What a Guard Is Actually Protecting Against

Why this matters

Almost every guard audit performed in the field is a presence check: is it there, is it fastened, is it undamaged. That check passes on the guards that hurt people. A guard is engineered against a specific hazard mechanism approached from a specific direction at a specific distance, and all three of those can change while the guard itself stays exactly as it was. The most common way a shop ends up with an unsafe machine is not a missing guard. It is a change to the machine that quietly moved the hazard while the guard stayed put, and a subsequent inspection that recorded the guard as present.

The action, before anything else

A guard comes off on a stopped, isolated machine only. Lock and tag the energy-isolating device under 29 CFR 1910.147, then watch the wheel, sheave or rotor until it has completely stopped, because the stored rotational energy in a coasting drive is exactly the mechanical hazard that standard covers and it outlasts the sound of the motor by a long way. Nothing enters an opening in a guard while the machine turns, at any speed, for any reason. If a reading has to be taken with the machine running, take it with a non-contact optical instrument aimed through an existing port from outside the plane of rotation, never with a contact instrument on a turning shaft and never with a hand inside an opening.

The mechanisms guards are built against

Naming the mechanism is what turns a presence check into a real one, because each mechanism has its own geometry.

• In-running nip point. Two surfaces converging and moving into each other draw in whatever meets them. On a belt drive the nip is where the belt runs onto the sheave, not where it leaves; on a chain drive it is where the chain meets the sprocket. Nips are directional, and reversing a drive's rotation relocates every one of them.
• Entanglement on a revolving part. A shaft, a coupling, a key, a set screw, a bushing flange, a projecting bolt. Smooth shafts entangle clothing and gloves; projecting keys and set screws do it far faster. 29 CFR 1910.219 addresses revolving collars and shaft couplings specifically, requiring that bolts, nuts and set screws not present a projecting hazard on revolving surfaces.
• Shear and trap points. A moving part passing close to a fixed one: a sheave rim near a bracket, a coupling near a base rail, a belt near a guard's own edge.
• Ejection. A failing belt, a thrown blade, a fractured coupling element, a burst wheel. This is the mechanism most often assumed and least often designed for.
• Contact with a hot or energized surface. Some barriers are thermal or electrical rather than mechanical, and they are rated for that and nothing else.
• Reach geometry. An opening in a guard is compliant as a function of its size and its distance to the hazard behind it, and both terms matter.

What a guard does not do

This is where the field errors live, so it is worth being blunt about each one.

It does not stop the machine. A guard is a barrier. Unless it is an interlocked barrier whose switch removes power when it opens, and unless that interlock has been verified and cannot be defeated with an ordinary tool, opening it changes nothing about the machine's state. A panel that comes off with a nut driver is not a control, and a machine under automatic control can start while it is off.

It does not contain a failure unless it was designed to. Expanded metal, perforated sheet and woven wire belt guards are access guards. They keep a hand out. They will not stop a parting belt, a thrown blade or a burst wheel, and the fix for a machine whose failure risk has risen is not a heavier guard, it is not running the machine in that condition. Containment is a rated design decision, not a material choice made on site.

It does not protect anyone inside its own opening. The safe-distance relationship between an opening's size and the distance from that opening to the hazard is tabulated in ANSI B11.19 and ISO 13857, and both terms are in the table. An opening that was compliant at one distance is not compliant at a shorter one, and the difference can be small enough that nobody notices it happened.

It does not travel with the hazard. Move a sheave, change a shaft extension, reverse a rotation, add a component, and the guard's geometry is now stale even though the guard is untouched.

It does not cover a hazard that was added after it was designed. A tension-check door, a grease line penetration, a sensor bracket, a tachometer slot: each one is a new opening or a new part, and each one needs its own answer.

Where the duty lives

29 CFR 1910.212 carries the general machine-guarding requirement for general industry, covering guarding methods, the point of operation, ingoing nip points, rotating parts and flying chips and sparks. 29 CFR 1910.219 is the specific standard for mechanical power-transmission apparatus - belts, pulleys, sheaves, shafting, collars and couplings - and it keys many of its requirements to whether the component sits seven feet or less above the floor or working platform, which is why a drive at head height and the same drive on a mezzanine can carry different obligations.

For a field-service shop working on a construction site rather than in an operating facility, the equivalent duty sits in 29 CFR 1926, Subpart I, at 1926.300(b), guarding of moving parts including belts, pulleys, chains and flywheels. Which Part you fall under depends on the character of the work, not on the machine, and it is worth settling before an inspection settles it for you.

