Dalian Oppgate Glass Windows & Doors Co., Ltd. 路 Exporting WorldwideVerified Alibaba Storefront

Structural Silicone Glazing on High-Rise Facades: Bite Geometry, Sealant Compatibility and the Adhesion Failures That Decide Service Life

Low-angle photograph looking up at a cluster of dark glass-clad office towers against a pale overcast sky, with a lower glass-fronted pavilion in the foreground showing a bronze lattice soffit behind its framing
Low-angle photograph looking up at a cluster of dark glass-clad office towers against a pale overcast sky, with a lower glass-fronted pavilion in the foreground showing a bronze lattice soffit behind its framing
Key Architectural Takeaways
  • Four-sided structural glazing makes the sealant the only load path from glass to frame, so the joint is a structural element and must be calculated as one rather than selected from a data sheet.
  • Bite is set by frame geometry long before a sealant is chosen: a 1.5 m lite at 2.4 kPa needs roughly 13 mm of bond width at an allowable stress near 0.14 MPa, which many adapted cap profiles cannot physically offer.
  • On coated aluminium the weak interface is often the finish rather than the silicone, so a late change of coating supplier or colour is a change to the structural joint.
  • Dead load belongs on setting blocks or mechanical supports; sustained shear is where structural silicone performs least well over decades, which is why the allowable stress for that case is an order of magnitude lower.
  • Aluminium moves about two and a half times as much as glass over the same temperature change, and a 6 m mullion turns that difference into a few millimetres the bead must absorb.
  • Adhesion failure is progressive: once water reaches an interface the bonded area shrinks and the stress in what remains rises, so batch records and adhesion tests decide how far a defect spreads.

1. What Four-Sided Structural Silicone Glazing Actually Is

Structural silicone glazing fixes glass to a frame by adhesive bond instead of mechanical pressure. In the version that dominates high-rise work, four-sided structural glazing, a bead of structural silicone is applied to the frame, the glass is set into it, and the joint is left exposed to the weather. There is no external pressure plate, no visible cap and no fixing through the glass. What the client buys is a flush facade of uninterrupted glass. What the engineer takes on is that a bonded polymer joint, typically a little over half a centimetre to just over one centimetre deep and a similar width, becomes the load path between the glass and the building.

Two-sided structural glazing is the cheaper relative and is frequently confused with the four-sided version in tenders. In a two-sided system the glass is bonded on two opposite edges and mechanically retained on the other two, usually by a captured frame or a pressure plate. Wind load is shared between the bond and the retainer, and the facade still reads as flush from most angles. In a four-sided system the bond carries the whole wind load, and the design, testing and quality control requirements step up accordingly. A specification that says "structural glazing" without naming which of the two is being bought leaves the most consequential decision on the project open.

A third distinction separates a joint that is genuinely structural from a joint that is decorative. Cosmetic silicone between a glass edge and a cover cap transfers nothing and is not a structural joint. If a sealant is doing work, three things have to be true at once: it is a structural-grade product supplied within a system evaluated as a kit, the bonding surfaces have been prepared to that system's own instructions, and somebody has calculated the joint for the loads it will actually see. When any one of the three is missing, the facade can still look correct on the day of handover, and the gap only becomes visible years later.

2. Bite Geometry: The Dimension That Governs Everything Downstream

Bite is the width of contact between the sealant and each of the two bonded surfaces, measured across the joint. It is the number that decides whether a structural joint works, and it is fixed by the frame design long before anyone chooses a sealant. The arithmetic is short. Wind pressure on the glass is collected along the joint at a tributary width equal to half the glass span on each side, giving a load per metre of joint; that load is divided by the allowable design stress of the sealant to give the minimum bite.

