Reading, history, twins, and screens stay on the edge server if the internet drops.
TwinEdge / Platform · Across every sitePlatform overview
Across every site
On-premise, on cloud, or both. Same product family.
TwinEdge OS is the edge server on-premise — including air-gapped. TwinEdge Platform can live in your rooms or in the cloud. Collectors and a local historian keep history on site. You can also run TwinEdge on top of the SCADA and CMMS you already have.
Pick a layout. Same product family.HYBRID
TwinEdge OS at the site. TwinEdge Platform in the cloud.
The edge server keeps the plant running. The cloud half compares sites and reports when the link is up.
On-premise
TwinEdge OSEdge server next to the equipment
- Equipment / PLCRuns the plant
- CollectorsRead the signals
- TwinEdge OSEdge server
- Local historianHistory on site
- Local twinsModel this site
- Local screensAlarms and views
Cloud
TwinEdge PlatformSame family, across every site
- TwinEdge PlatformCompare, learn, report
- HistorianEstate history
- Analytics & twinExplain the change
- AssetOps & FieldDo the repair
- Capital PlanningRepair or replace
Compare plants, spread what worked, and learn from more history than one site has.
OS, historian, analytics, AssetOps, and capital planning stay one product line.
Collectors and a local historian sit with TwinEdge OS. TwinEdge Platform can live in your rooms or in the cloud.
§ 01
How you connect. You can use more than one.
TwinEdge OS is the edge server on the plant network. Collectors and a local historian sit with it, so the site keeps history even when it is air-gapped. TwinEdge Platform reads SCADA, CMMS, and other systems you already run — in your rooms or in the cloud. You can use one path, or both.
§ 02
What the cloud half is for.
- 01Comparing sites honestly
Two plants doing the same job with the same equipment, and one uses 14% more power. That question can only be asked somewhere that can see both.
- 02Spreading what worked
A setting change that saved money at one site is checked against every similar machine elsewhere, and proposed where it applies.
- 03Learning from more history than one site has
Models trained on the failures of two hundred pumps are better than models trained on the failures of three. This is the one job genuinely better done centrally.
- 04Knowing which sites need attention
One view of every site: what is running, what is degraded, what has lost its connection, and where the work is piling up.
- 05Reporting somebody will actually read
Figures for the whole estate built from the same records the engineers use, so the board pack and the plant room do not disagree.
§ 03
Watching every site at once.
Running software across many sites is mostly the problem of noticing. A site with a fault will usually tell you. A site that has stopped talking looks exactly like a site with nothing to report.
| State | What it means |
|---|---|
| Healthy | Reading, storing, and reporting normally |
| Degraded | Running, but something is wrong: a dead sensor, a full disk, a slow link |
| Buffering | No route out. Still reading and storing; nothing is being lost |
| Quiet | Nothing heard for longer than expected. Treated as a fault, not as calm |
- 01Which sites are fine, and which are not
One list, ranked worst first. Not a wall of green squares.
- 02What version is where
Which software and which models each site is running, so a fix can be aimed at exactly the sites that need it.
- 03Updates that go out carefully
New versions reach a few sites first and only continue when those sites stay healthy.
§ 04
What deliberately does not happen here.
The central half never becomes something a site depends on to keep running. No alarm waits on it, no screen an operator uses is served from it, and no reading is lost if it is unreachable for a week.
That is a constraint rather than a limitation. It is what makes the whole thing safe to put into a plant, and it is why a site with no outbound connection at all is a supported configuration rather than an awkward conversation.
“If a site would notice this being down during a shift, it does not belong in the middle.”
The dividing line
Start with two sites that should be identical.
Everybody has a pair: same design, same equipment, different numbers, and an argument about why that has been running for years. That is the most interesting place to point this first.