Joe Jaroff
Jaroff Design, US

Steven Capri, PE
Steven Capri PE Consulting Engineer, US

Abstract
Designing, engineering, and fabricating glass projects for public use present unique safety challenges. Difficulties compound when traditional engineering methods do not apply to complex designs or materials with no historical data. This study reviews the engineering, modeling, and testing of two custom glass projects: nine-foot structural laminated glass treads for a staircase for the Jazz at Lincoln Center venue in New York City, and 56 vertical monolithic borosilicate monuments for the 9/11 Memorial in New Jersey. New and untested laminated glass materials were used.

Methods: Jazz at Lincoln Center included testing for glass optics, reflectivity, anti-slip, public traffic patterns, concentrated point load, and live load distribution on a full-scale mock-up, with 24-hour testing after breakage to measure deflection. Testing for 9/11 vertical glass panels included thermal shock, stability in extreme weather, concentrated loads, and wind forces on an engineering model and full-size mockup. Results: Jazz at Lincoln Center testing confirmed deflection and predicted engineering analyses for breakage and concentrated load. For 9/11, initial tests were mixed. Further testing resulted in modifying 9/11's adhesive to increase stability and control movement, and harder setting block material.

Conclusion: Engineering models cannot alone adequately predict performance with new laminated glass materials. Unique glazing projects in varied environments require physical testing and full-scale mockups.

1 Introduction

Combining the variables of laminated glass with unknowns inherent in designing public art creates complex problems.

Considerations include:

Manufacturing variables

Interlayer type
Interlayer thickness
Type of adhesive (UV, silicone, urethane, etc.)
Setting material
Glass impurities
Float-side adhesion vs. airside adhesion
Quality control of autoclave cycles and
Bond strength of glass into interlayer
Micro-fractures, scratches from drilling, cutting, polishing and other secondary manufacturing artifacts
New glass chemistries, interlayers, and adhesives with no historical test data

Site variables

UV exposure
Freeze/thaw cycles, moisture, and wind loads (if located outside)
Post-installation scratches, fractures, and fisheye damage
Unanticipated public use and abuse
Sustained dead loads over a prolonged period
Human interface and interactions with physical and emotional responses

Inadequate analyses can have implications beyond structural failure. For example, in 1989, despite in-depth structural engineering, our insurer dropped coverage after installation of a glass staircase in a Banana Republic clothing store in Chicago claiming no data existed for understanding glass staircase performance. Moreover, structural calculations do not cover emotional responses. On that same project, feedback was received that people didn't feel comfortable “walking on air,” causing the client to replace clear glass treads with frosted glass.

There is a long history of using tempered glass in architectural applications.[1] This paper presents case studies of two unique laminated glass structures, one horizontal, the other vertical. Both required new engineering and testing methods. Design constraints for each project included public use under a variety of known and unknown conditions and structural resistance to potentially unforeseen forces.

Assumptions were made and accompanied by digital prototypes, which were then transformed into full-scale mockups. Point and lateral load tests were developed and administered to understand stability and resistance against forces of use, nature, and public assembly.

2 Case Studies

2.1 Case Study 1 - Laminated nine-foot structural glass stair treads for a monumental interior staircase, Jazz at Lincoln Center, New York City, New York

The first project was a monumental public staircase with nine-foot-long, five-layer laminated treads [Figures 1 and 2].

Figure 2. Structural glass treads.

Glass is not 100% homogenous; it contains impurities that can result in spontaneous and random failure. Structural redundancy was created by choosing a standard low iron/soda lime float glass multi-layer laminate composite.

The top and bottom layers in a laminated glass structure are considered sacrificial (1) , i.e., design assumes the top layer will scratch and develop micro-fractures, and that the bottom layer will crack from tension. Therefore, treads must be designed such that mid-layers are capable of performing adequately for all loading conditions. The adhesive must be strong enough to prevent layers from sliding and act as tensile membranes without becoming brittle over time.

