Edge Protection and Fall Distance

The drop is a budget, and it is already being spent

A harness does not stop a fall. It ends one, some distance below where it began, and that distance was fixed by decisions taken long before anybody stepped over the edge. Working a leading edge, a scaffold return, a roof perimeter or an unformed deck penetration, one number matters more than any other: how much air sits between the attachment point on a worker's back and the first hard thing underneath. Every component in the system is a claim on that number.

Six line items spend it. Free fall, before the energy absorber does any work at all. Deceleration, while the absorber tears out. Harness stretch and the dorsal D-ring riding up the back. The length of the person hanging below that D-ring. Movement in the anchor and in whatever the anchor is fixed to. Then a margin at the bottom, which exists so the sum can be slightly wrong without being fatal. Add the six, compare against the measured drop, and the answer is either yes or a different system.

Ownership of those line items is scattered across the job. Procurement chose the lanyard, the rigger chose the anchor height, the temporary works designer signed the beam, and the groundworks crew left the surface the scaffold legs now stand on. None of them see the total. The operative clipping on does, usually with a poor view of the drop and no tape in his pocket.

Permitted free fall and required margin vary by jurisdiction. United States construction work sits under OSHA 29 CFR 1926 Subpart M, Fall Protection, with equipment requirements in the ANSI/ASSP Z359 series. European component standards include EN 355 for energy absorbers, EN 360 for retractable type fall arresters, EN 795 for anchor devices and EN 13374 for temporary edge protection systems. Australia and New Zealand work to AS/NZS 1891.4 for the selection, use and maintenance of industrial fall arrest systems. The arithmetic behaves identically everywhere; the values you are permitted to substitute into it do not, and the regulation governing the site decides which apply.

Free fall: the term you set with a tape

Free fall is the distance travelled before the absorber begins doing anything, and it is set with a tape rather than a calculator. Anchor overhead, connector short, no slack in the line: the term approaches zero. Anchor down at foot level with a two metre lanyard: the worker falls the length of the lanyard to reach the anchor, then the same length again before the pack opens. Two decisions about where to clip, separated by a couple of metres of steel, produce fall distances that differ by a factor of four.

Slack hides in the connectors. A beam anchor sling, an oversized rebar hook, an extra carabiner added because the first one would not close on the ring, a rope grab that has to travel before it bites, a lanyard clipped back to itself around a member — each contributes real centimetres, and they accumulate in one direction only. Crews rarely measure them, and the assembled system is routinely longer than the length printed on the lanyard.

Components are qualified for a stated free fall. Standard energy-absorbing lanyards are generally rated for a free fall no greater than about 1.8 m, with equipment marked for longer free fall sold as a separate product class. Exceeding the rating does two things at once: the absorber deploys further than the figure in your arithmetic, and the arrest force may climb beyond what the harness and anchor were qualified against.

What the absorber takes back

Absorber deployment is the only line item deliberately designed to consume distance. Its job is to convert a short violent stop into a longer, survivable one, capping the arrest force transmitted into the body and the anchor. That cap is a regulated number, and the price paid for it is stretch: the pack tears, elongates and adds length to the system exactly when the worker can least afford it.

Deployment length is not a single figure. It varies by standard, by the rated free fall of the product, and by the mass being arrested; a heavier operative with a full tool pouch deploys more absorber than a light one. The label carries the rated capacity range and the maximum deployment figure that belongs in your sum, and catalogue memory does not.

Self-retracting devices change the shape of the problem rather than eliminating it. They lock quickly, so free fall is small, but arrest distance still exists, and devices qualified for use over an edge behave differently again because the line drags across the edge before and during arrest. A device marked for leading-edge use carries its own arrest distance and its own minimum clearance, sometimes larger than a crew expects from something that looks like an inertia reel. Anchoring one below the D-ring, which is the whole point of a leading-edge unit, also reintroduces free fall that an overhead installation would not have had.

The body's share, and the person below the D-ring

Two terms come from the human body and neither is negotiable. Harness stretch and dorsal D-ring shift together typically account for around 0.3 m; webbing extends under load and the ring travels up between the shoulder blades toward the neck. Below that ring hangs roughly a metre and a half of person, boots included, and boots are what the arithmetic is protecting.

Nobody stops at exactly the calculated depth, which is why a margin sits at the bottom of the sum. Common practice puts it somewhere between about 0.6 m and 1 m, with the governing regulation and the equipment manufacturer's instructions setting the number for a given site. That margin absorbs measurement error, a slightly heavier user, a slightly longer connector than the one assumed, and the ordinary imprecision of a tape held over a windy edge.

Below the margin sits whatever the site put there this morning. A drop measured to a clean slab on Monday can be measured to stacked kerbs, a skip, a formwork table or a scaffold lift on Wednesday. The arithmetic does not go stale on its own — the site invalidates it, usually without telling anyone working above.

Anchors move, and the movement is distance

An anchor that moves spends clearance as surely as an absorber that deploys. A horizontal lifeline sags at midspan under arrest load, and that sag drops the worker further while multiplying tension into the end anchors; a long span with few intermediate supports can add more distance than the absorber does. Manufacturers of flexible line anchor systems declare a deflection figure for a given span and number of users, and that declared figure, not an assumption, belongs in the sum.

Rigid structure deflects too. An anchor bolted to a purlin, a light beam, a metal deck or a proprietary roof anchor plate loads that member with an arrest force in the kilonewton range applied in a fraction of a second. Roof anchors fixed into decking, insulation or corroded sheet can pull, tilt or dish the substrate before they hold, and each millimetre of that movement is distance the worker travelled. Anchor device standards such as EN 795 classify devices by type and require the supporting structure to be verified separately — the anchor being certified says nothing about the purlin being adequate.

