Fire Door Drop Test and Reset

Purpose

A rolling steel fire door spends its entire life doing nothing, and then has to close by gravity, at a controlled speed, on the worst day the building will ever have. The only way anyone knows it still will is to drop it on purpose, once a year, and write down what happened.

The failure this procedure prevents is not usually a door that refuses to drop. It is a door that drops correctly, gets reset wrong, and sits for another year looking perfectly normal while the release will no longer let go. That door passes every visual inspection, and the building only finds out during a fire. The second failure it prevents is a shop that drops a door it cannot reset, walks away, and leaves a rated opening either standing open or blocking an exit path.

Scope

Covers the annual drop test and reset of rolling steel and sliding fire door assemblies with automatic closing devices: pre-test condition check, the drop itself, closing speed verification, release reset, and the written record.

Does not cover routine service of non-rated rolling doors, which the commercial rolling steel door service SOP owns. Does not cover swinging fire door assemblies, fire dampers, or fire alarm and detection equipment - a fire alarm impairment in particular is not this shop's to declare. Does not cover repairs discovered during the test: field modification of a rated assembly, including welding, drilling or substituting hardware, voids the listing and is not performed on findings from this procedure.

Roles and responsibilities

Procedure

1. Confirm scope and clear the opening before touching anything. Read the listing label on the guide, hood or door, and identify the release type: fusible link in a cable run, a resettable release device, or an electrically held release tied to detection. Acceptance: label found and legible, release type identified, the full opening and both sides of the door clear of people, product, racking and vehicles, and the facility contact present and aware the door is about to close. Wrong looks like a pallet inside the opening or a forklift route nobody stopped. Stop rule: no legible label means the assembly's rating cannot be confirmed, and that is a finding to report, not a reason to skip the test. Hazard: this door is designed to close under its own weight with nothing holding it back, so the opening is treated as a live drop zone from this step until the door is reset and proven held.

2. Inspect the assembly against the standard's condition items before the drop. Check the curtain and slats, guides and their fastening, bottom bar, hood, release mechanism and cable run, and confirm no field modification, no painted-over label, no added auxiliary hardware, and no storage that would prevent full closure. Acceptance: nothing obstructing the closing path, guides and brackets sound and correctly fastened, no unlisted modification. Wrong looks like a chain and padlock through the guides, conduit run through the closing path, or a bottom bar drilled for a lock. Stop rule: an unlisted field modification stops the test and is reported as a written deficiency, because dropping a modified assembly proves nothing about a listed one.

3. Verify the drop can be initiated from outside the closing path. Confirm the release point, the step-clear path, and that the second person stands outside the path with a clear view. Acceptance: whoever initiates can reach the release and step clear without crossing under the curtain. Wrong looks like releasing a fusible link from a ladder set inside the door path. Stop rule: if the release is only reachable from inside the closing path, reposition the access equipment first; nobody initiates from under the curtain.

4. Drop the door and time the close. Initiate the release per the manufacturer's method for that device, and time the curtain from start of motion to fully closed against the measured opening height. Acceptance: the door closes fully and unassisted, and the average closing speed falls inside the range the applicable NFPA 80 edition specifies for rolling doors, commonly cited as not less than 6 in per second and not more than 24 in per second - confirm the figures against the edition your authority having jurisdiction has adopted rather than from memory. Wrong looks like a door that stalls part way, closes in fits, or slams because the governor is not controlling descent. Stop rule: a door that does not close fully, or that closes outside the speed range, is a failed test - the assembly is impaired from that moment, and step 8's impairment path runs. Hazard: nobody stands in the opening during the drop and nobody attempts to slow, catch or guide a descending curtain; a rolling steel curtain closing under gravity is several hundred pounds with nothing to stop it.

5. Reset the release exactly as its manufacturer specifies, and replace what is not resettable. A fusible link that has separated is consumed and gets replaced with the same listed link, matched by temperature rating and part number - it is never re-hung, spliced, or substituted. A resettable release is reset per its own instruction. Acceptance: the release is in its set position, the cable run has correct tension with no slack loops and no binding at pulleys, and the link temperature rating matches what the assembly was listed with. Wrong looks like a link of the wrong rating because it was the one on the truck, or a cable reeved back through a pulley with a twist. Stop rule: the correct link or release part is not available means the test does not get closed out today, and the door goes onto the impairment path rather than being left reset "close enough".

