Why Joint Sealants Fail

A failed joint sealant nearly always looks like one of six things at the joint face. It has peeled cleanly off the concrete. It has split down its own middle. It has torn at the base of the slot. It has been dragged out by traffic or by a plough blade. It has gone soft and tracked across the pavement. Or it never bonded at all, because the slot was damp when it was poured. Each of those has a mechanism behind it. Each has a cause that sits further back in the job than the pour.

The uncomfortable part is that the material is rarely the problem. Shepherd and Sons have been sealing joints for more than 40 years, and we are members of the Extruded Sealant Association. A good share of that work has been cutting out somebody else's failed seal and starting again. What we find in the slot is usually dust, water, a missing backer rod or a bead the wrong shape. This article works through the failure modes in rough order of how often we meet them. It explains the mechanism in each case, then sets out what a specifier can actually write down to prevent them.

Adhesion failure is the seal letting go of the joint face. The bead itself is usually intact. It has simply peeled away from one side, leaving a clean dark line down the concrete. You can often lift a metre of it out by hand. Water then runs straight down the open side and into the pavement. That is the one thing the seal was there to stop.

The mechanism is simple enough. Sealant bonds to whatever is on the face, not to the face. Leave saw slurry, laitance, dust, curing compound or old sealant residue in the slot, and the sealant bonds to that thin weak layer instead. It holds for a season or two. Then the first real thermal cycle peels it. A damp face does the same job, because water sits between the sealant and the concrete and no bond forms through it.

Prevention is mechanical. Faces are cleaned by wire brushing, grit blasting or routing, then blown out with oil-free compressed air. Where the concrete at the face is unsound or still carries old sealant, the slot gets widened until fresh, sound material is exposed on both sides. Cold applied systems add a primer, and the primer has its own window. In our experience the joints that come back to us are the ones where the slot was swept rather than blown. Or blown at a pressure that pushed the dust along the slot rather than out of it. Grit blasting and dry cutting both raise respirable dust, so HSE guidance on control measures covers the preparation too.

Cohesion failure looks different. The sealant is still firmly stuck to both faces. It has torn through itself, leaving a ragged split down the centre of the bead. That tells you the bond was fine and the sealant was asked to stretch further than it could.

The cause is usually geometry rather than movement. Strain in a sealant is the movement divided by the width it has to absorb it across. A narrow joint carrying the same slab movement as a wide one is under far more strain. Depth makes it worse again, because a deep bead is restrained by the faces along its whole height and cannot deform freely. A slot cut 6mm wide and 40mm deep is a sealant failure waiting to happen, whatever grade goes into it.

The shape of the bead is therefore part of the specification, not a site decision. As a working rule the sealant should never be deeper than it is wide. Hot applied pavement seals are commonly run at roughly twice the width of their depth. Two-part cold applied systems sit nearer equal width and depth, and Thioflex 555 is quoted at 1:1 to 1.5:1. There is a minimum sealant depth underneath the ratio as well, 10mm on that product, so the depth is a floor and not only a proportion. The backer rod is what sets that depth, which brings us to the next one.

Three-sided adhesion is the failure most people have never heard of, and it is common. The sealant bonds to both joint faces and to the bottom of the slot as well. Now it is held on three sides. When the joint opens, the bead cannot pull in at the waist the way it is designed to. All the strain concentrates at the base instead, and the sealant tears there. Sometimes it tears at the bottom and splits upward. Sometimes it rips a lump of the base out with it.

This is the entire reason a backer rod exists. People assume it is there to save sealant, and it does, but that is a side effect. Its real job is to be a bond breaker across the base of the slot so the sealant is anchored on two faces only. It also fixes the depth of the bead, which is what governs the shape factor above. Two jobs, one cheap component, and it is the one most often left out when a gang is behind schedule.

The detail matters. The rod is closed cell and slightly oversized for the slot, so it stays put under the pour rather than floating up. For hot applied sealant it must be a heat-resistant grade. Standard polyethylene rod simply melts when material at pouring temperature reaches it, and then you have neither a bond breaker nor a depth. Where a slot is too shallow to take a rod at all, a bond breaker tape along the base does the same job. What you never do is fill the slot to the bottom and hope.

This one presents as cohesion failure, so it gets misdiagnosed as a bad batch. The slot geometry is right, the preparation was right, and the bead still splits within a few seasons. The joint was simply moving more than the grade could take.

Hot applied sealants are classified under BS EN 14188-1, the hot applied part of the BS EN 14188 series. N1 is the elastic, high extension grade. N2 is the normal, low extension grade. Neither is fuel resistant; that is what the F1 and F2 grades are for, and it is a separate question from movement. N2 is a perfectly good product on closely spaced contraction joints in short concrete bays, where the movement at each joint is genuinely small. Put it on a long-bay pavement, an expansion joint or a bridge approach and it will tear. The comparison is set out in full in the difference between N1 and N2 sealant.

The fix is arithmetic, not judgement. Bay length, joint spacing and the service temperature range give you a calculated movement at each joint. The grade follows from that figure. What causes trouble is a specification copied from the previous scheme, where the slabs were a different length. It is worth saying that the error runs both ways. Specifying the high extension grade across a whole site regardless of movement spends money without buying performance.

The next group are surface failures. The seal is bonded, the shape is right, and traffic takes it out anyway. There are two versions.

The first is tracking, or pick-up, and it happens in hot weather. Bituminous materials soften as the surface temperature climbs. The overbanding grades OBS 45 and OBS 60, for instance, are quoted at a softening point of 85°C plus or minus 15 to BS EN 1427. They lose stiffness long before they reach it, and a dark surface on a still summer afternoon runs far hotter than the air above it. A seal finished flush, or standing slightly proud, then takes the full contact patch of every tyre. Warm sealant sticks to the tyre, comes out of the slot in stringy lifts, and gets tracked in black smears across the pavement. On an airfield that is also foreign object debris, which is a different order of problem.

The second is ploughing out. A snow plough blade running across a transverse joint catches any seal standing above the surface and peels it out in long strips. The same happens to a proud overband and to sweeper brushes at speed. Both versions have the same root cause, which is recess. The finished sealant should sit below the pavement surface, not level with it. Common practice on transverse joints under normal vehicle traffic is a minimum 5mm recess, more on wide joints and more again in cold weather. It is a measured dimension, and it is checkable on site with a straightedge and a ruler while the gang is still there.

Recess is also why exposure has to be stated up front. We have sealed joints across highway, airfield and port pavements, and the traffic on each behaves differently. Slow, heavily loaded, turning traffic in a container yard punishes a proud seal far harder than free-flowing motorway traffic does.

Every sealant system we install needs a dry slot. That is not a preference, it is what the manufacturers' data and the Extruded Sealant Association Code of Practice both require. The failure it causes is usually adhesion failure, but the timing is what makes it dangerous. The joint looks perfect when the possession hands back. It fails eighteen months later, long after anybody connects it to the weather on the night.

With hot applied material the mechanism is violent. Sealant arrives in the slot at pouring temperature and any trapped moisture flashes straight to steam. That blows the bond apart at the face and leaves voids through the bead. With a two-part cold applied system the vulnerable moment is the primer window instead. Sealant follows the primer within a set period, typically from around twenty minutes to a few hours. A shower inside that window costs you the preparation, not just the pour. A cured or contaminated primer has to be ground off and the slot primed again.

Temperature is a separate problem that gets confused with rain. Cold changes the numbers, and past a point it stops the work altogether. Most sealants want a substrate at 5°C and rising before anything goes in, and cold applied systems want it warmer still to cure. Cure and return to service both stretch out. The machine grade of a two-part polysulphide such as Thioflex 555 returns to service in about 30 minutes at 20°C. The hand grade of the same material wants hours rather than minutes at that temperature, and both stretch out sharply as the night gets colder. There is more on that material in the benefits of Thioflex 555. Cold also changes the target recess. Concrete contracts as it cools, so on a cold night the joint is at its widest as you seal it. Fill it to the usual level then, and the sealant is squeezed proud when the slab expands again in summer. Frost rules the work out completely. A damp slot can be dried with hot air lances. A frozen one cannot be sealed at all.

Read back through those six failure modes and a pattern comes out of them. Five of the six are decided before any sealant leaves the melter. That is the honest shape of this work. Pouring the joint is quick and it looks like the skilled part, because it is the part anyone watching can see. It is the last five minutes of a sequence that runs like this.

1. Cut or reform the slot to the specified width and depth, so the bead has a shape it can work in.
2. Remove the old sealant and any unsound concrete at the faces, widening until sound material is exposed.
3. Clean the faces mechanically, by wire brushing, grit blasting or routing.
4. Blow the slot out with oil-free compressed air and confirm it is dry, sound and free from frost.
5. Fit the backer rod to depth, or a bond breaker tape where the slot is too shallow, and prime if the system calls for it.

Then, and only then, the sealant goes in. On the M20 between junctions 8 and 9 we reformed and resealed 23,304 linear metres of joint. That took twenty night shifts on roughly 20km of concrete carriageway, in a 10pm to 3:30am possession. The pour was never the constraint on those shifts. The sawing, cleaning and rodding set the pace, and the programme was built around them. The same holds on a single bay in a warehouse floor, at a much smaller scale. It is also why crack chasing is a discipline of its own. Forming a clean reservoir is the work; sealing it is the closing operation.

Every failure above can be specified out. None of it requires exotic wording. It requires the right handful of dimensions and hold points written into the joint sealing specification, where they can be checked rather than assumed.

• Against adhesion failure: faces cleaned mechanically and blown out with oil-free compressed air, and unsound faces widened back to sound material. Add a hold point before sealing, so somebody looks into the slot.
• Against cohesion failure: a stated slot width and depth, with the width-to-depth ratio named rather than left to the gang on the night.
• Against three-sided adhesion: a backer rod of the correct size in every slot, specified as a heat-resistant grade wherever the sealant is hot applied.
• Against the wrong movement class: the calculated joint movement stated on the drawing. The sealant class then follows that figure, not the last scheme.
• Against tracking and ploughing out: a recess depth, with the traffic type and exposure stated. Measured and recorded on site, rather than eyeballed.
• Against wet and cold weather failures: a written stand-down trigger for rain, frost and low temperature. Name the person on site who makes that call.
• Behind all of them: sample retention and depth checks at a stated rate, so the record exists before anyone needs to argue about it.

One more question is worth asking, and it costs nothing. Ask who is doing the preparation, and what they are carded for. Over forty years we've found a fairly reliable pattern in that answer. The seals that last tend to be the ones where the same crew cuts, cleans, rods and pours. Nobody can then quietly hand a half-prepared slot on to somebody else. We are members of the Extruded Sealant Association, and work to its competence standards. We are SafeContractor and CHAS accredited, and have been sole contractor to the British Airport Authority since 2008.

If you are writing a joint sealing specification, we can help you get those details right first time. If you are looking at a seal that has already failed, the joint face usually tells us which of the six it is. For work across Kent, London, Sussex, Essex and the wider UK, get in touch with our team.