Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
rotatingMachine
Node-RED custom node for individual rotating-machine control — pumps, compressors, blowers. Part of the EVOLV wastewater-automation platform developed by R&D at Waterschap Brabantse Delta.
Models a single asset with an S88 state machine, curve-backed flow/power prediction, and parent/child registration for orchestration by machineGroupControl or pumpingStation.
Install
In a Node-RED user directory:
cd ~/.node-red
npm install github:gitea.wbd-rd.nl/RnD/rotatingMachine
Or consume the whole platform:
npm install github:gitea.wbd-rd.nl/RnD/EVOLV
Run node-red and the node appears in the editor palette under the EVOLV category.
Quick start
Drop a rotatingMachine onto a flow, fill the Asset menu (supplier, model — must match a curve in generalFunctions/datasets), and wire three debug nodes to the three output ports. Inject these in order:
| Topic | Payload | Effect |
|---|---|---|
setMode |
"virtualControl" |
allow manual commands |
simulateMeasurement |
{type:"pressure",position:"upstream",value:200,unit:"mbar"} |
seed upstream pressure |
simulateMeasurement |
{type:"pressure",position:"downstream",value:1100,unit:"mbar"} |
seed downstream pressure |
execSequence |
{source:"GUI",action:"execSequence",parameter:"startup"} |
start the machine |
execMovement |
{source:"GUI",action:"execMovement",setpoint:60} |
ramp to 60 % controller position |
execSequence |
{source:"GUI",action:"execSequence",parameter:"shutdown"} |
shut down |
Ready-made example flows are in examples/:
01 - Basic Manual Control.json— inject-only smoke test02 - Integration with Machine Group.json— parent/child registration withmachineGroupControl03 - Dashboard Visualization.json— FlowFuse dashboard with live charts
Import via Node-RED Import ▸ Examples ▸ EVOLV.
Input topics
| Topic | Payload | Notes |
|---|---|---|
setMode |
"auto" | "virtualControl" | "fysicalControl" |
mode gates which sources may command the machine |
execSequence |
{source, action:"execSequence", parameter} — parameter: "startup" | "shutdown" | "entermaintenance" | "exitmaintenance" |
runs an S88 sequence |
execMovement |
{source, action:"execMovement", setpoint} — setpoint in controller % |
moves controller position |
flowMovement |
{source, action:"flowMovement", setpoint} — setpoint in configured flow unit |
converts flow → controller %, then moves |
emergencystop |
{source, action:"emergencystop"} |
aborts any active movement and drives state to off |
simulateMeasurement |
{type, position, value, unit} — type: pressure | flow | temperature | power |
dashboard-side measurement injection |
showWorkingCurves |
— | diagnostic — reply on port 0 |
CoG |
— | diagnostic — reply on port 0 |
Topic case is preserved; sequence parameter and action names are normalized to lowercase internally (so "emergencyStop", "EmergencyStop", "emergencystop" all work).
Output ports
| Port | Label | Payload |
|---|---|---|
| 0 | process |
delta-compressed process payload; keys are type.variant.position.childId (e.g. flow.predicted.downstream.default). Consumers must cache and merge each tick. |
| 1 | dbase |
InfluxDB line-protocol telemetry |
| 2 | parent |
{topic:"registerChild", payload:<nodeId>, positionVsParent} emitted once on deploy for parent group/station registration |
State machine
idle ─► starting ─► warmingup ─► operational ◄─┐
▲ │
│ ▼
│ accelerating / decelerating
│ │
└──────────┘
│
▼
stopping ─► coolingdown ─► idle
│
▼
emergencystop ─► off
warmingupandcoolingdownare protected — new commands cannot abort them.acceleratinganddeceleratingare interruptible. If ashutdownoremergencystopsequence is requested mid-ramp, the active movement is aborted automatically and the sequence proceeds once the FSM has returned tooperational.- Timings come from the
Startup/Warmup/Shutdown/Cooldownfields in the editor (seconds).
Predictions
Flow and power outputs are curve-backed predictions driven by the controller position and the differential pressure across the machine. Inject both upstream and downstream pressures for best accuracy. With only one side present the node warns and falls back to the available side. With no pressure, predictions use the minimum pressure dimension (flow/power will look unrealistic).
The active curve is selected from machineCurve.nq and machineCurve.np, keyed by the closest matching pressure level. Curve units are declared in the Asset menu (default: mbar, m³/h, kW, %).
Units
Canonical units are used internally (Pa / m³/s / W / K). All inputs and outputs convert at the boundary via the configured unit for each measurement type. The speed field in the editor is a ramp rate in controller-position units per second (so speed: 1 → 1 %/s → a setpoint of 60 % from idle completes in ~60 s).
Testing
cd nodes/rotatingMachine
npm test
79 tests cover construction, mode/input routing, config loading, sequences, emergency stop, shutdown, interruptible movement, movement lifecycle, prediction health, pressure initialization, CoolProp efficiency, registration, negative/null guards, output format, listener cleanup. Run the full suite in ~2 seconds.
For end-to-end verification, see ../../docker-compose.yml — a Docker stack (Node-RED + InfluxDB + Grafana) that hosts the live node. The scripts in ../../../memory/ and examples/ document the E2E protocol used for production-readiness benchmarks.
Production status
Last reviewed 2026-04-13 — trial-ready. See the project memory file node_rotatingMachine.md for the latest benchmarks, known caveats, and wishlist.
License
SEE LICENSE. Author: Rene De Ren, Waterschap Brabantse Delta R&D.