by Anthony J. Barnes
Contributors: Melissa Szotkowski, structural glass systems manager for Oldcastle BuildingEnvelope, and Bryan Wedan, director of design engineering for Enclos.
The engineering of guardrails has historically been straightforward. The requirements are few and have not changed tremendously over the last 50 years. Glass guardrail provisions are also few in number and have remained fairly static for about 30 years.
Having been involved in the industry for much of this period, I’ve gained insight and experience that qualify me as a bit of a subject matter expert. I’ve also connected and collaborated with other experts on this topic, two of whom graciously contributed to this article, which takes an in-depth look at the history of code concerning guardrails and, more specifically, glass guardrails. It also will address how and why the code has evolved and some possibilities and projections for where it may go.
A Look Back in Time: Guardrail Code
First, let’s review the history of sources for structural requirements for guards, specifically for live load requirements. The first concise and authoritative document addressing live loads may have been the “Minimum Live Loads Allowable for Use in Design of Buildings,” published by the U.S. Department of Commerce in 1925. This document created standards for applying live loads based on collected data. Its introduction states that live loads assumed in designing many types of buildings were largely matters of tradition and had a scant scientific basis. There was a clear need to address a gap in safety. This document evolved over the next several decades. It was eventually published again under the supervision of the U.S. Department of Commerce’s National Standards Bureau and the American Standards Association as American Standard (ASA) A58.1 in 1945. The first mention of a railing load requirement of 50 pounds per linear foot applied at the top of the railing appears to be within this document. ASA eventually became the American National Standards Institute (ANSI), and this document eventually became ASCE 7, referenced in the International Building Code (IBC) for determining loads on structures. The 200-pound concentrated load applied to railings was added to railing requirements in the 1988 version of ASCE 7.
Glass Use in Guards
Now, let’s look at the background of glass used in guards. While building codes were evolving, building material technology was advancing, making glass a viable structural material. First, heat-strengthening methods that could quadruple the strength of glass were developed. Next, glass manufacturing methods drastically improved between the middle of the 19th century and the middle of the 20th century. Quality economically produced float glass was available by the early 1960s. Lastly, laminated glass became widely available by the early-to-mid 20th century, primarily for use in the auto industry. It was intuitively understood that the physical properties of glass differed from those of other traditional building materials such as steel, wood or reinforced concrete. Glass was brittle and failed suddenly. Since the 1960s, window and skylight glass used in the construction industry has generally been designed using a simple chart based on destructive glass tests. These tests included various aspect ratios and glass thicknesses but did not consider anything other than four-side supported rectangular plates. This method was widely accepted and likely resulted in safe and sufficient designs for nearly all applications of glass at that time (i.e., windows).
The industry wanted a more general method to predict glass failure. W. Lynn Beason published a study introducing a glass failure model that employed Ernst Hjalmar Waloddi Weibull, who offered a statistical failure analysis for predicting the strength of brittle materials in 1939. This became the basis for discussion within the American Society for Testing and Materials, out of which ASTM E 1300 Annealed Glass Thickness Selection Charts was born in 1989. Charts provided allowable loads on flat glass supported on edges. The IBC eventually referenced this document for determining wind-, snow-, seismic-, and dead-load resistance of glazing. E 1300 has been expanded and revised several times to address heat-strengthened glass, laminated glass, insulating glass, etc. While early versions of this standard did not explicitly prevent its application to glass guards, more recent versions have stated that the practice excludes balustrades. Appendices in the document allow for the derivation of permissible glass stresses. Engineers have commonly used these derived stresses to determine allowable stresses that could be used as limits for conditions outside the scope of E 1300, such as discrete support points on glass, non-rectangular-shaped glass, non-uniform loads, etc.
The guidance specific to glass used in guards was first published in the 1988 version of the United Building Code (UBC). This was published one year before ASTM E 1300 and certainly a few years before any model codes referencing E 1300. Bill Lingnell PE, of Lingnell Consulting Services, was the historian on this subject matter (1) and an original group member who created ASTM E 1300. He was also involved in empirical glass tests to justify design before E 1300. Bill graciously spoke to me and provided a wealth of historical knowledge. Not coincidentally, he was the person code authorities contacted to provide a safe limit for use in glass guards in the late 1980s. Based on what he knew from testing and his work with the probabilistic methods of E 1300, a simple statement regarding glass guard design was published in UBC 1988. This simple statement was, “A safety factor of four shall be used.”
Current Code Status – Clear or Confusing?
To this day, this language remains in the 2021 International Building Code, albeit with a few minor alterations that have not changed the spirit of the statement. The requirement, at first glance, seems simple enough. However, it is a bit ambiguous and results in some differing interpretations:
- To what is the factor of safety to be applied?
