1. Design Loads for Balustrades

In the United States glass balustrades are governed by IBC⁹, ASCE 7³ and ASTM E2358⁵, with ASTM E2353⁴, (although the ASTM standards are notable in their absence as referenced standards of IBC.) The design loading can be loosely summarized as:

• a distributed load of 50 lb/ft (0.73 kN/m) (IBC 1607.8.1, ASCE 7, ASTM E2358),
• 200 lb point load to the top rail, and
• use of laminated glass to prevent fall of glass to walkways below (IBC⁹ 2407.1.), however there is an exception for this rule if there is no walkway below or the walking surface is permanently protected from falling glass.

A comparison of relevant codes regarding design loads between the US and other countries is presented in detail in a separate paper for this conference (Green et al. 2024¹⁰) Those findings highlight that a majority of countries outside the US are adopting assembly loads about double (2x) the US loading and four times (4x) the load for areas subject to overcrowding. This paper looks at design practices for glass balustrades, particularly where the glass is the structural element providing the primary restraint for the occupant.

2. Retention, Redundancy and Residual Capacity

The requirement for laminated glass was introduced to IBC in 2015, after a series of glass balustrade failures occurred in Seattle, Toronto, Austin and elsewhere. (Headley 2011¹²) There were a number of close calls; it was considered that sooner or later a fatality or serious injury was likely, and the code was revised to restrain falling glass. While guardrails are provided for a purpose -- to restrain people from falling off a precipice -- the retention of the glass provided only addresses part of the risk. A robust solution needs to consider retention, redundancy and residual capacity so that brittle materials continue to serve their purpose of preventing falls.

Australian Standard AS1288² addresses all aspects of robustness for balustrade glass. It has required lamination since 2006 and it introduced a requirement for residual capacity and stiffness in 2021.

The primary risk addressed by lamination in the IBC is fall of glass due to spontaneous fracture, often due to Nickel Sulfide (NiS), however spontaneous fracture is not limited to the effects of NiS, as other inclusions and mounting system stresses can also participate. For an inclusion to cause fracture, its expansion must be significant and interact with a tension field to overcome the inherent tensile strength of the glass. Usually spontaneous fracture requires a ‘large’ NiS stone ( up to 0.55 mm, typically 0.22mm) near the center of the glass in fully tempered glass. (Figure 2)

Source: Fundamentals of Spontaneous Breakage Mechanism Caused by Nickel Sulfide - Dr Andreas Kasper¹³

^{The distribution of inclusions throughout the thickness of the glass will be random. The size of NiS stones is somewhat random, but larger stones tend to fall to the bottom of the melt pot and not be included in the float. For spontaneous fracture to occur, growth of the inclusion needs to be in a tensile stress field. While stones outside the tension zone will not cause spontaneous fracture, either in practice or heat soak testing, it can interact with tension fields due to bending to reduce the capacity below design levels. The weakened glass may interact with applied loads resulting in unexpected fracture, such as reported by the New Zealand Herald (Figure 3.) The requirements in IBC protect against falling glass but do not restraint the occupants in cases of failure.}

3. Fully Tempered ‘Safety Glass’

It is also important to understand that FT ‘safety glass’ does not always break into ‘harmless small dice’. The crack wave front is driven by tension stress; the tension is in the center of the glass, but the surface is in compression. The compression zone tends to remain as a ‘skin’ on the fractured glass, holding clumps together until the aggregate expansion of the wavefront causes tension or it impacts something hard – and if the ‘something hard’ happens to be a person, the outcome can be lethal.

4. Glass Risk Categories And Robustness Requirements

CEN/TS 19100⁶ goes part way to providing design guidance, incorporating the concepts of in-service loading during a fracture event and post-fracture analysis, but it does not include what levels of residual capacity to use, or when to use them.

Part of the challenge is setting suitable design targets for different circumstances: just as live loads are based on occupancy or building risk categories, so robustness levels also need to be based on occupancy (both sides of the barrier), thus the importance of the concept of ‘glass risk categories’. The Structural Glass Design Manual¹¹ (SGDM) provides a general proposal for glass risk categories and robustness is included for context.

4.1 Proposed Glass Risk Categories

The Structural Glass Design Manual¹¹ proposes four Glass Risk Categories.

The Glass Risk Category is a function of the occupancy (Table 1)

The Robustness requirement is a function of the application (Figure 6) defining the types of post-damage checks required (Table 2).

The combination of Glass Risk Category and Robustness Categories define the residual load factor (Table 3.)

4.2 Reasons Why Robustness is Important by Occupancy

Residential: - G-II

Young children do not recognize the difference between a solid barrier they cannot see and a missing barrier they cannot see. In the case of spontaneous fracture resulting in a glass infill barrier leaving an opening, young children have failed to recognize the opening or risk and have fallen to their death. As such retention of barriers are recommended throughout.

Light Commercial: G-II

These applications are typically over walkways, despite lower occupancy and risk of injury and typically adult occupants, fall of glass is recommended to be prevented.

