What a Hole Cover Has to Do, and What Marks It

Why this matters

A guardrail announces itself. A cover does the opposite: it looks like floor, it invites weight, and the only thing standing between a worker and a drop is a decision somebody made about a piece of material, possibly a piece of material they found. Covers fail quietly and they fail under people who had no reason to suspect anything. If you take one thing from this card, take the sequence: when you find an unmarked cover, stand clear of it, do not test it by stepping on it, put a barricade or a guardrail around it and keep everyone off until its capacity and its securing are established by someone who can state them.

The call

A mechanical mezzanine in a light industrial building. A pipe riser had been removed the previous year and the 24 in by 24 in floor opening it left was covered with a piece of half-inch wood panel, roughly 32 in square, laid loose. No stencil, no colour, no fastener. It had been there long enough to be dusty and long enough that everyone stepped over it out of habit rather than out of caution.

A service tech wheeled a hand truck with a small compressor on it across the mezzanine on the shortest line to the unit. The panel tilted into the opening as the front wheel crossed near its edge. He got a hand on a duct hanger and did not go down, and the compressor did, onto the floor below.

The first instinct in the room was that the panel was too thin. That instinct is wrong, and the way it is wrong is the point of this article.

What the standard asks a cover to do

Three separate requirements, and people remember one of them.

Under 29 CFR 1926.502(i), a cover on construction work must be capable of supporting, without failure, at least twice the weight of employees, equipment and materials that may be imposed on it at any one time. It must be secured against accidental displacement by wind, equipment or employees. And it must be colour coded or marked with the word HOLE or the word COVER, so that a person who does not know the history of that floor is warned by the floor itself.

Where a cover sits in a roadway or a vehicular aisle, the same paragraph raises the basis entirely: it must support at least twice the maximum axle load of the largest vehicle expected to cross it. That is a different and much larger number and it belongs to an engineer, not to a judgement call at the point of installation.

For general industry work, 29 CFR 1910.28(b)(3) requires holes to be protected, and 29 CFR 1910.29(e) sets the cover criteria, including the twice-the-maximum-intended-load basis and securing against displacement. Read your Part's paragraph for the marking language rather than carrying the construction wording across.

What counts as a hole is broader than people expect. 29 CFR 1926.500(b) defines a hole as a gap or void 2 in or more in its least dimension in a floor, roof or other walking surface. Two inches. A missing floor plate, a removed riser, a drain sump, a skylight opening and an unfinished duct penetration are all holes. The 24 in opening on that mezzanine cleared the definition twelve times over.

The load that crossed it

Weigh what was actually on the floor, then apply the standard's basis rather than the day's reality.

Tech in boots and gear, weighed              205 lb
Hand truck                                    30 lb
Compressor on the truck                      120 lb
                                             ------
Imposed on the day                           355 lb
Standard's basis: twice that                 710 lb

Now re-base it, because the paragraph does not say "twice what was on it," it says twice what may be imposed at any one time. Two techs work that mezzanine together on a change-out, and a second man steps in to steady a load as a matter of course.

Two techs plus truck plus compressor         560 lb
Required capacity, twice that              1,120 lb

That is a re-basing of the 710, not an addition to it. The 710 was computed on the wrong population and it stops existing the moment you correct the population. Adding them produces 1,830 lb, which is a number with no meaning, and it is the arithmetic mistake people make when a correction arrives late.

And the total-load figure only covers one of the two ways a panel gives up. A hand truck delivers its share of that load through two small wheels, so the panel is being asked to survive global bending across the 24 in span and local punching under a contact patch of a couple of square inches. The twice-the-weight arithmetic addresses the first. The second is a property of the specific panel and its rating. A wood panel's allowable span and allowable concentrated load come from its own span rating and the applicable panel design standard, and a sheet cut from a scrap pile carries no rating that anyone can read. If you cannot state where a cover's capacity number came from, the cover does not have one.

What actually failed

The panel came out of the opening intact. It was not broken, cracked or crushed. Bending capacity was never the mechanism, which is why the first instinct was wrong.

It was displacement. The panel overlapped the opening by about 4 in on each side. When the front wheel crossed just inside the opening edge, the panel pivoted about the floor bearing at that edge and the far side lifted.