A worked case makes the magnitudes visible. Take a 1.5 m square lite in a four-sided system at a design wind pressure of 2.4 kPa. The tributary width on one joint is 0.75 m, so the load per metre of joint is 1.8 kN/m, or 1.8 N/mm. At an allowable tensile design stress in the region of 0.14 MPa, the required bite is about 13 mm. Nothing in that result is exotic: bites of roughly 12 to 16 mm are normal for facades of that geometry, and the figure explains why a slim 50 mm cover profile cannot be turned into a structural system by filling it with sealant. There has to be a flat bonding surface of the required width on the frame, square to the glass and uninterrupted.

Depth behaves differently from width. Joint thickness has to be sufficient for the sealant to be applied as a continuous, void-free bead and is usually specified at not less than 6 mm; a joint that is very wide and very thin behaves as a constrained layer and cannot accommodate movement. The width-to-thickness ratio, the shape of the bead, the presence of a bond breaker at the back of the joint and the treatment of the frame corner all belong on the same drawing. An extrusion drawn for a captured system and adapted for bonding by deleting the cap frequently ends up with a bonding face that is too narrow, interrupted by a gasket groove, or not perpendicular to the glass. Those three defects present themselves as a short bite, a path for voids and an uneven stress distribution, and all three were decided at the die rather than on site.

3. Sealant Chemistry, Substrate Compatibility and the Coating That Fails First

Structural sealant for glazing is a one-part, neutral-cure silicone, supplied against a product standard such as ASTM C1184 and evaluated as part of a system rather than as a standalone tube of adhesive. Acetoxy-cure silicones, which are common in sanitary and general construction work, release acetic acid as they cure and are not appropriate against the metallic and coated surfaces of a facade. Choosing the right chemistry is not the difficulty, since it is stated on the data sheet. Keeping the specified chemistry on the project when a site runs short of cartridges is the difficulty.

The bond is only as good as its weakest interface, and on aluminium facades that interface is often the finish rather than the sealant. A powder-coated or PVDF-coated extrusion presents the sealant with a polymer film that is itself bonded to metal, so the adhesion of that film to the aluminium becomes part of the structural load path. A coating that passes a bend test and a coating that will hold a structural bead for twenty-five years are not the same claim. This is why systems are evaluated against a named finish system, and why a late change of colour or coating supplier is a change to the structural joint rather than only to the appearance. Anodised surfaces, mill-finish aluminium, stainless steel and glass each need their own preparation, and the system instructions normally pair that preparation with a specified primer.

Compatibility is the second place where projects come apart. Gaskets, setting blocks, glazing tapes, backer rods and weather seals all sit in contact with the structural bead, and plasticisers or other constituents can migrate into the silicone and change its properties. The screening test that exists for this, in the form of a compatibility assessment of liquid-applied sealants against accessories such as ASTM C1087, is a screen rather than a guarantee, which is why system suppliers publish an approved accessories list and why substituting a cheaper gasket is a structural decision. Adhesion testing on the delivered assembly, described later, remains the only reliable check that the combination works on these particular surfaces.

4. The Loads the Bond Must Carry, and the One It Should Not

The bond carries wind load in tension and shear, and it is designed for that with a large margin. The allowable design stress is a small fraction of the sealant's ultimate tensile strength, because the joint has to survive decades of load reversal, thermal cycling and movement rather than one peak event. Design pressures come from the project's own wind analysis and are demonstrated on the assembly by a structural performance test to ASTM E330, which applies a uniform static air pressure difference and records deflection and permanent set. That test qualifies the assembly as a whole; it does not prove the bond geometry for every panel size on the building, which is why the calculation and the test run in parallel instead of one substituting for the other.

Dead load is the load the bond should not be given. Glass weighs about 2.5 kg per square metre for every millimetre of thickness, so a 12 mm laminated unit is around 30 kg per square metre, and all of that weight acts in shear on the joint permanently. Structural silicone behaves well in tension and far less well in sustained shear, which is why design practice applies an allowable value for sustained shear an order of magnitude below the wind case, in the region of 1 psi against roughly 20 psi, and why the normal solution is to carry each lite on setting blocks or mechanical supports and let the sealant see only the share of load caused by frame deflection and movement. A design that relies on the bead alone to hold the glass up has a slow failure mode built into it.

