In short. A generator controller can publish its status, measurements, alarms, fuel level and hours to a plant SCADA, a building management system or a data centre’s DCIM over Modbus, SNMP, BACnet or IEC 61850. The useful integration is the one that sends the facility what it needs to act on, keeps read-only monitoring separate from remote control, and leaves the generator fully able to operate when the network is down. More data is not the same as more awareness.
Three supervisory systems, three audiences
SCADA belongs to the plant electrical and operations team; it supervises switchgear, transformers, generators and process equipment and expects high-resolution electrical data and event records. A BMS belongs to the facility team of a commercial building, hospital or campus; it watches HVAC, lighting, fire and power together and wants summarised status and alarms. DCIM belongs to data centre operations; it tracks capacity, redundancy and energy for every element of the power and cooling chain and wants generator availability, load and fuel autonomy in the same view as the UPS and the chillers. One controller can serve all three, but each needs its own point selection, update rate and alarm philosophy. The general case of generator connectivity is covered under communications and remote monitoring; the focus here is integration into a system the customer already runs.
What data is worth sending
| Data group | Typical points | Who needs it | Update rate |
|---|---|---|---|
| Status | Mode (auto, manual, off), running, ready, on load, breaker position | All three systems | On change, polled every few seconds |
| Electrical | Voltage, current, frequency, kW, kVAr, power factor per phase | SCADA, DCIM | 1 to 5 seconds |
| Engine | Oil pressure, coolant temperature, speed, battery voltage | SCADA, maintenance | 5 to 10 seconds |
| Fuel | Tank level, estimated runtime, leak and low-level alarms | DCIM, BMS | Minutes, alarms on change |
| Alarms and events | Warnings, shutdowns, time-stamped event log | All, with priorities | Immediate |
| Counters | Running hours, start count, kWh, maintenance due | Maintenance, DCIM | Hourly or on demand |
Alarms are the part most often done badly. A controller may expose two hundred alarm bits; a BMS operator needs perhaps a dozen, grouped as warning or shutdown, with a plain description and a defined response. Alarm floods during a start sequence, where every transient low-pressure or under-voltage bit lights up for a second, should be suppressed at the source or filtered by a start-in-progress flag. The alarm and event management page sets out how to prioritise.
Protocols and the hardware that speaks them
Most generator controllers from the major families offer Modbus RTU over RS-485 as standard and Modbus TCP over Ethernet as an option or through a gateway. BMS integration in buildings often expects BACnet/IP, and DCIM platforms commonly poll SNMP, so a protocol gateway sits between the controller and the supervisory network, mapping Modbus registers to BACnet objects or SNMP OIDs. MV switchgear and larger plants increasingly use IEC 61850, where the protection relays and the plant controller publish standardised data models. Whatever the protocol, the integration depends on a register map or points list from the controller manufacturer, a mapping document agreed with the SCADA or BMS integrator, and a test of every point at commissioning. Capability must be verified against the actual controller model and firmware. The architecture choices for larger plants are described in SCADA architecture for power systems.
Monitoring is not control
Reading the generator’s state over a network is low risk. Starting it, stopping it, changing its mode or operating a breaker from a remote screen is a different category. Remote control introduces questions of who is authorised, whether the person at the screen can see the machine and the people near it, how a remote command interacts with the local emergency stop and the maintenance lockout, and what happens if the command is issued twice or arrives late. Many facilities settle for read-only integration with remote start reserved for a defined test mode, enabled by a local key switch. Where remote control is genuinely needed, as in an unmanned telecom or pumping site, the controller’s own access levels, a logged command path and a local-remote selector that defaults to local are minimum requirements. These controls belong to the OT network architecture for the site rather than to the generator alone.
Security follows from the control decision
A generator controller or a protection relay exposed to the office network, or worse to the internet for convenient access, is an invitation. Segment the OT network from IT, use the gateway as the only point of contact, change default credentials, disable write access to registers that must not be written, and keep a record of configuration changes. Remote reset of shutdown alarms deserves particular caution, because it can restart a machine that tripped for a mechanical reason.
What happens when the link fails
Communication will fail: a switch loses power, a cable is cut during civil work, a firmware update changes a register map. The generator must not care. Its AMF logic, its protections and its load sharing must run entirely locally, with the supervisory link as an observer. The supervisory system, for its part, must know the data is stale. A SCADA screen that keeps showing yesterday’s fuel level as if it were live, because the last good value was never flagged, is more dangerous than a blank screen. Every integrated point should carry a quality flag or a heartbeat, and a loss of communication should itself be an alarm of moderate priority. For data centres, where DCIM feeds capacity decisions, the critical power monitoring and analytics page discusses how to keep the model honest.
Commissioning the integration
Integration is proven point by point. Each register is forced or observed at the controller and confirmed on the supervisory screen with the right scaling, units and sign; kVAr sign conventions and three-phase averaging catch out more projects than any protocol issue. Each alarm is raised and acknowledged end to end. Loss of communication is simulated and the stale-data behaviour checked. If remote start is in scope it is tested with a person at the generator. The signed points list and the test record become part of the handover, and where master controls exist they are tested as the plant’s single interface.
Related photographs and videos
Common mistakes
- Promising BMS integration before checking the controller model. The installed controller has no communication port and a retrofit module or a new controller is needed.
- Mapping every available register into the BMS. Operators face hundreds of meaningless points and miss the fuel-low alarm among them.
- Displaying the last received value without a stale flag. A tank reads full on the screen hours after the link dropped and the generator runs dry.
- Enabling remote start and stop without a local-remote selector. A technician working on the set is exposed to a start command from another building.
What to send us
Send the make and model of each generator controller with a photograph of its fascia and communication ports, the name and version of the SCADA, BMS or DCIM platform and the protocol it expects, a list of the points the facility wants to see, and whether remote control of any kind is required. A network diagram showing where the generator room sits relative to the supervisory network helps. Use the application engineering enquiry page.
Questions engineers ask
Can an older generator with a basic AMF panel be connected to a BMS?
Usually yes, at two levels. Dry contacts for running, common fault and mains failure can be wired to BMS digital inputs with no change to the panel. Full measurement data needs either a retrofit communicating controller or an external power meter on the generator output with its own Modbus port, which is often the cheaper route.
Which protocol should we ask for, Modbus, BACnet or SNMP?
Ask for what the receiving system natively supports. Generator controllers speak Modbus almost universally; BMS platforms prefer BACnet; DCIM tools often poll SNMP. A gateway converts between them, and specifying the gateway and its mapping as part of the generator scope avoids the common gap where neither the generator supplier nor the BMS integrator owns the conversion.
Is it safe to allow remote start of the generator from the control room?
It can be, with safeguards: a local-remote selector at the set that defaults to local, remote start permitted only in a test mode that does not transfer load unless intended, logged commands tied to a user, and a lockout procedure that disables remote commands during maintenance. Without these, read-only monitoring is the responsible choice.
What should the generator do if it loses communication with SCADA?
Nothing. The generator’s automatic mains failure logic, protections and load sharing must be complete without the supervisory link. The SCADA should flag the generator’s points as stale, raise a communication alarm and stop displaying old values as live. Any remote command queued during the outage should be discarded, not executed when the link returns.
Technical content reviewed by Srinivas Reddy, Managing Partner, Reddy Generators. Last updated 19 October 2026.