In many Gulf buildings the BMS knows about a fault well before the help desk does. A chilled water pump trips at night, the alarm sits on a screen in the control room, and a job is raised only when a tenant complains about a warm floor the next morning. Connecting the BMS to the CAFM system closes that gap, but only if the link is selective. This guide covers which alarms should open work orders, how to filter the rest, how alarm priorities map to SLA clocks, and what to ask the BMS contractor and the CAFM vendor before you build it.
Key takeaways
- Pass a short list, not the whole alarm log. Only alarms that need a person to attend should become work orders. Everything else is logged or shown on a dashboard.
- Filter before you create. Time delays, grouping of repeats and suppression during planned maintenance keep the job queue credible.
- Map each BMS point to a CAFM asset. Without a shared naming convention the work order cannot carry the location, history or contract.
- Agree priorities and the SLA start in writing. The alarm priority in the BMS and the job priority in the contract are two different scales until someone maps them.
Why Connect the BMS to CAFM
The two systems do different jobs. A building management system reads sensors and switches plant in real time, and raises an alarm when a value goes out of range. A CAFM system holds the asset register, the contracts and the SLAs, dispatches technicians and keeps the service history. The BMS knows that something has failed. The CAFM knows who should fix it, how fast, and whether it has failed before.
Without a link, the handover between them is a person: a control room operator who sees the alarm, judges it and phones or types it into the help desk. That works during the day shift in a staffed control room. It breaks down at night, in buildings with no full-time operator, and across portfolios where one team watches many sites. It also leaves no reliable record of when the fault started, which matters when the owner and the contractor later disagree about response time.
An integration replaces that handover for a defined set of alarms. The owner gets a time-stamped job for every critical fault, and the contractor gets the job on the technician's phone instead of on a screen nobody is watching.
Which Alarms Should Become Work Orders
The test is simple: does this alarm need a person to go and do something? If yes, it is a candidate for a work order. If it only informs, it belongs in a log or on a dashboard. Sort the BMS alarm list into three groups before any interface is configured.
| Group | Typical examples | What the CAFM should do |
|---|---|---|
| Act now | Chiller or pump trip, high temperature in a server or comms room, water leak detected in a riser or plant room, sump pit high level, generator failed to start, loss of a district cooling supply signal | Create a work order straight away with the highest reactive priority and notify the on-call trade |
| Act in working hours | AHU filter differential pressure high, a fan coil unit not reaching setpoint, a duty pump running with the standby unavailable, a sensor reading out of range | Create a work order with a normal reactive priority, or add it to the next planned visit for that asset |
| Inform only | Plant start and stop messages, mode changes, alarms that return to normal within the agreed delay, communication retries that recover | No work order. Keep it in the BMS log; show counts on a dashboard if useful |
Fire alarm signals need separate treatment. The fire alarm system is a life safety system with its own panel, procedures and authority monitoring, and the response to a fire signal is never a routine work order. What the CAFM can usefully receive is the fault and follow-up side: a detector fault, a panel fault, or a job to inspect and reset after an activation.
Running hours and meter readings are the other useful feed. They are not alarms, but they let the CAFM schedule planned preventive maintenance by use rather than by calendar, for example servicing a pump after a set number of running hours.
Filtering Nuisance Alarms
The fastest way to make an integration fail is to connect every alarm on day one. A queue flooded with jobs that need no action is soon ignored, and the one real fault is ignored with it. Industrial plants have dealt with this problem for years. The ISA-18.2 standard on alarm management is written for the process industries, not for buildings, but its central idea transfers well: every alarm should be rationalised, which ISA describes as "determining the need for alarms, justifying their use, documenting details, assigning priorities". For an FM team that means a written alarm list, reviewed line by line, before anything is passed to the CAFM.
In practice, five filters do most of the work:
- Agreed list. Only alarms on the approved list pass through. New points added to the BMS do not create jobs until someone adds them to the list.
- Time delay. An alarm must stay active for an agreed period before a job is created, so a value that drifts out of range and recovers on its own does not call anyone out.
- Grouping of repeats. If the same alarm fires again while its work order is still open, the new event is added to that job as a note. It does not open a second job.
- Parent and child. When a chiller trips, the high temperature alarms on the floors it serves will follow. Link them to the chiller job instead of raising one job per floor.
- Maintenance suppression. When an asset has an open planned maintenance job and is isolated, its alarms are held back until the job is closed.
