Legacy Drawing & Obsolescence Mining
The vessel will serve until 2050. Its drawings were filed in 1985 — by someone who retired in 2015.
No sector keeps hardware in service longer than defence, and none pays a higher price for it in paperwork. Naval platforms run forty to fifty years through refits that must trace back to original design data. Land systems get life-extended, re-configured per customer nation, and supported decades past their design era. And the Zeitenwende added a twist no other industry knows: production lines that were shut down a generation ago are being reactivated — and their entire documentation is paper, microfilm and unindexed TIFF.
Meanwhile the obsolescence clock ticks against every drawing in the pile. Suppliers vanish, materials get withdrawn, components go end-of-life — and each time, someone must answer fast: which systems, which vessels, which delivered units are affected, and what would a form-fit-function replacement need to satisfy? Today that answer lives behind days of archive digging by the few veterans who still read the old conventions. In defence, that is not an inconvenience — “no obsolescence management, no military business” is a contract reality.
Drawing mining turns the pile into data: AI reads every scan — title blocks, parts lists, materials, suppliers, key dimensions — resolves revision chains, and indexes it all as structured, queryable records. And because these are defence documents, everything runs inside your security perimeter, on the same sovereign stack that powers our air-gapped engineering AI. Nothing leaves the building. Ever.
// The defence archive problem, in numbers
Unfindable means unmaintainable — and in defence, unmaintainable is contractual
Most defence archives were “digitized” long ago: fed through a scanner, filed by drawing number, done. The result is a wall of pixels that answers questions only for people who already know the answer — which number, which era’s conventions, which superseded material norm. Those people are retiring, and the questions are multiplying.
Three forces drive the multiplication. Refits and life extensions demand original design data at exact revisions, per vessel, per configuration. Reactivated production — the Zeitenwende’s most concrete engineering task — means whole product documentation packages from the seventies and eighties must become manufacturable again: parts lists re-captured, materials re-mapped to current norms, suppliers re-sourced. And obsolescence never sleeps: every supplier insolvency and every end-of-life notice triggers the same fleet-impact question, and the contract expects an answer in days, not quarters.
Manual archaeology cannot scale to that load. Mining can: extract the data once, structure it, link parts to suppliers to systems to platforms — and every future question becomes a query instead of an expedition.
In — microfilm scan, 1983
Out — structured record, inside the perimeter
From archive to obsolescence radar: how the mining works
The pipeline is a batch process with a quality loop — and one step more than our civilian mining offers, because in defence the extracted data earns its keep when it is linked: parts to suppliers, suppliers to systems, systems to platforms. Everything runs inside your perimeter, with export-control markings honoured at every step.
The archive as it is — where it is
TIFFs, PDFs, microfilm scans, mixed eras and quality. Processing runs on hardware inside your security perimeter; classified and export-marked material never touches external infrastructure. No re-scanning campaign as a precondition.
Structure before reading
Assembly drawing, part drawing, schematic, parts list, process sheet — typed and grouped by programme, system and component class, so tens of thousands of sheets get an order before a single field is extracted.
The data locked in the pixels
Title blocks, revisions, materials, suppliers, parts lists, key dimensions — confidence-scored per field. Era conventions and superseded norms are learned from your samples and mapped, not guessed; ambiguities are flagged, never silently "modernized".
Humans check the doubtful — the PLM gets the result
High-confidence records flow into Teamcenter (or your PLM) as items with source scans attached and revision chains resolved; low-confidence fields queue for fast human review. The original drawing remains the authoritative record — the index makes it findable.
The obsolescence radar goes live
Extracted parts link to suppliers, materials, systems and platforms. From then on, an end-of-life notice or supplier insolvency is a query — affected systems, vessels and delivered units in minutes — and geometric similarity search proposes form-fit-function candidates from your own part history.
What a mined archive unlocks
The index is the foundation — on top of it, four defence-critical processes stop depending on veteran memory:
Obsolescence radar, fleet-wide
A supplier folds or a component goes end-of-life: one query answers which systems, vessels and delivered units carry it — the analysis that today takes weeks, before the actual engineering starts. Similarity search proposes form-fit-function candidates from your own history.
Refit & repair at enquiry speed
"Which drawings define this system on this vessel, at which revision?" becomes a search. Refit scoping and repair quotes start from referenced records — with the configuration evidence linked, not remembered.
