There is a wall on the road I take every morning from my house to the office in Rajasthan. It belongs to a three-storey residential building — flat terrace, south-facing, full sun by 9am. I first noticed it failing in October 2024. By April 2026 it had been “repaired” three times, and it is still failing.
I never spoke to the contractor. I watched it from the road. That is enough, because waterproofing failure has a grammar — a set of visual signs that tell you exactly what went wrong, in what order, and why. By the third failure I could read the whole story from a moving car.
What I saw: the three stages
October 2024 — the first failure. The original terrace had cracked along the parapet joint. Classic thermal movement crack: the concrete slab expands and contracts on a cycle the masonry parapet cannot follow. The repair was a grey coating painted over the crack and extended about 30 cm either side. By December the coating had debonded along the centre line. The crack was visible again, now with a white bloom of efflorescence beneath.
February 2025 — the second attempt. This time they went bigger. The entire terrace surface received a white coating — almost certainly an acrylic waterproofing product, based on the colour and sheen. They also patched the parapet crack with a cement–sand mortar fill before coating. By May 2025 — one pre-monsoon cycle — the coating had lifted from the crack perimeter in blisters the size of dinner plates. The efflorescence was back, darker.
October 2025 — the third attempt. This time they added a screed layer on top of the previous coating. I assume this was someone’s attempt to protect the membrane and add thermal mass. By March 2026 the screed had delaminated from the parapet edge and was curling upward, pulling the coating with it. The wall below the parapet is now stained with a regular vertical pattern of brown water tracks.
The chemistry they kept ignoring
Every single failure had the same root cause: they were coating over a substrate that was still moving and still wet. The crack at the parapet joint is an active thermal movement crack — the kind classified as Type 2 in any basic structural crack assessment. It opens in the cool months (November–February in Rajasthan), closes partially in the hot months (April–June), and cycles again. You can confirm this by marking the crack ends with a pencil on two dates three months apart. The measurement will be different.
An active crack has a specific treatment protocol. The sequence matters and cannot be skipped:
- Stop all water ingress first. The slab must be genuinely dry — not surface-dry, which can happen in two days of sun, but dry to depth. IS 456:2000 clause 13.5 specifies minimum curing periods for concrete (7 days for OPC, 14 days for blended cements) — the same principle in reverse tells you how long water takes to leave.
- Treat the crack as active, not cosmetic. Surface sealing a thermal crack with cementitious slurry is closing a door that is still moving. It will open again. The correct treatment for an active thermal crack is either an elastomeric crack-bridging membrane or a PU injection if the crack has width — followed by an elastomeric top system.
- Protect the membrane from mechanical damage. A coating on an exposed flat terrace in Rajasthan — direct sun, foot traffic, thermal shock — will not last 3–4 years without a screed protector. But the screed must be bonded to a dry, primed substrate, not to a previous failed coating.
Why acrylic failed
The second attempt used what appears to be an acrylic waterproofing membrane. Acrylics are water-based dispersion systems: they cure by water evaporation and form a flexible film. On a positive-pressure application — where water is pushing from outside the slab down through the structure — acrylics are chemically unsuited. The hydrostatic pressure from the monsoon above will lift the film from the substrate faster than the acrylic can resist.
Acrylics work in negative-pressure applications: applied from the dry side, with water pressure pushing the membrane against the substrate. On a flat exposed terrace, during monsoon, the pressure is always positive — water sitting on top of the slab seeks to migrate through. For this application, a cementitious crystalline system or a PU membrane (minimum 1.5mm DFT with screed protection) is chemically appropriate.
The crystalline mechanism is worth understanding precisely: calcium silicate hydrate crystals grow into the capillary pores of the concrete slab when activated by water. The crystals become part of the concrete matrix — they cannot debond, because they are the concrete. This is why crystalline systems work on active cracks below a certain width, and why they resist positive hydrostatic pressure that would lift any film-forming membrane.
What the third repair got right — and wrong
Adding a screed protector was the right instinct. A screed layer over a waterproofing membrane dramatically extends the membrane’s working life by absorbing the thermal shock and mechanical abrasion that would otherwise fatigue the membrane directly. On a flat Rajasthan terrace — where surface temperatures can exceed 70°C in June — an unprotected membrane is engineering a failure within 3–5 years.
But the screed was applied over a failed, debonded coating on a wet substrate. The bonding of screed to waterproofing membrane requires a properly cured, dry, and mechanically sound membrane. The third contractor built a new layer on top of a layer that had already lost adhesion. When the screed expanded in the April heat, it had no anchor. The delamination was visible by the following March.
What a correct repair looks like
A proper repair of this terrace would require, in sequence:
- Full removal of existing coatings and failed screed down to bare concrete
- Mechanical preparation of the slab surface (grinding, not acid wash)
- Assessment of crack type and width — is it still active?
- If active: PU crack injection or pre-formed movement joint at the parapet
- If stabilised: V-groove and fill with polymer-modified mortar
- Full drying period — minimum 28 days in this climate after any wet repair
- Crystalline waterproofing slurry applied in two coats to damp (not wet) concrete
- Screed protector: 40mm minimum, reinforced, with proper expansion joints at perimeter
- Surface finish: tiles or similar to reduce thermal shock on screed
The total cost of this correct sequence — materials and labour — would be higher than any of the three “quick fix” attempts. The total cost of the three failed repairs, plus the water damage to the structure below, is already higher than doing it right once.
The wall on my morning drive will fail a fourth time. Not because the materials are bad, and not because good waterproofing chemistry doesn’t exist in India. It will fail because the person making the decision does not understand what is being asked of the product — and does not know how to tell the difference between a product problem and a substrate problem.
Almost every waterproofing failure I have seen in 8 years in this industry was a substrate problem.
Construction chemicals professional based in Rajasthan, India. Writing about the chemistry behind what makes buildings fail — and last.
More about the author →
Update — The full breakdown of why Indian waterproofing specs don’t require a pressure test is in Edition 6 on the hydrostatic head test India doesn’t require.
Related editions: The same field-grammar approach applied to waterproofing failures across India. The substrate chemistry that lets coatings fail is the same one that lets tile adhesives debond. Walls bearing AAC have an additional failure mode covered in AAC blocks. Brick is losing. The hydrostatic performance gap in Indian waterproofing standards — and the pressure tests that BIS doesn’t require — is in edition 6 on the hydrostatic head test India doesn’t require.