Review the result every month for the first quarter. Any alarm that keeps producing jobs closed as "no fault found" either needs a better setpoint or delay in the BMS, or comes off the list. Any real fault that reached the help desk by phone and not through the link is a candidate to add.
Priorities and SLA Clocks
BMS alarms carry their own priority. In a BACnet system each alarm notification has a priority number from 0 to 255, where a lower number is more urgent. How those numbers are used varies by installer, but vendor documentation commonly repeats the same suggested bands; Siemens' engineering help, for example, lists them as specification suggestions. The table shows those bands and one reasonable way to map them to CAFM job priorities. The right-hand column is our suggestion, to be replaced with the priorities in your own contract.
| BACnet priority band | Suggested message group | Suggested CAFM handling |
|---|---|---|
| 0 to 31 | Life safety | Not a routine job. Follow the fire and life safety procedure; the CAFM records the follow-up inspection |
| 32 to 63 | Property safety | Emergency priority: flooding, loss of power or cooling to critical rooms |
| 64 to 95 | Supervisory | Urgent priority: main plant tripped or running without standby |
| 96 to 127 | Trouble | Normal reactive priority, attended in working hours |
| 128 to 191 | Miscellaneous, higher priority | Add to the next planned visit, or no job |
| 192 to 255 | Miscellaneous, lower priority | Log only |
Do not assume the BMS on your site follows these bands. Ask the BMS contractor for the priority actually configured on each alarm in your list, then set the CAFM priority per alarm type. The job priority is what starts the SLA clock, so three points need to be written into the contract or the operating procedure:
- Start. Does the response clock start at the alarm time, or when the work order is created after the delay filter? With an automated link the gap is small, but it should be stated.
- Hours. Which alarm types call out the on-call technician at night and on weekends, and which wait for the next working day?
- Stop. Is the job complete when the technician closes it, or when the alarm has also cleared on the BMS? The second is the stronger proof that the plant is back in range.
Once these are fixed, alarm-driven jobs feed the same response and resolution measures as any other reactive work. Our guide to work order KPIs sets out how those are calculated.
Mapping BMS Points to Assets
A BMS identifies things by point name, such as a pump status or a room temperature on a given controller. A CAFM identifies things by asset ID and location. The integration needs a table that joins the two, and building it is usually the largest piece of work in the project.
| Mapping field | Source | Why it is needed |
|---|---|---|
| BMS point or alarm name | BMS contractor | The key the interface listens for |
| CAFM asset ID | Asset register | Puts the job on the right asset, with its history and warranty |
| Location (building, floor, room) | Asset register | Tells the technician where to go |
| Fault category and trade | FM team | Routes the job to HVAC, electrical or plumbing |
| CAFM priority | FM team, from the contract | Starts the correct SLA clock |
| Delay and grouping rule | FM team with the BMS contractor | Applies the nuisance filters above |
| Out-of-hours rule | FM team, from the contract | Decides whether the on-call technician is notified |
If the asset register and the BMS were built by different parties with different naming, expect gaps: BMS points with no matching asset, and assets with no points. Fix the register first. Our guide on building an asset register covers the hierarchy and tagging that make this join possible. Then agree with the BMS contractor that point names will not be changed without notice, because a renamed point silently breaks its mapping.
Worked Example: A Pump Trip at 02:10
An office tower in Dubai has two chilled water pumps serving the upper floors, one duty and one standby. The hard services contractor provides an on-call technician at night. The times below are illustrative, to show how the records move.
- 02:10, alarm. The duty pump trips on overload. The BMS starts the standby pump and raises a pump trip alarm.
- 02:12, filter. The alarm is on the agreed list and has stayed active past its two-minute delay, so it passes. No open job exists for this pump, so it is not grouped with anything.
- 02:12, work order. The mapping table links the point to the pump's asset ID on level 32. The CAFM creates a reactive job in the urgent priority, because the floors are now running with no standby, and starts the response clock.
- 02:13, dispatch. The on-call HVAC technician receives the job on the mobile app with the asset, location and alarm text.
- 02:45, on site. The technician accepts and arrives. The response time is recorded against the contract target.
- 03:30, repair. A failed overload relay is found and replaced from van stock. The technician records the cause, the part and photos, and returns the pump to duty.
- 03:32, close. The pump trip alarm returns to normal on the BMS, and the job is closed with that time as confirmation.
In the morning the FM manager sees one closed job with a full timeline, not a line in a BMS alarm log. No tenant called. If the same pump trips again next month, the history is on the asset, and the repeat is visible. The technician's photos and times also give the owner the evidence described in our guide to technician proof of work.