Assembly and Heavy Commercial: G-III

These applications are typically over walkways, hence fall of glass is required to be prevented.

Crowd – G-IV

In a crowd situation, there will be continued thrust from the crowd after the initial breakage. As such, the barrier needs to continue to provide restraint immediately following the loss of one ply to prevent sudden collapse of the barrier.

5. Proposed Design Loads for Handrails

The Structural Glass Design Manual proposes adopting loads based on AS1170.1¹ and EN1991-1⁷ (Table 4.) A separate paper at this conference "Balustrade Design Loads: Failures, Fatalities, Research and Global Design Practices" (Green et al. 2024²⁹ ) addresses the logic and history of these loads in greater detail.

6. Design Practices

For structural glass balustrades, AS1288² requires laminated glass for fall heights greater than 5m for infill glass and lamination with residual capacity for systems where the glass is providing the structural restraint. With one ply broken, AS1288 requires residual capacity to resist serviceability loads (unfactored (ASD) load vs. ultimate design strength) with one ply broken. AS1288, since 2021, also requires testing for both plies broken with a maximum deflection of 250mm (10in) under a load of 200N (45lbf).

It should be noted that AS1288 references design loads from AS1170.1¹ (imposed loads/Live Loads) AS1170.2 (Wind Loads) and load combinations from AS1170.0. The load factor for live load is 1.5, thus the design residual capacity load is 67% of the design load. The design load includes assembly loads of 1.5 kN/m (~100 lbf/ft) and crowd loads of 3.0 kN/m (~200 lbf/ft).

ASCE-7³ Section 2.5 includes load combinations for ‘extraordinary events’. If the glass is well designed, well fabricated and well installed, then it should not break under design loads and the causation of breakage is likely due to some other extraordinary cause. This section includes two tests, during the damaging event and residual capacity. In both cases ASCE-7 has a load factor of 0.5 L as opposed to 1.6 L in the design situation, however during the damage event there is an additional load affect that must be accounted for:

• During the event: (0.9 or 1.2D) + A_k + 0.5L + 0.15S (ASCE 7-22 eqn 2.5-1)

in which A_k is the load or load effect resulting from the extraordinary event, A.

• After the event: (0.9 or 1.2) D + 0.5L + 0.2(Lr or 0.7S or R) (ASCE 7-22 eqn2.5-2)

Although ASCE 7 is requiring L as a gravity load, it is perhaps a good analogy for the live load during the extraordinary event that glass fails at below design levels of loading.

The ASCE 7 combination is 31.25% of the design live load but the sudden loss of stiffness due to breakage results in a dynamic effect for the component of stiffness loss. While the theoretical dynamic factor is 2.0, it is common in crane codes, etc. to use dynamic load factors of 1.8, making allowance for some damping.

Before looking into the implications of the continued load during the extreme event, consider the type of events that cause glass fracture in balustrades and whether load may be present.

Spontaneous Events – fracture without load application or fracture well below design loads.

Grab/Impact Load – the impact load which breaks the glass also stops the person and there is no sustained load. (G-II) Unlikely to have sustained load after initial impact load.

Assembly Loads – (G-III and G-IV) Possible continued thrust after damage event, but it is likely that there will be some recoil after the damage event and the load will decrease as the balustrade deflects away.

Crowd Loads – (G-IV) Potential/probable continued thrust from crowd after damage event and barrier is required to provide a safety function. Damage event calculation is highly recommended.

Examining some of the options, let us consider a hypothetical balustrade without top rail subject to assembly loads in a G-III environment with a crowd load of 3.0 kN/m (~200 lbf/ft)

7. Design Examples

Example 1 Design to Australian Standard AS1288-22

AS1288: 2021 requires a live load factor of 1.5 for the unbroken state and "The glass shall be designed in accordance with serviceability limit states" (unfactored live load and deflection less than cantilever height /30)"in the event that any one layer of the laminated panels is broken.

"In addition, the laminated glass design shall demonstrate the suitability of the post-breakage behaviour
of the selected glass type by limiting outward deflection, after failure of all layers, to a maximum of
250 mm when a serviceability load of 200 N concentrated load is applied."

In the following example, we see that post-damage controls the design with one ply broken.

Thickness required after one ply broken is greater than half initial thickness – post-damage capacity controls. Likely outcome 16+1.52+16mm (5/8" + 0.060"+5/8")

Example 2: A two ply balustrade with sustained load and dynamic failure redistribution

If the logic of sustained load during the damage event is applied, breakage of the second lite is likely in assembly and crowd loading for 2 ply systems. Alternative design following extraordinary loads case, consistent with ASCE7, offers another option with superior outcomes. In this case the strength model is based on EN16612, CEN/TS 19100 and the Structural Glass Design Manual.

First look at the behavior of a 2-ply laminate with damage-event loading using the ASCE7 extraordinary loads case with a thickness based on the in-service design load.

