Back to all articles
    Questions Answered

    Why Do Glass Railings Sometimes Shatter on Their Own?

    Spontaneous glass breakage is almost always nickel sulfide inclusions, edge damage or installation stress — and all three are preventable with the right specification.

    Katena TeamSeptember 28, 20269 min read
    Why Do Glass Railings Sometimes Shatter on Their Own?

    Glass guards that appear to break for no reason are almost never breaking for no reason. The overwhelming cause is a microscopic nickel sulfide inclusion trapped inside fully tempered glass, which slowly expands over months or years until the panel's own internal tension tears it apart. The second and third causes are edge damage during handling and installation-induced stress — a panel squeezed by an over-torqued shoe, sitting on the wrong setting block, or in direct contact with metal. All three are preventable. None of them are bad luck, and all of them are addressable at specification stage for a small fraction of what a replacement campaign costs.

    Nickel sulfide: the microscopic time bomb

    Float glass is made by melting raw materials in a furnace. Traces of nickel — from stainless steel contaminants, refractory materials or the raw batch — can combine with sulfur to form nickel sulfide particles typically well under a millimetre across. In annealed glass these particles are harmless. In fully tempered glass they are not, because tempering puts the core of the panel into permanent high tension.

    Nickel sulfide exists in two crystal forms. The high-temperature form is frozen in place by the rapid quench of the tempering process. Over time, at ambient temperature, it slowly converts to the low-temperature form and expands by a few percent in volume. If the particle happens to sit in the tensile core of the panel, that expansion eventually exceeds what the glass can accommodate and the panel disintegrates instantly. The fracture pattern is diagnostic: two joined cones radiating from a single origin, often described as a butterfly or hourglass, with the offending particle visible at the centre under magnification.

    The rate of occurrence is low per panel but not negligible per building. Published estimates vary widely by glass source, thickness and processing, and anyone quoting you a single confident percentage is guessing. The useful way to think about it: on a tower with a thousand glass panels, a rate that sounds trivially small still means a maintenance conversation somewhere in the building's life.

    Heat soak testing: the mitigation that works

    Heat soak testing is a destructive proof test. After tempering, panels are loaded into an oven and held at an elevated temperature — around 290 degrees Celsius under the European standard, for a defined soak period — which accelerates the nickel sulfide phase change by years. Panels containing critical inclusions break in the oven. The panels that come out the other side have been proven, not inspected.

    It is not free and it is not absolute. Heat soaking adds cost and lead time, and it greatly reduces but does not entirely eliminate the risk, because the process is statistical rather than perfect. It is still the single most effective specification line you can write for a project where panel replacement means a swing stage, a road closure or an unhappy owner on the twenty-second floor. Specify heat-soak testing to a recognized standard and require the mill certificates with the shipment.

    Heat soak testing does not inspect glass for defects. It proves the glass by trying to break it in a furnace, where breaking costs nothing.

    Edge damage: the cause nobody photographs

    Tempered glass is strong across its face and vulnerable at its edge, where the compression layer is thinnest. A chip, a nick from a hand tool, a panel set down on a concrete slab without protection, two panels rubbing in a crate during transport — any of these can create a flaw that survives installation, sits dormant through a season, and then propagates when thermal cycling or wind loading adds the last increment of stress. The breakage looks spontaneous because the damage happened months earlier and nobody saw it.

    • Specify polished or ground edges rather than as-cut, and require edge protection in shipping
    • Require panels to be stored vertically on padded racks, never leaned against structure or laid flat on a slab
    • Reject panels with visible edge damage at delivery rather than installing them and hoping
    • Keep hand tools, metal shims and fasteners away from glass edges during setting
    • Photograph panel condition at delivery — the record settles arguments later

    Installation stress and hard contact

    Glass must never touch metal or concrete directly. Every point of hard contact is a stress concentration, and stress concentrations plus thermal movement plus wind cycling equals a crack that appears three winters later. The engineering answer is boring and consists of correctly durometer-rated setting blocks, gaskets sized for the actual glass thickness tolerance, adequate edge clearance in the shoe, torque values applied with a calibrated wrench rather than an impact driver, and a base shoe installed genuinely plumb so the panel is not being levered into position by its own glazing system.

    This is also where installation quality stops being an abstraction. A crew that installs glass guards every week develops the habits — checking shoe alignment before glazing, using the right blocks, not forcing a panel that does not want to go in — that a general labour crew borrowed for the week does not have. It is one of the concrete reasons we do not subcontract installation.

    Why laminated construction changes the consequence

    Everything above is about reducing the probability of breakage. Laminated construction changes the consequence of breakage, which matters more. A monolithic tempered panel that breaks becomes a shower of fragments and an open edge. A laminated panel that breaks stays in the opening — the interlayer retains the fragments, the guard continues to function as a barrier, and the replacement happens on a schedule rather than as an emergency. Canadian guard-glass requirements have moved firmly in this direction: laminated construction is the option accepted in essentially all guard conditions, while monolithic tempered glass is restricted to panels set back from the edge of the floor. On a guard over a meaningful drop, laminated is the answer, and it is not one to negotiate around. Confirm the specific requirement against the standard and code edition in force for your project.

    Combine the two strategies and the problem largely goes away: heat-soaked heat-treated plies to lower the probability, laminated construction with a quality interlayer to control the consequence, careful handling and glazing to eliminate the avoidable causes. That specification costs modestly more per panel and eliminates the scenario nobody wants to manage from a boardroom.

    Frequently Asked

    Specifying glass for a tower?

    Katena specifies, sources and installs glass guard systems with the processing requirements written in, not assumed. Call (514) 821-0842 or email info@katena.ca.

    Get a Quote
    Tagged:
    glass railings
    tempered glass
    heat soak
    quality control
    specification
    Share this article: