Building a Rated Shaftwall

The face you will never touch again

A shaft enclosure is the only rated wall on a typical building where one face is permanently out of reach. Elevator hoistways, mechanical risers, trash chutes, stair pressurisation ducts — once the slab edge is formed and the void is open, nobody is standing inside it with a screw gun. Every component has to be positioned, captured and fastened from the occupied side, and it has to stay where you left it for the life of the building.

That restriction is not a detail hanging off the assembly. It is the reason the assembly exists in the form it does. An ordinary rated partition gets board on both faces because a crew can walk either side of it. A shaftwall substitutes a thick gypsum liner, captured in the web of a purpose-rolled stud, for the far-side layer nobody can reach. The stud is open on one flank so the liner can be fed in edgewise; the face layers of Type X board then go on the corridor side and do the rest of the rating work.

Everything downstream inherits the same limitation. Layout cannot be corrected by reaching around. Bracing cannot be added behind. A missed backing plate cannot be retrofitted. Sequence stops being a preference and becomes a constraint, and the sections below are ordered the way that constraint actually propagates through the work.

Layout has no second pass

Layout carries more weight here than on any other partition because there is no second pass. The controlling dimension is almost never the wall line on the architectural plan — it is the clear inside dimension of the shaft, driven by elevator running clearances, duct outside dimensions plus insulation, or the chute manufacturer's required void. Set the J-track to that clear face and let the tolerance fall on the corridor side, where a skim of mud or a shim can still absorb it.

Plumb transfer is the second discipline. A hoistway has to hold its clear dimension for the full rise because car guide rails, door sills and toe guards all reference it, and the elevator contractor will survey the shaft before setting rails. Pull each floor's layout from the same building control points rather than shooting up from the wall below; stacked error is how a fifteen-storey shaft ends up out of tolerance at level twelve with all the liner already installed.

Orientation of the runner matters for the same reason. The J-runner has a long leg and a short leg, and the long leg belongs on the shaft side, because that is the leg you can never reach again to straighten, close or re-fasten. Fix the track before the fireproofing crew and the deck-flute closure work arrive, and confirm the head condition — the head runner has to receive the deflection movement the structural drawings call for, not sit tight to the underside.

Liner goes in blind, and it goes in first

Liner installation is the point where the one-sided rule becomes physical. The panels are nominally one inch thick, commonly twenty-four inches wide, with bevelled long edges and a fire-rated core, and they are cut to a length short of the structural clear height so the head can move. They are set into the floor J-runner, tipped up, entered into the head runner, then slid sideways into the groove of the stud already standing.

Note what does not happen: in most tested designs the liner takes no fastener at all. It is captured — bottom runner, head runner, and the two stud webs either side of it. That is a good system and a fragile one. A panel that has not fully seated into the groove looks identical from the corridor once the face board goes on, and the first thing anyone learns about it is a failed inspection or, worse, a fire test the wall never had to pass on paper.

Handling costs real time. A one-inch liner panel is heavy for its size, wants two people above about ten feet, and snaps along the long edge if it is walked flat. Store it dry and off the slab; gypsum liner that has taken standing water is scrap, and it will not be obvious until it is already in the wall. Cut with a knife and snap where possible, and keep field cuts to the top, where the bevel is not doing anything.

Quantity is where estimators get caught out. The panel count is not shaft perimeter divided by panel width. Openings consume partial panels and generate offcuts that rarely find a second home, closures at the end of a run take a cut piece by definition, and every corner condition eats a panel edge. Count the run, then count the interruptions.

1. Set and fix floor and head J-runner to the shaft-side layout line, long leg toward the shaft.
2. Stand and fix the first stud at the start of the run, working from a jamb, corner or abutment.
3. Enter the liner panel bottom into the floor runner, tip it up into the head runner, slide it into the standing stud groove.
4. Set the next stud over the free panel edge and fix it top and bottom.
5. Repeat in one direction only, and stop at the planned closure — never leave the last panel to be inserted between two fixed studs.

Studs follow the panels, not a spacing habit

Studs follow the panels rather than leading them, which reverses the habit of every other framing task on the job. In a shaftwall the C-H, C-T or I-stud is not laid out to a nominal spacing and then sheeted; the spacing is set by the liner width, because each stud has to land on a panel joint and receive both panel edges. Get that relationship wrong at the first stud and the error walks the whole run.

Beyond the field studs there is a whole population of members that has nothing to do with spacing arithmetic. End studs where the wall meets structure or an abutting partition. Corner assemblies, which usually take a different member and often a back-up angle. Jamb studs at hoistway door openings, frequently doubled and frequently a heavier gauge than the field. Closure studs where a run terminates. None of these can be slipped in afterwards, because inserting a stud between two captured liner panels means pulling the panels back out.

Gauge is a separate decision from spacing, and it is driven by height and load rather than by convenience. Nonstructural steel framing members should meet ASTM C645, and installation follows ASTM C754; the tested fire-resistance design will also name a minimum member, and that minimum is a floor, not a target. Substituting a lighter stud to make a delivery date changes the assembly the rating was issued against.

Height, with nothing behind you to brace against

Height is the constraint that quietly kills shaftwall jobs. Shafts run tall by nature — full building rise for a hoistway, several floors at a stretch for a riser — and the wall has no far face to brace against. There is no blocking, no strongback, no cross-bracing, no opportunity to add any of it later. The stud and the liner do the entire job across the unsupported span.