As an example of dynamic effects, imagine a weight initially hanging on 4 similar parallel springs, it has a natural deflection of "d", and the same weight is on a single spring has a deflection of 4d, but if the weight is released on a single spring from d it will still be moving when it gets to 4d and will have a maximum deflection of ~7d; the neutral position plus the drop height above the neutral position (4d + (4-1)d = 7d)

Example 3: A three ply balustrade with sustained load and dynamic failure redistribution

Consider an alternate configuration of 6mm + 1.52mm + 6mm + 1.52mm + 6mm (18mm glass, 21.04mm overall). For simplicity it is assumed the interlayer is effectively rigid for the duration of the load redistribution.

Note that glass with typical strength will not break under the design load and a weakened glass is likely to fracture before developing full load, hence the reduction in load during the damage event. Further, as the glass fractures and the balustrade deflects, there will be some recoil from the occupant and some release of load (crowds excepted.)

The residual capacity with the 3-ply system and the most critical ply broken is greater than a 2-ply system and the portion of the load that is dynamically redistributed is much less. As a consequence, it is far less likely that progressive failure of further lites would occur if the handrail was to fracture under load.

All plies broken

Should all plies fail, the stiffness of the interlayer becomes a critical aspect of the design. Typical PVBs were initially designed for car windscreens to absorb energy if a rock hit the glass and the extension of the PVB allows the glass to fold over. To resist this, AS1288 has proposed a residual capacity test of <250mm (10 in) deflection under a load of 0.2 kN (45 lbf).

The use of a structural grade interlayer with a suitable stress-strain curve, combined with a 3-ply configuration, offers a much stiffer system with significant residual capacity.

Interestingly, due to the cost efficiency of 6mm (1/4in) glass, a review of 6+6+6mm glass configuration with glass suppliers has indicated that, where lamination is already occurring, 3 plies of 6mm may be cheaper or similar cost relative to 2 plies of 10mm.

Taking this into consideration, adopting the principle of Extraordinary Events in ASCE-7 to glass balustrade design results in glass of similar thickness, similar cost and far superior robustness.

8. Conclusion

Global design practice for balustrades indicates that loads of 1.5kN/m (100 lbf/ft) are justified for assembly locations and 3kN/m (200 lbf/ft) for areas subject to crowd loading.

Glass may have rare imperfections or inclusions that reduce the capacity of the glass.

The design of glass balustrades (without structural handrails) should provide post-damage capacity and such practices are required by AS1288, CEN/TS 19100 and the Structural Glass Design Manual. In another approach, considering extraordinary events per ASCE-7 and CEN/TS 19100 take into account the potential for load during a damage event, this study finds this to potentially be a critical event. The Structural Glass Design Manual proposes that appropriate level of load during and after a damage event is a function of the occupancy.

A configuration of 3 plies of 6mm glass (6mm + 1.52mm+ 6mm+1.52mm+6mm=21.04mm) was found to be significantly more robust than 2 plies of 10mm (10mm+1.52+10mm = 21.52mm) forming a similar overall thickness but with far superior redundancy characteristics. Surveying manufacturers found that superior performance can be achieved without significant cost increase.

AS1170.0 - Australian Standard: Structural design actions

Part 0: General principles

Part 1: Permanent, imposed and other actions

AS1288 – Australian Standard: Glass in buildings—Selection and installation, Standards Australia

CEN/TS 19100:2021 "Design of glass structures"

EN1991.1 Eurocode 1: Actions on structures - Part 1-1: General actions -Densities, self-weight, imposed loads for buildings

EN16612 Glass in building - Determination of the lateral load resistance of glass panes by calculation

ASCE7 (ASCE/SEI 7) Minimum Design Loads and Associated Criteria for Buildings and Other Structures (American Society of Civil Engineers

Structural Glass Design Manual (Public draft 2024) Green R, Crosby A, McDonnell T www.structglass.org

A Tale of Three Cities – and Lots of Broken Glass – Megan Headley US Glass, Metal and Glazing October 2011

Balustrade Design Loads: Failures, Fatalities, Research and Global Design Practices Green R., Crosby A., McDonnell T. Facade Techtonics World Congress 2024

Fundamentals of Spontaneous Breakage Mechanism Caused by Nickel Sulfide - Dr. Andreas Kasper Glass Processing Days 2003

Factors That Influence Spontaneous Failure in Thermally Treated Glass – Nickel Sulphide - Dr Leon Jacob Glass Processing Days 1997

NiS in HS Glass - Dr. Andreas Kasper; John Colvin; Frank Rubbert; Francis Serruys https://www.saint-gobain-glass.com/nis-in-heat-strengthened-glass

Design of glass balustrades to CEN/TS 19100 – Chris O’Reagan and Peter Lenk, The Structural Engineer January 2024.

A simplified approach to the dynamic effective thickness of laminated glass for ships and passenger yachts Vergassola and Boote, International Journal on Interactive Design and Manufacturing DOI: 10.1007/s12008-019-00614-2