Finding Out Why Block Paving Sank Before Relaying It

Measure the dish before forming an opinion about it

Work from a datum that cannot have moved. The door threshold, the kerb line at the footway, the rim of a chamber cover set in concrete, the edge course butting the house wall. Pull a line between two of them, run a straightedge across the low ground on a grid of roughly half a metre, and write down the offset at every intersection. Then mark on the paving, in crayon, the line where the offsets come back to the plane the rest of the drive sits on. That chalk line is the honest extent of the movement, and it is almost always wider than the eye reports.

Separate design fall from settlement while you are down there. A drive built to shed water reads low at the far edge because that is what it was built to do, and a straightedge laid along the fall will confirm a defect that does not exist. Take one run parallel to the fall and one across it, and judge the dip only against the plane the surrounding paving actually occupies. Minimum crossfall for modular paving is set out in BS 7533 and in Interpave's guidance; a drive that ponds all over rather than in one hollow has lost its fall or never had it.

The shape carries most of the diagnosis. A broad, soft-edged saucer a metre or two across is a layer consolidating over an area, which points at the laying course or the sub-base. A sharp-edged trough with a defined rim and something close to a vertical drop at its edge is a void that collapsed, and voids have causes. Two parallel troughs following the wheel path mean the pavement as built is under-designed for what parks on it, and the dip you were called about is only the first place it showed. A straight-sided step running across the drive is a trench line or a chamber, not a pavement failure at all. And where blocks have rotated out of the pattern, or joints open progressively toward one side, the movement has a horizontal component and belongs to the perimeter rather than to depth.

Photograph it with the straightedge in shot, keep the sketch with its offsets, and note the date and the weather of the preceding week. If the answer turns out to be a drain, that record is what a sewerage undertaker or an insurer will ask for. If the answer turns out to be nothing much, the same sketch repeated in six months is the only way anyone will know whether it is still moving.

Joints first, because it is the cheapest thing that might be true

Run a screwdriver down a dozen joints, inside the dip and well outside it, and note how far the sand stands below the block surface. Full joints are the mechanism by which one block hands load to the next; an empty joint leaves each unit to deflect alone, which is why block paving that has lost its jointing chips at the arrises and rocks long before it dips. Kiln-dried sand is expected to need topping up in the first year and after any heavy cleaning, so half-depth joints across a whole drive are usually a maintenance story rather than a fault — most often a decade of close-range pressure washing.

The diagnosis is in the distribution, not the depth. Joints low everywhere and a dip in one place is two separate problems: refill the joints and keep looking, because loose jointing does not produce a 30 mm hollow on its own. Joints empty only inside the dip while the surrounding field is full is the interesting case. Sand does not evaporate, and a joint that has emptied locally has a route down through the laying course into somewhere that will take it. That is the first real evidence of a void, and it moves the drain check up the list.

Refill to match what is already down. Kiln-dried sand stays reworkable, which suits a surface you may open again. Polymeric and resin jointing compounds resist washout and weeds but are unforgiving of process: dry blocks, the correct fill depth below the chamfer, every grain swept off the faces before wetting, and a watering method that activates the binder without flushing the joint out. The temperature and rainfall limits in the manufacturer's instructions are conditions, not advice. Mixing types across one drive shows for the life of the surface, so a bay repointed in a different material reads as a patch from the pavement whether or not the level is right.

Whether the edge has let go

Crouch at one corner and sight along each perimeter in turn. On a domestic drive the restraint is usually a soldier course or a kerb bedded and haunched in concrete, and it fails quietly: the haunching cracks below ground where nobody looks, the kerb rotates a degree or two outward, the last course tilts with it, and the field behind gains a few millimetres of room every season. Blocks that were cut tight to fit start to sit loose in their gaps.

The confirmation is horizontal rather than vertical. Joints widening progressively as you work toward one boundary, blocks turned slightly out of the herringbone, a whole field that has walked toward a lawn — all of that is lateral movement. Measure across the paving between two things that genuinely cannot have moved, the house wall and the far kerb, and compare the answer to the block module. A field of 200 mm units that has gained 40 mm across its width has lost the confinement that lets one block share a wheel load with its neighbours, and it will keep settling in the middle without any void being involved at all.

