Almost everything published about expansion joints is about choosing one. Very little is about the thing that decides whether the one you chose can actually be installed — the recess in the structure, its width, its squareness, and what is already sitting in it.
That gap in the literature matches a gap on site. By the time anyone asks us for a joint, the slot has usually been cast. Every figure below comes off our own datasheets and method statements, and the reason we publish them is that they are needed months before we are.
The sequence problem, stated once
A movement joint is a fit-out purchase. The blockout it sits in is a frame-stage pour. On a typical programme those are separated by a year and by two different contractors, and the second one inherits whatever the first one cast.
You cannot value-engineer a blockout at fit-out. By then it is concrete.
1. The blockout grows faster than the gap
This is the figure that surprises people most, and it is the one worth putting in front of a structural engineer. Doubling the movement gap does not double the recess — but the recess is always far larger than the gap, and it grows steadily with it.
From the SeismAtec 450 datasheet, floor-to-floor (A01), blockout dimension C:
| Movement gap | Blockout C | Ratio |
|---|---|---|
| 50 mm | 387 mm | 7.7× |
| 100 mm | 461 mm | 4.6× |
| 150 mm | 536 mm | 3.6× |
| 200 mm | 612 mm | 3.1× |
| 300 mm | 763 mm | 2.5× |
| 500 mm | 1,063 mm | 2.1× |
A 500 mm seismic gap needs a metre-wide blockout. That is slab edge, rebar detailing and formwork — decided at frame stage and effectively permanent. The floor-to-wall (A02) figures are smaller, 212 mm at a 50 mm gap rising to 923 mm at 500 mm, but the same point holds.
If your building has a wide seismic separation — and in Zone IV or V it very likely does — the blockout belongs on the structural drawing, not on a fit-out schedule.
2. Gaps come in steps, not on a slider
The SeismAtec 450 is rated at 50, 75, 100, 150, 200, 250, 300, 400 and 500 mm. There is nothing in between.
So if the movement calculation lands at 130 mm, somebody is rounding — and which way they round is a decision, not a formality. It changes the blockout, and as the next section shows, it changes the load rating too. Rounding is fine. Rounding silently, at fit-out, by whoever is placing the order, is not.
3. Load rating falls as the gap widens
Same system, same datasheet. The Eurocode load ratings are not constant across the range — a wider joint spans further, so it carries less:
| Gap | Pallet truck | Forklift | Car |
|---|---|---|---|
| 50–150 mm | 20 kN | 69 kN | 30 kN |
| 200–400 mm | 10 kN | 46 kN | — |
| 500 mm | 5 kN | 36 kN | — |
The pallet-truck rating falls to a quarter across the range. So rounding the gap up "to be safe" quietly spends load capacity — and on a warehouse floor, a hospital service corridor or a podium that takes a fire tender, that capacity is the thing you actually needed. Widening a gap is not a conservative move. It is a trade.
4. The rebate: depth is forgiving, width is not
Where the finishes zone is too shallow for a standard detail, a rebate is cut or cast into the slab. Our method statement treats its two dimensions in opposite ways, and this asymmetry is the single most useful thing on this page for anyone supervising the pour.
Depth — deliberately over-cut
The rebate is made to the full height of the profile plus 5–10 mm for tolerance, or, put the other way, to the frame height plus a minimum 10 mm levelling bed, minus the floor finish depth. The extra is intentional: it is taken up by a high-strength mortar levelling bed of at least 40 N/mm² compressive strength, which is what actually brings the frame to the right level.
So an over-deep rebate is recoverable. The bed absorbs it.
Width and squareness — no tolerance at all
The method statement is blunt about this: the rebate must be square, and "do not make the rebate bigger than the blockout dimension". There is no levelling bed for width. An over-wide or out-of-square rebate leaves the frame sitting where its anchors have less concrete than the design assumed, right at a slab edge — which is where concrete quality is usually worst anyway.
One line on a pour checklist prevents most of this: depth may run over, width may not.
5. Torque is set by the anchor, not by the gap
A reasonable-sounding assumption is that a 500 mm joint needs a heavier fixing than a 50 mm one. It does not.
The fixing spec is a property of the anchor and the substrate, not of the joint width. On the MegAtec 340 the frames are fixed with M6 × 75 mm countersunk concrete screws, into holes drilled 6 mm diameter to 80 mm depth, tightened to 25 Nm — and that figure does not scale with the gap. What scales with the gap is the number of fixings and the width of the frame, not the torque on each one.
This matters because "bigger joint, tighten it harder" is exactly the improvisation that strips a fixing in a slab edge. Torque is a specification, and it should be on the method statement the installer is actually holding.
6. The one exception — and it is easy to miss
There is a case where the fixing spec changes, and it is not about size at all.
Two things must be in the structural gap before the visible joint system: the fire barrier and the waterproofing membrane. They are installed first, at structural level, under everything.
Which means that when the frames are anchored, the drill goes through a slab that already has a membrane in it. Our method statement handles this explicitly: where waterproofing has been fitted, each hole takes a gun-applied resin anchor rather than a plain concrete screw — so the anchor seals the penetration it has just made.
Get that wrong and you have a building where every component was installed correctly and the joint leaks anyway, because the fixings perforated the membrane that was put there to stop exactly that. It is the most avoidable leak we are called to look at, and it is decided by one line in a method statement. The wider waterproofing picture is in where the joint meets the waterproofing.
What to put on the structural drawing
- The blockout dimension, per joint location, taken from the datasheet of the system actually intended — not a nominal allowance.
- The gap, rounded to a rated step, with the rounding direction recorded and its load-rating consequence checked.
- Rebate width and squareness called out as critical, and depth called out as over-cut to a levelling bed.
- Whether a membrane or fire barrier crosses this joint — because it changes the anchor.
- A survey requirement: the cast gap measured at three points along each continuous run before manufacture. If it varies by more than the system's own tolerance, the number on the drawing was never calculated.
Points 1 to 4 cost nothing at design stage and are close to unrecoverable afterwards. Point 5 is the one that catches the difference between the drawing and the building — and on essentially every project we survey, there is one.