- Should the design load be applied four times in a test?
- To what value of glass strength or modulus of rupture should the factor be applied?
- The sentence preceding this also mentions the glass and support system. Does a factor of safety apply to the system supporting the glass?
To an engineer familiar with glass design, intuitively, the answer is to obtain a glass strength, divide that number by four and use this value as the allowable glass stress. To the glass railing manufacturer seeking product approval, the directive seems to apply four times the load in a test. And finally, to the code official, the instruction is, conservatively, to use a factor of safety of four and apply it either in test or analysis to both the glass and the supports.
When we began working on glass guards in 2014, we used the most commonly known value for the modulus of rupture of glass, taken from AAMA CW-12-84, and divided that value by four. From experience, it was obvious that this safety factor should only be applied to glass, not the support system for the glass. Some jurisdictional authorities thought differently and asked the International Code Council (ICC), those responsible for the publication of the IBC, for clarification. The response was the following:
- Glass need not be tested with four times the design load (this was later clarified in code commentary); and
- Supports directly supporting glass need to be designed with a factor of safety of four.
The folks at ICC agreed that the language could use some clarification, and if I wanted, I could submit a code change proposal. This began my naïve journey into modifying the code.
In 2018, I wrote a simple proposal for the 2021 code cycle, seeking to clarify to what the factor of safety of four should be applied. Eventually, through negotiation within ICC committees and consultation with a group of technical experts, the proposal was modified to provide an allowable stress level for guard rail glass derived using the methods of E 1300. The changes seemed logical in my engineering brain, but that didn’t translate to changes that were easy to digest and accept by the glass industry and the body of ICC. What sounds like the end of my foray into code work is the beginning.
Current Code Work and What the Future Might Hold
What is the future of the glass guard code? With the 2021 code cycle behind us, the next deadline would be for the 2024 code cycle. I learned from my first attempt that the ICC comprises municipal building officials, industry lobbyists, manufacturers and technical experts. Most decision-makers are code officials who sit on various committees, but representatives of industry interests generally keep a keen eye out for things that could negatively impact their industries.
A code provision change first needs the support of the specific committee, so you must convince the members that your idea is palatable (in the case of guardrail provisions, the structural committee). Then, you need to convince appropriate industry representatives that the impact is positive, so they don’t speak against your proposal. Through communication with people in the glass industry and those on the ICC structural committee, I started connecting the dots. I determined that the best action would be to connect with people in ASTM. A workgroup within ASTM was already working on writing a glass guard standard. The idea was to publish a standard in time for it to be referenced in the 2024 IBC. I felt extremely positive with three years ahead of us. I prematurely and naively assumed we had achieved a panacea that could easily be implemented in time for the 2021 deadline for code change proposals.
I was wrong, however, and missed the deadline to adopt an ASTM standard for the 2021 building code. Instead, in the 11th hour, technical folks and industry advocates came together to enter a code change proposal that suggested allowable stress values for glass used in guards. This proposal eliminated any mention of the factor of safety of four and clearly decoupled glass and the supporting structure. Although this proposed language suffices to provide clarity of intent, the most reasonable technical approach to the design for glass guards is to use a method similar to that of ASTM E 1300 (i.e., a probabilistic mathematical method). Ironically, without any mention of glass guard capacity in the building code, an engineer would use ASTM E 1300 methods to derive allowable glass stress and compare that to resultant stresses generated by the application of the code-prescribed loads. Of course, the issue is that the code language developed in 1988 specifying guard capacity (factor of safety of four) was published before ASTM E 1300.
The alternative to the current process would be to include a reference to an ASTM standard in the building code. This method has advantages. It reduces the amount of prescriptive language in the code, and adopting a standard by the building code implies a certain amount of trust in the standard organization to provide future guidance responsibly. The ASTM committees can be more agile than building code bodies and adopt future changes as technological developments occur or design considerations need to be addressed.
Ultimately, I believe it would be ideal to have a glass construction manual referenced by building code –i.e., a document similar to AISC 360 or ACI 318 – a specification covering all glass design. For this to occur, an organization or group of people need to offer their support. Currently, a 2018 document published by the National Council of Structural Engineers Associations (NCSEA), “Engineering Structural Glass Design Guide,” could serve as a foundation for this purpose. Based on current code/industry precedent (having material-specific trade organizations sponsor these publications), this type of effort would be best led by industry organizations that have worked on similar resources. In any case, this will take an ambitious and dedicated effort.
Anthony (Tony) Barnes is the director of structural engineering for Minneapolis-based Sightline Commercial Solutions.