The third load is movement, and it receives the least attention at tender. Glass and aluminium do not expand at the same rate: aluminium moves roughly two and a half times as much as glass over the same temperature change, in the region of 23 against 9 parts per million per kelvin. On a 6 m mullion that difference amounts to a few millimetres across a seasonal cycle, and the bond has to absorb it without being worked to failure. Joint thickness, bead geometry and the freedom of the glass to move in its own plane decide whether that movement is absorbed or converted into stress at the glass edge, where the bond has least area to redistribute it.

5. Glass Selection: Heat Treatment, Edge Work and Breakage Mode

Structural glazing puts the glass into a load path it would not otherwise occupy, and that changes the glass specification. Heat-treated glass is the normal requirement: toughened glass for strength, or heat-strengthened glass where the flatter surface and the different breakage pattern justify the lower strength. Guidance for bonded systems, including the assembly rules of the EN 13022 series, addresses the glass products that may be used with structural sealant, and a specification that inherits an annealed build from a captured facade sits outside that guidance.

Choosing between toughened and heat-strengthened glass is a decision about how the pane fails. Toughened glass breaks into small dice across the whole lite; heat-strengthened glass breaks into larger fragments that tend to remain in the opening. On a high-rise facade where falling glass is the governing risk, the second behaviour is often preferred, and where toughened glass is used, the risk of spontaneous breakage from nickel sulphide inclusions is managed by heat soaking to EN 14179. That subject has its own arithmetic and is treated separately, but the point for a bonded facade is that a pane which shatters for reasons unconnected with the bond will still be attributed to the bond.

Edge work matters more in a bonded system than in a captured one. The bond is made to the face of the glass, but the edge sits close to the joint, and a shell chip or an unpolished arris on a heat-treated lite is a stress raiser in the zone where the joint transfers load. Edges should be ground and, for exposed work, polished, and any cut-out or hole has to be specified before toughening because it cannot be added afterwards. Dimensional tolerance matters for the same reason: bite is measured from the edge of the glass, so a lite that is 3 mm short on one edge has 3 mm less bond width. On a joint designed with a modest margin that is not a cosmetic matter.

6. Frame Stiffness, Movement and the Joint With Nowhere to Go

A bonded joint is stiff, and it makes the frame stiffer, but it also couples the glass to every deflection the frame makes. The serviceability limit for mullion deflection on a glass facade is commonly written as span divided by 175, or about 20 mm, whichever is less, and the reason it is written at all is that excessive deflection both opens the weather joint and works the structural bead. Under wind a mullion that deflects visibly rotates the joint at each end and changes its thickness; over a season of gusts the sealant experiences that as fatigue rather than as a single event.

Frame design for a four-sided system therefore begins with stiffness and tolerance rather than with appearance. The bonding face has to run continuously around the panel, in one plane, at a defined width, with no fastener or frame joint interrupting it. Where a transom meets a mullion, the bead must either continue through the intersection or the geometry must be arranged so the two runs are bonded separately with no gap at the corner. Water must not be able to stand behind the bead: a joint that traps water against the bond line combines continuous wetting with freeze-thaw in cold climates, and the tide mark it leaves at the glass edge is often the first visible symptom of a joint that has been leaking for a year.

Drainage and ventilation of the cavity behind the facade belong on the same drawing as the bond. In a four-sided system the structural bead frequently doubles as the weather seal, which is why the sealant has to be UV-stable and why bead shape and continuity are not optional details; some systems add a separate weather seal in front of the structural bead, which brings its own compatibility questions. The detail that fails most often is the base condition, where the facade meets a waterproofing membrane and each trade expects the other to close the interface. Water entering there does not show as a leak in the room for months. It shows as a dark tide mark at the bottom of a panel, and by then the bond line has been wet for a season.