Loading hazards included the unknowns of public traffic patterns, the possibility of every tread simultaneously occupied, and the potential of unexpected live loads (e.g., the entire New York Jets football team doing jumping jacks on the stair). The location was also evaluated, as the stair was situated above an escalator, creating a worst-case fall distance of 25 feet. Testing included a 2,700-pound force load distributed evenly over a sample tread [Figure 3]. This was followed by point load testing of a 3,200-pound concentrated load directed to the center of the tread with a hydraulic ram, which confirmed a calculated movement of 0.412 inches deflection.

2.1.1 Engineering and Test Methodology

The time dependency of the deflections and stresses in laminated glass is well documented and is dependent upon the shear coupling capabilities of the polymeric laminate layer. The effectiveness of the polymeric laminate sheet is based upon its inherent shear stiffness (i.e. its modulus of elasticity in conjunction with its thickness), its temperature and the amount of time in which it is subject to stress. It is the ability of the laminate layer to transmit shear stress between the layers of glass, which ultimately determines the deflections and stresses in the glass. The stiffer the laminate is and the greater the ability of the laminate to transmit shear in between the glass layers without shear deflection the more effective it is in making the overall laminate composition behave as a monolithic glass piece. The time dependency of the glass and its “strength” is nothing more than creep deflection.

While there are software packages which are able to analyze laminated glass compositions, this project presented an opportunity to analyze the stresses and deflection of the glass via an analytical method presented in two papers by Laura Galuppi and Gianni Royer-Carfagni.[2][3] These analytical methods were proposed as an alternate to traditional laminated glass solutions which seemed to give generally inaccurate results as the number of layers is increased. A requisite of this analysis is that it is linear and as such it is prescribed that the total deflection is to be less than 1⁄2 the thickness of the overall glass thickness.

In the case of the treads we used a total of six layers of annealed glass. A spreadsheet was developed in order to calculate the effective thickness of the treads for both stress and deflections according to Galuppi and Carfagni. While we were certainly interested in the stresses in the glass at the time of failure and how they compared to an allowable stress—it was not practical to measure it - and so it was the deflection of the glass, which we were able to ultimately correlate and compare to an analytical solution.

The test of the glass treads needed to account for several different load conditions. There are two New York City code-loading requirements for stair treads: [4]

100 lbs psf., on full area of tread
Minimum concentrated load of 300 lbs. on an area of 4 square inches.

What seems to be lacking in the code based requirements is an impact load condition. Whether or not this is accounted for in the 300 lb. concentrated load is not known but it was ultimately decided that a 200 lb. person bounding down the stair with an impact factor of 3 on the tread was deemed to be sufficient. Thus a 600 lb. concentrated test load was decided upon as a minimum base load.

There were three primary goals for the test program conducted on the longest tread on the stair:

Correlate the measured load- deflections with those calculated by the “Enhanced Effective Thickness” method.
Measure the actual failure load for the longest tread.
Measure the post-failure strength of the tread

Two primary load configurations were selected:

Concentrated load at center of tread – 600 lb. minimum
Uniform loading – 100 lbs psf. minimum

2.1.2 Factor of Safety Calculation

The factor of safety for glass breakage can be assessed relative to the probability of breakage:

Using the analytical method proposed by Galuppi & Carfagni with a concentrated load of 600 lbs we calculated a theoretical maximum glass stress of 970 psi.
Using a probability of breakage of 1 lites/1000 for a 120 second load duration (the approximate length of the test) we calculated an allowable stress of 3750 psi for annealed glass (via ASTM E- 1300).[5] Knowing this stress we then calculated the concentrated load, which would result in that stress. That concentrated load was calculated to be 2,775 lbs.
Using a probability of breakage of 8 lites/1000 for a 120 second load duration (the approximate length of the test) we calculated an allowable stress of 5046 psi for annealed glass (via ASTM E- 1300). Knowing this stress we then calculated the concentrated load, which would result in that stress. That concentrated load was calculated to be 3,800 lbs.
Using a probability of breakage of 20 lites/1000 and a 120 second load duration the allowable stress is 5752 psi. The concentrated load corresponding to a predicted stress of 5752 psi is 4,360 lbs.
Using a probability of breakage of 50 lites/1000 and a 120 second load duration, the allowable stress is 6550 psi. The concentrated load corresponding to a predicted stress of 6550 psi is 4,990 lbs.
A probability of breakage of .00005 lites per 1000 or 5 lites per 100,000,000 with a load duration of 120 seconds results in an allowable stress of 970 psi as calculated with ASTM E1300.

