18 Millimetres of Lippage Corrected across a Porcelain Terrace with a Raimondi Levelling System

Lippage is the height difference between two neighbouring slab edges. On this terrace, the survey found a worst reading of 18 mm across a diagonal run of 20 mm rectified porcelain, checked with a 2 metre spirit level and feeler gauges. The BS 7533 family of standards treats a step of more than a few millimetres on foot-traffic paving as a trip hazard, and porcelain manufacturers including Marshalls and Bradstone publish laying tolerances around 0.5 mm for their rectified ranges. Moving from 18 mm to below one millimetre was the practical measure of success.

The Raimondi system came in at the final setting stage, after the slabs were already close to the intended plane on a wet bed. The 18 mm reading traced back to a cumulative fall in the sub-base, which meant the correction had to start below the porcelain.

Where the 18 mm error began

Porcelain slabs are dimensionally flat to within a fraction of a millimetre. Warping was not the cause of the survey reading; the fault sat in the layers below the slab.

On this job, the MOT Type 1 sub-base had been compacted as a single 150 mm lift. There was no plate compactor pass between thinner layers. Guidance from the Association of Interlocking Concrete Pavement and the UK paving trade converges on compacting granular sub-base in layers of 100 mm or less, with each layer whacked with a vibrating plate before the next layer goes down.

A deep lift can leave the lower half loose. After a wet winter, that loose material settles unevenly, so a terrace that looked level on the day of laying can develop falls of 10 to 20 mm within two seasons.

The bedding method added its own problem. The original slabs had been laid on five dabs of mortar per slab: one at each corner and one in the middle. Porcelain has near-zero water absorption, so a dab bed leaves voids that never develop a reliable bond. Point loads then let the slab rock, and rocking drives the edges out of plane.

The replacement bed had to give the levelling caps something firm to pull against. A full mortar bed, screeded across the field and buttered to the slab, removed the voids that had made the old surface move under load.

The bed that gave the caps a chance

Correction started with a full wet bed of 4:1 sharp sand and cement, laid at roughly 40 mm compacted thickness across the whole field. Bare mortar gives porcelain very little bond, so every slab was brushed on the back with a slurry primer made from SBR bonding agent and cement mixed to a thick paint. Without that primer, the slab may sit on the mortar with no real grip, and freeze-thaw cycles can lift it within a year.

The bed was screeded to a consistent fall of about 1:80 away from the house. Once that plane was set, individual slab height varied by only a few millimetres either way, which sits inside the working range of the levelling hardware. The base clips slid under the slab edge before the slab was placed, the cap threaded onto the exposed spigot, and turning the cap drew adjacent slab edges to a shared plane while the mortar was still soft.

A low slab still had to be lifted and re-bedded. The cap force is useful for pulling a proud edge down and holding a joint flush while the bed cures. That made the screed accuracy of the wet bed the ceiling on the final finish. Areas reading 2 to 3 mm before capping could be brought below 0.5 mm; an 8 mm discrepancy was outside the cap range, so those slabs were taken back up, reset on fresh bed, and capped again.

Raimondi caps, clips, and the count on this terrace

The Raimondi RLS uses a reusable cap with a disposable base clip. For 20 mm porcelain, the correct base clip is the version rated for 12 to 20 mm slab thickness. Using the thinner-tile clip on a 20 mm slab is a common error, and it tends to snap the clip under torque.

Clip count followed slab size. On a 600 by 600 mm slab, two clips per long edge is the working minimum. Once shared edges are counted, that gives four to six clips per slab. Across 42 square metres of 600 by 600 slabs, the job used a little over 600 base clips and around 40 reusable caps, rotating the caps forward as the mortar set behind the laying front.

The caps came off after the bed had gone off enough to hold the plane, usually the next day. The base clips then snapped at the bed line after a sideways strike with a rubber mallet along the joint.

Clean slab edges mattered as much as cap pressure. Mortar squeeze rising into the joint can foul the clip and give a false flush reading. The laying front stayed narrow, four to six slabs ahead of the capping, which left time to clean the joints before each cap went on. Setting the whole field and capping later fails because porcelain bedding mortar starts to skin within the hour.

Jointing after the caps came off

The joints were filled with a two-part resin slurry grout, brushed across a damp surface and washed off before it cured on the porcelain face.

That resin holds a 3 to 5 mm porcelain joint against frost and weeds far better than a sand-cement slurry. In narrow porcelain joints, sand-cement slurry tends to crack out within a season.

Porcelain beside Fairstone sandstone

Fairstone is Marshalls’ natural sandstone range, and it sits in a different part of the paving world from rectified porcelain. The bedding principles are similar, yet the price, finish, and tolerance behaviour are very different.

A Fairstone sandstone flag is calibrated to a nominal thickness while still varying by a millimetre or two across a pack, because it is a split and sawn natural product. Rectified porcelain is much more dimensionally consistent. That consistency is exactly why porcelain shows lippage so clearly and why a levelling system earns its cost on porcelain while it is often skipped on riven sandstone.

Laying Indian sandstone flags to a good finish depends on the layer reading the calibration variation in the pack and adjusting the bed under each flag by feel. The usual control is a taut string line, with the flag tapped down using a rubber mallet until the surface sits right. A 3 mm step can read as part of the texture on riven sandstone, while the same height on polished porcelain catches the light and every visitor’s toe.

The cost gap runs in the opposite direction to what many buyers expect. Porcelain slab prices per square metre often sit above mid-range sandstone, although porcelain fitting is faster once the bed is right and it needs no sealing. Sandstone wants a penetrating sealer against staining. Over a decade, the running totals can converge. Bedding depth, primer, and levelling hardware have more influence on the finished terrace than the slab material printed on the pallet.

Drainage at the rendered wall

A 1:80 fall carries surface water to the terrace edge, but on this site the edge met a rendered wall with a damp-proof course only 150 mm above the finished paving. Falls toward a wall need interception, so an ACO channel drain was set along the wall line. Its grating sat flush with the porcelain, and its invert was cut into the bed so the channel collected run-off before it reached the render.

The channel sat on its own concrete haunch, separate from the paving mortar. A linear channel under foot traffic and occasional light vehicle load needs a stiffer support than a slab bed provides. The outfall ran to a soakaway sized on the terrace area and the local infiltration rate, which on clay subsoil means a larger crate volume than the same area over free-draining sand.

Leaving out the channel and relying on slab falls alone would have sent water toward a DPC that already sat too low. Levelling accuracy also cannot fix water tracking behind a render line.

The channel exposed another fault after the lippage was corrected. Once the field read flat to half a millimetre, the ponding that the uneven surface had scattered became visible as a thin sheet near the wall, leaving one awkward observation: some terraces appear to drain only because their lippage breaks the water before it can gather.