How the Link Is Built
There are four common ways to get alarms from the BMS to the CAFM. Which one is possible depends mostly on the BMS: how old it is, whether it uses open protocols and what the supervisory software can export.
| Method | How it works | Watch for |
|---|---|---|
| BACnet/IP | The CAFM side, or a gateway, reads alarm notifications and point values over the building network using the open BACnet protocol | Needs a network route and the BMS contractor's agreement; keep the connection read-only |
| Modbus | Values are read from registers on plant controllers such as chillers, generators and meters | No built-in alarm concept: thresholds and alarm logic have to be defined on the receiving side |
| API or event feed from the BMS software | The supervisory software pushes alarm events to the CAFM through a web interface | Depends on the BMS product and licence; ask what is included |
| Alarm email | The BMS emails each alarm and the CAFM reads the mailbox and creates jobs | Works on older systems, but carries less detail and fails quietly if the mailbox or format changes |
Network and security
The BMS runs the building's plant, so IT and the BMS contractor will, rightly, be cautious. Three rules keep the conversation short. The link is one-way for control: the CAFM reads alarms and values and never writes setpoints or commands. The connection goes through a controlled route agreed with IT, not an open port on the BMS network. And where the CAFM is hosted outside the building, the data leaves through a gateway that only sends, so nothing outside can reach in to the controllers.
Questions to Ask Both Vendors
Ask these before the integration is priced. Several of them belong in the FM contract or the BMS maintenance contract, not only in the software scope.
For the BMS contractor
- Which protocols and interfaces does the installed BMS expose, and is any extra licence needed to use them?
- Can you export the full point and alarm list, with the priority configured on each alarm?
- Will you agree a point naming convention and give notice before renaming or removing points?
- Who adjusts setpoints and delays in the BMS when an alarm proves to be a nuisance, and is that covered by the maintenance contract?
For the CAFM vendor
- Which BMS platforms and protocols have you connected in production, and can we speak to that client?
- Can we set the alarm list, delays, grouping and priorities ourselves, per alarm type, without a change request?
- How are repeats of the same alarm handled while its work order is open?
- Can alarms be suppressed automatically while a planned maintenance job is open on the asset?
- Does the work order record the alarm time, the creation time and the time the alarm cleared?
- What happens when the link itself fails: who is alerted, and are missed alarms recovered when it comes back?
Our CAFM RFP template has a section for integration requirements where these can be scored.
How Reflexion handles BMS alarms
Connecting your BMS to CAFM?
Bring your BMS make and your alarm list. We will walk through which alarms should open jobs, how they map to your assets and how the SLA clock runs.
Frequently Asked Questions
What is BMS and CAFM integration?
It is a link that passes selected alarms and readings from the building management system to the CAFM system, so a fault detected by the BMS becomes a work order with an asset, a priority, an SLA clock and a named owner. The BMS keeps controlling the plant; the CAFM manages the response and keeps the record.
Should every BMS alarm create a work order?
No. Only alarms that need a person to attend should open a job. Status messages, alarms that clear themselves within a short delay and alarms caused by planned maintenance should be logged or shown on a dashboard instead. Start with a short agreed list of critical alarms and extend it once the team trusts the jobs it produces.
Which protocol is used to connect a BMS to a CAFM system?
Most links use an open protocol the BMS already speaks, usually BACnet/IP or Modbus, or an API or event feed from the BMS supervisory software. Where none of these is available, some sites fall back to alarm emails that the CAFM reads, which works but carries less detail and is harder to keep reliable.
When should the SLA clock start for a BMS alarm?
Agree it in the contract. The cleanest rule is that the clock starts when the work order is created, which for an automated link is within moments of the alarm passing the filter. If the alarm is first reviewed by a control room operator, write down whether the clock starts at the alarm time or at the operator's dispatch.
Who is responsible for the integration, the BMS contractor or the CAFM vendor?
Both, with a line drawn between them. The BMS contractor exposes the points and alarms on an agreed interface and keeps point names stable. The CAFM vendor reads them, applies the filter and the asset mapping, and creates the jobs. The owner or FM team owns the alarm list and the priorities, because those are operational decisions.
Related reading
What Is a BMS?
What a building management system monitors and how it differs from CAFM.
CAFM ERP Integration
The other major integration: record ownership and the five data flows.
Work Order KPIs
Response, resolution and repeat-fault measures for alarm-driven jobs.
FM Help Desk
How requests from every channel become prioritised work orders.