Reactivated production, documented
When a dormant line restarts, its documentation package is mined first: parts lists re-captured, materials mapped to current norms, missing sheets identified — the paper-to-production gap closed in weeks instead of years.
Knowledge that survives retirement
The veterans' ability to navigate the archive — which era, which conventions, where to look — is encoded into structure and search. New engineers become productive on legacy programmes in weeks, not years.
When does mining pay off?
The case scales with platform age, archive size and obsolescence exposure. You are a candidate if most of these apply:
- You support naval platforms or land systems with decades of remaining service — refit, repair and life extension are core business.
- You are reactivating or scaling production of products whose documentation predates your PLM — or your CAD system.
- Obsolescence management is a contractual obligation, and every case starts with days of manual archive research.
- The archive holds tens of thousands of paper, microfilm or TIFF drawings with little searchable metadata.
- The engineers who read the old conventions and superseded norms retire within the next five years.
- Your drawings are VS-NfD or export-controlled — processing outside your perimeter is not an option, full stop.
What the mining pipeline handles
Built for the messy reality of a decades-old defence archive — and for the security reality around it.
Before and after: the supplier insolvency letter
A component supplier files for insolvency. Their parts sit somewhere in thirty years of your delivered systems. Here is what happens next, on each side of the change:
Before — archaeology under deadline
- Which of our systems use their parts? Nobody knows — weeks of digging
- Veterans pulled off programme work to read 80s parts lists
- Material specs in norms withdrawn twenty years ago
- Replacement search starts from a blank sheet
- The customer's deadline for the impact report passes — twice
After — a query, not a crisis
- Supplier query: affected parts, systems, platforms — minutes
- Impact report drafted from linked records — same day
- Materials pre-mapped to current norms — engineering starts immediately
- Similarity search proposes form-fit-function candidates from your own parts
- Veterans verify the plan — instead of excavating the data
Why inteliScience
Generic OCR reads clean title blocks; defence archives are anything but. The hard part is the whole surround: title block layouts that changed four times since 1970, materials in withdrawn norms that must be mapped — verifiably, with your materials engineers releasing the mapping table — revision and change-note chains across sheets, and the linking that turns records into an obsolescence radar. And in your sector, one more thing decides everything: where the processing runs. Ours runs inside your perimeter, on the sovereign stack, with export-control markings honoured per document and per user.
The last mile is our home ground: mined drawings become proper Teamcenter items with attached source scans, ready for the refit workflows and targeted NX reconstructions that follow. The quality loop stays honest — confidence scores per field, human review where the AI is unsure, accuracy measured on your scans in the pilot. The original drawing remains the authoritative record; the mining makes it findable and connected. AI does the reading. Your people stay the authority on what goes into a warship.
Fixed price — no day rates, no surprises
Drawing Mining Pilot
- 1,000 drawings from one programme or system family, mined end-to-end inside your perimeter
- Extraction tuned to your era-specific title blocks, conventions and norms
- Measured accuracy report per field type — your scans, your numbers
- Obsolescence radar demo: part–supplier–system linking on the pilot corpus
- Cost & timeline projection for the full archive + Teamcenter integration concept
8–10 weeks · processed entirely on your infrastructure · all mined data is yours, whatever you decide afterwards.
For comparison
Manual re-indexing
- Typing metadata for 50,000 drawings by hand — with cleared personnel only
- A multi-year project that every ramp-up deadline will interrupt
- Meanwhile: every obsolescence case still starts with weeks of digging
Frequently asked questions
Our drawings are VS-NfD and export-controlled. Where exactly does processing happen?
Can this really make a dormant production line's documentation usable again?
How automated is the obsolescence radar really?
The material specs on 1970s drawings don't exist anymore. What happens to those?
Do we get 3D models out of this?
Can we start with one vessel class or one programme?
The fleet will outlive its paperwork by twenty years. Close that gap while the veterans are still in the building.
Show us 20 sample scans from one programme — inside your perimeter, on your terms — and we'll show you what the extraction and the radar get out of them.
Talk to a Drawing Mining EngineerAlways here to support you – reach out to us with your questions, and we’ll guide you every step of the way.
+40 774 463 429
+40 774 463 429
Romania
info@inteliscience.net
info@inteliscience.net
