Control islands
Continuous duty wears
Energy per tonne
Live-capable workWhere the control systems are, and what they carry
| Area | Typical assets | What the control system is holding |
|---|---|---|
| Raw material handling | Ship unloaders, long conveyor runs, transfer houses, silos | Sequencing, interlocks, blockage and belt protection, feed rate to the potline |
| Potroom | Rectifiers, busbar, feeders, crust breakers, tending cranes | Current stability, feed control, alumina dosing, hood and duct integrity |
| Gas treatment | ID and booster fans, filters, alumina injection | Differential pressure, fan drive health, emission compliance |
| Anode and rodding | Presses, bake furnace, rodding shop, cleaning and casting stations | Batch and station sequencing across many small controllers that rarely talk to each other |
| Cast house | Furnaces, casting lines, handling and despatch | Recipe, temperature, weighing and traceability |
| Utilities | Power distribution, compressed air, cooling water | Availability, and energy per tonne across the whole site |
Three problems that recur across smelters
Control islands
A rodding shop or an anode area typically grew as a dozen separate controllers, each supplied with its own machine, each with its own HMI and its own vintage. Individually they work. Collectively nobody can see the process end to end, so a stoppage at one station is only understood after it has propagated.
Consolidating that is not, usually, replacing twelve controllers. It is a supervisory layer that reads them, plus the discipline of a single operating picture. Far cheaper, far less risky, and it can be done without a shutdown.
Conveyor and fan fleets under continuous duty
Material handling and gas treatment run continuously, so they wear continuously, and their failures stop production immediately. These fleets are the strongest candidates for condition monitoring in the plant. The reporting has to be on an operating-hours basis, because duty and standby rotation on a calendar report will show a rested machine as healthy and hide how hard the working one has been driven.
Energy per tonne is the number everything is judged against
In a smelter, energy is the dominant cost and the dominant environmental metric. Making it visible per area, per line and per shift, alongside condition and availability, is usually the thing that converts a maintenance conversation into a management one.
Sequencing work in a plant that never stops
The engineering is rarely the hard part. The scheduling is.
- Scope is frozen early. A change requested two weeks before a window is a change that will not be ready, and pretending otherwise is how windows get blown.
- Materials land before the window opens, all of them, checked, not en route.
- A rollback point is written down with an abort criterion agreed by someone with the authority to call it, before anything is disconnected.
- Work is split into what can be done live and what genuinely needs the window. Most surveys, most monitoring installation and most supervisory work is live-capable. Reserving the window for what truly needs it is the cheapest scheduling decision available.
- Recommendations carry their slot. A four-hour job proposed into a period the plant cannot give is not a recommendation, it is an observation.
What we do in metals plants
- PLC and DCS migrationSequenced against shutdowns, with a rollback point
- Supervisory and operator systemsOne picture across control islands, without replacing them
- Conveyor and fan drive fleetsStandardised so one spare covers many machines
- Rotating equipment monitoringOn an operating-hours basis, scheduled against production
- Switchgear and busduct surveyUnder load, graded to a criterion
- Panels, MCCs and consolesBuilt locally in Bintulu
Questions we get asked
Can work be done without a shutdown?
More than most plants assume. Survey, condition monitoring installation, supervisory and reporting layers, and much panel preparation are live-capable. What genuinely needs a window is anything that breaks an existing I/O connection or takes a controller out of service, and reserving the window for only those items is usually worth more than any efficiency in the work itself.
We have a dozen separate controllers in one area. Do they all need replacing?
Usually not. If each one is doing its job, a supervisory layer that reads them all and presents one operating picture solves the actual problem, which is visibility, at a fraction of the cost and with far less risk. Replacement is justified when a controller is itself unsupportable.
Where should condition monitoring start in a smelter?
Material handling and gas treatment fans, because they run continuously and their failure stops production directly. One nominated asset, baselined and proven, before anything scales.
Can you work with our existing maintenance system?
Yes. The objective is a work order in the system you already run, raised from a diagnosis that names the action and the window. A monitoring platform that requires a separate parallel process tends to be abandoned within a year.