Construction joint sealant: movement, geometry and failure

A construction joint sealant is selected by movement, not by strength. Get the movement class and the joint geometry right and almost any competent product works; get them wrong and the best product on the market fails within two winters.

Movement is expressed as a percentage of width

A 20 mm joint that opens 5 mm is asking for 25% movement. Sealants are classified on that basis — ISO 11600 groups them by movement capability and modulus (for example 25LM, 25HM, 20LM, 12.5E), and ASTM C920 uses a similar class system.

The practical consequence: a product rated for 12.5% movement installed in a 25% joint will fail, no matter how good its adhesion. This is the most common specification error in façade and precast work, and it is invisible on the data sheet unless you know which line to read.

Width-to-depth ratio sets the demand

For a moving joint, depth is normally kept at 0.5–0.7 of width. Two reasons:

• Too shallow: not enough material to absorb strain, so the bead tears at the bond line.
• Too deep: the sealant is being stretched from a thin section and the strain concentrates at the interface.

A backer rod controls depth and prevents three-sided adhesion. A bead bonded to two faces can stretch; a bead bonded on three faces cannot, and it fails at the weakest bond line — usually the one nobody inspected.

The four failures, in the order they actually occur

1. Three-sided adhesion — no backer rod, or a rod compressed so far it no longer controls depth.
2. Substrate preparation — dust, form release agent, or a damp surface. PU does not bond to dust.
3. Wrong movement class — a 25% joint sealed with a 12.5% product.
4. Bead geometry — tooled to a feather edge, or applied too thin for the joint.

None of these is fixed by buying a more expensive sealant. All four are visible before the cartridge is opened.

The two catalogue grades and where the line sits

Property	KA340A	KA340B	Why it matters
Elongation at break	615%	550%	Strain the bead can take before tearing
Elastic recovery	75%	80%	Whether the joint closes again after movement
Tensile strength	2.5 MPa	2.0 MPa	Bulk strength of the cured sealant
Shore A	35	48	Softer grade follows movement; harder grade resists pressure
Water pressure	not specified	high resistance	Below-grade and drainage joints

The counter-intuitive part: KA340B is the higher-modulus grade with lower elongation and higher hardness, and it is the one for wet, narrow, static joints. KA340A is the softer, more extensible grade for joints that move. Choosing “the stronger one” for a moving façade joint is the wrong direction.

Substrate condition decides more than the product

Concrete must be cured, dry and dust-free. Moisture-curing chemistry refers to the bead after application, not to a wet substrate: sealing green concrete traps water at the interface and the bond fails weeks later, when nobody connects it to the installation.

Porous or dusty concrete needs a primer. Steel and aluminium need degreasing. Stone needs a test for staining before a full run.

Cold weather is a joint problem, not only a cure problem

Below about 10 °C two things happen at once: the sealant cures more slowly, and the joint is at its widest because the structure has contracted. A bead installed at maximum extension has no reserve left when summer closes the joint — it goes into compression and can buckle or debond.

The practical rule is to install near the middle of the joint’s movement range, which in many climates means avoiding the coldest and hottest part of the year for exposed joints.

When this is not the right choice

• The joint is a crack that is still moving. Cracks need routing and a different repair strategy, not a bead over the surface.
• The structure is subject to hydrostatic pressure from a wet side. Use a water-pressure-rated system and a designed waterstop, not a surface bead alone.
• The joint is a movement joint in a fire-rated assembly. Fire performance is a system property; the sealant alone does not provide it.

Reading a joint drawing

Four dimensions decide everything, and they are usually on the drawing in a detail box that nobody reads.

Dimension	What it sets	Common error
Joint width	Movement class (movement ÷ width)	Built narrower than drawn
Joint depth	Bead thickness and cure rate	Filled solid, no backer rod
Backer rod diameter	Depth control and bond break	Omitted to save time
Recess depth	How much sealant sits below the surface	Too deep, bead too thin

Construction joint vs movement joint

A construction joint is where two pours meet; it may be static or may become a movement joint once the structure is loaded. A movement joint is designed to move and its sealant must be rated for that movement.

Treating a construction joint as static because it was static on the day of sealing is how a bead ends up cracking two summers later.

What to measure on site

Before sealing, measure the actual width at three points along the joint. A joint that varies by more than a few millimetres needs a different detail, not a thicker bead. Record the installation temperature as well: a joint sealed at its widest has no compression reserve.

The movement arithmetic behind the class

Movement class is not an abstract label; it is arithmetic you can do before choosing a product. Thermal movement in a panel joint is roughly:

movement (mm) = α × panel length (mm) × temperature swing (°C)
α concrete    ≈ 10 × 10⁻⁶ per °C

Take a 30 m concrete panel in a climate with a 40 °C swing between installation and the hottest day:

30,000 mm × 40 °C × 10 × 10⁻⁶ = 12 mm
12 mm ÷ 20 mm joint           = 60% movement

No ISO 11600 class covers 60%. The joint as drawn cannot be sealed with an elastomeric sealant at all — the design needs shorter panels, wider joints, or both. Run the same building with 12 m panels:

12,000 mm × 40 °C × 10 × 10⁻⁶ = 4.8 mm
4.8 mm ÷ 20 mm joint          = 24%  → 25LM class

That is the whole selection logic in two lines: panel length and joint width are chosen together, and the sealant class follows from them. When a specification says “25LM” without stating the panel length, it is describing a product, not a joint.

