Every atrium reaches a point where the repairs stop being repairs and start being instalments on a replacement. The awkward part is that nothing announces it. There’s no failure, no date in a log, just a slow accumulation of call-outs that each looked reasonable at the time.
Most buildings go past that point without noticing, and then spend two or three more years paying reactive money on a roof that needed a project.
The signals that matter
Frequency is the first and the clearest. One leak in five years is a defect. Three in eighteen months, in different places, is a system telling you something about its age.
Dispersal is the second. A single recurring leak usually has one cause and can be found and fixed. Leaks appearing at different bays, different corners and different junctions mean the problem isn’t a defect, it’s the condition of the seals and gaskets across the whole roof. Fixing them one at a time is the expensive way to replace all of them.
Then there’s the sealed unit question. Misting shows that an edge seal has gone, and edge seals across one roof tend to be the same age, the same product and exposed to the same conditions. When units start failing in numbers rather than singly, the rest are on the same clock.
Framework condition is the one that forces the decision. Glass can be replaced in a sound framework almost indefinitely. Corroded steel, failed thermal breaks, distortion that stops pressure plates clamping evenly, these are not pane-by-pane problems, and a new unit fitted into a framework that has lost its geometry won’t seal properly however carefully it goes in.
Run the arithmetic honestly
The comparison that decides it isn’t repair cost against refurbishment cost. It’s the whole cost of carrying on for the next decade against the cost of doing the work once.
Reactive repairs carry the access charge every time. Mobilising a platform or a rope team for a single pane costs most of what mobilising for twenty panes costs, and that charge lands on every call-out. Five visits in three years is five access bills to fix a fraction of a roof.
The costs that don’t appear on the invoices belong in the sum too. Floor space out of use, out-of-hours premiums, damage to finishes below, the staff time spent arranging it all, and the tenant goodwill spent each time reception gets wrapped in sheeting.
Energy performance is worth putting a number against as well. Older atrium glazing runs single-glazed in some buildings and early double-glazed in many more, with no coating or an early one, and a large area of overhead glass with poor performance drives both winter heat loss and summer cooling load. A refurbishment is the only practical moment to change that, since the glass has to come out anyway. The thinking is the same as looking at the operating cost of a system rather than its purchase price, and over the life of a roof the operating side is the larger figure.
The thermal upgrade decision deserves its own survey rather than being bolted onto a repair, and it sits alongside the access planning, since the permits and method statements a glazing project needs have lead times of their own.
What refurbishment actually means
It isn’t always a full reglaze. There’s a middle ground that gets overlooked, and on the right building it’s the sensible answer.
Reseal and regasket keeps the glass and the framework and renews everything that perishes. It’s substantial work with full access, it addresses leaks across the whole roof at once, and it buys years where the glass itself is sound. Worth considering when units aren’t failing but the seals are tired.
Reglazing into retained framework replaces the units and keeps the structure. Cheaper and faster than a full replacement, and constrained by what the existing rebate depth and bar capacity will take.
Full replacement takes framework and glass out together. It costs the most, it’s the only option when the framework has gone, and it’s the only one that lets you specify from scratch.
Which of the three applies is a survey outcome, not a preference, and the survey needs close access rather than binoculars from the floor.
The planning stage people skip
Refurbishment is a project with a design stage, and treating it as a large repair is how it goes wrong.
Overhead glazing has requirements about how it behaves when it breaks, and those are set out in the gov.uk Approved Documents. Loading needs checking, because modern high-performance units are usually heavier than what’s coming out and the increase across a whole roof is a real number. Smoke ventilation built into the atrium needs a temporary variation to the fire strategy while bays are open.
Contractors handling atrium refurbishment with the access capability in house tend to give firmer programmes, since nothing waits on a second firm’s diary.
The useful discipline is to put a figure on the last three years of reactive spend, access charges included, before deciding anything. Most building owners have never added it up, and the number is often larger than the quotation they’ve been avoiding.