Compatibility between Building Movements and Curtain Walling – A Dark Art…?
There seems to be a growing demand for curtain walling systems integrating single piece glazing, extending almost full storey height. This puts a greater demand on façade engineers and contractors….which is worthy of discussion.
For façades up to 3-4 storeys, it is normal for the curtain walling to be “stack loaded”, usually from ground floor level. (Each mullion acts as a column carrying all the load from the façade above, down to the slab or foundation.)
Above this height, where the compressive load in the mullions becomes higher, stack loading becomes less economical, and it becomes necessary to change the loading principle. This is often achieved by “splitting the screens” (with the provision of a gap at the mullion joint) so that the weight of screen can be divided and allocated to the desired level.
Structural engineers must design the building structure to some reasonable deflection criteria. At the time of scheme design, the exact type of façade system is rarely defined. Codes of practice do give some outline recommendations for limiting deflection of edge beams. This does vary but values of span / 360 under total load, and span / 500 under live load are not uncommon.
On face value, limiting deflections to these criteria may seem reasonably stringent and one could expect this to be compatible with the systems that will be used. However, that is often not the case.
To demonstrate this, we need to consider a typical situation in more detail.
It is important to identify that, as façade engineers, we are generally only interested in movements that occur both during and after we install the curtain walling, on the assumption that the façade will be “set level” upon installation.
If we then consider what happens as the supporting structure moves under live loading after installation:
Fig 2 - Glass Movements in Frame for Simply Supported Edge Beam
Fig 1 - Typical Glass Panels in Curtain Walling before any Slab / Floor Movement
Fig 3 - Glass Movements in Frame for Continuous Edge Beam
With reference to Fig 2, for a theoretical “simply supported” edge beam condition (usually associated with a steel frame under gravity loading) the panels near the supports undergo the highest level of rotation.
With reference to Fig 3, for a theoretical “continuous” edge beam condition (usually associated with a concrete / moment frame) then it more likely that the panels further in from the ends will undergo a higher degree of rotation.
To give this some relevance, let us now consider a simply supported edge beam spanning 6m with 6 bays of glass. If this beam is designed to a maximum mid span live load deflection of span / 500, then this equates to a maximum deflection of only 12mm. Deflections of this order are barely visible. However, assuming some parabolic deflection, over any bay, the “relative deflection” could be in the order of 6mm to 7mm. Again, this does not sound significant.
However, this is when the issues with tall, narrow glass panels become apparent. If the glass panels are 4m high, then the high aspect ratio leads to lateral movement at the top of the glass of between 24mm-28mm.
Under floor deflection the curtain walling doesn’t “tilt”, instead staying vertically aligned. In most curtain walling systems, the glass movement would be restricted by the mullion nosing, where metal / glass contact can occur. This gives an increased theoretical risk of glass breakage. In some situations, the full weight of the glass panel could be transferred to only one side, leading to higher loading on the glass support bracket as well as the mullion / transom connections. In most cases, movement of this order is likely to significantly reduce the “bite on the glass”, leading to a compromise in the integrity of the system.
Different manufacturers produce different systems that can accommodate different amounts of movement. However, it is generally true that the greater the tolerance of the system, the wider the sections will be. 50mm wide systems cannot accommodate as much movement as an 80mm wide system. There is usually a trade-off, and the Geordie saying of “you get nowt for nowt” is apt.
Quite often the compatibility issues become apparent late in the design process, so some compromises are necessary.
An easy solution is to avoid tall glass panels, and instead split them with additional transoms. It is quickly acknowledged here that this can compromise the aesthetic appearance. Once the suggestion has been made and the Architect has finished hyperventilating into a paper bag, another option must be sought.
Increasing the stiffness of the edge beams is usually the next suggestion. However, if floor loads are high, spans are large or there are spatial constraints, then this may not be practical. Furthermore, this usually requires a substantial uplift on steelwork, at significant increase in cost.
Introduction of specially designed “façade beams” (that support the weight of the façade and isolate it from vertical deflection of the floor beams) is another option, but again this can introduce spatial constraints that need to be addressed early in the design process or require modifications to be made.
A further option often considered to make the problem go away involves hanging curtain walling screens from stiff roof beams. This can overcome issues with mullion buckling, but installation often becomes more difficult, and due to various features common to the roofs on most commercial buildings, is often not practical.
One further (and often last) option is to try and “work with the movement” by introducing packing within the system to control the tilt and movement of the glass (historically known as “top and tailing” the glass). However, this is only possible in some situations and requires an expert understanding of all the relevant issues for it to be executed successfully.
This puts the façade engineer and contractor under considerable pressure, designing and installing to much more stringent tolerances than should be necessary.
So how can such situations be avoided in future?
The most obvious way is to appoint a façade consultant at concept design stage to advise the architect and engineer of general constraints. (We would of course say that!)
Engineers should read guidance from the SCI (which is applicable for other structures) that outline some of the principles of movement and tolerance, so they can advise the contractor of the need to reduce deflections in critical areas.
System manufacturers should be able to offer advice to architects earlier in the design process and let them know what is and isn’t possible with their curtain walling systems.
It is interesting to note that there are a significant number of curtain wall screens supported from structures that are (in theory) not compatible with these limited movements. Why don’t they fail?
We suppose that depends on what failure looks like. There may be a lot of existing situations where the glass is leaning in the frames, but not sufficient as to be obvious or cause stress capable of cracking the glass.
It is also possible that suspended slabs are not being loaded to their theoretical maximum value. Work done by Mitchell and Woodgate in CIRIA report 24 goes into the statistical derivations – but this is now 50 years old, and general building usage has changed since then.
Some respected organisations suggest that maximum theoretical deflections do not occur due to oversimplifications in analysis models which among other factors, neglect nominal fixity in connections. Whilst we agree that this is likely to be the case, to quote W. Edwards Deming, “Without data, you’re just another person with an opinion”.
Until further research and design guidance is available, then façade engineers must work to the information provided, and therefore compatibility between curtain walling tolerance and building movement remains a complex and troublesome issue.
We are currently commissioning our own research into this. If any other organisations or individuals feel able to contribute to this, then please get in touch with us.
This is a blog post, not a full technical paper. Assessment of movement and tolerance is a complex issue and involves many other factors; this is a simplified example to demonstrate a general problem. If you have any queries or we can assist you then please don’t hesitate to get in touch.