Worked example: the guard that passed every check and stopped protecting

A belt-driven fan in a mechanical room. The guard is fabricated expanded metal with a hinged access door for belt tension checks and a small slotted opening through which a non-contact tachometer is aimed at the fan shaft during service. It has been in place for years and has passed every inspection.

The change. A sheave change was made to raise airflow: the driven sheave on the fan shaft went to a smaller pitch diameter, and to take up the resulting slack the motor was moved on its slide base, shortening the center distance. All of that was done on a locked-out drive with the motor supported before its base bolts were loosened, and the belts were re-tensioned to the manufacturer's force-deflection figure with the drive stopped and locked.

What the presence check reported afterwards: guard present, fastened, undamaged, opening unchanged. Pass.

What the mechanism check reported. Three findings, none of which involve the guard being different in any way.

1. The nip moved. Shortening the center distance and shrinking the driven sheave moved the point where the belt runs onto the driver sheave roughly 2 inches closer to the slotted tachometer opening. That distance is one of the two arguments in the safe-distance table, and the opening's compliance was established against the old distance. Nothing about the opening changed; the answer it produces did.
2. A new entanglement point appeared. The replacement sheave is mounted on a tapered bushing whose set screws project beyond the hub face, sitting near the same opening. Revolving projecting hardware is a named hazard in 29 CFR 1910.219, and it did not exist on this machine the day the guard was designed.
3. The failure risk rose while the containment did not. The smaller driven sheave raised the effective belt tension substantially, and it may sit below the belt section's minimum recommended diameter, both of which raise the probability of a belt parting in service. The guard is expanded metal. It is an access guard. It was an access guard before the change too, but before the change nobody was relying on it for anything else.

What was actually done. The set screws were replaced with hardware that does not project beyond the hub, which is the cheap fix and closes finding two outright. The tachometer slot was closed permanently and readings moved to a shaft-end reflective target visible through the existing access door with the door latched, which removes the opening rather than trying to re-qualify its distance. Finding three was not solved by the guard at all: the drive ratio was reconsidered against the belt section's minimum diameter, because a drive that is more likely to throw a belt is a drive problem, and adding steel around it is treating the wrong thing.

The failure mode this avoids. The version where nobody runs the mechanism check is not dramatic. The guard stays on, the inspections keep passing, and eighteen months later somebody puts a tachometer through the slot the way they always have and finds the nip two inches closer than their muscle memory expects. Nothing in the record would have predicted it, because every entry in the record said the guard was present.

What changes the call

An interlocked barrier changes the analysis but not the isolation. If the guard's switch reliably removes power and cannot be defeated with an ordinary tool, opening it is a different act from opening a bolted panel. It is still not a substitute for lock and tag when a hand goes into the machine, because an interlock is a control-circuit function and the isolation requirement is about the energy source.

A machine under automatic control raises the stakes on every opening, because the guard is the only thing between a person and a start command nobody in the room issued. On those machines, an access door used routinely for service is the single item most worth converting to an interlocked design or eliminating entirely.

A guard on a reversible drive is a different guard. Reversing rotation swaps which side of every sheave and sprocket is the in-running nip, so a guard designed around one direction of travel has to be re-evaluated as if it were a new machine.

How to verify you got this right

• For each guard, write down the mechanism it addresses, in words, before you record its condition. "Nip at the driver sheave, entanglement at the shaft end and bushing hardware" is an audit entry. "Present" is not.
• Check the openings against distance, not just size. Both arguments go in the safe-distance table, and only one of them is visible from across the room.
• Trigger a guard review off every mechanical change, not off the calendar. Sheave swaps, motor moves, coupling changes, rotation reversals and added instrumentation are the events that invalidate geometry, and none of them appears on an annual schedule.
• Ask what the guard would do in a failure and answer honestly. If the honest answer is that a parting belt goes through it, then the guard is not the mitigation and something upstream needs to change.
• Verify any interlock by function, opening the barrier and confirming the machine cannot be started, rather than by the presence of a switch.

References

• 29 CFR 1910.212, general requirements for all machines, and 29 CFR 1910.219, mechanical power-transmission apparatus, for general-industry guarding duties including revolving collars, couplings and projecting hardware
• 29 CFR 1926, Subpart I, at 1926.300(b), guarding of moving parts, for the construction counterpart
• 29 CFR 1910.147, control of hazardous energy, for isolation and stored rotational energy before any guard is removed
• ANSI B11.19 and ISO 13857 for safe-distance relationships between guard openings and the hazard behind them
• See related: How a Belt Drive Transmits Torque; How a Fan Law Changes What the Drive Has to Deliver; How to Return a Machine to Service and Verify It