So you think you're safe? Part 3

So you think you're safe? Part 3

Glazing used as a smoke baffle

Wow! Another pretty intense week. I don’t know about you, but for my part, between the training given at ÉTS for glass walls acting as guardrails and the engineering mandates, it’s been a pretty busy week! It should be mentioned that I live in the Quebec City region, so traveling to Montreal complicates things a bit. Fortunately, as I now travel by train, I can work while traveling.

Despite this busy schedule I was able to find some time to write another article that should interest you. It must be said, with a Canadian standard for the calculation of glass (CAN/CGSB-12.20-M89) which was written in 1989 and never revised since, it is not always easy to design glass elements in a way that is safe. It is often necessary to refer to standards elsewhere. This is what we are currently doing within the committee responsible for updating standard 12.20. We are looking at the American and European standards on which we will base ourselves for the overhaul of the Canadian standard which should be published in 2 to 3 years.

Today, I want to talk to you about an error that is sometimes found in certain specifications. This is about the thickness of the glazing used as a smoke baffle. In fact, the error does not come from the specifications, but rather from certain hardware suppliers who have developed smoke baffle baseshoes for thick... very thick glass in fact. When I talk about a baseshoe, I'm talking about the thick aluminum 'U' extrusion which is used to connect the glass to the structure. In fact, many baseshoes are designed for 12mm thick glass.

To begin with, something needs to be clarified regarding safety glass. It is wrongly believed that tempered glass breaks into thousands of small pieces when they shatter, but in fact, they will instead form pieces of cracked glass (glass clumps) which will shatter when they hit an object (fragmentation), usually the ground or if you are less lucky, a person. There have been cases of injuries from railings in Ontario (glass falling on a passerby from a balcony railing). See figure 2.7 and the text in yellow:

Also look at the following image where part of a tempered glass is still hanging on a railing. This shows that even broken, tempered glass can remain clumped together and fall into large pieces:

This image is taken from the document produced by the committee responsible for analyzing broken guardrail glass in Ontario. It is this study which gave birth to the SB-13 supplement and the CSA A500 standard for guardrails… we will talk about it a little further.

Another problem with tempered glass is the fact that the surface stress required by the ASTM C1048 standard is not high enough (69 MPa). For those who are less familiar, this surface stress is the compression found on the surface of tempered glass and which gives it its additional resistance.

By definition, a "safety" tempered glass should break into thousands of small, non-sharp pieces, but in reality the standard that governs safety glass in Canada (CAN/CGSB-12.1-2017) allows for pieces of glass up to 4in (100mm) in length.

On the other hand, according to a study presented at Glass Performance Day 2009, we could sometimes end up with even larger fragments. Below, we can see the size of the pieces of 4mm thick “tempered” glass. The stress is in theory “compliant”, i.e. 72 MPa:

The black square is 50 x 50mm. We can therefore see that the piece of glass in which the square is located is approximately 180mm long. We can also see that this piece is very sharp at the bottom. According to this same study, here is the size of the particles that can be found in tempered glass depending on the surface stress:

It would certainly take more testing to draw precise conclusions, but it is still worrying when we look at the size of the fragments on the left on the vertical axis. Some tests give fragments of more than 300mm… for surface stresses greater than 90 MPa.

Remember that the American standard (ASTM C1048) which is also used by Canadian glass manufacturers recommends 69 MPa as a minimum surface stress. It should be noted that most glass manufacturers will apply heat treatment according to parameters allowing surface stress to be significantly higher than 69, but all the same, we can see certain gaps in the standards.

For the skeptics, I should mention that as part of the glazing technical committee for the update of the Curtain Wall Guide, a member of the team working for a glass supplier did a test to validate the fragment dimension of glass with a surface stress of around 69 MPa and he obtained results similar to what we found above for 4mm glass with 72 MPa.

As for the graph above, the data is missing for surface stresses of less than 90 MPa, because on the European side the surface stresses are generally higher, but we can see that according to the standards here you can end up with quite large pieces of glass. In fact, it depends on how the glass supplier does its quality control, because it should normally detect this anomaly and correct the parameters since the standard for safety glass (CAN/CGSB-12.1) does not allow fragments of glass greater than 100mm (4in).

On the other hand, glass manufacturers must pay attention to the size of the samples used for quality control. The smaller the sample, the smaller the glass pieces will be. So, if the quality control is carried out on very small samples, it is possible that the fragment dimension is consistent, but in reality, this does not necessarily guarantee that this is the case if the glass panes produced are very large, because the size of the fragments varies with the size of the glass:

Above we can see that small samples will give fragments approximately 2 times smaller than the largest samples.

When glazing ends up above people's heads for applications like smoke baffles, I believe this information becomes even more relevant.

Otherwise, in terms of spontaneous breakage, the risk is quite low for smoke baffles because they are inside and the heat gain is generally low... to be validated depending on the presence of curtain walls, windows and skylights..

On the other hand, the risk of spontaneous breakage varies with the weight of the glass... it works per ton of glass produced, so glass of the same size, but thicker, will be more at risk. It should also be mentioned that there are other possible causes of breakage... glass having been hung before installation and which breaks under low stress later, glass which is too tight (tolerance problem around the holes) and which breaks following movement of the structure, etc.

On another note, Ontario adopted supplement SB-13 for guardrails. This follows the numerous glass guardrails that fell from condo balconies in Ontario around 2010. This document concerns guardrails, but it mainly aims at one thing, to protect people below from pieces of glass that could fall from a guardrail in the event of breakage.

According to this document, guardrail glass that were not treated with a "heat soak test" can be installed if it is located more than 150mm from the edge of the slab (inwards). In this way the glass will fragment on the 150mm of slab in front and then fall into a thousand pieces on the ground. On the other hand, if non-heat soaked tempered glass is used, it must be 6mm or less in thickness... less risk of spontaneous breakage and less mass of glass, therefore less risk of injury for the people below in the process.

For those who are less familiar, the heat soak treatment is a secondary heat treatment through which the glass is heated for a certain time following the tempering. The goal is for the nickel sulfide inclusions to expand and shatter the glass before it is delivered and installed. The risk is not completely eliminated, but greatly reduced.

Here is an excerpt from the SB-13 supplement:

In the same vein, according to the document called “Guidance for European Structural Design of Glass Components”, there is mention of a standard concerning glass roofs. This is a British standard, BS 5516. According to this standard, in the United Kingdom, it would be possible to make a monolithic tempered glass roof if this glass roof is less than 5m from the ground. Otherwise, from 5-13m, monolithic tempered glass is permitted, but only if its thickness is less than or equal to 6mm and its surface area less than 3 square meters. Then, if the drop height is more than 13m, laminated glass is required. I cannot confirm whether this standard is applicable to residential or commercial buildings or both, however.

These recommendations are in the same direction as Ontario's SB-13 supplement since in both cases we are talking about a glass of a thickness of 6mm or less. The difference is in the heights, because the SB-13 supplement does not specify heights. On the other hand, the CSA A500-16 standard for guardrails mentions that the SB-13 table should be applied from a fall height of 4.2m.

Here is an extract from the British document:


We discussed the fact that tempered glass tends to stick together (glass clumps), the fact that fragments of tempered glass can be larger than we think... the standard for safety glass which allows up to 100mm length for fragments and the ASTM C1048 standard which recommends too small a surface constraint which can lead to even larger fragments. It has been discussed that the size of glass fragments can be influenced by the size of the piece of glass. Small tempered glass will tend to break into smaller fragments and large glass will tend to break into larger fragments.

We also discussed Ontario's SB-13 supplement for guardrails which only allows 6mm tempered glass (max.) when the guardrail is set back 150mm from the edge of the slab. And finally, we talked about the British standard which allows the use of 6mm monolithic tempered glass when positioned less than 13m from the floor if the surface area is less than 3 square meters.

It therefore appears clear that the use of thick glass should be avoided for a smoke baffle.

While waiting for the overhaul of the Canadian standard for the calculation of glass, here are some possible solutions related to the discussions above:

Possibility of falling less than 4.2m: Tempered glass 6 or 8mm**

Possibility of falling 4.2-13m: 6mm tempered glass**

Possibility of falling more than 13m: laminated glass

**For tempered glass (non-laminated), limit the size to 3 square meters or less

Please note that this information is for information only since each project must be verified by a qualified engineer in order to assess the risks specific to each project.

However, you must pay attention to the height of the glass itself. To my knowledge, there is no load specified in Canadian codes and standards for glazing acting as a smoke barrier. On the other hand, CR Laurence engineering documents recommend considering a load of 50 lbs on a surface of 305 x 305mm or a distributed load of 5 pounds per square foot. Which makes sense for maintaining the glasses, because the person who is going to clean them will certainly apply pressure on them. It is therefore necessary to ensure that the glass panes are able to withstand this type of effort depending on the height and the method they are anchored.

Hope you enjoyed!

Well done to those who made it to the end!

© Les Solutions de Verre et Mur-rideau inc., 2023. All rights reserved. Any unauthorized reproduction of this article or its contents is prohibited. This article is based on our interpretation of the codes and standards. This article and its contents do not constitute professional opinion and are provided for informational purposes only. This is an overview and does not necessarily cover all technical aspects and/or special cases. LSVM, MCi VSA or the author of the article cannot be held responsible for decisions taken in connection with this article. Reuse of this article or its contents for professional purposes, including engineering purposes, is prohibited. Please note that each project has specific conditions and must be validated by an engineer. Also note that this is valid for all articles published previously and subsequently.
Back to blog