2.1.3 Post Breakage Factor of Safety

The results of the concentrated load test indicated that the post breakage strength of the overall laminate is approximately what one would calculate with the one less layer of glass. The margins between measured and calculated deflections of treads were fairly consistent, remaining at about 20%. The presence of the interlayer at the bottom (above the broken laminate layer) does not seem to add any appreciable strength over what one would expect without it.

After the initial breaking of the bottom layer the tread was not stressed further to see at which point the second layer of glass would break. After the initial break the tread was able to still resist a 2400 lb. concentrated load at the center without a significant deflection, which would be cause for alarm. In fact – after the initial breakage of the bottom layer it would still take a 1000 lb. load to deflect it 1/8”.

2.1.4 Predicted Failure Load

We did not have a predicted failure load at the outset of the test. The probability of breakage at any point in the loading program is difficult to predict. The primary impetus for the program was to measure the load-deflection and compare it to an analytic method. The pre and post failure strength of the tread was also of interest.

2.1.5 Test Results

Physical testing confirmed the glass breakage at 3,200 pounds of concentrated load. Because glass strength is relative to time [Figures, 4,5,6], [Table 1], [Figures 7,8] an incremental force at two-minute intervals was applied until breakage of the bottom layer. Twenty-four-hour duration testing after breakage measured the amount of deflection from carrying a uniform 2,700-pound force load.

Figure 6. Gauge to measure deflection at mid-span of tread. — **Figure 6.** Gauge to measure deflection at mid-span of tread.

Table 1. Measuring results — **Table 1.** Measuring results

2.1.6 Installation

While engineering and destructive testing was necessary to understand how the glass would perform in catastrophic failure, the design-build process and fieldwork were equally important. After 18,000 pounds of glass was installed, stringer deflection required treads to be individually adjusted to maintain equal spacing. A system to make adjustments and evenly distribute loads was designed and masked by back-painting glass bearing surfaces. This solution was incorporated into the final load testing. For safety, facets at tread leading edges were created to increase depth perception and reflectivity. An anti-slip wearing surface was also developed.

2.2 Case Study 2 -Laminated exterior seven-foot high monolith glass for 9/11 Memorial, City of Hoboken, New Jersey

The second project was a 9/11 Memorial comprised of 56 inscribed and illuminated glass monoliths. Each panel represents local lives lost during the 9/11 attack. Monoliths were supported by a stainless steel structural base and bracket system. The memorial, aligned with the World Trade Center Site in Manhattan, was unveiled at an interfaith ceremony on September 11, 2017 (the 16th anniversary of the attack) at Pier A Park on the Hudson River. Location at water’s edge exposed the materials to varying weather cycles [Figures 9,10].

Design considerations included vertical glass panels withstanding thermal shock and maintaining stability in extreme weather conditions (standard glass thicker than 0.75 inches cannot accommodate this). Because of its low coefficient of expansion, borosilicate glass was chosen and designed to resist multiple forces and varying instabilities. Concentrated loads of public assembly were a critical design consideration. Three layers of one-inch thick annealed borosilicate glass (Pyrex) were laminated to yield three-inch thick monoliths [Figure 11].

As designers, we faced with both time and material challenges. Hired in mid-January 2017 (one month before the project was slated to be cancelled) we were advised the installation date was a memorial event on September 11, 2017 (less than eight months away). And we were working with borosilicate glass laminated glass (Pyrex) that had never been engineered, tested, assembled, or installed for this use. The material was in short supply and new material no longer available, which restricted destructive testing.