Transverse pressure is real and it is not static. Stack effect loads a shaft continuously in tall buildings, and a moving elevator car generates a piston pressure that reverses as it passes. The design pressure for the assembly comes from the project documents; interior partition minimums are set by the governing building code, and designers routinely specify above that minimum for hoistways. Whichever value applies, it has to be checked against the system manufacturer's limiting height tables for the specific stud, spacing and liner arrangement — not against a general partition height chart.

When the required height beats the table, the honest fixes are all decided before framing starts: a heavier or deeper stud, tighter spacing, a second liner, or intermediate horizontal support tied back to structure. Discovering the shortfall after the liner is in means demolition, because the remedy always lives on the shaft side.

The head of wall deserves its own attention. Shaft walls terminate at structure that deflects, and the joint has to accommodate that movement while maintaining the rating, using a joint system tested to ASTM E1966 or UL 2079. The commonest field error is simple: someone screws the face layer, or the stud, through the deflection track legs and turns a slip joint into a rigid one. The wall then cracks along the head on the first load cycle and the fire-resistive joint no longer matches anything tested.

Every screw lands on one flange

Fastening is a one-sided proposition, and the detail library reflects it. Type X face layers screw to the corridor flange of the stud at the spacing named in the tested design, with application and finishing per ASTM C840 and the board itself meeting ASTM C1396/C1396M. Multi-layer build-ups stagger their joints so no joint stacks through the thickness, and the base layer takes a different screw length than the face layer.

Screw depth is a quality issue with a fire consequence. A screw driven through the paper face has lost its holding power and its contribution to the assembly, and on a two- or three-hour wall the face layer is what holds the base layer in place once the gypsum starts calcining. Set the gun clutch, check it after every board change, and back out and re-drive rather than adding a second screw beside a broken one.

Conditions that would normally be solved with a fastener from behind need a different answer here. Outside corners, closures against existing construction, and abutments to columns are all detailed with clip angles, back-to-back studs or continuous angle fixed from the accessible side. Anywhere the standard detail assumes two-sided access, stop and find the manufacturer's one-sided variant rather than improvising, because an improvised attachment is outside the tested design.

Anything that lands on the shaft face goes in now

Penetrations have to be solved before the wall closes, not after, and that inverts the usual trade sequence. Duct risers, sleeves, sprinkler drops where the hoistway requires them, hall call and lantern boxes, hoistway door frames and sills, and any backing plate that lands on the shaft face — all of it either goes in as the liner rises or it does not go in at all without demolition. Walk the shaft elevation with the mechanical, electrical and elevator contractors while the framing is still open.

Elevator work carries its own governing document. What may project into a hoistway, how door frames are supported, and what surfaces are required inside the shaft are matters for ASME A17.1/CSA B44, and the local jurisdiction's adopted edition is what counts. Guide rail brackets are the recurring flashpoint: they load into structure, never into the shaftwall, and their embeds or clip angles have to be coordinated with the concrete or steel package long before framing.

Firestopping closes the loop. Through-penetrations are sealed with a system tested to ASTM E814 or UL 1479 for the specific wall type, penetrant and annular space, and the tested system is specific enough that a substitution is a new submittal. Membrane penetrations on the corridor face — boxes, hangers, brackets — carry their own limits in the design listing. Photograph every penetration and every firestop before the face layer covers it, because that photograph is the only evidence anyone will ever have.

Runs travel one way, and the far end is a detail

Runs proceed in one direction, and the far end of the run is a detail, not an afterthought. Because each liner panel enters sideways into a standing stud, a run builds like a zip: stud, panel, stud, panel. The last panel in a run cannot be entered the same way once both bounding studs are fixed, so the manufacturer's closure detail applies — typically a cut panel and a closure member installed in a specific order, sometimes with an attachment made through the face.

Openings interrupt the zip entirely. A hoistway door opening breaks the run into two, and each side of it restarts as a new run with its own jamb condition, its own header or lintel arrangement, and its own closure. Plan the direction of travel so that closures land somewhere sensible — beside a jamb, at a corner — and not in the middle of a clean stretch where the detail is awkward and the inspector's eye goes first.

Corners repeat the same problem at ninety degrees. Inside and outside corners each have a defined assembly, and both are places where a crew reaching for a normal partition detail will build something that has never been tested. Where two crews work toward each other from opposite ends of a shaft, agree the meeting point in advance and make it a closure, because two zips arriving face to face leave a gap that nothing in the system is designed to fill.

Sign it off before the shaft disappears

Closeout is inspection of something that is about to become permanently unreachable, and the window is narrow. Continuity to the underside of the deck is the first item — the rating is worthless if the wall stops at a ceiling line, and flute closure at metal deck has to match the tested design. Walk the wall above the ceiling grid before the hard lid or the tile goes in, with a light, looking for unseated liner, missing closures, unfirestopped penetrations and screws that have gone through the paper.

Identification is a code requirement in most jurisdictions and a practical one everywhere. Fire and smoke barriers are marked with permanent stencilling above the ceiling stating what the wall is and the rating it carries, in the language the adopted building code specifies. That marking is what stops a future contractor from cutting a hole through your shaftwall for a cable tray in five years.

Where the jurisdiction requires special inspection of fire-resistant construction, book it while the assembly is still open on the corridor side, and keep the tested design listing on site with the design number that the submittal was approved against. The fire-resistance rating comes from a full-scale test to ASTM E119 or UL 263 of a specific assembly, and an inspector's only tool for judging your wall is whether what is standing matches what was tested.

Then the elevator contractor sets rails and the shaft is gone. Anything left undone behind the liner stays undone, or comes out through a wall that somebody has to rebuild from one side, again.