Sugar Australia: From a 1.59 GB folder to a searchable drawing register
Sugar Australia handed us a folder of drawings, documents and archives built up over decades of plant operation, some of it dating back to 1990. It is filed by drawing number, which works well if you already know the number. What was missing was a register to search it by anything else. Three weeks later there was one: 1,762 drawings, and a DMS decision they could get on with.
A decision Sugar Australia could now make
Everything is filed by drawing number. There was no register to search it any other way: not by title, not by revision, not by what the drawing is of. Sorting that out in house was a job they did not have the spare resource for.
The drawings are counted, the duplicates are mapped and the gaps are known. Sugar Australia could go ahead with the system of their choice, knowing exactly what was going into it.
Fewer than one file in four was a drawing worth keeping, and barely a quarter of the volume. The rest was duplicate copies, leftover software and files that were never drawings. Nothing of engineering value was dropped to get there.
A duplicate is not just wasted disk. Every extra copy is another result to open, read and rule out, so the effort of finding the right drawing climbs faster than the pile itself. Cutting the set down to the drawings that are actually distinct does not just make the archive smaller. It makes every search that follows shorter.
When a drawing exists in seventeen places, a change written on one copy is a change the other sixteen never receive. The next person opens a different file and works from a drawing that is quietly wrong. Knowing there is exactly one copy is what makes a markup trustworthy, because the next person to open that drawing sees the change.
The folder was bigger than it looked
Drawing archives grow one file at a time, and the people who filed them have long since moved on. Every year adds another copy, another revision, another folder that made sense to whoever created it. Without a register, the obvious questions have no quick answers. How many drawings are there? What metadata is there to carry across? Is this drawing already here, or does it need to be redrawn?
Opening the folder up produced a few surprises. More than half the content was hidden, either compressed inside archives or repeated across folders, and a large share of what remained was not drawings at all.
One of them turned out to hold copies of 24 of the others, so the same drawings sat nested inside the archive more than once.
58% of the DWG and DXF files were repeat copies of the same drawing filed in different folders. A single drawing, MAC-B1, existed in 17 separate locations.
Alongside the real content sat an entire AutoCAD and Adobe installation, fonts, plot logs, installer fragments, Windows thumbnail caches and CAD backup files. None of it had engineering value.
Forty-nine drawing numbers existed in more than one version. Same number, genuinely different content. Every version was kept rather than guessed at, because that call belongs to Sugar Australia’s engineers.
Nothing was deleted
Nothing was removed from what Sugar Australia sent us. The delivered folder is securely backed up and held intact, exactly as received. Duplicates were identified and mapped, not deleted. What to keep is their decision, and they can now make it later with the full picture in front of them.
Duplicates were matched by comparing the contents of the files themselves, byte by byte, never by filename or date. Filenames lie in both directions: identical files turn up under different names, and different drawings under the same name. Two files are only treated as copies of each other if their contents are an exact match.
What came out of that is a map showing every copy of every drawing and exactly where it sat in the original folder. It tells Sugar Australia what they have and where, so the decision about what to retire can be made deliberately rather than by whoever tidies up first.
- 01Verify the delivery against its manifest
- 02Extract every archive so nothing stays hidden
- 03Identify duplicate copies by file contents
- 04Separate drawings from non-drawings
- 05Split multi-page documents into sheets
- 06Extract metadata, with a human in the loop
1,762 drawings, each with a record
Every active drawing came out with a unique ID and a full metadata record. Title, drawing number and revision were the core set. Drawn by, revision date, project number and contract number were added at the engineering team’s request, because those are the fields their people actually search on.
The lower figures reflect what the drawings themselves carry, not a limit in the extraction. Only 8% of these title blocks carry a contract number at all, and roughly half record a project number. A register can only hold what was written on the sheet in the first place. Knowing which drawings lack a project number is itself useful, and is now something Sugar Australia can filter for.
Sugar Australia get their drawings back in their original file types: DWG, DXF and PDF, exactly as delivered. Nothing was converted or flattened on the way through. The DWGs remain editable, which matters the first time someone needs to revise a drawing rather than just look at one.
The register is built to load straight into a document management system, with metadata populated in bulk rather than every document being indexed by hand.
Sugar Australia went from a folder they could not search to a register their engineers can rely on. The work is complete: every file processed, every drawing catalogued with a metadata record, every duplicate mapped and every gap known.
The drawings are ready when the operation needs them, and the people who need them can trust what they find.
Turn a folder of drawings into a register
Most archives are filed by drawing number, which only helps if you know the number. A register is the first thing worth having, and it takes weeks, not years.