A WPS library isn't a binder of forms — it's the operational backbone of a fab shop's QC program. Shops that organize it well run smooth audits, onboard new CWIs in days instead of weeks, and respond to AHJ requests in hours instead of days. Here is the structure that works.
Library organization
A few organizing dimensions, ordered by priority:
Process — SMAW, FCAW-G, FCAW-S, GMAW, SAW, GTAW
Base metal Group — Group I (A36), Group II (A572-50, A992), Group III, etc.
Position — 1G/2G/3G/4G groove, 1F-4F fillet
Joint detail — B-U2, B-U4, T-joint, etc.
A typical structural shop's library might break down as:
4–6 SMAW WPSs (different positions on Group I and II)
Plus 2–4 specialty procedures for non-standard work.
Naming convention
A consistent naming scheme makes the library navigable:
WPS-{YY}-{NNN}
Where YY is the two-digit year of initial release and NNN is sequential. So WPS-25-001, WPS-25-002, etc.
Some shops add a process suffix: WPS-25-001-SMAW. Useful but increases the chance of inconsistency. Pick a scheme, stick to it.
Storage architecture
Three options, in order of operational maturity:
1. Word + network drive. A folder of .docx files on a shared drive. Cheap, familiar, but invites revision drift. Audit prep takes longer because there's no central index.
2. PDF library with version control. A folder of signed PDFs with explicit revision-numbered filenames (WPS-25-001-rev-3.pdf). Better than Word because PDFs aren't accidentally editable. Still requires discipline to keep the floor copy current.
3. Software-driven library. Database-backed; revisions are first-class; current revision is rendered on demand. Floor terminals always show current. This is what AISC-certified shops at scale use.
The cost difference between option 1 and option 3 is roughly $0 to $100/seat/month. The audit-day difference is enormous.
Document index
A library needs an index card per WPS:
WPS number
Process, base metal Group, position, thickness range
Supporting PQR number (or Clause 5 citation)
Current revision number and date
Prequalified vs qualified-by-test status
CVN scope (yes/no)
Notes (any active restrictions, special use cases)
An index of 30 WPSs fits on one page and saves enormous time during audit prep.
PQR linkage
Each qualified-by-test WPS must link to its supporting PQR. Best practice:
PQR archive is separate from WPS library
Each WPS lists supporting PQR number in the header
PQR archive is organized by PQR number, not by WPS that uses it
One PQR can support multiple WPSs
When the PQR archive is well-organized, audit prep becomes pulling-from-a-shelf instead of searching.
Welder qualification linkage
A WPS authorizes the procedure. A WPQ authorizes the welder. The library should support cross-matching:
For each WPS, list the welders currently qualified to follow it
For each welder, list the WPSs they're qualified to follow
Track WPQ continuity (last weld in process X, due-date for requalification)
A spreadsheet works. Software is better.
Onboarding a new CWI
A well-organized WPS library lets a new CWI hit the ground in days:
Day 1 — read the library index. Understand the shop's process coverage.
Day 2 — review 5 sample WPSs in detail. Trace each to its supporting PQR.
Day 3 — walk the shop floor; confirm each active weld traces to a WPS.
Day 4 — review the WPQ register; confirm welder coverage.
Day 5 — observe a daily inspection cycle.
By end of week one, the new CWI knows the library cold. With a disorganized library, this takes a month.
The audit-day shortcut
When the auditor arrives:
One-page library index
Tabbed PDF archive (or a software export of all current revisions)
PQR archive with lab certifications attached
WPQ register, current
CVN supplementary documentation, if applicable
Hand the auditor that package on arrival. Audit time drops from days to hours.
FAQ
Frequently asked.
How many WPSs does a typical fab shop need?
Most structural shops carry 15–30 WPSs covering their common process × material × position × thickness combinations. Specialty shops doing pipe, pressure vessels, or stainless will add 5–15 more for those scopes.
Should one WPS cover everything, or many narrow WPSs?
Narrow WPSs are easier to maintain and audit. A WPS that tries to cover too wide a range (multiple processes, all positions, broad thickness) usually has ambiguous parameters and fails review. Build the library wide, not deep.
How long does it take to build a new WPS from scratch?
A prequalified WPS using existing templates: 30–90 minutes of engineering time. A qualified-by-test WPS with new PQR: 2–4 weeks calendar (mostly waiting on the lab) plus 4–8 hours of engineering.