Every system has the same well. None of them is right.
Accounting has it as a cost object. Production accounting has it under a different number. SCADA has it as a tag prefix, GIS as a point, the work order system as an equipment record. Somebody maintains a spreadsheet that maps a few of them to each other, and that spreadsheet is the closest thing you have to a master record.
The problem is not that those systems are bad. It is that none of them is trying to be authoritative. Each holds the fields it needs and treats the rest as reference data somebody else keeps current, so a well gets renamed in one place and corrected in another and they drift. You find out when a report does not reconcile, or when a callout goes to a well you sold eight months ago.
Keep your identifiers. Register them.
Nothing here asks any system to change its keys, which is the requirement that kills most master data projects. A node in your tree carries one identifier per system, and Field Warden is the place that knows they are the same physical thing.
- Field Warden
- 100/12-14-038-01W4The UWI, validated at entry
- Accounting
- WELL-1234Their key, unchanged
- SCADA
- PLANT1.WELL14A tag prefix, not an asset id
- GIS
- FID_88213A feature id
That is the part that is genuinely different, and it is worth being plain about why. Everybody sells integration, which moves data between two systems and hopes. Almost nobody offers to be the place that holds the mapping, and the mapping is the thing you actually cannot buy today. It exists in your company already, in somebody’s head and a spreadsheet column, and nowhere holds all of it.
Whose number wins, decided once
Authority cannot sit with one system, because none of them deserves all of it. So it is set per field, and a write to a field the writing system does not own is refused with the owner named in the error. Not silently dropped, and not accepted with a warning nobody reads.
- Hierarchy and node type
- Field WardenIt is the structure everything else hangs off
- UWI and licence
- Field WardenValidated at entry, and the identity of the thing
- Cost centre
- AccountingAn accounting construct, and they change it
- Tag path
- SCADAOurs would be a stale copy of theirs
- Surface coordinates
- GISSurveyed there, not typed here
This is a default rather than a rule, and it is configuration rather than something baked into the software. If your survey data is more reliable than your GIS, own coordinates here. The point is that a disagreement becomes a question answered once instead of an argument had every quarter.
Reading it out without asking us
The people who actually consume an asset list are an analyst in Excel, a manager in Power BI, and an integrator in their own host. None of them wants an API key.
So the registry is published as OData, which Power BI and Excel both read natively with no driver to install and no connector to buy. It is also what commercial connector tooling speaks, so if you already standardise your data access through something like that, this sits underneath it rather than beside it.
Alongside it, the same REST, MQTT and bulk CSV access every part of the platform has. One asset list, whichever shape your tool wants it in.
What this is worth on an ordinary Tuesday
A well is sold, or shut in, or renamed. Today that is a change in six systems, four of which nobody remembers until something breaks. With a registry it is one change, and the systems that care read it.
The obligation that comes with it is ours and worth stating: a registry that is authoritative and wrong is worse than no registry, because everything downstream now trusts it. So every write is attributed and timestamped, every change to an attribute is kept rather than overwritten, and a refused write is visible to the system that attempted it rather than only in our logs.
Where this actually stands
Designed and specified, not built. The identifier crosswalk, per field ownership and the OData endpoint are described in ADR 0023 in enough detail to build from, and none of them is shipping. What exists today is the asset tree itself, typed attributes with validated UWIs, and REST, MQTT and CSV access to everything in it.
It is on this site because it changes whether the tree you build now is worth building, and because if you are weighing this against a project of your own we would rather you knew the plan than found out later.