Production documentation: the working papers at the station and keeping them current

The set a workstation genuinely needs

In most plants it comes down to a handful of items: the work instruction covering method and sequence, the drawing or specification extract for the features made here, the setup sheet with machine parameters and tooling, the inspection requirement including frequency and what to do on a failure, safety information specific to this station, and the packaging or labelling specification where the station handles finished product. Anything else — full assembly drawings, quality manuals, procedure documents — belongs somewhere accessible rather than at the station. A folder containing the whole management system is a document that is present and unread.

Written for the person doing the job, not for an auditor

Instructions fail because they are written by an engineer describing what the process does rather than telling a person what to do. Useful ones use short steps in the order they happen, the words the floor actually uses, photographs of the correct and incorrect condition, and explicit criteria rather than adjectives — not securely tightened, but the specific torque figure and the tool to use. Where the workforce spans several first languages, images and short sentences carry more than translation quality. The test is simple: hand it to a competent person who has not done the job and see whether they can follow it.

Revision control fails at the point of use, not in the system

Master documents are usually well controlled; the copies at the machine are where control breaks. A revised instruction is issued and the old one stays in the drawer, so two versions coexist and the operator uses the one that is easier to read. The mechanism has to close the loop physically: a named person removes and destroys the superseded copy when issuing the new one, and the station holds a list of the documents it is supposed to have with their current revisions so that any audit or supervisor walk can verify it in a minute. Issuing without retrieving is the single most common documentation finding anywhere.

Uncontrolled copies, phones and the laminated sheet from years ago

Every plant accumulates informal documentation: a photocopy in a toolbox, a photograph of a setup sheet on somebody's phone, a laminated card made by a supervisor who left, handwritten settings taped to a machine. These exist because the controlled documents were inadequate, unavailable or unreadable, so removing them without fixing the cause simply drives them out of sight. Treat each one as a requirement statement — somebody needed information the system did not provide conveniently — and then absorb the useful content into the controlled document rather than merely confiscating the copy.

Who writes it, who reviews it, and who keeps it alive

Instructions written entirely by engineering describe an idealised process; instructions written entirely by operators encode current habits including the bad ones. The workable arrangement is drafted by engineering, walked through and corrected by the people who do the job, then approved by both plus quality where the document controls a product characteristic. Ownership afterwards matters more than authorship: name the role responsible for each document, set a review trigger tied to change rather than only to a calendar date, and make updating the instruction part of implementing any process change rather than a task done later.

Frequently asked questions

Should work instructions be on paper or on a screen at the station?

Screens solve the revision problem, since everyone sees the current version and obsolete copies cannot linger. They introduce others: a screen that is dirty, glare-washed or occupied by another application is not read, and an outage removes every instruction at once. Paper is readable anywhere and fails quietly instead of loudly. Many plants use screens for the frequently changing content and paper for stable safety and setup information, with a defined fallback when the system is unavailable.

How detailed should a work instruction be?

Detailed enough that a trained person can perform the task correctly without asking, and no more. Over-specification is its own failure: instructions that describe every motion are not read, go out of date quickly and create nonconformances whenever reality diverges. Concentrate detail on the steps where error is likely or consequential — sequence that matters, settings that must be exact, checks that catch a defect — and summarise the parts any competent operator will do correctly regardless.

Do we need documented instructions for every operation?

For operations where the method affects a product characteristic, safety or a regulatory requirement, yes. For genuinely simple tasks a competent person cannot get wrong, a documented instruction adds control effort without reducing risk. The judgement should be recorded rather than left implicit, so that the absence of an instruction is a decision somebody made rather than an oversight. Certification schemes generally expect that reasoning to be visible, not that every task has a document.