Two numbers get confused constantly and they are not the same thing. The gap is how wide the slot through the structure is. The movement is how much that width changes. A joint system has to be selected against both, and a specification that names only one of them is incomplete.
This piece works the movement calculation through with numbers from the cities we actually build in. The gap itself is a structural decision and belongs to the engineer of record — what follows is how to check that the gap you have been given, and the system being put into it, agree with each other.
The calculation
Thermal movement is first-principles physics and there is nothing clever in it:
For reinforced concrete, α is taken as about 10 × 10−6 per °C. IS 456 gives a range for it — roughly 6 to 12 × 10−6 depending on the aggregate — and 10 is the value normally used for design. Limestone aggregate sits at the bottom of that band and quartzite near the top, so on a project where the aggregate source is known and unusual, it is worth checking rather than assuming.
The important subtlety is that L is the length of one block, not the length of the building. A block free to expand moves outward from its own centre, so each end travels ΔL/2. A joint between two equal blocks therefore takes ΔL/2 from each side, and the joint's own movement range comes back to α · L · ΔT with L the bay length between joints.
What that gives for Indian conditions
Thermal movement in millimetres, at α = 10 × 10−6:
| Bay length between joints | ΔT = 20 K Chennai, Bengaluru | ΔT = 30 K Kolkata, Hyderabad, Guwahati | ΔT = 45 K Delhi NCR, Jodhpur |
|---|---|---|---|
| 30 m | 6.0 mm | 9.0 mm | 13.5 mm |
| 45 m | 9.0 mm | 13.5 mm | 20.3 mm |
| 60 m | 12.0 mm | 18.0 mm | 27.0 mm |
| 90 m | 18.0 mm | 27.0 mm | 40.5 mm |
Two readings of that table are worth making explicitly. The first is that the same building moves more than twice as far in Delhi as in Chennai — which is why a catalogue-default gap carried from one project to the next tends to be about right in the south and too narrow in the north. The second is that these are ranges, not one-way movements. A 45 m bay in Delhi swinging 20 mm is opening and closing by roughly ±10 mm about its mid position, and that is the figure to compare against a system's rated movement capacity.
Choosing ΔT honestly
This is where most of the error lives, and it is not in the formula.
ΔT is not the weather report. It is the effective range the structural element itself experiences, and that depends entirely on where the element sits:
- A shaded internal frame in a conditioned building sees a much narrower range than ambient — the envelope is doing exactly what it was built to do.
- An exposed roof slab or open podium deck sees a much wider one. Surface temperature on a dark exposed slab in a north Indian June runs well above air temperature, and the night-time low is close to ambient. The roof is routinely the governing case in a building whose internal frame would have been comfortable.
- A multi-storey car park is structure with no envelope at all. Treat it as external throughout.
The practical consequence: a single ΔT applied across a whole project is a simplification, and on any building with a significant exposed deck it is the wrong simplification. Split it.
The three things the formula does not include
1. Shrinkage, which only ever opens the joint
Drying shrinkage is permanent and one-directional. At the order of 0.0003 total shrinkage strain that IS 456 uses for design, a 45 m bay shortens by roughly 13 mm over its life, and a joint between two such bays sees the sum from both sides.
That sounds alarming next to a 20 mm thermal range and in practice it is much less so, for a reason worth knowing: most drying shrinkage happens in the first year, and the joint system goes in at fit-out — often six to eighteen months after the pour. A large share of the movement has already occurred by the time the frames are anchored, and creep relieves some of the rest. What matters is that the residual is one-directional and opens the joint, so it eats into the opening half of the system's capacity and not the closing half. On a young structure being fitted out fast, that asymmetry is real and should be allowed for in where the joint is set.
2. Where the joint sits when you install it
This is the single most common avoidable failure we see, and it costs nothing to prevent.
A system rated at ±10 mm assumes it is installed at the middle of its travel. Install the same system on a 45 °C afternoon in June, when the structure is at its most expanded and the gap at its narrowest, and you have effectively used up the closing half of the capacity before anyone has walked on it. Every millimetre of the winter movement then has to come out of the opening half, and the system is asked for 20 mm of travel in one direction that it was only ever rated to give 10 mm of.
The fix is procedural, not technical: record the ambient temperature and the measured gap at the time of installation, and set the centre plate accordingly rather than centring it by eye. Any competent system allows the setting to be adjusted. Almost no specification asks for it to be recorded.
3. The gap on site is not the gap on the drawing
On essentially every project we survey, the built gap differs from the drawn gap — sometimes by a little, occasionally by a great deal, and rarely uniformly along the run. Formwork moves, shuttering is struck early, a filler board gets compressed, and a nominally 100 mm gap turns out to run 85 mm at one end and 130 mm at the other.
This matters more than it sounds, because a joint system is a manufactured assembly with a rated gap. Order to the drawing, measure at fit-out, and you get either a system that will not go in or one that goes in with the frames sitting where the anchors have no concrete. Every one of our jobs is manufactured against a site survey, not against the drawing, and where the survey comes back non-uniform we would rather split the run into two gap sizes than average it.
The calculation gives you a movement. The survey gives you a gap. Specify against the first and manufacture against the second, and the two failure modes that account for most joint complaints never happen.
Putting it together: a worked case
An exposed podium deck in Delhi NCR, joints at 45 m, aggregate ordinary:
- Thermal range. α = 10 × 10−6, L = 45,000 mm, ΔT taken as 45 K for an exposed deck. ΔL = 10 × 10−6 × 45,000 × 45 = 20 mm, i.e. roughly ±10 mm about mid position.
- Shrinkage residual. One-directional, opening. Allowed for by biasing the installed setting slightly closed rather than by widening the gap.
- Seismic. Delhi is Zone IV. If this line also separates two blocks, the structural engineer's separation figure will be larger than 20 mm and governs outright — the thermal number stops being relevant. See what the seismic zone changes.
- System selection. A rated capacity of ±10 mm with nothing to spare is not a selection, it is a coincidence. Choose a system whose rated movement comfortably exceeds the calculated range, sized to the surveyed gap, and rated for the traffic the deck actually takes.
For a case like that we would typically be looking at a MegAtec 340 for its self-centring action on a trafficked deck, or a SeismAtec 450 where the separation runs wide — the 450 is designed to take seismic movement of the order of half the gap width, which is a different order of problem from thermal cycling.