Deflection limits & movement
A facade fails twice: once when it breaks, and long before that when it bends too far or cannot follow the building as it moves, swells and sways.
The glass on a curtain wall almost never breaks because it is too weak. It breaks because the frame holding it bent 30 mm and the glass could only take 20.
A facade lives in a moving world. The wind pushes its mullions into a bow. The afternoon sun heats a dark spandrel by 40 degrees and it expands. The concrete frame behind it shrinks and creeps for years. And in a storm or an earthquake the whole building leans, so the floor above shifts sideways relative to the floor below. None of these break the facade by overstressing it - they break it by _moving_ it: bending a mullion until the glass edge crushes, or shoving one floor past another until a stiff panel shatters in its frame. This is why facade design is governed as much by **serviceability** - how much it may move - as by strength. The skin must bend, swell and sway with the building, and still keep the weather out.
Deflection limits and the three movements a facade must absorb
span/175, L/240 and the 19 mm rule - serviceability limits that protect glass and seals
A mullion that is strong enough not to break can still bend far enough to be a problem. Too much deflection cracks the glass it holds, breaks the weather seals, makes doors and vents bind, and simply looks alarming. So we cap deflection at a serviceability limit, checked under the 1.0x (unfactored) wind pressure from Lesson 4.1 - not the 1.5x ultimate load.
The industry limits, widely used on Indian and international curtain wall, are:
- span/175 for a single-glazed framing member (a common base limit), or - L/240 for members supporting brittle finishes / insulated glass, with - an absolute cap of 19 mm (about 3/4 inch) regardless of span - because past ~19 mm even a long mullion stresses the glass edge and the gaskets too much.
The governing limit is whichever is smaller. For a 3.5 m mullion, span/175 = 20 mm but the 19 mm cap rules; for a 4.5 m mullion, span/175 = 26 mm, so the 19 mm cap rules hard. The lesson: on tall floor-to-floor mullions the 19 mm cap, not the span ratio, is usually what sizes the section.
Strength asks 'will it break?'. Serviceability asks 'will it move too much first?'. On a curtain wall, serviceability almost always answers first.
Thermal, building and seismic movement - and the joints that absorb them
Beyond wind bow, a facade must absorb three kinds of movement, and each gets a detail.
Thermal movement. Aluminium expands about 0.000023 per degree C - roughly 2.3 mm per metre per 100 degrees C. A 3 m dark mullion swinging through a 60-70 degree C surface-temperature range moves ~4-5 mm; over a full storey-height unitised panel that adds up, and it is taken up at the stack joint between units. Glass, aluminium and the concrete frame all expand at different rates - differential thermal movement - which is why dissimilar materials are never rigidly locked together.
Building movement. The concrete frame shrinks, creeps and deflects under load over years; columns shorten differentially. The facade's connection to each slab must allow vertical and horizontal slip so this slow movement is not transferred into the brittle skin.
Seismic and wind drift. Under lateral load the building sways, and each floor moves sideways relative to the one below - inter-storey drift. Codes limit storey drift to about h/250 to h/500 of the storey height. For a 3.5 m storey, h/250 ~ 14 mm of horizontal racking the facade must follow at every level. Unitised systems absorb this by letting panels rock or slide at the stack and the anchor; stiff, rigidly-fixed panels (and stone) crack instead - which is why drift, not force, is the earthquake's real threat to a facade.
Build the movement in: slip anchors, stack joints and the rule of dissimilar materials
Movement is designed in, not resisted. Three habits do most of the work.
First, never rigidly fix two things that move differently. Glass floats in its frame on setting blocks and edge clearances; panels connect to the slab through slotted or slip anchors that lock the load path but free the movement direction. A bolt through a round hole carries load both ways; the same bolt in a slotted hole carries load one way and slides the other - that single detail absorbs thermal and building movement.
Second, the stack joint is the facade's expansion joint. Between vertically stacked unitised panels a deliberate gap (often a few millimetres, with interlocking gaskets) lets each panel grow, shrink and rock without loading its neighbour, while staying weathertight via a labyrinth/pressure-equalised seal.
Third, size every joint to the sum of the movements it must take: thermal + building + the share of drift at that location, plus an installation tolerance. Undersize it and the panels grind and the glass cracks in the first hot, windy season; oversize it and the joints leak and look crude. Movement accommodation is the quiet half of structural facade design - the half that decides whether the skin survives its first decade.
