Glass has no yield point, so it is designed to a probability
Every other structural material here has a strength: a yield stress, a characteristic cube strength, a bending grade. Glass does not. It is perfectly brittle, it does not redistribute stress around a flaw, and its strength is governed by microscopic surface damage rather than by composition — so two panes from the same furnace break at different loads.
The consequence is that glass design is STATISTICAL. The accepted basis is a probability of breakage, conventionally eight lites per thousand at the design load, and the published load resistance of a pane is the load at which that probability is reached. A result expressed this way is not hedging: it is the only honest form the answer has.
Strength is also raised by what is done after the float line rather than by the glass itself. Heat strengthening roughly doubles the allowable stress of annealed glass and full tempering roughly quadruples it, both by locking a compressive skin into the surface so that an applied tension has to overcome it before any flaw sees tension at all. That is also why tempered glass cannot be cut or drilled afterwards — releasing the skin releases the whole pane.
Load duration changes the answer, and by a lot
Glass suffers STATIC FATIGUE: a flaw under sustained tension grows slowly in the presence of moisture, so the same pane carries a three-second wind gust far better than it carries a load applied for a month. Design charts are therefore published against a stated duration, conventionally three seconds for wind, and a longer duration requires the load to be converted to its three-second equivalent before the chart can be read.
This is why a balustrade, which carries a crowd load that may stand for hours, and a window, which carries a gust, are not checked the same way even at the same pressure. It is also why snow and dead loads on sloped glazing and skylights are the demanding cases rather than wind.
The practical trap is a calculator answering for one duration being used for another. Every page here states the duration its result assumes, because a three-second resistance used for a sustained load is unconservative in a direction the number itself does not reveal.
The insulating unit: the cavity works, and the edge does not
A sealed unit insulates with the GAS in its cavity, not with its glass — glass itself is a poor insulator and two panes of it are barely better than one. What the cavity does is stop convection and conduction across the gap, and a low-emissivity coating on one of the cavity-facing surfaces then blocks most of the radiant transfer that remains.
The cavity does not keep helping indefinitely. Below roughly ten millimetres conduction across the gas dominates and the unit is poor; above roughly sixteen the gas begins to convect in a loop, carrying heat from the warm pane to the cold one, and performance flattens or falls. There is an optimum width rather than a rule that wider is better, and it moves with the fill gas because argon and krypton have different viscosities.
The edge is the weak point. The spacer bar bridges the two panes with a conductive path the centre of the unit does not have, and the heat it carries is concentrated in a strip a few centimetres wide. This is accounted for as a LINEAR thermal transmittance along the perimeter — a term proportional to length rather than to area — which is why a window's overall U-factor depends on its size and shape and not only on its specification. Many small panes perform worse than one large one of identical make-up, because they have far more perimeter per unit of area.
- U_g, A_g
- centre-of-glass transmittance and the glass area
- U_f, A_f
- frame transmittance and the projected frame area
- ψ
- linear transmittance of the glass edge and spacer, per unit length
- L_g
- total perimeter of the glazing — the term that penalises many small panes
U-factor and solar gain are different questions and often conflict
A window loses heat by conduction and gains it by admitting sunlight, and the two are described by separate numbers: the U-factor for the first, the SOLAR HEAT GAIN COEFFICIENT for the second. A coating that improves one frequently worsens the other, so there is no single best glass — only a best glass for a climate and an orientation.
Which surface a low-e coating sits on decides which way it works. In the conventional numbering, a coating on the inner face of the outer pane rejects solar gain before it enters; one on the outer face of the inner pane keeps interior heat in. The same coating on the wrong surface produces a unit that performs as specified in the wrong season.
External shading multiplies the gain rather than adding to it, which is why an overhang or a fin is so effective on a high-SHGC glass and so nearly pointless on a low one. The effective coefficient is the product of the glass's own and the shading's, and shading placed OUTSIDE the glass is worth substantially more than the same device inside, because interior blinds absorb the energy after it has already entered the room.
Condensation is a surface-temperature question, not a humidity one
Condensation forms when a surface falls below the dew point of the air touching it. So the question a glazing calculator actually answers is not whether the room is humid but how cold the INTERIOR GLASS SURFACE gets, which follows from the outdoor temperature, the indoor temperature and the assembly's thermal resistance.
