was reported
• what you found, backed by evidence such as trends, readings and photos
• what you did
• what’s still outstanding, with a recommendation and, ideally, a price

That last part matters commercially. A clear report that says “valve actuator on AHU 2 heating coil has failed; replacement recommended, quotation attached” is how one call-out turns into the next job, and how the client starts to trust you with more.

Common traps

• Changing several things at once, so you never know which one fixed it.
• Trusting the BMS reading without checking it against a reference.
• Assuming the drawings match the installation.
• Leaving plant in hand or overridden “just until tomorrow”.
• Not writing anything down, so the next engineer starts from scratch.

For the panel side of things, see hand-off-auto, motor starters and interlocks.