Movement is why facades have visible joints - and why a 'seamless monolithic' skin is usually a fiction. If you want long uninterrupted lines, understand you are fighting thermal and drift movement, and the engineer will need either generous concealed joints or expensive special details. Dark colours and large panels both increase thermal movement; a deep west-facing dark spandrel is a thermal-movement problem you are designing in. Accept the stack joint as part of the language of the facade rather than something to hide, and the skin will last.
Run both checks on every mullion: strength under 1.5x wind, deflection under 1.0x wind against the smaller of span/175 (or L/240 for IGU) and 19 mm. Then build a **movement budget** for each joint: thermal (alpha x deltaT x length, with the differential between materials), building (shrinkage/creep/column-shortening from the structural engineer), and the storey-drift share (h/250-h/500). Detail slip anchors so the load path is determinate while the movement direction is free, and verify the stack joint stays weathertight at full racking - that is exactly what the dynamic-movement test on the performance mock-up proves.
Those small gaps and slotted holes are not sloppy fabrication - they are the movement joints, and they must be installed as drawn. If you pack a stack joint solid with sealant, bolt a slip anchor tight in its slot, or omit the setting blocks under the glass, you have just removed the facade's ability to move - and it will crack in the first hot or windy spell. Check that slotted anchors can actually slide, that glass edge clearances are kept, and that expansion gaps are clean and the right width. The facade is meant to move; let it.
AAMA TIR-A11 / curtain-wall practice
Deflection limits
The source of the span/175 and L/240 framing deflection limits and the 19 mm absolute cap used on most curtain wall. They are serviceability rules, not strength - checked under unfactored wind.
IS 1893 (Part 1): 2016
Inter-storey drift limits
Caps building storey drift (commonly ~0.004h, i.e. about h/250) under design seismic load - the racking the facade must accommodate. It limits the structure's drift, not the facade's own detailing.
IS 875 (Part 3): 2015
Serviceability wind
Provides the 1.0x (unfactored) wind pressure used for the deflection check, distinct from the 1.5x case used for strength. It gives the load, not the deflection limit.
CWCT Standard / EN 13830 (global)
Curtain wall performance
Define serviceability deflection and movement-accommodation requirements and the dynamic-movement test that proves a unitised system stays weathertight while it racks. Limits vary by spec; always read the project's own value.
“If the glass and frame are strong enough not to break under the design wind, the facade is structurally fine.”
Strength is only half the design. A member can be far from breaking yet deflect enough to crack the glass it carries, burst the weather seals, or fail to follow the building's drift - all serviceability failures that occur well below the ultimate load. Curtain wall is routinely governed by the deflection limit (span/175 or the 19 mm cap) and by movement accommodation, not by strength. A facade that passes strength but ignores deflection and drift will leak and crack long before it would ever collapse.
Worked example - deflection check on a curtain-wall mullion
We sized the load in Lesson 4.1. Now check whether a chosen aluminium mullion is stiff _enough_ - the serviceability check that usually governs curtain wall.
The SLS wind pressure from Lesson 4.1, the mullion's section properties (I, from the extruder's catalogue), the deflection formula, and the project's stated deflection limit.
GIVEN (a vertical mullion spanning floor-to-floor, simply supported):
Span (storey height) L = 3.5 m = 3500 mm
Mullion tributary width w = 1.5 m (panel centre-to-centre)
SLS wind pressure p = 2.0 kPa = 0.0020 N/mm2 (1.0x wind, Lesson 4.1)
Aluminium modulus E = 70,000 N/mm2
Trial mullion I = 9.0 x 10^6 mm4 (second moment of area)
Line load on mullion q = p x w
Mid-span deflection (UDL, simply supported):
d = 5 q L^4 / (384 E I)
LIMIT = lesser of L/175 and 19 mm- 1Convert the pressure to a line load. The mullion carries the wind over its tributary width: q = p x w = 0.0020 N/mm2 x 1500 mm = 3.0 N/mm along the mullion.
- 2Compute the deflection. d = 5 q L^4 / (384 E I). Numerator = 5 x 3.0 x (3500)^4 = 5 x 3.0 x 1.5006 x 10^14 = 2.251 x 10^15. Denominator = 384 x 70000 x 9.0 x 10^6 = 2.4192 x 10^14.
- 3Divide. d = 2.251 x 10^15 / 2.4192 x 10^14 = 9.3 mm of mid-span bow under serviceability wind.
- 4Find the limit. L/175 = 3500 / 175 = 20.0 mm; the absolute cap is 19 mm. The governing limit is the smaller: 19 mm.
- 5Compare. Actual 9.3 mm < limit 19 mm -> the mullion passes, with healthy margin (it uses only ~49% of the allowance). Good - we want margin, because the limit also protects the glass edge and the seals.