7. Factory Bonding, Cure Control and Adhesion Verification

Bonding is a factory operation in most four-sided systems, and it should be. A dedicated area, controlled temperature and humidity, clean handling, a defined cure period before the panel is moved, and a record of which cartridge went into which panel are what separate a bonded facade from glass that has been glued. Cure of a one-part silicone depends on moisture from the air and progresses from the surface inwards, so a thick bead in a dry, cold plant cures far more slowly than the same bead in a warm, humid one. Panels transferred to a stillage before the bead has developed handling strength are the origin of joints that look perfect and carry nothing.

Verification is mechanical, inexpensive, and the part of the process a buyer should insist on witnessing. Tensile adhesion testing to ASTM C1135 establishes whether the sealant bonds to the actual glass and the actual coated aluminium in use, and the result is reported as the proportion of cohesive failure within the sealant against adhesive failure at an interface, with the system's own criteria defining a pass. Peel testing to ASTM C794 offers a second, more sensitive view of the same interface. In production, a destructive test on a sample from the run, or a witness bead cured alongside the panels, provides the continuing check. These are standard tests reported pass or fail against published criteria; none of them requires a laboratory the size of the plant, which is precisely why their absence is hard to justify.

The record that travels with the panels matters as much as the test. A bonded facade should be traceable batch by batch: sealant product and lot, primer and lot, date and time of bonding, ambient conditions, cure time before handling, and the adhesion result for that batch. When a defect appears three years later, that record is what allows one batch to be investigated instead of the whole facade being condemned. A supplier who cannot produce it is asking the project to accept an unverifiable process, which is a different proposition from accepting an untested product.

8. Specification Checklist and the Failures That Appear in Service

What is specified at tender decides what can be verified at handover. The table sets out the items that belong in a specification for a bonded facade and what each one actually demonstrates.

ItemWhat to requestWhat it demonstrates
System evaluationEvidence that sealant, primer, glass and finish are evaluated together as one kit, with the coating system namedThat the bond is a tested combination rather than a product chosen on price
Joint calculationBite and thickness for each panel size, with design pressure and allowable stress statedThat joint geometry was designed for the loads on this facade
Glass specificationHeat-treated build, edge work and size tolerance tied to the joint designThat the glass matches what the joint calculation assumes
Substrate and finishCoating system and supplier, with the adhesion of that coating to the extrusionWhere the weakest interface in the load path will be
Dead load supportSetting blocks or mechanical supports, with the sealant excluded as the sole vertical supportThat the slow failure mode has been designed out
Assembly testStructural performance to ASTM E330 at the project design pressureThat the assembly meets the specified pressure, and how it deflects
Adhesion testingTensile adhesion to ASTM C1135 and peel to ASTM C794, per batch, with pass criteriaThat the interface works on these actual surfaces
Production recordSealant and primer lots, bonding date, ambient conditions, cure time before handlingTraceability if a defect appears in service

Failures in service cluster in three places, and none of them is mysterious. The first is the interface: adhesive failure at the glass or at the coated aluminium, usually traceable to surface preparation, a missed primer, or a coating changed after the system was evaluated. Adhesion failure is progressive, because once water reaches an interface the bonded area shrinks and the stress in what remains rises. The second is geometry: a joint thinner than designed because the frame was not square, a bite reduced because a lite was cut short, or voids behind a bead that was tooled over a gap rather than applied as a solid. The third is movement, where a frame deflects more than assumed or a glass is restrained in its own plane and transfers load into a bead that was never sized for it.

  • Which system is being supplied, and is each condition two-sided or four-sided?
  • What bite and thickness are calculated for the largest panel on the facade, and against which design pressure?
  • Is the coating on the aluminium the one the sealant was evaluated against, and who confirms adhesion to it?
  • How is the dead load of each lite carried, and what happens to the joint if a setting block compresses over time?
  • Where are the batch records and adhesion results for the panels delivered to this project?

Planning a Commercial or Villa Project?

Send your window schedules or architectural CAD drawings to the OPPGATE engineering team for a prompt, certified quotation and structural review.

Request an Engineered Quotation
Request a QuoteSend Drawings