Perimeter First: What a Paving Edge Actually Has to Do, and Why So Many Give Way
Photo by Photo by Teddy Yang on Unsplash on Unsplash
When a driveway starts to go, it rarely goes in the middle. The blocks in the centre of a drive that's ten years old are usually still sitting there, tight and level. The trouble is at the edges — the perimeter line where the paving meets soil, kerb, wall or lawn.
That's not a coincidence. The edge of a pavement is where every load gets concentrated. It's where the weather gets in. It's where the haunching either does its job or quietly turns to crumbs under a few winters of freeze-thaw. If you only ever look at the middle of a driveway, you're looking at the part that's least likely to tell you anything.
What an edge is actually for
Paving is a system that wants to spread sideways. Block paving, flags, resin bound — they all behave a bit like a dry stone wall in that respect. Load one part and it wants to push out at the boundary. The edge restraint is what stops that happening.
So an edge has three jobs:
- Hold the paving in place laterally, so units don't creep, open up or fan out.
- Contain the bedding and jointing material, so it doesn't wash or blow out.
- Transfer load down into something solid rather than into soft ground.
When the edge does all three, a driveway can last decades. When it does one and a half of them, you get the familiar list: gaps opening at the perimeter, jointing sand disappearing, blocks spreading, kerbs rocking when you step on them, and a mossy black line where water has been sitting in a void.
What a proper edge looks like
For most block paving and flag work, a concrete kerb is set on a compacted, level bed and then haunched — a wedge of concrete packed against the outer face, typically something like 100mm deep and 150mm wide, tapering down. The kerb sits slightly proud of the paving, and the paving is cut tight against it rather than tucked under with a loose fillet of sand.
Two details separate a decent edge from a token one:
- The haunch is on the outside, and it's continuous. Gaps in haunching are where water gets in and where frost starts prising.
- The kerb itself is bedded, not just placed. A kerb resting on soil with a smear of concrete at the back is decoration.
Resin bound and gravel surfaces need a physical edge just as much as block paving — often more, because the aggregate has nothing interlocking to hold it. A resin edge, timber edging or kerb line isn't optional there; it's structural.
Where it gets harder: slopes and tight plots
An edge on flat ground is straightforward. An edge on a slope is where the real thinking happens.
On a gradient, water running down the surface hits the bottom edge at speed and with volume. If the driveway falls toward a kerb line that isn't level or hasn't been stepped, you get washout behind the kerb and a slow collapse. On steeper drives, kerbs often need to be stepped or set level in sections, with the flow managed — sometimes with a channel drain before the boundary rather than relying on the kerb to act as a dam.
Tight plots are the other headache. A terraced front garden where the drive runs boundary to boundary leaves nowhere to put a haunch. In those cases you're looking at options rather than a standard detail:
- A recessed edge, where the restraint is buried slightly below the paving line and the haunch sits under the paving rather than beside it.
- A concrete raft or widened bed under the edge, effectively thickening the base along the perimeter.
- A pinned or bolted kerb where there's a retaining element, such as a low wall or a step up to a lawn.
And where the paving is holding back soil — a raised patio, a drive cut into a slope — the edge is a retaining structure, not a kerb. That means proper design: mass, depth, drainage behind it, and often a structural engineer's input on anything over around a metre. A garden wall with a driveway on top of the retained side is not a DIY detail.
Reading the warning signs
Walk your perimeter once a year, ideally after a wet spell. Look for:
- Rocking. Stand on the kerb. Any movement at all is a sign the bedding or haunch has failed.
- Open joints at the edge. A widening line between paving and restraint means the paving is moving away from the edge.
- Sand loss. Jointing sand washing out along one line only is a drainage or haunch problem, not a sand problem.
- Settlement dips. A slight hollow just inside the perimeter usually means the sub-base wasn't compacted at the edge, where compaction is hardest to reach.
- Moss and algae lines. A dark green streak at the boundary says water is sitting there.
- Soil or gravel escaping. If you can see the bedding, something has failed.
Repairing it before it spreads
Good news: edge failures are usually cheaper to fix than they look, provided you catch them early.
If one kerb is rocking and the paving either side is sound, the repair is a lift-and-rehaunch: take out the kerb, clear the old material, re-bed on fresh concrete, re-haunch properly, and reinstate the paving against a clean, tight line. A morning's work.
If a run of edge has gone — spreading blocks, a wavy line, sand loss along ten metres — you're into lifting a strip back, re-compacting the sub-base and bedding, and rebuilding the restraint. Still a contained job, and far better done at that stage than after the movement has travelled into the main field of paving.
If the edge is a retaining one, or if the paving is on a slope and water is the driver, get it assessed rather than patched. Repairing the kerb without dealing with the water is a repeat visit waiting to happen.
What you shouldn't do is fill the gap with sand and hope. It lasts a season, it hides the cause, and it lets water in at exactly the point where you least want it.
Order of operations, every time
The perimeter gets built first and it gets built properly — before the bedding, before the paving, before anyone admires the pattern. It's the least photogenic part of the job and it's the part that decides whether the driveway is still flat and tight in fifteen years.
If you're comparing quotes and one of them doesn't mention edge restraint, haunching or how the boundary details will work on your slope, that's the question worth asking. It's a cheap conversation now and an expensive one later.