Run the moments, treating the panel as rigid, which is the optimistic assumption and still fails:

Overturning: 355 lb applied about 3 in inside
  the near opening edge
  355 lb x 3 in                            1,065 lb-in

Restoring: the panel's own weight at its
  centroid, 12 in from that bearing line.
  A half-inch wood panel runs roughly
  1.5 lb per square foot, so a 32 in square
  sheet is about 11 lb.
  11 lb x 12 in                              132 lb-in

Ratio, overturning to restoring                8 to 1

Eight to one against. Nothing about the panel's thickness changes that, because thickness barely changes its weight and the restoring term is its weight. A heavier cover helps only in proportion to how much heavier it is, and you would need roughly eight times the panel weight to reach equilibrium with no margin at all.

Treating the panel as rigid also flatters it. A real sheet flexes, the near edge crushes into the floor, and the load line moves further inside the opening as it goes, which makes the overturning arm longer while the restoring arm stays put. The failure accelerates once it starts.

So the securing requirement is not a tidiness clause. For any cover a person or a wheel can approach off-centre, securing is the requirement that governs, and capacity is the one that gets all the attention. The standard names both because both are load paths, and this one failed on the one nobody checks.

What it was replaced with

A fabricated steel plate, sized and stated by the fabricator, with a written capacity on the work order rather than in someone's head. Cleats welded on the underside on all four sides so it cannot slide in any direction, plus two fasteners into the deck so it cannot be flipped or lifted by a wheel or by wind at an open bay door. HOLE stencilled on the top face in a colour that reads against a dusty floor.

Two decisions in that list are worth naming. Cleats resist sliding; fasteners resist lifting. They are different failure directions and one does not cover the other, which is why the paragraph says displacement by wind, equipment or employees. And the capacity is written down at the point the cover is made, because a capacity that lives in the fabricator's memory is not available to the tech who arrives eighteen months later.

The other decision was about what the cover was for. The original panel was a debris cover installed by a different trade to keep dust out of the shaft. It was never a personnel cover, nobody had ever assigned it a load, and there was no marking on it to prompt anyone to ask. That is the whole failure in one sentence: a cover that looks like floor and carries no marking transfers the entire safety decision to whoever walks on it next, who has no information at all.

How to verify you got this right

This is an audit walk, and it is the only reliable way to find these, because a cover in good condition is invisible.

Walk the building with a tape and a marker. Anything 2 in or more in its least dimension in a walking surface is a hole, whether or not anyone calls it one. Look specifically at removed risers, abandoned penetrations, drain sumps, floor boxes, hatch surrounds and skylights, and look under stacked material, because the reliable place to find an uncovered opening is beneath something someone parked on it.

For each covered opening, ask three questions, and do not step on anything to answer them:

1. What is its capacity and who set it? A name or a document, not an opinion. No answer means no cover.
2. How is it secured, and in which directions? Push it laterally with a foot from the floor beside it, not from on top of it. Try to lift an edge with a bar while standing clear of the opening. A cover that moves in either test is not secured.
3. Can a person who has never been in this building tell it is a cover? If the stencil has worn off or the colour has faded into the floor, the marking requirement has quietly lapsed even though the cover is sound.

Then check for the condition that produces most of these: a cover installed as temporary that has become permanent. Ask when it went in and what the plan was. A cover with a two-week story and a three-year age has outlived whoever knew what it was for, and its capacity is now folklore.

Where an answer comes back short, the immediate action is a barricade or a guardrail meeting the criteria in the related card, not a note on a list. A cover that is under review is an open hole for the purposes of everyone walking past it.

References

• 29 CFR 1926.502(i), covers on construction work: at least twice the weight of employees, equipment and materials imposed at any one time, secured against accidental displacement, and colour coded or marked with the word HOLE or COVER; twice the maximum axle load where vehicles cross.
• 29 CFR 1926.500(b), which defines a hole as a gap or void 2 in or more in its least dimension in a walking or working surface.
• 29 CFR 1910.28(b)(3) and 29 CFR 1910.29(e) for the general-industry hole protection and cover criteria.
• See related: universal-what-a-guardrail-has-to-withstand-and-where-it-has-to-be for the barricade that replaces a failed cover, and universal-what-fall-restraint-is-and-why-it-is-the-better-answer for why an uncovered skylight is an edge.