Making Sense of PUWER Defect Grading

Writing grade definitions that survive a hostile reading

A grade definition earns its place by being usable at seven in the morning by a fitter with a torch, and defensible three years later in front of someone paid to find fault with it. Both audiences reward the same style: short, concrete, and built on observable facts rather than judgement words.

Define the stop grade by consequence and immediacy: a condition that could injure someone during normal operation or foreseeable misuse before the next scheduled inspection, examples printed alongside, guard missing or defeated, interlock inoperative, brake or emergency stop not functioning, structural cracking at a loaded point. Define the dated grade by trajectory: a condition not dangerous today that will or may become the stop grade before the next inspection if untreated, wear approaching limits, damage to secondary parts, leaks and chafing on the way to failure. Define the monitor grade by margin: deterioration noted, measured where possible, comfortably inside limits, recorded for trend.

Then armour the definitions with three rules. Every grade carries one sentence of reasons, because the reasoning is what an investigator actually audits. Ambiguity resolves upward: an inspector unsure between two grades applies the more severe and says so, which converts uncertainty from a risk into a record. And grades never move for operational convenience: the definition mentions no busy seasons, and neither may the grading.

Test the draft against your own history before publishing it. Take the last dozen findings the estate can remember and grade them on paper with the new definitions: where two people land differently, the definition is loose at exactly that point, and tightening it now is free.

Calibration: making two inspectors grade the same defect the same way

A grading system's real test is not whether one inspector applies it well but whether three inspectors apply it identically, because drift between graders is invisible day to day and glaring in an audit. The same worn guard fixing graded dated by the day inspector and monitor by nights is a system quietly failing, and neither inspector knows it.

Calibration is built with small, regular habits rather than a training event. Keep an example bank: photographs of real findings from your own estate, each with its agreed grade and the reason, growing by a few entries per quarter. New inspectors grade the bank blind as induction; existing inspectors re grade a sample annually, and disagreements go not to a vote but to the definitions, because a split decision means the definition is loose, and the fix is a sharper sentence in the published grades.

Add a moderation walk each quarter: two inspectors, one machine line, grading independently and comparing at the end. The walk finds drift while it is still small, doubles as the estate's most honest condition survey, and costs an hour. Where the estate uses contractors or multiple sites, send them the definitions and the example bank before their first inspection, and moderate their first round against your own graders, because imported grading habits are the fastest way to fracture a calibrated system.

Watch one metric between calibrations: the grade distribution per inspector. Graders drift toward signatures, the one who never issues a stop, the one whose findings are all monitors, and a simple quarterly count per grader surfaces the signature before the audit does. The point is never to punish the outlier; it is to find the loose definition or the missing confidence that created them.

The reasons line: one sentence that carries the whole system

Every practice in this guide converges on a single habit: the sentence of reasoning beside each grade. It looks like bureaucracy and is the opposite, the one line that converts a tick into a judgement and a judgement into evidence. Grade: dated, fourteen days. Reason: cover chafed through, pressure layer intact, progression certain but not imminent. Twelve words, and the finding now explains itself to every future reader.

The reasons line does four jobs at once. It disciplines the grader, because a grade you must justify in a sentence is a grade you actually think about. It transfers knowledge, because the fitter reading the record learns what mattered about the finding, not just its urgency. It enables audit, because consistency between reasons and grades is checkable in a way bare letters never are. And it protects the estate, because after an incident the question when did you know and what did you decide is answered in the finding's own words, contemporaneously, which is the strongest form any answer can take.

Keep the discipline light so it survives contact with real shifts. One sentence, plain words, present tense, stating the condition and the trajectory. Ban boilerplate by policy: as per inspection and see above are not reasons, and a record where the same phrase appears beside forty findings has stopped thinking. Where a measurement exists, the reason includes it, wear at 60 per cent of limit, because numbers are the only reasons that trend.

The line matters most on monitor grades, the ones that ask nothing today. A monitor without a reason is noise; a monitor with a measurement is the first point on a curve, and estates run on curves.

Mapping grades to actions: the response half of the system

Grades exist to trigger responses, and a grading scheme without a published action map is a thermometer with no medicine cabinet. Beside the three definitions, print the three response paths, so that the grade and its consequences arrive together and nobody negotiates the response finding by finding.

The stop grade's path: equipment isolated and physically secured immediately, tagged with reason and authority, planning informed, sibling equipment checked where the failed component is common, and return to service only on a named competent person's tested confirmation recorded in the record. The dated grade's path: a task with an owner and a deadline inside the grade's date, any use conditions written at the machine, closure evidence filed, and a standing decision rule for dates at risk, stop or re justify in writing before the date, never silence. The monitor grade's path: logged with its measurement, reviewed at maintenance planning, and promoted without ceremony the moment duty or condition changes.

Publish who holds each power, and push the stop authority down, not up: supervisors and shift leads must be able to stop equipment at three in the morning without phoning anyone, because a stop grade that waits for office hours has a defined period during which the estate knowingly runs dangerous equipment. The record then shows the stop at the time of the finding, which is the single line investigators look for first.

Wire the map into the systems that already run the site, maintenance planning for dated tasks, shift handover for stops and conditions, procurement for the spares that closures keep needing, so the grades drive machinery that exists rather than a parallel bureaucracy that will not survive its first busy month.

One estate, four regimes: keeping your grades honest beside the statutory ones

A PUWER grading scheme never operates alone. The same estate holds LOLER reports with their A, B, C conventions, PSSR reports with dated repairs and Regulation 10 stops, and LEV reports with failed hoods and remedial lists, and the people applying your PUWER grades read those documents daily. The schemes should rhyme, or the estate's people will translate between four urgency languages on the fly, and translation under pressure is where mistakes live.

Align the logic, not the labels. Your stop grade should mean what a LOLER Category A and a PSSR Regulation 10 mean: danger now, equipment or system down, senior attention today. Your dated grade should mean what a Category B and a 9(5)(b) entry mean: a legal quality deadline with required work and closure evidence. Your monitor should mean what observations mean everywhere: recorded, measured, trended, nothing today. Print that equivalence as a one line key on the record template, and a supervisor who understands any one regime now understands the estate.

The alignment pays twice more. It lets the defect log merge: one log, every finding from every regime, sorted by the shared urgency logic, which is the single document that shows an estate's true condition and the one an inspector or insurer is most impressed to receive. And it keeps your grading honest by comparison: if your PUWER inspections have produced no stop grades in three years while the LOLER reports on the same machines carry Category A entries, the gap is not the machines, it is your grading, and the merged log makes that visible before an investigator does.