2.2.1 Background

In 2002, the City of Hoboken hired architect Demetri Sarantitis to design a memorial for local lives lost at the World Trade Center. The city purchased one-inch Pyrex slabs from a factory in upstate New York for use in the memorial. However, no one with knowledge and experience using Pyrex for this purpose was found, so the glass was stored in an unconditioned warehouse for over a decade in New Jersey. Since borosilicate glass melts at a higher temperature than ordinary silicate glass, its manufacture requires different ovens than required for float glass with extremely high heat and melting points; double the requirement for normal glass manufacture. When the Pyrex factory burned to the ground, no other facilities existed that could produce more of the material, and therefore became irreplaceable. In 2015, the stored glass was moved to a laminator for testing, mockups and a feasibility study.

2.2.2 Testing Methodology

The chemistry and properties of borosilicate glass are different from soda-lime glass. Because its use is not in glazing, there was little data on its use for the memorial project. An engineering model was created, as was a full-size mockup to analyze the structural stability of the individual monoliths, type and amount of adhesive required, and glass movement.

To simulate forces of wind, public assembly, and people climbing, each monolith was inserted into a mockup jig and pull-tests performed to measure deflection [Figures 12, 13, 14]. Samples were also sent to a testing lab to measure interlayer and adhesive performance.

2.2.3 Test Results

Initial results from testing were mixed. Some components passed, while others only marginally passed. For example, preliminary engineering indicated 0.375 inches thick adhesive with 0.375 inches deep caulk glue joints would be adequate [Figures 15,16]. After testing and peer review, the depth surface area of the adhesive was tripled and a thinner application used to increase stability and control movement.

Mockup pull-tests demonstrated standard setting blocks allowed too much deflection [Table 2]. We adjusted the setting block to a harder durometer material, leading to greater stability [Figures 17,18,19]. We also tested for adhesion.

Table 2. Measuring results — **Table 2.** Measuring results

3 Conclusion

Due to internal material variables and external site conditions—and the multitude of problem combinations they can create—unique glazing projects in varied environments require physical testing and full-scale mockups. These are critical to assessing performance, materials, visual effect, optics and esthetics, and provide a level of assurance not obtainable through virtual engineering models. When projects include complex engineering problems, only full-scale testing can confirm, refine, or redirect creative problem-solving.

The industry should consider establishing standards and protocols for testing complete applications. These would not only prove the engineering but ensure that quality control in the float, tempering and laminating processes are within specifications.

Engineering alone cannot adequately predict performance in catastrophic failure. Mockups can aid in visualizing not only structural but emotional and other considerations not considered initially.

4 Acknowledgements

The authors wish to thank Gloria Jaroff, AIA and Richard Buday, FAIA for their assistance reviewing this paper.

References

[1] ASTM E2751/E2751M-13 Standard Practice for Design of Laminated Glass Walkways.

[2] Laura Galuppi, Gianni Royer-Carfagni. The effective thickness of laminated glass plates. July 2012, Journal of Mechanics of Materials and Structures 7(4):375-400. DOI: 10.2140/jomms.2012.7.375

[3] Laura Galuppi, Gianni Royer-Carfagni, Enhanced Effective Thickness of multi-layered laminated glass. Composites Part B: Engineering,. Volume 64, 2014, Pages 202-213. ISSN 1359-8368, DOI: 10.1016/j.compositesb.2014.04.018.

[4] New York City Building Code 2010, Chapter 16, Structural design/Sections 1601-1613.

[5] ASTM E1300 Standard Practice for Determining Load Resistance of Glass in Buildings.

Theory vs. Reality: Engineering, Modeling, and Testing Two Laminated Glass Public Projects

1 Introduction

2 Case Studies

3 Conclusion

4 Acknowledgements

References

Art Glass: Processes & Possibilities

More than Skin Deep: The Right Materials Create Lasting Impact