Who this guide is for
- Owners of buildings with flat roofs, terraces or ballasted build-ups
- Readers planning work that involves lifting or moving roof surfacing
- Anyone comparing how different coverings and facades are held in place
- Renovators altering a roof edge, verge or perimeter
- Readers preparing questions about uplift before a roof or facade is chosen
Wind acts on the surface and is resisted through the stack
The records describe restraint as a chain rather than a component. On a warm deck the adhesive, mechanical fixings or applied weight hold the insulation to the deck and the membrane to the insulation, and the outermost layer is only as secure as the layer it is fastened to. The single ply record puts it even more directly: wind acts on the assembly, not on the sheet.
That is why the choice of how a layer is held reaches down through the build-up. Mechanical fastening drives fixings through everything between the membrane and the structure, so the structure has to be able to receive them and each layer passed on the way has to tolerate being perforated. Adhesion demands a prepared, compatible and dry substrate. Ballasting demands a structure and a build-up that can accept weight above the membrane. Changing the attachment method means reconsidering the stack beneath it.
- Fastened: fixings pass through every layer to reach something that will hold them
- Adhered: the layer immediately below has to be sound, compatible and prepared
- Ballasted: the structure has to accept weight, and the layer below goes out of sight
- In each case the outer layer inherits the weakest hold beneath it
The restraint path passes through the vapour and thermal layers
The chain does not respect the other duties in the build-up. The warm deck record notes that a fixing holding the insulation down passes through the vapour control layer to reach the deck, so the layer that is meant to be continuous is deliberately perforated by the thing keeping the roof on. That trade-off is why deck type, restraint method and vapour control are settled together, and why swapping one of them on site is not a small substitution.
On a wall the equivalent is the batten or bracket. A timber frame wall carries wind load through the membrane into the frame at every batten fixing, which is why the membrane is expected to close around a compressed fixing rather than remain unpierced. A rainscreen bracket transfers load, creates a path around the insulation and perforates the sheet at one point, and easing one of those consequences generally worsens another.
Edges and corners are not the field
Uplift concentrates. The single ply record states that it concentrates at edges and corners, so a membrane held as though the roof were uniform will be underheld exactly where it is loaded hardest. The tiled covering record says the same about a discontinuous covering: uplift concentrates at eaves, verges, ridges and around anything that interrupts the surface, so the restraint pattern is not uniform even though the covering looks uniform.
The trafficked deck record extends it to loose-laid surfacing, noting that units near the perimeter, at corners and around any upstand face conditions the field does not, which is why the edge of a terrace is often arranged differently from its centre even though it looks continuous. The inverted roof record asks directly how the restraint assumed for a build-up changes towards the edges and corners of the roof.
Insulation surrounded by water wants to float
The inverted roof is the case where restraint and water meet. Its record describes buoyancy as a real condition rather than a theoretical one: water standing on the waterproofing surrounds the underside of loose-laid boards, and boards surrounded by water want to rise. The weight above is therefore resisting wind from one direction and flotation from the other.
That is why ballast in such a roof is a mechanical component rather than a finish. The record states that it cannot be reduced, thinned or partly removed on aesthetic grounds without changing the roof, and that stripping surfacing from part of a roof to carry out unrelated work leaves the layers beneath it restrained by nothing. Where designed storage is added on top, the attenuation record notes that held water sustains the same flotation the ballast exists to resist, so restraint becomes a larger question rather than a smaller one.
- Loose-laid boards under standing water are being lifted from below as well as above
- Ballast is resisting two directions at once, not decorating a surface
- Lifting surfacing for unrelated work removes restraint from everything under it
- Areas intended to be openable, and how restraint is restored, belong in the design
Restraint that is concealed and never inspected
Most of the restraint in an envelope is invisible once the building is finished. The veneer record notes that a veneer that appears sound can be inadequately anchored, because the anchors are invisible and the leaf will stand until it is loaded. The masonry cavity record describes the ties as the only structural connection between the leaves and also the commonest accidental bridge across them, and observes that a wall can look entirely sound while the connection between its leaves has deteriorated.
The same is true above. In a seamed metal roof the clip holds nothing unless the seam is folded closed over it, and a length of seam left unfolded or a clip missed at a change of direction produces a local failure with no visible cause on the surface. Condition in all these cases is described as judged by opening up rather than from the elevation.