6. Repeat the drop test to prove the reset. Run the release and the drop again after the reset. Acceptance: the door releases on the second attempt and closes fully and within the same speed range. Wrong looks like a door that will not release the second time, which is exactly the condition this whole procedure exists to catch, because it is invisible from the ground. Stop rule: a failed second drop is a failed test regardless of how clean the first one was - do not record the first result as the outcome.

7. Return the door to normal operating condition and prove it is held. Raise the door on its normal means of operation, confirm the release holds the curtain up, and confirm the automatic closing device is armed rather than merely quiet. Acceptance: the door sits fully open, held by the release, and the operator or hand chain runs the door normally without disturbing the release setting. Wrong looks like a curtain held up by a clamp, a strap, or the operator's brake while the release sits unset. Stop rule: a curtain that is not held by its own release is not returned to service, it is secured and declared impaired.

8. Record the test, and if it failed, put the impairment in writing before you leave. Acceptance: a completed written test record with every field from the section below, signed by the technician and countersigned by the facility contact, delivered the same day. Where the test failed, the record names the deficiency, the date and time the impairment started, the compensating measure the owner is taking, and who owns getting the assembly back in service. Wrong looks like a verbal "it needs a part" and a record that follows next month. Stop rule: an impaired rated opening is never left without the owner's written acknowledgement, because the duty to maintain the assembly sits with the owner and the shop's record is what tells them it is on them now.

The record this produces

A written inspection and test record per assembly, retained by the building owner and available to the authority having jurisdiction, with a copy in the shop's equipment record:

• Assembly identity: building, opening designation, door make and serial, listing label information, and rating.
• Release type, and the fusible link temperature rating found and the rating installed.
• Pre-test condition findings from step 2, including any field modification or obstruction.
• First drop: released yes or no, closed fully yes or no, measured travel time against opening height and the resulting closing speed.
• Second drop after reset: same fields.
• Deficiencies found, parts replaced, and whether the assembly was returned to service or left impaired.
• Date, technician name and qualification basis, and the facility contact's signature.

The authority having jurisdiction reads this record on inspection and its absence is the citation. Next year's technician reads the link rating and the prior closing speed, because a door trending toward the slow end across three annual tests is a governor or guide problem developing in plain sight. The owner reads the impairment line as their own action item.

Worked pass: two rolling steel fire doors, one warehouse, annual cycle

Opening 3, 10 ft wide by 12 ft high, fusible link release. Step 1 found a legible label and a link in a cable run across the header. Step 2 found the closing path clear and no modification. Step 4 dropped clean: the curtain released, closed fully, and covered the 12 ft opening in about 11 seconds, which is 144 in over 11 s, roughly 13 in per second, inside the 6 to 24 in per second range from the adopted edition. Step 5 replaced the separated link with a matched listed link of the same temperature rating. Step 6's second drop released and closed fully in about the same time. Step 7 raised the door and confirmed the release held it; step 8 recorded both drops and the link change.

Opening 7, 12 ft wide by 14 ft high, fusible link release. Step 1 and step 2 passed on label and condition, though the record noted the guide fasteners were newer than the door. Step 4 failed. The curtain released but stalled with roughly the bottom 2 ft still open, and it never reached the floor. Under step 4's stop rule the assembly was impaired from that moment, and the technician did not attempt to pull the curtain down.

The cause was at the guides: the newer fasteners had been driven with heads standing proud into the guide throat near the floor, and the bottom bar was hanging up on them. That is the unlisted field modification step 2 exists to catch, and it was missed because the fasteners read as maintenance rather than modification. Reworking a rated assembly is out of scope, so no correction was attempted on the spot.

Step 8 ran the impairment path: the deficiency, the start time of the impairment, and the owner's compensating measure were written on the record and countersigned by the facility contact that afternoon, and the manufacturer was engaged for a listed guide repair. The honest reading of this pass is that opening 7 had almost certainly been in that condition for months, and the only reason anyone knows is that somebody dropped it. The visual inspection alone would have passed it again.

References

• NFPA 80, Standard for Fire Doors and Other Opening Protectives, in the edition adopted by your authority having jurisdiction, which is what makes it enforceable on the building owner and reaches your shop through that owner's maintenance obligation. Take the annual test requirement, the closing speed range and the record contents from the adopted edition in hand.
• NFPA 72, National Fire Alarm and Signaling Code, in the adopted edition, which is where detection and alarm impairments belong; a fire alarm impairment is not filed under the door standard.
• Door and release device manufacturer's instructions, the authority for the reset method in step 5 and the listed link part number.
• See related: the commercial rolling steel door service SOP for non-rated doors, and the rule that any door found with a label or release hardware during that procedure routes here.