- 6Stress-test the section. Suppose the architect lengthens the storey to L = 4.5 m. Then d scales with L^4: d = 9.3 x (4.5/3.5)^4 = 9.3 x 2.73 = 25.4 mm - now FAILS the 19 mm cap. You would deepen the mullion (raise I) or add a transom to shorten the span. This is exactly why taller storeys force deeper facade framing.
- 7Note the drift you must also absorb. Separately, at this 3.5 m storey, an h/250 inter-storey drift = 14 mm of horizontal racking - which the stack joint and slip anchors, not the mullion, must accommodate. Stiffness and movement are two different checks on the same panel.
You’ll walk away with
A clear pass/fail on mullion stiffness: 9.3 mm actual against a 19 mm limit - the serviceability check that, with the strength check from 4.1, fully qualifies a framing member. Plus the insight that storey height drives deflection by the fourth power, so it is the most expensive geometric change you can make.
Two quick movement reflections.
- 01Re-run the deflection for the same mullion but doubling its second moment of area I (a deeper section). Deflection halves to ~4.7 mm - showing that stiffness, not strength, is what depth buys you, and why curtain-wall mullions are deeper than they 'need' to be for strength alone.
- 02Estimate the thermal movement of a 3 m dark aluminium mullion over a 60 degree C swing: alpha x deltaT x L = 0.000023 x 60 x 3000 = ~4.1 mm. That is the movement the stack joint above and below it must quietly absorb.
A facade is governed by serviceability as much as strength: deflection is capped at the lesser of span/175 (or L/240 for IGU) and 19 mm under unfactored wind, and the skin must absorb three movements - thermal, building and inter-storey drift - through slip anchors and stack joints. Strength asks 'will it break'; serviceability and movement ask 'will it move too far first', and on curtain wall they usually answer first.
Deflection limit = lesser of span/175 (L/240 for IGU) and 19 mm, checked under 1.0x wind. Three movements: thermal (alpha=23e-6/C for aluminium, ~2.3 mm/m/100C, differential between materials), building (shrinkage/creep/column shortening), and inter-storey drift (~h/250-h/500). Absorbed by slotted/slip anchors and stack joints. Deflection scales with span to the fourth power - taller storeys force deeper framing.
What is the deflection limit for a curtain-wall mullion?
A common serviceability limit is the lesser of span/175 (or L/240 where the member supports brittle/insulated glass) and an absolute cap of about 19 mm, checked under the unfactored (1.0x) design wind pressure - not the factored strength load. On tall floor-to-floor mullions the 19 mm cap usually governs because span/175 of a 3.5-4.5 m member already exceeds it. The limit exists to protect the glass edge and the weather seals, not to prevent collapse.
What movements must a facade accommodate?
Three. Thermal movement, as aluminium, glass and concrete expand at different rates with temperature (aluminium ~0.000023 per degree C). Building movement, as the frame shrinks, creeps and columns shorten over years. And inter-storey drift, as the building sways under wind and earthquake and each floor moves sideways relative to the one below (codes cap this around h/250 to h/500). These are absorbed by slip/slotted anchors and stack joints, not resisted rigidly.
Why is inter-storey drift, not seismic force, the main earthquake risk to a facade?
A curtain wall is light, so the seismic inertial force on its own mass is small and rarely governs. But the earthquake makes the building sway, and each floor racks sideways relative to the floor below by several millimetres to a couple of centimetres. A facade rigidly fixed to two floors cannot follow that racking and the glass shatters. Unitised systems are detailed to rock or slide at the stack joint and anchors, absorbing the drift - which is why drift, not force, drives seismic facade detailing.
Peer-reviewed journals & authoritative standards
- 01Bedon, C. et al. Performance of structural glass facades under extreme loads - design methods, existing research, current issues and trends. Construction and Building Materials, 163. — Construction and Building Materials (Elsevier), 2018.
- 02Review on Glass Curtain Walls under Different Dynamic Mechanical Loads: Regulations, Experimental Methods and Numerical Tools. IntechOpen. — IntechOpen (peer-reviewed chapter), 2023.
- 03Wind Resistance Performance Evaluation of a Cable-Type Curtain Wall System on Reinforced Concrete High-Rise Buildings. — (peer-reviewed; University of Bath research portal), 2021.
_We have loaded the panel and limited its movement. Now we must size the most fragile thing it holds - the glass itself - as a structural plate that must resist the wind pressure without breaking or deflecting too far. Glass thickness, next._