Sharing matters too. A second worker arriving on a line rated for one changes both the deflection and the load path, which is why anchor selection is a temporary works decision with a designer's name against it.

The ground under the legs

Everything measured from a scaffold platform assumes the scaffold has not moved. Standards bear onto base plates, base plates bear onto sole boards or onto whatever the ground crew left, and the load path ends in soil whose bearing capacity nobody on the erection gang has checked. Soft made ground, a backfilled service trench under one leg, a saturated verge after a week of rain, or a plate resting half on slab and half on fill will settle. Uniform settlement lowers the whole platform and quietly shortens the drop, which sounds harmless until the guardrail geometry, the tie loads and the standard's plumb all change with it.

Differential settlement is the dangerous version. One leg drops, the frame racks, the platform slopes, the toe board opens a gap and the anchor mounted to a standard is no longer where the designer put it. A scaffold that has moved is a scaffold whose edge protection and whose anchor points both need re-checking before anyone clips on, and the movement is usually visible first at the base plate rather than at the platform.

Checks at the base are quick and worth doing every time the perimeter changes. Plates fully seated with no rocking, no packing with brick, offcuts or timber wedges, sole boards spanning across soft spots rather than sitting inside them, and a look at whether groundworks have excavated near a leg since the last handover. Where the ground is questionable, the bearing pressure under the plate is the number to calculate rather than eyeball.

Swing fall: a horizontal error with a vertical bill

Swing fall is a horizontal error that arrives as a vertical bill. Work well to one side of the anchor and the fall becomes a pendulum: the worker drops, then swings, and the lowest point of that arc sits below the point at which the system arrested him. Required clearance rises with the offset, and the arithmetic that closed comfortably directly under the anchor may not close at all thirty degrees away from it.

Impact is the second cost. A swinging body travels fast and horizontally into columns, formwork, stacked material, scaffold standards and the face of the structure itself, and the injury from that strike does not care that the harness held. On a leading edge the line also drags sideways across the edge under load, which is precisely the condition that leading-edge rated devices are tested for and standard ones are not.

Limiting offset is a planning job, not a discipline problem. Move the anchor, add anchors, or run a horizontal line with a travelling trolley so the attachment tracks the work. Telling a crew to stay within a cone while the work plainly extends outside it produces exactly the shortcut everyone predicted.

Spending nothing: the passive edge

The cheapest way to win the arithmetic is to make the total zero. A compliant guardrail, a secured cover over a penetration or a properly installed net removes the fall rather than managing it, and no clearance calculation is needed for a fall that never starts. Temporary edge protection systems in Europe are classified by EN 13374 according to the loading and the slope they are intended for, and United States guardrail geometry and load requirements sit within OSHA 29 CFR 1926 Subpart M. Class, height, infill and load path all have to match the actual edge rather than the one in the standard detail.

Passive protection still deflects, which is the part crews forget. A rail loaded by a body, a wheelbarrow or plant movement moves outward before it holds, and how far it moves is governed by post spacing, post fixing or embedment, and rail stiffness. Barriers along a highway works interface or a site perimeter behave the same way: increase the spacing between posts and the system becomes more flexible, deflecting further under the same impact. A barrier that deflects past the edge it is protecting has spent distance it never had.

Detailing decides whether the edge stays at zero. Toe boards to stop material becoming a falling object, no unguarded gaps at returns and terminations, and a written rule about who is allowed to remove a panel for a load-in and who signs it back. Most guardrail incidents involve a section that was legitimately removed and never reinstated, not a section that failed under load.

When the sum does not close

Sometimes the sum does not close, and the honest response is to change the system rather than the assumptions. Raising the anchor overhead is the single most effective move available, because it collapses the free fall term and often the swing term at the same time. Swapping an energy-absorbing lanyard for a self-retracting device anchored overhead does the same thing with less deployment. Both are cheaper than discovering the shortfall at the bottom.

Restraint beats arrest wherever geometry permits it. Configure the anchor position and lanyard length so the D-ring physically cannot reach the edge, measure that distance on site rather than assuming it, and remember that restraint is only restraint while the anchor stays where you left it and the lanyard stays the length you specified. A restraint system that reaches the edge is an arrest system with no clearance behind it.

Where neither works, the answer moves up the hierarchy: a mobile elevating work platform, a proper working platform with edge protection, a net installed close beneath the work, or a resequenced task done from the deck below before the edge exists. Rescue planning belongs to the same decision, because tolerance for suspension is limited and a successful arrest still ends with someone hanging.

Re-walking the arithmetic every shift

Numbers written on Monday describe Monday's site. The drop below an edge changes as decks are struck, material lands, scaffolds rise, skips arrive and excavations open, and each of those changes rewrites a line in the arithmetic without anybody being told. Re-measuring the drop is a two minute job at the start of a shift and belongs in the same conversation as the permit, not in a file.

Record what was actually assumed rather than a generic figure. The anchor used, its height relative to the D-ring, the assembled connector length, the absorber's rated deployment from the label, the declared anchor deflection and the measured drop give a chain anyone can audit later. Entries reading "harness and lanyard" describe nothing and defend nobody.

Near misses in this trade read deceptively well. The system worked, the worker was arrested, everybody went home — and the boots stopped a hand's width above a stack of scaffold boards nobody had included in the sum. Treat that as the failure it is, because the margin that saved it was the last line item in the budget, and it was never meant to be spent.