Two corrections matter in practice:

• The base temperature is the installation temperature, not the annual average. A joint sealed in July at 35 °C will be near its closed position all winter and near its widest the following July, not the other way round.
• Different materials move differently. Aluminium is about 23 × 10⁻⁶ per °C and PVC can be 50–70 × 10⁻⁶, so a metal façade on a concrete frame moves mostly at the fixing, not in the panel. Sealant joints at material changes are where the differential shows up.

Backer rod: the part that decides whether the bead can move

A sealant bead has to be able to stretch. It can only do that if it is bonded to two faces and free on the third. The backer rod provides the free face and controls the depth of the bead.

Joint width	Rod diameter	Sealant depth	Note
10 mm	12–13 mm	6–7 mm	Below 6 mm the bead tears at the bond line
20 mm	24–26 mm	10–14 mm	The common façade detail
30 mm	36–40 mm	15–21 mm	Deep joints need a slower-cure discussion
40 mm+	1.25 × width	20–28 mm	Consider a two-stage or a different joint design

Rules that come out of the table:

• Rod diameter is 20–30% larger than the joint width, so it compresses and grips the faces instead of sliding.
• Depth is 0.5–0.7 of width. Deeper than that and the strain concentrates at the interface; shallower and there is not enough material to absorb the movement.
• Minimum depth is around 6 mm. Below it, the bead is a membrane and fails in tension rather than stretching.
• Closed-cell rod for exterior work. An open-cell rod absorbs water and holds it against the bond line.

A bead bonded on three faces cannot stretch; it can only tear or pull off. This is why “we used plenty of sealant” is not a defence when the joint fails — filling the joint solid is the mechanism of failure, not a margin of safety.

Primers, by substrate

Primer is not a universal good. It is a chemical bridge for substrates that need one, and a waste of a process step on those that do not.

Substrate	Primer needed?	What to watch
Concrete, dry and dust-free	Usually no	Dust and form release are the real problem
Concrete, damp or green	Yes, and often a moisture-tolerant system	PU will not bond to a wet interface
Brick and block	Yes on porous faces	Test for staining before a full run
Bare steel	Yes after degreasing	Mill scale flakes under the bead
Galvanised steel	Yes, after removing oil	Oily zinc is a classic debond
Aluminium	Yes	Oxide layer forms within hours of cleaning
Glass and glazed tile	Yes or an activator	Silicone contamination defeats both
PVC and plasticisers	Test first	Plasticiser migrates into the bead
Bitumen and asphalt	No — usually incompatible	Use a compatible system or a physical break

The test that settles it is a peel test on an offcut of the actual substrate, cured for the full cure time and pulled. A primer that improves adhesion on a test coupon but adds 20 minutes per joint on site is a decision the contractor should make with data, not habit.

What the sealant is not doing

A joint sealant keeps water out of a joint. It does not waterproof a structure, and it does not replace a waterstop.

• Below grade, hydrostatic pressure will find any discontinuity. The waterstop is the waterproofing; the sealant is the joint finish.
• On a façade, the sealant is the first line and the cavity drainage is the second. A rainscreen without drainage relies on the bead for everything, which is a design decision with a maintenance cost attached.
• On a roof, a movement joint needs a dedicated joint cover or a proprietary system, not a deeper bead.

When a leak appears at a joint, the first question is not which sealant was used but whether the joint was designed to be sealed at all.

Checking the work before the scaffold comes down

The inspection that catches most failures is cheap and happens before handover:

1. Measure the joint width at three points. A joint that varies by more than a few millimetres needs a different detail, not more sealant.
2. Check the bead section at a stop end. Cut one and look: is the depth within 0.5–0.7 of width, and is the rod visible below the bead?
3. Look for three-sided adhesion. At corners and around penetrations, this is where it happens.
4. Record the installation temperature. It tells you where the joint sits in its movement range and predicts which season will stress it.
5. Re-inspect after the first wet-and-cold cycle. A preparation failure declares itself then, and it is still early enough to fix without cutting out the joint.

How long the bead is expected to last

A PU joint sealant in an exposed façade is usually expected to give 10–15 years before replacement, and the number depends far more on movement and UV exposure than on the brand. The failure that ends the life is rarely cohesive tearing; it is adhesion loss at the bond line, driven by water ingress behind the bead.

That has a design consequence: joints that drain and joints that stay wet have very different service lives with the same product. If the detail traps water against the sealant, no movement class will save it.

The practical position

Specify the joint before the product: width, depth, movement class and substrate condition. Then choose the grade with the recovery that matches the movement, and budget the backer rod and primer as part of the system rather than as extras. Inspect the bond line after the first wet cycle — that is when a preparation failure declares itself, and it is still early enough to fix without cutting out the joint.

Standards referenced

• ISO 11600:2002 — classification and requirements for jointing products; the source of the 25LM / 25HM / 20LM / 12.5E classes used here: iso.org.
• ASTM C920 — standard specification for elastomeric joint sealants, the US class system that runs in parallel: astm.org.
• ASTM C1193 — standard guide for use of joint sealants, which covers joint design and installation rather than product classification: astm.org.
• EN 15651-1 — sealants for non-structural use in joints, façade elements; the harmonised standard behind CE marking in the EU: CEN listing.
• ISO 9047:2001 — adhesion/cohesion at variable temperatures, the cyclic test that separates movement capability from elongation: iso.org.
• ISO 10563:2023 — change in mass and volume, which is how shrinkage of a curing bead is measured: iso.org.