Elements that catch the wind rather than resist it
Anything standing clear of the envelope adds a surface that wind can act on in both directions. The shading record notes that wind on a bank of louvres both pushes and lifts rather than simply pressing, and that the reversal is often the case the connection has to be designed around. It describes the wind action delivered to the supports as following from area, spacing and geometry rather than from how heavy the element looks.
That is why the records treat adding such an element to a finished building as a structural question rather than an appearance one. An arm that lands on cladding quietly makes the cladding's own fixings structural, which is described as a change nobody records and no one inspects.
What later work does to the chain
Ordinary building work interrupts restraint without anyone intending it. Lifting paving for access, removing a strip of ballast to trace a leak, renewing fascia and soffit boards, refixing a gutter, replacing a covering with one of a different weight, or re-laying a verge all act on the chain. The tiled covering record notes that replacing a covering is not a like-for-like job, because a change of unit weight or fixing method reaches back into the battens, the structure and the wind restraint.
The practical step the records point to is asking, before the work, what is being relied on to hold the assembly down in the area being opened, and how that will be restored. Where a record of the restraint arrangement exists, this is what it is for; where it does not, the question belongs with the designer of the roof or facade rather than with whoever is on site.
Questions about how an envelope is held down
- 1Ask what holds each layer down, and what that layer is in turn held to
- 2Establish whether the attachment is by adhesion, mechanical fixing or weight
- 3Ask what the chosen method requires of the deck or backing behind it
- 4Ask which layers the fixings pass through on their way to an anchorage
- 5Establish how the arrangement changes at edges, corners and around upstands
- 6Ask whether any surfacing is acting as restraint rather than only as a finish
- 7On an inverted or ballasted roof, ask what happens if part of it is lifted
- 8Ask which areas are intended to be openable and how restraint is restored
- 9Establish what restrains loose-laid units at the perimeter of a terrace
- 10Ask what anything standing clear of the surface is fixed back to
- 11Check whether a proposed covering change alters weight or fixing method
- 12Ask what record exists of the restraint arrangement and where it is held
- 13Ask who assesses uplift for this building and its exposure
Common mistakes to avoid
- Treating restraint as a property of the outermost layer alone
- Holding a roof or facade as though it were one uniform surface
- Removing ballast or paving for unrelated work without restoring restraint
- Reading a ballast layer as decoration that can be thinned on appearance
- Assuming a sound-looking leaf or covering is adequately anchored
- Fixing a projecting element to cladding rather than to a checked backing element
- Replacing a covering with a heavier or differently fixed one as though it were like for like
When to involve a professional
- Ask how wind uplift has been assessed for this building and this exposure
- Ask how the restraint arrangement changes at the perimeter and the corners
- Ask which layers the restraint fixings perforate and how that has been accounted for
- Ask what the surfacing is doing mechanically as well as visually
- Agree how restraint is reinstated wherever surfacing is lifted
- Ask what will be recorded about the restraint arrangement at handover
Frequently asked questions
Questions readers ask about this topic
Why does the edge of a roof need a different arrangement from the middle?
Because uplift concentrates there. The records state that wind uplift concentrates at edges and corners, at eaves, verges and ridges, and around anything interrupting the surface, so an assembly held as though the roof were uniform is underheld exactly where the loading is highest. The pattern is not uniform even where the covering looks it.
Can I lift a section of roof paving to do other work?
It depends what the paving is doing. On an inverted build-up it is the restraint holding the stack down against wind and against the buoyancy of insulation surrounded by water, and the records state that removing it even temporarily and in a small area removes wind restraint from the layers beneath and should not be done without advice.
What does buoyancy have to do with a roof?
In an inverted build-up the insulation sits above the waterproofing and is laid loose, so water standing on the membrane surrounds the underside of the boards. The records describe boards surrounded by water as wanting to rise, which means the weight above is resisting flotation from below as well as wind from above.
Does the way a membrane is held affect anything else?
Yes. The records describe attachment as a build-up decision that reaches down to the deck and back through the vapour control layer. A fastened arrangement perforates every layer it passes on the way to an anchorage, an adhered one depends on the layer below being sound and compatible, and a ballasted one puts the membrane out of sight.
Is changing a roof covering a like-for-like job?
The records say not necessarily. A change of unit weight or fixing method is described as reaching back into the battens, the structure and the wind restraint, which makes it a question for the structural and roofing designers rather than a substitution made at the point of ordering.
Keep reading