Flat Roof with a Single-Ply Membrane
A flat roof is not flat, and must not be: it is a surface shaped so that water finds its own way to the outlets. The membrane is only the last link. The two questions that decide the outcome are asked long before: where does the water go, and what holds the covering down when the wind stops pushing and starts pulling.
Overview
TAV. 00Waterproofing with a single-ply membrane in welded sheets has replaced multi-layer built-up systems almost everywhere, for a simple reason: continuity does not rely on an adhesive but on a weld, that is, on the material fused to itself. The strengths and the weak points both follow from that. The covering is light, repairable and testable; in exchange it is thin, and anything that pierces it or holds it down becomes part of the watertightness. The design plays out on three distinct planes that should be kept separate in your head: the hydraulic plane, meaning falls and outlets, which must keep working even when one outlet blocks; the hygrothermal plane, meaning the order of the layers and the vapour barrier, which decides whether condensation forms inside the build-up; and the mechanical plane, meaning how the covering resists wind, which is not a load pressing down but a suction tearing upward, and which counts for far more at edges and corners than out in the middle of the roof.
The site score
TAV. HThe detail of the craftthe hydraulic shape of the roof
Falls: deciding where the water goes
The slab's actual levels are surveyed, the outlets positioned, and the sloping plane that serves them is formed, either with a screed to falls or with pre-tapered insulation. The valleys along which water converges are checked, and every point is verified as draining towards the outlet that serves it rather than away from it.
«A flat roof does not leak: it ponds. And once it ponds, sooner or later it leaks.»site notebook — editorial synthesis
click a beat · arrow keys ← → walk the site · dashed pauses are the chemistry at work
The force nobody draws
Wind does not push on a roof: it sucks it up
When wind meets a building, the flow above the roof accelerates and separates, and the local pressure drops below that of the air inside. The result is that the covering is pulled upward, not pressed down. That suction is not uniform: it is modest at the centre of the roof, rises along every edge and peaks at the corners, where the flow curls into vortices. Hence the rule governing the fixings: the pitch is not constant but follows three zones, and the corner zone, which occupies a tiny share of the area, takes the largest share of the fasteners.
A qualitative diagram of the load zones; the extent of each band and the number of fasteners come from the wind load calculation for the specific building, according to height, exposure and geometry.
Three ways of staying down
Fixed, bonded or ballasted: the choice is not only about cost
A single-ply covering can be held down in three ways, and each moves the problem somewhere else. Mechanical fixing is the cheapest and quickest, but it transfers every gust into the structure at concentrated points and requires piercing the insulation and the vapour barrier. Bonding spreads the pull across the whole surface and pierces nothing, giving a quieter covering with no rippling, but it demands a sound, dry, clean substrate and tolerates no improvisation. Ballast fixes nothing: it holds the covering down by weight, shields the membrane from sunlight and greatly extends its life, but it must be verified against the structure and substantially changes the permanent load. That choice is made with the structural engineer, not on site.
A qualitative comparison of the three systems; the choice depends on the building, its wind exposure and the available load capacity, and must be verified case by case.
How it ages (and what betrays it)
TAV. PQuestions from the site
TAV. QWhy does a flat roof need a fall at all?
Because "flat" describes the appearance, not the behaviour. The membrane really is watertight even under standing water, so the fall is not there to prevent an immediate leak: it is there to prevent three slower consequences. Ponding adds permanent load exactly at midspan, where the slab has already deflected, and the phenomenon tends to worsen on its own because more water means more deflection. Standing water concentrates dirt, leaves and deposits, which in time become a seedbed for vegetation and block the outlets. Finally it exposes the joints to permanently wet conditions that no covering enjoys, accelerating ageing precisely at the most delicate points. That is why the fall is verified on the finished surface and not on the drawing: it is the only figure that counts.
TPO or EPDM: how do you choose?
They are two families with different logics, and the most important practical difference is how the sheets are joined. TPO is a thermoplastic: it is hot-air welded, so the joint is a fusion, it is executed quickly with automatic machines and it can be probe-tested. It is commonly available in a light, reflective colour, which keeps the covering cooler and helps in summer. EPDM is a cured rubber, not thermoplastic: it is not welded but joined with dedicated tapes and adhesives, it has very high elasticity which makes it forgiving on substrates that move, and in large-format sheets it drastically reduces the number of joints. As a practical rule: where there are many details, many penetrations and you need objective verification of the joints, the thermoplastic is easier to govern; where the surface is large, regular and free of obstacles, a single large sheet has fewer weak points by definition.
How long does it last, and how do you notice it ageing?
Good-thickness single-ply membranes have service lives in the region of twenty-five to thirty years, but that number depends on two factors more than on the material: ultraviolet exposure and ponding. A ballasted covering, or one under a green roof, is shielded and lasts appreciably longer than an exposed one of the same product. Ageing does not announce itself as a sudden leak: the surface loses elasticity and becomes dull and slightly chalky to the touch, lap edges start to show localised debonding, and small star-shaped stresses appear around the fixings. The useful inspection is not hunting for a hole, it is looking for those three signals, and doing so after a serious storm and before winter, while any problems are still small. A membrane stays repairable until very late: a welded patch genuinely restores continuity, and it is almost always more sensible than premature replacement.
Can photovoltaics go on a membrane roof?
Yes, and it is one of the most natural places for it, but the decision belongs before rather than after, because it changes the roof design instead of leaning on it. There are two routes. The first uses ballasted frames, which pierce nothing and rely on weight for stability: it is the better option for watertightness, but it adds permanent load the slab must be able to accept, and it must also be checked against wind, because an array of tilted modules is a sail. The second uses mechanically fixed supports, which means as many membrane penetrations, each one turned up and welded as a detail in its own right. In both cases two questions get forgotten with great regularity: under the modules the covering can no longer be inspected or easily repaired, so its remaining service life should be compared with the system's before covering it up; and panel maintenance means repeated foot traffic on the covering, which needs protected walkways designed in from the start.