It appears first at the edges and in the corners, for the reason in the previous section: the spacer conducts, so the perimeter strip is the coldest interior surface in the window. A unit that is clear in the middle and wet at the bottom edge is behaving exactly as its physics says, and improving it means a warm-edge spacer rather than a better coating.
The counter-intuitive case is that better windows can make condensation appear ELSEWHERE. Replacing leaky windows removes the ventilation they were providing and raises indoor humidity, and the coldest surface in the room is then a wall corner or a lintel rather than the glass. The moisture did not appear; it moved to the new coldest place.
Structural silicone, setting blocks and clearance
In structural silicone glazing the sealant carries the wind load from the glass into the frame in TENSION and shear, with no mechanical capture on the structural edges. The bite — the width of the bond — follows from the wind pressure, the shorter span of the lite and an allowable stress that is deliberately low, conventionally around 138 kPa (20 psi), because the material is a rubber bonded to two dissimilar substrates in the weather for thirty years.
The bite and the GLUE LINE THICKNESS answer different questions and are sized separately. The bite resists wind; the thickness accommodates differential movement between glass and frame as they expand at different rates, and a joint that is too thin tears itself apart in service without ever seeing a design wind.
Setting blocks carry the pane's dead weight and belong at the quarter points rather than at the corners, because a block at a corner puts a point load into the most highly stressed part of the lite. Edge clearance exists so that the frame can expand without contacting the glass — and in fire-rated assemblies that clearance is a listed value from the test report rather than a calculation, because the qualifying evidence is the test, not the arithmetic.
Where it fails: thermal stress, deflection limits, and what only a test can say
THERMAL STRESS BREAKAGE is the failure mode that catches people, because it happens with no wind and no impact. A pane partly in sun and partly shaded — by a reveal, a spandrel, a blind held close, a sticker — expands unevenly, and the hot centre pulls against the cool edge that is still held by the frame. The edge is also the weakest part of the pane, because cutting leaves flaws there. Heat-strengthened glass is specified for this reason far more often than for wind.
Mullion deflection limits are commonly misread as protecting the glass. They protect the SEALS and the gaskets: a mullion that bows takes the glazing pocket with it, and the joint that was weathertight at rest opens. That is why the limits are tight — spans over a stated length are often held to a fixed absolute deflection rather than a span ratio — and why they apply to the serviceability wind rather than the ultimate one.
Acoustic and blast performance are TEST results, not calculations. An STC rating comes from a laboratory measurement across sixteen one-third-octave bands and is defeated by the frame, the perimeter seal and any flanking path; a blast rating is a performance condition observed in an arena test, describing where the fragments went. The estimators here place an assembly in the right region before a specification is written, and every one of them says on the page that the rating itself comes from a report.
Calculators that use this method
Basis
- ASTM E1300, Determining Load Resistance of Glass in Buildings — the probability-of-breakage basis, the glass type factors for heat-strengthened and tempered, and the load-duration conversion.
- ASTM C1036 and C1048 for float and heat-treated glass; ASTM C1172 for laminated glass.
- ASTM E2190, Insulating Glass Unit Performance and Evaluation — durability of the edge seal that the linear transmittance term describes.
- NFRC 100 (U-factor) and NFRC 200 (SHGC and visible transmittance), including the edge-of-glass and frame contributions to a whole-window rating.
- ISO 10077 and EN ISO 12631 for the linear thermal transmittance of the glazing perimeter used in the formula above.
- ASTM C1401, Guide for Structural Sealant Glazing, and sealant manufacturers' published design stress for structural silicone — the basis of the bite calculation.
- AAMA 501 series, ASTM E283, E331 and E330 for air, water and structural testing of curtain wall assemblies — performance is established by mock-up rather than by calculation.
- ASTM E90 and E413 for sound transmission loss and the single-number STC rating; GSA and ASTM F1642 for the blast performance conditions referred to here.
- IBC Chapter 24 and the Glass Association of North America Glazing Manual for setting block position, edge clearance and safety glazing locations.