Site interface
The site node that reads the plant and sends it home safely.
Many energy centres have controllers and meters but nothing that can send. is the site node for them: a small industrial unit for the DIN rail in the plant room. It reads what is there, keeps every reading until the broker has acknowledged it, and sends over one encrypted connection that only ever points outwards.
Every controller and meter already in the plant room.
Building management controllers
Trend IQ, Niagara, Schneider, Siemens, Honeywell and Distech controllers over BACnet/IP, and the Trend IQ web interface where that is what the site has.
Plant with registers
Pumps, inverters, boiler controls and anything else that speaks Modbus, over the network or over a serial line.
Heat and water meters
Energy, volume, flow and return from the meters and M-Bus gateways on site, read directly rather than through a third party.
Nothing is lost between the plant room and the broker.
Queued until acknowledged
Every reading goes into an on-disk queue first and leaves it only when the broker has confirmed it. A power cut or a dropped link loses nothing. A very long outage keeps the newest readings up to a cap.
Change of value with a heartbeat
A point is sent when it moves by more than its deadband, or when its heartbeat falls due. That keeps a mobile-data link within budget and still proves the point is alive.
One message per plant item
Each cycle sends one message per boiler, pump, meter or store, carrying its readings together with their units, times and quality, in the shape the brokerage expands into single readings.
One encrypted connection out
The node dials out to the brokerage with its own credentials and holds one session. Nothing listens on the site side and nothing comes in.
Works on any link
Designed for the rules of narrow-band mobile data, so it runs even better on wider mobile data or Ethernet. The radio is a setting, not a different product.
Reports its own health
Uptime, queue depth, readings sent, temperature, disk, signal and last error are published as a plant item like any other, so the node is watched the same way the plant is.
The control logic, graphics and alarms stay where they are.
- No supervisor is replaced. The site node reads alongside the building management system; it does not take over strategy, graphics or alarms.
- It never changes plant on its own. In its reading role it has no path to write to a controller at all.
- Locked down. It runs as a service under its own user on a read-only system with a watchdog, and exposes only the serial and network ports it needs.
Direct control returns the same way, and only with an engineer on site.
A catalogue, not a free hand
Control states are chosen from a catalogue agreed for the network. Each one carries an envelope of allowed values and limits; nothing outside it can be asked for.
Signed, with an engineer checked in
A state is called by name, signed, and carried through the brokerage to the site node. It is accepted only while an engineer is checked in at that site.
A heartbeat fail-safe
The node expects a heartbeat for as long as a state is held. If the heartbeat stops, the plant returns to its own strategy. Nothing changes plant unattended.