Where a drive meets a lawn with nothing but topsoil behind the edge, or meets the footway at a crossover where the kerb line takes every wheel, the restraint is being asked to do structural work with nothing behind it. A dip within a metre of a perimeter that has moved usually belongs to that perimeter. Relaying the middle and leaving the edge reproduces the same hollow within a couple of years, because the sub-base has to run past the last block and the restraint has to bear on it — not on whatever was shovelled back against the outside of the dig.

Rebuilding from formation upward

A reinstatement bay is a small confined excavation, and the two things that go wrong in one are compaction and edges. A plate compactor cannot work within its own width of the vertical face of existing paving, so the perimeter of a repair is always its least compacted part, and that is exactly where the settlement will reappear if it is allowed to. Thinner lifts than you would use in the open, and a rammer where the plate physically cannot reach, is what closes that gap.

The sub-base goes back as an unbound aggregate placed in layers and brought up to the underside of the laying course, not to the underside of the blocks. Order it by weight, because that is how a merchant sells it, and work the tonnage off the finished thickness at the material's compacted density rather than off a loose bulk figure — a layer swallows noticeably more loose material than its finished volume suggests, and getting that wrong is a second delivery charge on a job with the drive open and the car on the road.

Nobody is running a nuclear density gauge on a domestic drive, but the principle behind BS 1377-4 still applies at this scale: a layer is compacted when further passes stop producing measurable settlement. The way to know is to sit a level on the surface and watch it between passes rather than count passes and hope. Two extra passes on a layer that has stopped moving cost nothing; two too few is the whole failure.

1. Trim the excavation to vertical faces on the pattern line, and take out every loose block and all contaminated laying course rather than working around them.
2. Prove the formation with a bar and dig out soft material where it moves under hand pressure, instead of bridging it with extra sub-base.
3. Repair and re-test any drainage defect first, backfilling the pipe trench in compacted layers up to the underside of the sub-base before the pavement is touched.
4. Place the sub-base in lifts thin enough for the machine on site, compacting each one fully out to the vertical faces with a rammer where the plate cannot reach.
5. Check the finished sub-base level off a straightedge bridged onto the sound paving, so that the laying course can go down at one constant thickness rather than as a wedge.
6. Screed the laying course uncompacted to that thickness, keep off it, and screed only as much as will be laid before the weather or the end of the day gets to it.

The laying course, the level, and tying the bay in

The laying course is a screeded, uncompacted, constant-thickness layer of grit sand to the grading BS 7533 calls for. It is not a levelling medium, and on a repair the temptation to use it as one is enormous: the sub-base is 15 mm low in one corner and there is a heap of sand right there. Give in and the hollow returns in precisely that corner, because a thicker layer of sand consolidates by more than a thinner one, and does it under the first few wheel loads.

Tie the level in by measurement rather than by memory. The bay finishes proud of the surrounding paving before compaction by the amount your sand actually consolidates, and you can measure that in the trial opening instead of guessing it: screed a known thickness, bed two blocks on it, compact them and take the difference. Finish flush and the repair reads low after a month. Finish too proud and you have built a ridge that the first shovel of the winter will find.

Lay back in off the sound paving, never off the screeded sand. Clean the underside of every reused block, because a skin of old bedding stops a unit seating and rides it high. Run the joint width the surrounding field is running rather than the width that feels tidy — the joints are the load path between units, and a bay laid tight into a drive laid to its spacer bars is both structurally different and visible from the pavement.

Compact through a sole plate or a rubber mat, in overlapping passes across the bay and onto the sound paving around it, with the joints filled before the first pass and topped up after the last. Expect to top them up again after a week of the car going over it. A joint that is still taking sand a month later is not thirsty; it is telling you that something underneath is still moving, and that the diagnosis missed something. It is worth going back for that at week four rather than at month eighteen.